UNIVERSIDADEDELISBOA

FACULDADEDELETRAS

DEPARTAMENTODELINGUÍSTICAGERALEROMÂNICA

TheAcquisitionofPrimaryWordStressinEuropeanPortuguese

SusanaCorreia

DoutoramentoemLinguística

Especialidade:LinguísticaPortuguesa

2009

ii

iii

UNIVERSIDADEDELISBOA

FACULDADEDELETRAS

DEPARTAMENTODELINGUÍSTICAGERALEROMÂNICA

TheAcquisitionofPrimaryWordStressinEuropeanPortuguese

SusanaCorreia

DoutoramentoemLinguística

Especialidade:LinguísticaPortuguesa

Dissertaçãoorientadapor:

ProfessoraDoutoraMariaJoãoFreitas

2009

iv

v

The present work was sponsored by a Doctoral Grant (Ref. SFRH/BD/21696/2005) of the

Science and Technology Foundation (FCT), Portugal, funded by the PortugueseMinistry of

Science,TechnologyandHigherEducation.

vi

vii

ToRui

viii

ix

Aknowledgements

Firstofall,IhavetosaythatIamhappytohavearrivedhere,totheendofthislong

path,inwhichIhavelearnedalot.Tothosewhotaughtmeandsupportedmeinsuchatough

journey,Inowwishtosayaprofound'thankyou'.

My first words go to my supervisor. I deeply thank Prof. Maria João Freitas, my

professor since my undergraduate studies, for having taught me so much, first about the

phonologyofPortugueseandlateraboutphonologicalacquisition.Ithankherforbeingwith

mefromthebeginningofmyacademiccareerandforhavingdiscussed,readandcommented

onmyworkwithsuchcarefulattention.

IalsothankProf.PaulaFikkertforhavingmeinNijmegenforafewmonthsandfor

providingmewithagreatscientificenvironmentattheMaxPlanckInstitute.

Agradeço, também,àquelessemosquaisesta tesenãoexistiria.AosgémeosClarae

João, ao JoãoL., à Lumae tambémàEva, eu agradeço terem‐meoferecido tantaspalavras,

cheiasdebrincadeirasexi‐corações.Foicomelesquepercebi,pelaprimeiravez,comopode

ser‐se pai emãe, e foi com elas que relembrei amaravilha de ser criança. Aos pais,mães,

familiaresepessoaspróximasdascriançasqueacompanhei,agradeço,também,apaciênciae

agenerosidadecomquemereceberamao longodopercursodegravaçõesedepoisdele.À

JoanaeàInês,asduascriançascujasproduçõeseunãorecolhimasquefazempartedocorpus

queutilizei,aosseuspaiseàssuasmães,agradeçotambém.

To Fundação para a Ciência e a Tecnologia (FCT) I thank for the funding that was

given tome through a PhD grant, aswell as for the financial support for conferences and

tuition.

IamalsothankfultoFundaçãoCalousteGulbenkianandtoFundaçãoLuso­Americana

paraoDesenvolvimento,forhavingfundedmyparticipationintheLSASummerInstitute2007,

payingformytuitionandhousing.

Agradeço à D. Arlete e à D. Fátima, funcionárias dos Serviços Académicos da

FaculdadedeLetras,oprofissionalismo,acompetência,asimpatiaeacompreensão.

I thank zillions of times Prof. YvanRose, aswell asGregHedlund and all the great

research and technical team of Phon database. I suppose that was the number of times I

bugged Yvan with all kind of more or less insignificant problems related to the database

implementationandimproving.Yvanalwaysdemonstratedanendlesspatience,anefficient

responsiveness,acontagiousenthusiasmandawisemanagementofstress!Yvan,yourhelp

waspriceless!Thankyousomuchforeverything.

Ialsothankthewiseadvicesofmoreexperiencedcolleaguesandfellows:AnaLúcia

Santos,NéliaAlexandre,SóniaFrotaandMarinaVigário.ToAnaLúciaandNélia,Ithanktheir

x

optimism,pragmatism,andthewisdominmakingmeseewhatreallymatters. Iverymuch

appreciated their words of friendship and support. To Sónia and Marina, I thank the

discussionsonphonologyandtheacquisitionofphonologyand,also,thewordsofsupport.I

admiretheirvisionofwork,theircapacitytoworkandtoimplementnewideas.

IwishtothanktheLSASummerInstitute2007organization,forthegreatwest‐coast

bias seminars. For the first time inmy academic life, I looked at language froma different

angle.And,assaidbyP.W.Bridgman,"thereisnoadequatedefense,exceptstupidity,against

theimpactofanewidea."Inparticular,IwanttothankProf.MarilynVihmanandProf.Rory

DePaolis, for having discussed with such enthusiasm many issues on Phonological

Development that intrigued me for years. I hope this will be the beginning of a fruitful

collaboration.

To the remainder colleagues/friends I thank the support and the productive

discussions. I verymuch thank Letícia for the discussions on phonological acquisition and

datacollection.Also,Ithankherthemotivationalsupportinthetoughestmoments.

To Fátima Baia and Cristiane Silva, I am thankful for having shared withme their

ideasonlanguageacquisition,ontheacquisitionofphonologyandtheacquisitionofprosody.

ToFátima,inparticular,Ithanktheenthusiasm,thesharedbeliefsandthefiercediscussions

viaSkypeonthechallenging issueof theacquisitionofwordstress inPortuguese. Ibelieve

the topic of this dissertation ended‐up being a great excuse to start a beautiful and long‐

lastingfriendship!

Beforeproceeding,Iwouldliketoremember,inmemoriam,TiagoFreitas,acolleague

and,mostimportantly,afriendwhoweallmissverymuch.Istillfeelthatourworldwithout

youisfarlessinteresting,dearTiago.

Awordofgratitude isalsoduetoSharoneBackers, thenative‐speakerwhohadthe

tough responsibility of reviewing theEnglish in this dissertation.Also, I thank Sharone for

herwordsofunderstandingandmotivation.

ToProf.LuísaLoura,Ithankthehoursdedicatedtothediscussionofthedataonthe

acoustic analysis, in understanding them, in 'dismantling' them and putting them back

togetheragain,withaCartesianthoroughness.

Also, I thank Prof. FernandoMartins, for the classes on acoustic phonetics and the

precioushelpontheacousticanalysisconductedinthisresearchwork.

IalsothankmydearfriendsfromtheLSAgang(Ana,Geoff, Jette,RuiandYonatan),

for the discussions about everything and nothing (Linguistics included) and for the great

moments spent on the top of a mountain, somewhere in Yosemite National Park, in the

middleofthewilderness.

xi

Ialsowanttoaddresssomewordstothe'researchfamily'towhichIammoreorless

relatedtoby'bloodbonds',theSOCASresearchteam.IthankLeonelandTammy‐theheads

ofthegroup‐fortheirwiseadvice,fortheirfriendshipandsupport.TothoseIamcloserto,

Margarida,Sara,SofiaandTomás,Iwouldliketosaythatisanimmensepleasuretocounton

yourfriendship,intelligenceandgoodhumor.Itisfantastictosharelifewithpeoplelikeyou,

whodonotseelifefromoneangleonly.Ourconversationsaboutscienceandanimmensityof

otherlessrelevantstuffmademylifemuch,muchinterestinginamomentwhenourvisionof

theworldnarrowsconsiderably. Ibelievemoreandmore thatour life ismadeof thegood

peoplethatarepartofitandallofyouare,definitely,agoodpartofmylife.

ToAnaCastro,my'friendatfirstsight',thepersonwithwhomIsharedsuchgoodand

badmomentsallalongthisproject,Ithankforthepatienceandthewisdom.Anita,youwere

astrongpillarwhosustainedmefromfallinginveryparticularmomentsandyourwordsof

supportwerecrucialtofinallyreachtheendpoint.Yourexampleinspiredmeandyourwise

advicehasledmetotheplacewhereIwantedtobe.

To Teresa, my colleague and friend, I am immensely thankful for the knowledge

transfer, the discussions and opinions. I do not know how to thank you, however, for the

endlesshoursonthephone,sharingoursilliestandhappiestthoughts,andalsosharingour

anguishes. I thank,especially, theemailschangedonadailybasis inthetoughestmoments,

and the constant support and belief. I also thank Teté for having called me 'aquelas‐

personagens‐que‐são‐passadas‐a‐ferro‐mas‐que‐depois‐se‐põem‐em‐pé'.WithyouIlearned,

from the beginning, that the force is definitely with us, for we are capable of amazing

achievements.And if, throughout thiswholeprocess, Ihavecontributed tohelpyouwitha

tenthofwhatyouhelpedme,Iwillbetremendouslyhappy,forthatisalot.

Finally,Iwouldliketothankthemembersofmyfamily,thisgreatfamilythatisone

only.

ToMarianaManaBanana,IapologizefornotbeingabletoplayWiiandridethebike

forsuchalongtime.Ialsoapologizefortemporarilyhavingpreventedyoufrombeingwith

usthewayyoudeservetobe,playinggamesorsimplynotdoinganything.Thoughyou,aten

year‐old,donotknowaboutthis,youtoohavebeenagreatexampleofstrengthtome.

To Tita, Aurora and all the crazy family I am proud to be part of, I thank for the

patiencewithwhichtheyobservedusfromadistance.Weknow(meandRui)thatwearean

important bond in this immense 'Kusturica family' and we were touched with the global

mobilizationto'leaveusalone',withoutneverreallyhavingleftusalone.

I thank Manuela, to whom I cannot simply call 'mother‐in‐law' (ah, what an ugly

name!), for her unconditional support, for the role model and the right words, in every

moment.

xii

Tomyparents,Ithankthesupport,understandingandpatience.Inthemiddleofthis

'crazy life'ofmine,youalwaysunderstoodme, lovedmeandstoodbyme,nomatterwhat.

That,indeed,wasagreathappinessinthegoodtimesandagreatcomfortinthehardones.I

thankyouforhavingtaughtmetonevergiveup,toberighteousandfair.Thatwas,definitely,

thegreatestheritage that I got fromyouand, for all that, Iwant to tell you that I loveyou

very,verymuch.

Also, I thank my lovely brother, Simão, and my sister‐in‐law, Cláudia, for their

patience.Iknowyoumissedme,Si,butbelieveme,Imissedyoumore.Tomytwoyearsold

niece, Carolina, I thank the smiles, kisses and hugs offered uninterestedly, and for having

amazed me with such great achievements. When I look at you, very often I find myself

wonderingaboutwhatreallymattersinlife.

Finally,you,my love,mypartnerofallhours, themost importantperson inmy life,

thepersonwhosupportedandinspiredmethemostinthisparallelpaththatwehavetraced

andpursuedtogether,thepersonIchosetolivewithwithoutsigninganypaperoranything,

forthatdoesnotmeanathingwhencomparedtothe immensityofmylove. Idedicatethis

work to you, for you deserve it. For having showed me your love. For the extraordinary

person that you are. For your unshakable principles. For your immense humanity, your

infinitegenerosityandyourcommitmenttolife.Youmakemebetter,youmakemegood.You

areallthatmatters.Amidsttheadversitiesthatwehavebeenthrough,Iwanttotellyouthat

thetruthisalwayswhatreallytransformsandmakesusstronger.Unbreakable.

xiii

RESUMO

Na presente dissertação pretende fazer‐se uma descrição do percurso de

desenvolvimentofonológicodecincocriançasportuguesas,noquedizrespeitoàaquisiçãodo

acentoprimáriodepalavra(aquireferidocomoacentodepalavra).

Naperspectivadaprodução,queéaquelaqueadoptamosnestadissertação,oestudo

daaquisiçãodoacentodebruça‐se,sobretudo,sobreomodocomoascriançasproduzemas

relaçõesdeproeminência(i.e.,assílabas fortese fracas)existentesnaspalavrasprosódicas

dalínguaalvoe,também,sobreomodocomoascriançasadquiremoalgoritmoquesubjazà

posiçãodoacentonessamesmalíngua.

EmPortuguês, o algoritmodoacentodepalavranãoé consensual.Andrade&Laks

(1992), Lee (1995), Mateus (1983) e Pereira (1999) defendem um acento de natureza

estritamentemorfológica,sendoodomíniodasuaaplicaçãooradicalderivacional,nosnão‐

verbos, e a palavra lexical, nos verbos. Bisol (1992, 1993) eWetzels (2006) defendemum

acentodetiporítmico,sensívelaopeso.Wetzels(2006)assumeaindaqueapenasnosnão‐

verbos o acento é sensível ao peso silábico. De acordo com Bisol (1992, 1993) eWetzels

(2006),domíniodoacentoéapalavralexical,emnão‐verboseemverbos.

As duas análises fazem diferentes predições para a aquisição. Caso se assuma um

algoritmodebase morfológica,oacentorecainaúltimasílabadoradicalderivacional, nos

não‐verbos, favorecendoproduçõescujasílaba finalé forteepodendoconduzirascrianças

portuguesas para uma tendência inicial iâmbica (Santos, 2007). Pelo contrário, caso se

assumaumalgoritmodeacentodebaserítmica,oacentorecai,emgeral,napenúltimasílaba

da palavra lexical, favorecendo produções cuja penúltima sílaba é forte, conduzindo as

criançasqueadquiremPortuguêsparaumatendênciainicialtrocaica.

O estudo da aquisição do acento tem sido um tópico abordado na literatura sobre

aquisição da fonologia,muitas vezes relacionado com a aquisição do formato das palavras

(e.g.,Demuth,1995;Fikkert,1994).Aobservaçãodoformatodaspalavrasproduzidaspelas

crianças, em termosdenúmerode sílabas –mono‐, di‐, tri‐ oupolissílabos – e de tiposde

sílabas – fortes ou fracas –, tem sido um dos objectos de análise para inferir acerca do

processamento do algoritmo do acento na língua alvo. Adicionalmente, a observação das

estratégias de truncação (e.g., toMAto produzido como [»mA˘do]), de reduplicação (e.g.,

chaPEAU produzido como [po»po]) e de epêntese (e.g., door produzidos como [√»do˘]) tem

fornecidoevidênciaparaoestudodaaquisiçãodoformatoedoacentodepalavra.

Umoutromeiousado,ainda,noestudodaaquisiçãodoacentodepalavraedeoutros

aspectos rítmicos e entoacionais das línguas é a análise acústica, onde se medem,

objectivamente, os parâmetros responsáveis pelas proeminências, i.e., a frequência

xiv

fundamental, intensidadeeduração(e.g.,Pollock,Brammer&Hageman,1993;KehoeStoel‐

Gammon&Buder,1995).Osresultadosapontadosnaliteraturasugeremque,antesdo2;0,as

criançaspodemnãocontrolarosparâmetrosacústicosresponsáveispelaproduçãodoacento

depalavranalíngua‐alvo,equenemsempreascriançascolocamoacentodeacordocomo

dapalavra‐alvo.Aanáliseacústicaconduzidanumaamostradediscursodeduasdascrianças

que fazem parte do corpus desta dissertação mostrou que, no início, há uma grande

variabilidadenaproduçãodosparâmetros acústicos edo formatodaspalavras, quer entre

crianças,quernasproduçõesdamesmacriança.Osresultadosdaanáliseacústicaconfirmam,

ainda, que os parâmetros acústicos usados na produção do acento podem não estar

dominadosantesdos2;0.

Em algumas das línguas estudadas (Alemão, Castelhano, Catalão, Francês, Grego,

Hebreu,Holandês, Inglês,PortuguêsdoBrasil,Quiché,Sesotho,etc.),aaquisiçãodoaspecto

prosódicoacentodepalavratemencontradoresultadosvariados.Namaiorpartedaslínguas

germânicas, uma tendência para monossílabos é observada nas produções iniciais das

crianças (e.g., Demuth, 1996b; Fikkert, 1994; Kehoe, 1998). Mais tarde, amaior parte das

produçõesdascriançastêmumformatodissilábicoetrocaico(e.g.,BUcketéproduzidocomo

[»b√kIt] e book como [»bUk´]). Neste estádio, os dissílabos iâmbicos são truncados para

monossílabos (e.g.,giRAFFE é produzido como [»RQf]) e apenasmais tarde são produzidos

como iambos (e.g., [dZ´»rQf]). Também neste estádio, palavras de maior extensão, são

truncadas (e.g., avoCAdo é produzido como [»kAdo], Elephant é produzido como [»Qf´] e

kangaROO como [wu˘]). Em estádios subsequentes, as palavras de maior extensão são

produzidas comopadrão acentual do alvo.Note‐se, no entanto, que as línguas germânicas

estudadas são trocaicas e possuidoras de um acento que é sensível ao peso silábico. O

domíniodoacentoé,nestaslínguas,apalavralexical.

Durante muito tempo, a hipótese de um viés trocaico universal (Allen & Hawkins,

1979,1980)prevaleceunosestudossobreaquisiçãodoacentoedoformatodepalavranas

várias línguas.Porém,em línguasdeorigemromânicaouatéem línguasdeoutras famílias

(e.g.,Hebreu,Francês,EspanholouGrego),osresultadosnãosãoconsensuais(Bat‐El,2009;

Braud,2003;Demuth,2001b;Tzakosta,2004,respectivamente).Porexemplo,emPortuguês

doBrasil(PB),variedadequepartilhasemelhançasevidentescomoPortuguêsEuropeu(PE),

osresultadossão,desdeosprimeirosestudos, intrigantes, jáqueoPortuguêstemumforte

ritmo trocaico, e uma tendência iâmbica inicial foi verificada em alguns dos estudos

conduzidosatéagora(e.g.,Santos,2007a;Stoel‐Gammon,1976).

Osresultadosdaanálisedospadrõesdeacento,nosdadosdas5criançasPortuguesas

observadas nesta dissertação,mostram que, de facto, reduplicações e epênteses sãomuito

xv

frequentes nas produções iniciais e contribuem grandemente para uma tendência iâmbica,

que é aparente, já que os resultados tambémmostram que os monossílabos predominam

numa fase inicial. A simultaneidade de produções monossilábicas, onde reduplicações e

epênteses são possíveis, e a frequência de truncações em formas‐alvo de tipo trocaico e

iâmbico, levam‐nos a propor que a representação inicial das palavras em PE seja

monossilábica,comumasílabaopcionalàesquerda,semqueumpéestejaaserprocessado.

Maistarde,osdissílabos(troqueuseiambosnãoreduplicados)emergem,semqueosdados

demonstrem um predomínio de formas trocaicas ou iâmbicas. Num estádio subsequente,

umapreferênciaporformastrocaicastendeaemergir,querpelapreferênciadetruncaçãode

palavrasdetipoiâmbicoparamonossílabos,querpelapreferênciadetruncaçãodeformasde

tipofraco‐forte‐fraco(e.g.,'saPAto')paratroqueus.Estesresultadosmostramque,apesarda

variabilidade inicialedatendênciaaparenteparaumpédetipo iâmbico,oPEcomporta‐se

comooutraslínguasderitmotrocaico(nomeadamente,oHolandês,oInglêseoEspanhol),no

quedizrespeitoàaquisiçãodospadrõesdeacento.

Nesta dissertação testámos, ainda, a sensibilidade das crianças portuguesas para

aspectos relacionados com a aquisição da flexão nominal (nomeadamente, analisando a

aquisição da vogal temática) e verbal (analisando as formas verbais mais produtivas).

Complementarmente,analisámospalavrasondeasensibilidadeaopesopoderiaserevidente,

emconcreto,palavrascomo'aMOR'e'LÁpis',comsílabaspesadasfinais,tónicaseátonas.

Osresultadosdaanáliseda interacçãoentreamorfologiaeaaquisiçãodospadrões

deacentosãoinconclusivos.Porumlado,osdadosmostramque,inicialmente,oscontrastes

degéneronãosãoproduzidos,sugerindoque,defacto,aflexãonominaldegéneroaindanão

emergiu,motivandoumamaiorproduçãodeformasatemáticas(iambos).Poroutro,osdados

não fornecem evidências claras para um pé inicial iâmbico. As crianças podem produzir

reduplicaçõeseepêntesescomasílabafinal forte,masiambosetroqueus‐alvosãotambém

sujeitosatruncaçõesemqueasílabaátona–enãoapenasavogaltemática–éapagada.Este

padrãopareceinfirmarumainteracçãodoacentocomamorfologia,nasproduçõesiniciaisde

nãoverbos.

Os dados mostram ainda que, inicialmente, verbos polissilábicos são evitados. O

primeiropadrãoaseradquiridoé,emgeral,umtroqueuquecorrespondeaotemadoverbo,

i.e.,à3ªp.sg.doPresente/Imperativo.Ossufixosflexionais,nomeadamenteosdoInfinitivoe

da 3ª p.sg. do Pretérito Perfeito, que revelariam a emergência da morfologia verbal, são

produzidos mais tarde. Os dados obtidos mostram, no entanto, uma sensibilidade das

criançasparaaclassedepalavrasduranteaaquisiçãodoacento,jáquenoiníciodaprodução,

não‐verbos mono– e polissilábicos estão presentes no léxico seleccionado pelas crianças

observadas,masapenasverbosmonossilábicossãoseleccionadosnomesmomomento.

xvi

Osresultadosdaanáliseàsensibilidadeaopesosilábicomostram,porumlado,queas

palavras como 'LÁpis' raramente são seleccionadas pelas crianças e que o seu percurso de

aquisição é diferente do de palavras como 'aMOR'. Esta assimetria parece confirmar o

carácterexcepcional emarcadodasprimeiras relativamenteàs segundase,poroutro lado,

indicaque,de facto,ascriançasportuguesaspodemsersensíveisaopesosilábiconasílaba

final durante a aquisição do acento de palavra. De facto, num dado momento do

desenvolvimento, apesar de produzirem correctamente palavras como 'aMOR', as palavras

como'LÁpis'sãotruncadasouproduzidascomsílabafinalleve.

Emsuma,ascriançasportuguesasmostramsersensíveisaumalgoritmodeacentode

base rítmica, com um comportamento semelhante ao das crianças que adquirem línguas

trocaicas,masumainteracçãocomamorfologianãofoiexcluída.

Palavras­chave:aquisição,acento,algoritmo,troqueu,palavra

xvii

ABSTRACT

In this dissertation, we aim at describing the acquisition pathof five Portuguese

children,inrespecttoprimarywordstress(henceforthwordstress).

Portuguese word stress has been widely discussed, some authors arguing for a

morphology‐based algorithm (Andrade & Laks, 1992; Lee, 1995; Mateus, 1983; Pereira,

1999)andothersarguingforaweight‐basedone(Bisol,1992;Wetzels,2006).

Cross‐linguisticinformationontheacquisitionofwordstresssuggestedthatchildren

acquiringGermaniclanguagestendtomirrorthetrochaictendencyofthetargetsystem,asa

result of the application of a rhythmic stress algorithm. However, in other languages,

conflicting results were attested. Acoustic analyses of early words' production in some

languages additionally suggested that, before 2;0, children might not master the acoustic

parametersforwordstress,andthattheymightnotplacestresstarget‐like.

The acoustic analysis conducted on a speech sample of two children in our corpus

showed a great initial variability in the production of the acoustic parameters and in the

word shapes,bothwithin the samechild andbetween children.The resultsof theacoustic

analysis further confirm that the acoustic parameters to derive word stress might not be

mastereduntil2;0.

The results on the production of stress patterns in the speech of 5 Portuguese

monolingualchildrenshowthat,despiteaninitialdifference(inwhichweobservedaneutral

emergence of trochees and iambs, and a tendency for reduplications and epentheses), EP

resembles other trochaic languages (namely Dutch and English). The predominant initial

production of monosyllables, along with reduplications and epentheses at the left‐edge of

words, which heavily contribute to an apparent iambic tendency, lead us to propose a

monosyllabicrepresentationforearlywordsinEuropeanPortuguese(EP),withanoptional

syllableatitsleft‐edge.Later,non‐reduplicateddisyllablesemerge,withoutfavoringtrochaic

or iambicwords.Afterwards,apreference for trochaicwordswasnoticeable,both through

thepreferentialtruncationofiambicwords,andthroughthepreferentialtruncationofweak‐

strong‐weakwordsintotrochees,ratherthaniambs.Trisyllableswerethelaterwordshape

tobeacquired,especiallyiftheyhadanextrametricalsyllable(/SWW/).

Finally, we tested the Portuguese children's sensitivity to aspects related to

morphology and syllable weight. The results did not bring indisputable evidence for

morphological constraints interacting with word stress acquisition, early non‐verbs are

produced similarly, irrespective of their morphological constituency and the target stress

pattern.Earlyverbformsemergelaterandgenerallyconformtoatrocheeandaverbtheme.

Theresultsonweight‐sensitivityshowedthatwordswithfinalheavystressedsyllables(e.g.,

aMOR'love')areacquiredearlierthanwordswithfinalheavyunstressedsyllables(e.g.,LÁpis

xviii

'pencil'), and the latterare initially truncatedand laterproducedwitha light final syllable.

Theseresultssuggestthatchildrenrecognizethatstressin/'CV.CVC/wordsisnotinthefinal

syllable,which,therefore,cannotbeheavy.

Overallresultsprovideevidenceforanalgorithmforwordstressthatissensitiveto

therhythmicpropertiesofthetargetlanguage,thoughaninteractionwithmorphologyisnot

categoricallyrejected.

Keywords:acquisition,stress,algorithm,trochees,word.

xix

Symbolsandabbreviations

EP–EuropeanPortuguese

BP–BrazilianPortuguese

W‐Weak

S–Strong

[CV1CV2]‐Non‐reduplicateddisyllables

[CV1CV1]/[S2]‐Disyllabic(simple)

reduplications

[CVCV...]‐Multiplereduplications

[fS]–Fillersound+syllable

ω‐Prosodicword

Σ‐Foot/Feet

µ‐Mora

σ‐Syllable

C‐Consonant

V‐Vowel

G‐Glide

VN‐Nasalvowel

VGN‐Nasaldiphthong

N‐Noun

V‐Verb

dim.‐Diminutive

augm.‐Augmentative

fem.‐Feminine

masc.–Masculine

pl.‐Plural

sg.‐Singular

inf.‐Infinitive

subj.‐Subjunctive

pres.‐Present

simp.past‐SimplePast

pastimperf.‐PastImperfect

pluperf.‐Pluperfect

fut.‐Future

cond.‐Conditional

imp.‐Imperative

WM‐WordMarker/ClassMarker

Nr‐Numbersuffix

TV‐ThemeVowel

TM‐Tense/Moodsuffix

PN‐Person/Numbersuffix

*‐Ungrammatical/Notattestedinthe

language

]‐morphologicalboundary

fam.‐familiar/informalspeech/baby‐talk

xx

xxi

IPAsymbols(usedinEuropeanPortuguese):

Consonants:

Plosives‐[p]pata,[b]bata,[t]fato,[d]fado,[k]cola,[g]gola

Fricatives‐[f]faca,[v]vaca,[s]caça,[z]casa,[S]chá,[Z]já

Nasals‐[m]mota,[n]nota,[¯]manha

Liquids‐[l]mala,[R]caro,[¥]malha,[…]caldo,[{]carro

Vowels:

Oral‐[a]pá,[E]pé,[ç]pó,[å]cama,[e]cena,[o]bolo,[i]fita,[ˆ]pente,[u]fumo

Nasal‐[å‚]canto,[e‚]lente,[o‚]ponte,[i‚]tinta,[u‚]mundo

Glides:

Oral‐[j]pai,[w]pau

Nasal‐[j‚]mãe,[w‚]pão

Primarywordstress:[»]casa[»kazå]

xxii

xxiii

TABLEOFCONTENTS

0.INTRODUCTION 1

1.ADESCRIPTIONONWORDSTRESS 7

1.1.Generalobservationsonwordstress 71.1.1.Wordstressandtheprosodichierarchy 91.1.2.Metricalaspects 111.1.3.Quantity‐sensitivityandrhythmicpatterns 131.1.4.Acousticfeaturesofwordstress 16

1.2.Portuguesewordstress 171.2.1.GeneraldescriptiononPortuguesewordstress 171.2.2.AcousticdescriptionofEPwordstress 231.2.3.Metricalanalyses 241.2.4.Syllabletypes,wordshapesandwordstressinEP‐frequencyinformation 401.2.5.IssuesraisedinthedescriptionsonPortuguesewordstress 431.2.5.1.ThepertinacityoffeetinPortuguese 431.2.5.2.Weight‐sensitivityormorphology‐dependence. 45

2.THEACQUISITIONOFWORDSTRESS 51

2.1.Acousticstudiesonwordstressacquisition 52

2.2.Theacquisitionofwordstressacrosslanguages 552.2.1.Earlywords'representation 552.2.2.Theacquisitionofstresspatterns 59

2.3.Thenatureandroleoffillersoundsinearlyprosodicacquisition 77

2.4.Thenatureandroleofreduplicationsinearlyprosodicacquisition 84

2.5.TheacquisitionofwordstressinPortuguese 862.5.1.Theacquisitionofwordshapesandstresspatterns 862.5.2.FillersandreduplicationsinthepathofprosodicacquisitioninPortuguese 992.5.3.AcousticstudiesonwordstressacquisitioninPortuguese 103

2.6.Summaryofthemainissues,researchquestionsandhypotheses 104

xxiv

3.METHODOLOGICALASPECTS 109

3.1.Thedata­subjectsandsessions 109

3.2.Databaseproperties,transcriptionandnotationcriteria 1143.2.1.Databasedescriptionandfunctioning 1143.2.2.Criteriaoftranscriptionandinsertioninthedatabase 1163.2.2.1.Criteriaforthetranscriptionofthetargetwords 1163.2.2.2.Criteriaforthetranscriptionofthechildren'sproductions 1183.2.2.3.Codingandanalyzingfillersoundsandreduplications 119

3.3.Criteriaforemergenceandacquisition 120

4.APRELIMINARYACOUSTICSTUDYONTHEACQUISITIONOFWORDSTRESS 123

4.1.Method 125

4.2.Results 1294.2.1.Qualitativeanalysis 1294.2.1.1.Fundamentalfrequency 1294.2.1.2.Intensity 1324.2.1.3.Duration 1344.2.1.4.Summaryforthequalitativeanalysis 137

4.2.2.QuantitativeanalysisoftheacousticparametersforEPwordstressacquisition 137

4.3.Discussion 141

4.4.Summary 144

5.THEACQUISITIONOFSTRESSPATTERNSINEP 147

5.1.Results 1495.1.1.Productionpatternsinearlywords 1495.1.1.1.Wordshapepreference 1505.1.1.2.Theshapeofearlywords 1565.1.1.3.Summaryfortheproductionpatterns 166

5.1.2.Faithfulnesstothetarget 1675.1.2.1.Monosyllables 1675.1.2.2.Disyllables 1735.1.2.3.Trisyllables 1925.1.2.4.Summaryforthefaithfulnesstothetarget 197

5.1.3.Productionstrategies 198

xxv

5.1.3.1.Monosyllables 1995.1.3.2.Truncationindisyllables 2045.1.3.3.Truncationintrisyllables 2135.1.3.4.Stressshiftindisyllables 2245.1.3.5.Summaryfortheproductionstrategies 230

5.1.4.Summaryofresults 231

5.2.Discussion 2345.2.1.DevelopmentalpathfortheacquisitionofwordshapeandstresspatternsinEP 2345.2.2.Ontheearlystages‐theapparentiambictendency 255

5.3.Summary 267

6.TESTINGTARGETANALYSESONWORDSTRESSWITHACQUISITIONDATA 271

6.1.Results 2796.1.1.Ontheinteractionofmorphologyintheacquisitionofwordstress 2806.1.1.1.Thedistributionofwordclassesperwordshape 2806.1.1.2.Lookingcloseratthenounparadigm 2876.1.1.3.Lookingcloserattheverbparadigm 3016.1.1.4.Summaryformorphologyinteraction 322

6.1.2.Ontheroleofweightintheacquisitionofwordstress 3236.1.2.1.Deletionofunstressedsyllables 3266.1.2.2.Wordswithheavystressedandheavyunstressedsyllables:emergencepath 3276.1.2.3.Wordswithheavystressedandheavyunstressedsyllables:stressshift 3396.1.2.4.Summaryforweight 346

6.1.3.Summaryofthemainfindings 347

6.2.Discussion 3496.2.1.Discussingtheinteractionofwordstresswithmorphology 3496.2.2.Discussingweight 361

6.3.Summary 369

7.SUMMARYANDCONCLUSIONS 373

REFERENCES 383

xxvi

1

‘Iwishlifewasnotsoshort,’hethought.

‘Languagestakesuchatime,andsodoallthethingsonewantstoknowabout.’

J.R.R.Tolkien

0.Introduction

In this dissertation we will describe the acquisition of primary word stress in

EuropeanPortuguese1,withanemphasisontheearlystagesofwordproduction.

Language acquisition and development has been the subject of great interest for

linguists,psychologists,psycholinguists,philosophersandbiologistsover thepast70years

(Jakobson, 1941/1968). Either within a continuist (e.g., Locke, 1983; Davis & MacNeilage,

2000) or a maturacionist perspective (e.g., Smith, 1973; Stampe, 1979), the study and

systematization of the children's early linguistic behavior and its development across time

have provided empirical arguments for theoretical claims on language acquisition and

development, and, more generally, on how language is cognitively represented. The

observation of children's linguistic behavior, as well as the study of cross‐linguistic

differencesandsimilaritiesduringtheprocessof languageacquisitionisoneofthewaysto

investigatehowthe'knowledgeoflanguage'isstructured(Chomsky,1986).

Studying languageacquisition ingeneral, andphonologicalacquisition inparticular,

may provide useful insights on the nature of phonological processes, not only at the early

stagesofwordproduction,butalsoacrossdevelopment.

Theacquisitionofwordstressisanaspectofphonologicalacquisitionthatinteracts

withwordshapeandwordstructure.Thestudyofwordstressacquisitionprovidesuswith

cuesonhowchildrengofromastagewherewordsarehardlyrecognizabletoastagewhen

full words are uttered. The investigation on word stress acquisition is not limited to the

acquisition of this supra‐segmental feature in a language ‐ stress itself ‐, but it is also the

study of how children build words in the language they are acquiring. Besides the

'constructionwork' that underlies the acquisitionofword stress, the challengeof studying

word stress acquisition relies, indeed, on the fact that stress often depends andmight be

relatedtootheraspectsoflanguageacquisitionsuchassyllables,feet,wordstructure,word

size and morphological information. In fact, studying the acquisition and development of

wordstressisalsoinvestigatingasetofphonologicalandphonological‐relatedissues,which

can provide insights on the structure of the language across development. Furthermore,

1HenceforthEP.

2

investigatingwordstressacquisitioninEPisparticularlychallenging,giventhecontroversial

statusofstressinthelanguage.

Inrecentyears,manyissuesontheacquisitionofwordstressandwordshapehave

been introducedanddeveloped,andseveralassumptionshavebeenchallenged.Oneof the

topics thathavebeendebatedover thepast3decades,which is relative to thenature and

representationofearlyprosodicwordsinthelanguages,isthe'universaltrochaicbias'.The

'trochaic hypothesis' was proposed by Allen & Hawkins (1979, 1980), and suggested that

children have, by default, a trochaic (SW) word pattern (e.g., tiger). The claim from the

authors was based upon three types of evidence: (i) children's early lexicon, (ii) from

children'struncatedproductionoftrisyllabicwords(specifically,/WSW/),and(iii)languages

nurseryrhymes.Theauthorsfoundcross‐linguisticevidencesupportingauniversaltrochaic

constraintinphonologicalrhythm.However,theprimacyofthestudyofGermaniclanguages

(mostly English, German and Dutch) ‐ which have a clear trochaic pattern ‐ may have

shadowed the study of prosodic acquisition in other language families, namely Romance,

Semitic and other language families. Indeed, data from non‐Germanic languages (French,

K'iche', Brazilian Portuguese2 and even Spanish3) have challenged that assumption,

suggestingthatthe'universaltrochaicbias'mightnotbeuniversal.

Someoftheissuesthathavebeenaddressedintheliteratureconcerningwordstress

andwordshapeacquisitionreferto:

(i) howandwhendochildrenacquireprimarywordstress;

(ii) whether children's early words match a default monosyllabic or disyllabic

template;

(iii) whetherchildren'searlydisyllabicproductionsmatchadefaulttrochaicfoot.

Theseissueshavebeendiscussedintheliteratureofseverallanguages,butnotEP.

InPortuguese(EPandBP),thereisanongoingdebateonthenatureofstressinadult

language. The assumption of a strictly morphology‐based algorithm (e.g., Andrade, 1992;

Mateus,1983;Pereira,1999),bothforverbsandnon‐verbs'systemhasbeenconflictingwith

theassumptionofaweight‐basedrulefornon‐verbsandamorphology‐basedruleforverbs

(Bisol, 1992, 1999; Wetzels, 2003, 2006)4. The issue at stake ‐ the nature of word stress

algorithm‐hasbeendebatedforbothvarietiesofthelanguage(bothEPandBP),thoughvery

2HenceforthBP.3 Despite the general assumption for the universal trochaic bias in early phonological development, Allen &Hawkins(1979)refertoFrench,SpanishandBPaslanguagesthatdonotconformatrochaicpattern.4 A detailed description on the different approaches to word stress in Portuguese will be given in Chapter 1,section1.2.

3

oftenauthorslimittheiranalysestoasingleone(EPorBP).Thedifferentanalysesonword

stress make different predictions for acquisition: if we assume a morphology‐dependent

algorithm, stress fall, in general (in non‐verbs), on the domain‐final syllable (the stem),

favoring word‐final stress and driving Portuguese‐speaking children into an initial iambic

tendency (Santos, 2007a,b5). Conversely, if we assume a strict rhythmic, weight‐sensitive

stressalgorithm,stressfalls,ingeneral,inthepre‐finalsyllableofthelexicalword,favoring

earlyproductionsinwhichthepenultimatesyllableisstressed.

TheworksfromSantos(2001,2007)onBPwordstress,raisedimportantquestions

andsuggestfutureresearchpathsforthestudyofwordstressacquisitioninPortuguese.The

authorclaimsforaniambicearlybiasinBP(themajorityofearlywordsinBrazilianchildren

havea[WS]shape)anddefendsthatBrazilianchildrencannotproduceandacquiretrochees

due to the interactionwithmorphology,since trochaic–butnot iambic–non‐verbsbeara

final word marker. Until the moment when children are able to produce morphological

contrasts (gender and diminutive suffixes), they are not able to produce trochaic words.

Indeed, the influence of morphological information in the acquisition of word stress and

wordshapehasalsobeenthesubjectofanalysis insomelanguages,suchasDutch,English,

German, Spanish, French, Greek, Hebrew or BP. However, the research on the topic

(phonology‐morphologyinterface)hasmainlyfocusedonprosodicdomainsabovethelexical

word (e.g., Genari & Demuth, 1997; Demuth & Tremblay, 2008), namely on sequences of

function words (determiners and auxiliaries, mainly) followed by a noun. Though many

languages studied from aword stress andword shape acquisition perspective areweight‐

sensitive ‐ namely, Germanic languages ‐, no Romance language has been tested for the

potentialeffectofweight‐sensitivityduringacquisition,evenifthetargetlanguagedoesnot

showevidenceforitsrelevanceinthesystem.

Inthisdissertationwewilldescribeanddiscusstheacquisitionofwordstresswithan

emphasisintheearlystages.Inparticular,withthisworkweaimto:

(i) identify thedevelopmentalpath for theacquisitionof stresspatternsandword

shapeinEP;

(ii) provide empirical evidence for the ongoing discussion on the nature of word

stressinthetargetlanguage;

5Throughoutthisdissertation,wewillrefertoSantos(2007a)asSantos(2007).Otherwise,wewillspecifySantos(2007b).

4

(iii) compare the acquisition of stress patterns in EPwith the acquisition of stress

patternsandwordshapeinother languages(namelyDutch,EnglishandBP), in

order to contribute for a discussion on the nature of this prosodic aspect in

humanlanguages.

Basedonthespontaneousspeechproductionsof5Portuguese‐speakingchildren,we

aimat identifying their acquisitionpath in theprocessofmasteringword stress andword

shape, by means of developmental stages. In addition, we will compare the acquisition of

word stress inEP,with theacquisitionofword stress inother languages (namely,English,

Dutch,French,German,Greek,Spanishand,inparticular,BP),anddiscusspreviousmodelsof

word stress acquisition. Finally,we intend to provide evidence for or against the different

targetanalysesforwordstress.

Acousticstudieswereundertakensincetheearlyworksonwordstressacquisitionin

production(Allen&Hawkins,1980).However,onlyfromthemidnineties(19936)veryfew

studieswerededicated to theanalysisof theacousticpropertiesofword stressduring the

early stages of phonological acquisition. Being a phonetic and phonological aspect of a

language,wordstresshasspecialinterestinthestudyofphonologicalacquisition,asitisnot

easily identified in the early period of word production. Childrenmay not yet control the

acoustic parameters necessary to utter word prominence and/or children may not have

acquired the phonological rule to deriveword stress. The acoustic study ofword stress is

thusnecessary(i) tounderstandwhatarethephoneticcorrelatesofwordstress inagiven

timeofearlyphonologicaldevelopment,(ii)whatcanweinferfromthedataontheshapeof

early words and (iii) to provide further empirical evidence for the claims based on

perception‐basedanalyses.

Thisdissertationwillbeorganizedasfollows.

InChapter1,wewillintroducestressbyfirstlypresentingthegeneralprinciplesthat

governit,firstcross‐linguisticallyand,secondly,inPortuguese.Wewillmentionthefunctions

and properties of word stress in the languages of the world and present some prosodic

aspectsthatarerelatedtoit,suchastheprosodichierarchy,thewordmetricalorganization

and the basic foot governing laws. Also, in Chapter 1, we will furthermore describe the

Portuguese stress system. We will provide information on the descriptions conducted for

wordstress in the languageand,secondly,wewillpresent themetricalproposals forword

stress in Portuguese. In the third section of this chapter we will provide frequency

information about word stress position and words shapes. In the fourth section of this

6Cf.Pollock,Brammer&Hageman(1993).

5

chapter,wewillreviewtheissueofthepertincityofthefoot,andthediscussiononaweight‐

sensitiveoramorphology‐basedstressalgorithminPortuguese.

In Chapter 2, wewill review themainquestions regarding the acquisition of word

stress, both cross‐linguistically and in Portuguese. Special attention to the shape of early

wordswillbegiven.Also,areviewontheacquisitionofstresspatternswillbeconductedon

a vast array of languages (Catalan, Dutch, English, French, German, Greek, Hebrew, Kich'e,

Sesotho,Spanish),aimingatlistinggeneralregularitiesacrosslanguages,withrespecttothe

acquisitionofstresspatternsandthepotentialearlypreferenceforagivenearlywordshape

(monosyllables,trocheesoriambs).

InChapter3,wewillpresentthedatabaseusedinthepresentstudy,inparticular,the

informationaboutthechildren,thedatacollectionandthephonetictranscriptions.

In Chapter 4, we will investigate the acoustic properties of word stress and word

prominenceinaspeechsampleofthecorpusconsideredinthisdissertation.Ontheonehand,

the results of the acoustic analysis enabled us to reliably identifyword prominence in the

early speech of two Portuguese children, and, on the other hand, the acoustic analysis

providedusefulcuesontheearlyprocessingofwordprominenceandstresspatterns.

InChapter5,wewillinvestigatetheacquisitionpathoffivePortuguesechildren,with

respect to the different word shapes and stress patterns. We will analyze children's

productions of mono‐, di‐ and trisyllables, as well as the acquisition path for target

monosyllables,targetdisyllables(trocheesandiambs)andtrisyllables(/WSW/,/WWS/and

/SWW/).Additionally,wewillanalyzethestrategiesundertakenbythefivechildrenineach

wordshape.Specialattentionwillbegivento theproductionofreduplications,epentheses,

truncations and stress shift. Moreover, we will compare the acquisition path pursued by

Portuguesechildren,withthestagesofwordshapeandwordstressacquisitionestablished

forother languages (namely,Dutch, FrenchandSpanish).The earlywords' representation,

the 'trochaic bias hypothesis' and constraints on word size during acquisition will be

discussed,asameanstoinvestigateonwhichprosodicconstituentsPortuguesechildrenrely

onduringwordconstructionandwordstressacquisition.

InChapter6,wewill address thediscussiononword stress in the target language,

based on the acquisition data fromPortuguese children.Wewill discuss (i) the phonology

morphologyinterface,byinvestigatingtherelationbetweenstresspatternsandinflectionin

the noun and verb paradigms during acquisition, and (ii) the potential effect of syllable

weightinthefunctioningofwordstressinEP,bystudyingtheacquisitonpathofwordswith

stressedandunstressedheavysyllables.

Finally,wewillpresentasummaryofthemainfindingsandcontributions.

The results brought up by this dissertation will hopefully contribute for the

6

understandingofthestressphenomenoninthelanguageandwillprovidefurtherempirical

supportforphonologicaltheoriesandforatheoryoflanguageacquisition.

7

1.Adescriptiononwordstress

Inthischapterwewilldescribewordstressfromacross‐linguisticperspective,with

special reference to Portuguese. In section 1.1., we will present the general properties of

word stress from a cross‐linguistic perspective, by making a general description on the

distributionofstressandtonal languages7 intheworldandbyreferringtotherelationship

betweenwordstressandtheconstituentsthatarepresumablypartoftheuniversalprosodic

structure(1.1.1.).Secondly,wewillputforwardsometheoreticalinstrumentsofthemetrical

theoryaccountingforwordstress(1.1.2.).Thirdly,wewillpresenttherelationshipbetween

feet structures and stress systems (1.1.3.). Finally we will hand over a description on the

mainacousticfeaturesthatarerelatedtowordstress(1.1.4).

In section 1.2., wewill describe the Portuguese stress system. This sectionwill be

dividedintofiveparts:section1.2.1.willfocusonthetraditionalapproachestowordstress,

aswellasonthepreliminaryworksonthephonologyofwordstress;insection1.2.2.,wewill

describewordstress inEPfromanacousticperspective; insection1.2.3.,wewillpresenta

number of proposals forword stress in Portuguese, based onmetrical analyses; in section

1.2.4.,wewillprovidefrequencyinformationofwordshapesandsyllablestypesinrelationto

wordstress.Andfinally,insection1.2.5.,wewillhighlightsomeofthemainquestionsraised

bytheliteratureonwordstressinPortuguese.

1.1.Generalobservationsonwordstress

Asubsetoflanguagesintheworldusestresstoorganizeprosodicunitsinspeech,as

opposedtolanguagesthatusetonesandintonation(orpitchpatterns)todistinguishwords

and sentences. Tonal languages aremainly represented in Africa and Asia, whereas stress

languagesarespreadmostlythroughoutEurope,AmericaandOceania.

7Aswewill see further in this section,notall languagesare stress languages.Given that themain topicof thisdissertationiswordstresswewillneglectinformationonnon‐stresslanguages.

8

Figure1.Tonal(purpleandredcircles)andnon­tonal(whitecircles)languagesworldwide(Goedmans&vanderHulst,2008)

MaddiesoninGoedemans&vanderHulst(2008)showsthatapproximately58%of

theworld languagesusestress,whereas42%of languagesusetones,asshowninFigure2,

below:

Figure2.Percentageoftonalandstresslanguagesworldwide(Goedmans&vanderHulst,2008)

Based on the seminal works by Liberman (1975) and Liberman & Prince (1977),

Hayes(1995:8)definesstressas"thelinguisticmanifestationofrhythmicstructure".Instress

languages,rhythmingeneralandstressinparticularservesasanorganizinginfrastructureto

words,phrases and sentences in speech. In general terms, and irrespectiveof thedifferent

types of stress that we find in stress languages, stress is related to the way speech is

organized and it can be defined as "a matter of relative strength between 'stronger' and

9

'weaker'syllables"(Kager,2007:195).Astressedvowel isnot intrinsicallystrong; it isonly

strongerthantheother(unstressed)vowel(s)withinaspecificdomain.

The fact that not all languages have stress and the fact that it is not a regular

phenomenonacrosslanguagesmayindicatethatstressisnotanabsoluteLinguisticUniversal

(e.g., Greenberg, 1963; Ferguson, 1978), i.e., it is not a default feature of grammar like

grammaticalnumber,negationorvowelsandconsonants.Also,manyunspecializedspeakers

of stress languageshaveweak intuitionsor showwrong judgmentswhenasked to identify

stressinsomewords8.

Althoughtherearegeneralregularitiesonhowstress isrealizedcross‐linguistically,

there is not a strict characterization of its functioning. Stress varies from language type to

languagetype(forinstance,GermaniclanguagestendtobedifferentfromSlaviclanguages‐

cf.vanderHulst,1999)and,withinthesametypeoflanguage,itmayalsovary(forinstance,

FrenchisdifferentfromanyotherRomancelanguageinrespecttostress,sinceithasphrase‐

final accent and not word stress9). In some languages, only one stress per lexical word is

accepted,whereasinothers,twooreventhreelevelsofstressarepossible.Secondarystress,

however, isweaker than primary stress, and tertiary stress isweaker than secondary and

primarystress(Goldsmith,1989;Hayes,1995;Kager,1995,2007,amongothers).

1.1.1.Wordstressandtheprosodichierarchy

When studying speech in general and prominence in particular, researchers early

perceivedthatthespeechstreamwasorganizedintospecificdomainsthatwereusefulinthe

description, explanation and understanding of some language phenomena. The speech

streamisnotanuninterruptedsoundchain.Instead,spontaneousspeechismostlyorganized

into 'chunks' that obey and are the result of specific organization (phonological,

morphological,syntacticandsemanticorganization).Asfarasthephonologicalaspectsofa

language are concerned, the speech chain is organized into universal prosodic categories

(Hayes, 1989a; Nespor & Vogel, 1986; Selkirk, 1980a). In (1) we present the hierarchy of

prosodicconstituentsthatwillbeadoptedinthisdissertation.

8Forabriefreviewontheperceptionandidentificationoflexicalstress(andaccent)byadultnativespeakersinotherlanguages(English,Chinese,Norwegian,Japanese,etc.),cf.Beckman(1986:27‐31).ForinformationonEP,cf.Delgado‐Martins(1986)andAraújo(2004).9Cf.Delattre(1965)andDell(1984).

10

(1) Prosodichierarchy(Nespor&Vogel,1986):

Utterance(PhonologicalUtterance‐U)g

IP(IntonationalPhrase‐Ι)g

PP(PhonologicalPhrase‐φ)g

CG(CliticGroup‐C)g

PW(ProsodicWord‐ω)g

Ft(Foot‐Σ)g

S(Syllable‐σ)

Even if in some languages evidence for some constituents is not provided, these

categories needed to be evoked to explain phonological and phonetic phenomena cross‐

linguistically10. Specifically, in a wide set of languages stress falls on specific prosodic

constituents,suchastheprosodicword.Theprosodicwordis,inthiscase,thephonological

domainforstressassignment.Wealsoknowthatsyllablesmaybeorganizedinaveryfixed

manner(intofeet),ratherthaninanarbitraryfashion,inordertoprovidedifferenttypesof

stress(primary,secondaryortertiary)withinwordsorphrases.Truncations,hypochoristics

and reduplications11, for instance, constitute empirical evidence for the need of fulfilling a

prosodic (andoftenmorphological12) template inagiven language (Broselow,1982;Kager,

2007;McCarthy&Prince,1986/1996,1993/2001,1995).For instance, inEnglish,abinary

trochaicfoot(i.e.,afootwithtwosyllables,astrongonefollowedbyaweakone)ispreferred

(e.g.Anthony‐>TonyorRobert‐>Bobby).ContrarytoFrench,wherewordsdonothaveto

be preferably andminimally a binary foot (V and CVmonosyllables are frequent ‐ Plénat,

1993),inEnglish,prosodicwordsareminimallyabinaryfootandfeetarepreferablybinary

(CVC,CV.CVorCV:)(McCarthy&Prince,1986/1996,1993/2001,1995).

10Foradetailedmotivationoftheprosodicconstituents,seeNespor&Vogel(1986),Chapter2.11Hypochoristicsarereferredtoasnicknamesorshortnames.12 On ProsodicMorphology, cf.McCarthy& Prince (1995) and references therein, aswell as Kager (2007:223‐227).

11

1.1.2.Metricalaspects

Asmentioned in Hayes (1995:1), "[m]etrical stress theory is a branch of theory in

generativephonologythatdealswithstresspatterns".Itprovidestheformalrepresentation

for stress, in layers or grids,where the rhythmic principles of prominence are formalized.

Many authors have proposed several different types of grids. In all of them, a leveled or

hierarchical representation of the relevant constituents for word stress is established, in

ordertoassignthestrong/weakalternationsthatgoverntherhythmicwavewithinwords

andphrases.

Kager (2007) presents a representation of stress within a Constituentized grid

(followingHalle&Vergnaud,1987).Inthisgrid,thehierarchicalelementsare,vertically,the

different prosodic levels (the syllable, the foot, the prosodic word, and so forth) and,

horizontally,informationthatassignstherhythm,i.e.,thepeaksalternationsineachlevel.In

(2)wewillshowthemetricalgridproposedbyKager(2007)fortheword'Apalachicola'.

(2) Constituentizedgridfortheword'Apalachicola'13:

PW‐level x

Foot‐level x x x

Syllable‐level x x x x x x

(A pa la chi co la)

In(2)wefirstsignalthesyllables(atthesyllablelevel).Inthenextlevel(foot‐level),

we assign foot‐headedness. In the last level (PW‐level), wemark theword‐headedness. In

bounded systems, the rhythmic and metric information must be grouped, according to

language‐specificprinciples.Thecorrespondingboundedgridtothewordin(2)ispresented

below.

13Inalltheliteratureonstress,*(star/asterisk),Sorxmarkstrongsyllables;.(dot),Wor∅(nothing)markweaksyllables.Also,strongsyllablescanbedesignatedas'peaks',andweaksyllablescanbedesignatedas'trough'.

12

(3) Metricalconstituency‐groupingofgridelements:

PW‐level x

Foot‐level (x x x )

Syllable‐level (x x) (x x) (x x)

(A pa la chi co la)

InthisgridweseethatsyllablesaregroupedintofeetandthatthesefeetareSWor

left‐headed. In thiscase, footdominance is trochaic (SW).Right‐headed feetarepossibleas

well,and,whenright‐headedfeet(WS)arepredominantinalanguage,wecallthatlanguage

iambic.Hayes(1995)claimsthat:

(i) Elements contrasting in intensity naturally form grouping with initial

prominence;

(ii) Elementscontrastingindurationnaturallyformgroupingwithfinalprominence.

TheseclaimsconstitutethegroundoftheIambic/TrochaicLawandweremadeupon

asetofperceptionexperimentswheresoundsweremademoreprominentonintensitybasis

andondurationbasis. Inthesestudies, listenerswereaskedto judgehowthesoundswere

mostappropriatelygroupedinpairs.Basedontheresultsoftheperceptionexperimentsand

after an analysis of a few iambic languages, Hayes claims that "in iambic languages

lengthening is frequent and robust, based on its function in fulfilling a rhythmic target"

(Hayes,1995:84). Itdoesnotmean that lengtheningcannotbe found in trochaic languages

(DutchandGerman,forinstance,arebothtrochaicandhavevowellengthcontrast),onlyitis

lesscommon.

Kager (2007) presents an inventory of metrical feet, in relationship with syllable

weightinformation,asshownin(4).

(4) Inventoryoffeettypes(Kager,2007:201)14:

Licit Degenerate

Syllabictrochee: σσ σ

Moraictrochee: LL H L (quantity‐sensitive)

Iamb: LH H LL L

14Inthisrepresentation,'H'standsfor'heavy'and'L'standsfor'light'syllables.

13

Thescheme in(4) indicates that trochaic languagescanbeeitherquantity‐sensitive

orquantity‐insensitive.Iftheyarequantity‐insensitiveandleft‐headed,thedisyllabictrochee

is the dominant foot (and, in that case, monosyllables are interpreted as degenerate

trochees).Inthiscase,σσarelicitandσisdegenerate.Iftheyarequantity‐sensitiveandleft‐

headed, the moraic trochee is the dominant foot (LL) and both heavy (H) and light

monosyllables(L)areaccepted(althoughlightmonosyllablesareconsidereddegenerate).If

theyarequantity‐sensitiveandright‐headed,thedominantfootistheiamb(adisyllabicone,

likeLHorLL,thoughamonosyllabicHisacceptedaswell)and,inthiscase,thedegenerate

footisalightmonosyllable(L).

1.1.3.Quantity­sensitivityandrhythmicpatterns

Manyofthelanguagesoftheworldareweight‐sensitive,thatis,heavysyllablestend

toattractstress.Inthiscase,longvowels(CV:),diphthongs(CVG)andclosed‐syllables(CVC)

bear twomoras.Morasmaybedefinedasabstractunitsof time(Trubetzkoy,1939),which

are, simultaneously, "a measure of syllable‐weight" (Zec, 2007). In quantity‐sensitive15

systems,bimoraicsyllables tendtobeheavyand,possibly,stressed.GermanandDutchare

amongst the languages that are considered to be weight‐sensitive (Jessen, 1999; Kager,

1989). Spanish, on the contrary, is assumed to be partly weight‐sensitive, as syllable‐

heavinessisarequirementforstressassignmentinthenounssystemonly(Harris,1991).In

languageslikeFrench,ModernHebrewandGreek,aBranchingRhyme(withintheNucleusor

theCoda)doesnotimplystressand,therefore,heaviness.Thoselanguagesareconsideredto

beweight‐insensitive16.InFigure3wepresentamapofthedistributionofstress‐timedand

syllable‐timedlanguagesacrosstheworld.

15Quantity‐sensitivelanguagesmayalsobedesignatedweight‐sensitiveorstress‐timed.16Nonweight‐sensitivelanguagesmayalsobedesignatedquantity‐insensitiveorsyllable‐timed.

14

Left‐edge:Firstorsecond 37

Left‐oriented:Oneofthefirstthree 2

Right‐edge:Ultimateorpenultimate 65

Right‐oriented:Oneofthelastthree 27

Unbounded:Stresscanbeanywhere 54

Combined:Right‐edgeandunbounded 8

Notpredictable 26

Fixedstress(noweight‐sensitivity) 281

Themappresentedaboveshowsthat,contrarytotonallanguages,weight‐sensitivity

is spread out all over theWorld. In a total of 500 stress languages, 281 (56.2%) are not

weight sensitive (white circles) and 219 (43.8%) are weight‐sensitive in variousmanners

(othersymbols)(Figure4).

Figure4.Quantity­sensitiveandquantity­insensitivelanguagesworldwide(Goedmans&vanderHulst,2008)

Figure3.Weight­sensitivityacrosslanguagesworldwide(Goedmans&vanderHulst,2008)

15

With respect to stress patterns, languages can mainly be trochaic (if the stressed

syllableisthepenultimateofthestressdomain‐SW),iambic(ifthestressedsyllableisthe

last syllable of the stress domain ‐ WS) or dual (languages that are equally trochaic and

iambic). Figure 5 shows the number of languages with trochaic, iambic and dual rhythm

types. Information on languages with no rhythmic stress and undetermined rhythm is

providedaswell.

Figure5.DistributionofrhythmicpatternsacrossthelanguagesoftheWorld(fromGoedemans&vanderHulst,2008)

Aswepresentedpreviouslyinthischapter(section1.1.),althoughtrochaiclanguages

maybequantity‐sensitiveornot,iambicsystemstendtobequantity‐sensitive(Hayes,1981,

1995;Kager,1995:396).Thisisillustratedbelow(Figure6),wheredataonthedistributionof

trochaic,iambiclanguagesandinformationonweight‐sensitivityarepresented.

Figure6.Trocheesandiambsinweight­sensitivesystems(fromGoedmans&vanderHulst,2008)

16

The first generalization drawn from the table above is that the vast majority of

trochaic languages is not weight‐sensitive17 (43 are weight‐sensitive, 108 are not; 2 are

unpredictable with respect to stress location18). The second generalization is that most

iambiclanguagesaresensitivetoweight(19contra12).

Both the prosodic and metrical phonology frameworks provided some theoretical

instruments that enable us to emulate a representation of words and prominences in the

adultlanguage.Forinstance,thenotionsoffootandprosodicword,proposedbytheProsodic

Phonology, are helpful to describe many processes in the languages. Likewise, these

structures ‐ feet and words ‐ are not built or organized randomly. They obey a given

hierarchy, which, for stress purposes, relies on organized and alternating strong‐weak

positions.Thesepositionsmakethemetricandtherhythmofphrasesandutterances.

1.1.4.Acousticfeaturesofwordstress

Stress is not only an abstract representation of a relative prominence. It is a

phonologicalphenomenonwithspecificacousticcorrelates.

Kager (2007) mentions that, cross‐linguistically, stressed syllables have in general

higherpitch, longerduration andgreater loudness.However, phonetic or acoustic cues for

stress vary across languages andwithin context.Higher pitch, longer duration and greater

loudness are all phonetic cues for stress, but they do not always work together. Stress is

opportunistic or, in Hayes' (1995:7) words, parasitic, in the sense that it uses phonetic

properties for phonological purposes (like duration contrast or tonal distinctions, in

languageswhere those aspects arephonemic). For instance, in a tonal language, tones and

intonationdependonpitch.Ingeneral,highpitchcuesthestressedortheaccentedsyllable,

butitisnotthatstraightforward,especiallywhenphrasesandsentencesareconcerned.

Research on perception showed that pitch and duration, rather than loudness, are

relatedtostress(bothinperceptualandproductiveterms).However,anycloserrelationship

between two or three of the acoustic parameters for word stress must be established in

language‐specificterms19.

17However, somewell‐studied languages, suchasEnglishorDutch,donotconformto thisgeneralization,bothbeingtrochaicandquantity‐sensitive.18 "Here [in the 'not predictable' category] we group together languages that have no predictability in stresslocation whatsoever (cf. Burushaski, Spokane, Usan) and languages in which words are said to have severalstresses(rhythmicorweight‐determined),withnoneofthesebeingprimary,orallbeingequal"(Goedmans&vanderHulst,2008:Chapter15,section2.3).19ReferencestotheacousticcorrelatesofwordstressinPortuguese(EP)willbeintroducedinsection1.2.2..

17

1.2.Portuguesewordstress

TheanalysesonPortuguesewordstressarecomplex,oftenconflictingand,therefore,

require a detailed explanation. For that reason,we dedicate an entire section to the issue,

aimingtoprovidethoroughinformationonthematter.

Inthissection,wewilldescribewordstress inPortuguese.Wewill firstpresentthe

general information on Portuguese word stress (section 1.2.1.), by mentioning the

descriptionscarriedoutbythetraditionalanddescriptivePortuguesegrammars,aswellas

the earlyphonological approaches. In1.2.2.,wewill present the studies carriedout on the

acousticcorrelatesofwordstressinthelanguage.Insection1.2.3.,wewillpresentthemain

analysesofPortuguesewordstresswithinthemetricalphonologyframework.Section1.2.4.

will show the frequency information related to stress inEP, suchaswordshape frequency

andstressposition frequency.Finally,wewillmakeageneraloverviewon themain issues

raisedbytheliteratureonPortuguesewordstress(section1.2.5.).

1.2.1.GeneraldescriptiononPortuguesewordstress

Word stress in Romance languages in general and in Portuguese in particular has

beenwidelydiscussed(Bisol,1992,1993,1999;Carvalho,1987,1988;Lee,1995,2001,2006,

2007;Mateus,1983,Mateus&Andrade,2000;Pereira,1999;Roca,1999;Wetzels,2006).

In languages like Portuguese, Spanish or Italian, deriving from a weight‐sensitive

language (Latin),where stresswas fullypredictableon thebasisof syllableweight20,word

stress poses several problems froma synchronic perspective. In its evolution to the actual

languages passing throughVulgar Latin, Ancient Latin lost and gained a number of stress‐

relatedfeatures,likethelossoftheshort/longvowelcontrastandthegainofasimpler(and

shorter)word structure (Classical Latin is known to be a synthetic language, whereword

orderwasnotrelevant,whereasVulgarLatinisknowntobemoreanalytic,withpreposition

multiplying andword order becoming necessary for syntactic purposes ‐ Lee, 2006; Roca,

1999).

Thestudyonwordstress inPortuguese isnotrecent.Since thenineteenthcentury,

severalauthorshavebeenaccountingforthebehaviorofwordstressinthelanguage,oftenin

20AccordingtoRoca(1999),thegeneralrule(s)forstressinClassicalLatinis/are:(a)stressaheavypenultimate(syllable), if there isone; (b)otherwisestress theantepenultimate, if there isone; (c)otherwisestress the firstsyllable.

18

amerelyprescriptiveordescriptivemanner,butalsoinamoreexplanatoryperspective21.

In this section (1.2.1.), we will focus on two main approaches: the more recent

descriptiveperspectivesofthePortuguesetraditionalgrammars(Cuesta&Luz;1971/1988;

Cunha & Cintra, 1984) and the also descriptive phonological approaches to word stress

(Andrade,1977;Barbosa,1965/1983;Mateus,1975/1982,1983andMateus,Brito,Duarte,

Faria, Frota,Matos, Oliveira, Vigário& Villalva, 200322).Wewill leave section 1.2.2. to the

metricalanalysesonPortuguesewordstress.

AccordingtoCunha&Cintra(1984),wordsinPortuguesecanhavestressinthelast,

penulimate or antepenultimate syllables. That holds for the classification of oxytonic (or

acute),paroxytonic(orgrave)andproparoxytonicwords,respectively,asshownin(5):

(5) ClassificationofwordsinPortuguese,withrespecttostressposition:

Lastsyllable

(Oxytonics)

Penultimatesyllable

(Paroxytonic)

Antepenultimatesyllable

(Proparoxytonic)

café'coffee'[kå»fE]

funil'funnel'[fu»ni…]

rapaz'boy'[{å»paS]

gato'cat'[»gatu]

sapato'shoe'[så»patu]

rapazes'boys'[{å»pazˆS]

árvore'tree'[»aRvurˆ]

pássaro'bird'[»pasåRu]

pêssego'peach'[»pesˆgu]

InPortuguese, themajorityofwordswith twoormoresyllablesarestressed in the

penultimatesyllable23.

Portuguese words are divided into stressed and unstressed, according to their

grammatical class. Open class words (nouns, verbs, adjectives and some adverbs) are

stressed, whereas closed‐class words may or may not be stressed (determiners are

unstressed, whereas some pronouns24, some prepositions and some conjunctions are

stressed25).

Within early phonological (structuralist) approaches to Portuguese phonology,

Barbosa(1965/1983)assumesthatthedomainforstressinthelanguageistheword,withits

encliticandprocliticelements26.Primaryandsecondarystressesarepossible,thelatterbeing

21ForareviewontheearlyapproachestoPortuguesewordstress(fromtheXIXthcenturyto late1980'sof theXXthcentury),cf.Pereira,1990.22ForanEnglishversionof thedescriptionmadeforwordstress inMateusetal.(2003),cf.Mateus&Andrade(2000:109‐117).23Forfrequencyinformationonstresspatternsdistributioncf.section1.2.3.,furtherinthischapter.24Cunha&Cintra(1984:56)refertotheunstressedor'obliquous'personalpronounssuchas'‐me','‐te','‐se','‐o','‐a',etc.25Onstressedandstressedfunctionwords,cf.Vigário(2003:173‐182;255‐258).26 The author says: 'Chaquemot comporte, en principe, un accent et un seul [principal]' (Barbosa, 1983:215).Additionally,theauthorsstatesthat:"L'unitéaccentuelleestenportugais,danslaplupartdescas,le“mot”,avec,

19

presentincompounds(6),adverbsofmannerendingin­mente(7),wordswithz­suffixes(8)

andFutureandConditionalTenseswithamesocliticpronoun(9)27.

(6) quebra­nozes'breaknuts' [«kEbRå»nçzˆS]

(7) somente'solely' [«sç»me‚tˆ]

(8) homenzinho'mandim.' [«çmå‚j‚»zi¯u]

homenzarrão'manaugm.' [«çmå‚j‚zå»{å‚w‚]

(9) contá­lo­iamos'wewouldtellit' [ko‚«talu»iåmuS]

With respect to the principles behind stress assignment, the author says that only

traditioncanexplainitslocation.AccordingtoBarbosa(1965/1983),'[e]ncequiconcernela

place de l'accent dans le mot, rien sinon la tradition ne la détermine dans la synchronie

actuelle' (Barbosa,1983:218)28.Theauthor further states that stress inPortugueseplays a

distinctiverole,i.e.,stress‐orstresspositioninparticular‐allowsforadistinctionbetween

two words, which, otherwise, would not be different. In (10) we present some instances

wherewordstressplaysa'contrastive'role(Barbosa,1965/1983):

(10) amaram'theyloved,simp.past'‐amarão'theywilllove'

[å»maRå‚w‚]/[åmå»rå‚w‚]

amara's/heloved,pluperf.'‐amará's/hewilllove' [å»maRå]/[åmå»Ra]

dúvida'doubt'‐duvida's/hedoubts' [»duvidå]/[du»vidå]

Within the linear generative phonology framework and following Chomsky&Halle

(1968),Mateus (1975/1982)andAndrade (1977)proposed that stresswasaphonological

feature([+stress])assignedbyphonologicalruleswithintheworddomain.

Mateus (1975/1982) provides a rule accounting for the general accentuation in

Portuguese,thatis,theauthorpresentsageneralruleforwordswiththepenultimatestress

((11a.)), words with final stress ending in […], [R], [S]29 ((11b.)), nasal and oral diphthong

s'il y a lieu, ses enclitiques et ses proclitiques." (Barbosa, 1983:216). Vigário (2003) proposed that encliticelements are incorporated into the previous Prosodic Word, whereas proclitic elements are adjunct to thefollowingProsodicWord(Vigário,2003:182‐203).However,sincestressisassignedonthelexicalcomponentandcliticsareprosodified,inEP,inthepostlexicalcomponent(Vigário,2003:182),wordstressdoesnot'regard'cliticwords.27Again,Vigário(2003:218‐247)furtherdemonstratedthatwordssuchastheabove‐mentionedare,infact,twoprosodicwords,bearingtwoprimarywordstresses.28Aswewillseefurtherinthissection,earlyphonologicalapproacheswill,however,establishgeneralprinciplesthatwillbeabletopredictstressassignment.29Thisfricativedoesnotcorrespondtothepluralmarker.

20

((11c.)).

(11) a.sapato'shoe'[så»patu]

b.hospital'hospital'[çSpi»ta…],rapaz'boy'[{å»paS]ourumor'rumor'[{u»moR]

c.canção'song'[kå‚»så‚w‚],carapau[kåRå»paw]

In (12) we present the formalization of the general rule proposed by Mateus

(1975/1982)forPortuguesewordstress.

(12) Generalstressruleforwordswithpenultimatestressendingin[…],[R],[S],nasaland

oraldiphthong(Mateus,1982:241):

V→[+stress]/[_](G)C0VC0#30

In the cases ofhospital, canção and carapau, an underlying final vowel is assumed

(hospital_,canção,carapau).

Additionally, the author presents two other rules for words with final and

antepenultimatestress((13)and(14),respectively).

The diacritics [+A] and [+E] stand for oxytonic and proparoxytonic words, in the

followingstressrules.

(13) Stressruleforwordswithfinalstress(Mateus,1982:241):

V→[+stress]/[_]C0# +A

(14) Stressruleforwordsstressedintheantepenultimatesyllable(Mateus,1982:241):

V→[+stress]/[_]C0VC0VC0# +E

In(15a.)and(15b.),wordswithfinalstress(café'coffee'anditspluralform‐rulein

(13)) and antepenultimate stress (pássaro 'bird and its plural form ‐ rule in (14)) are

presented,respectively:

30Mateus&Andrade(2000:119)laterevokethisrulewithslightchangesintheformalization:V→[+stress]/[_]C0 V C0 #]N,A. Both account for non‐verbswith penultimate stress, like casa [»kazå] 'house', caixa [»kajSå] 'box',cabelo[kå»belu]'hair'.

21

(15) a.café'coffee'[kå»fE],cafés'coffees'[kå»fES]

b.pássaro'bird'[»pasåRu],pássaros'birds'[»pasåRuS]

Theverbalsystem,however,hadamoreirregularbehaviourtowardsstress.Inorder

toaccountfortheproparoxytonicverbforms,theauthorproposesarulewithmorphological

boundaries,wherethethemevowelismadevisible.Thisrulepredictsthethemevoweltobe

stressed,ifitisnotinword‐finalposition,asshownin(16)(Mateus,1982:241):

(16) V→[+stress]/+[_]+C0V(C0V)C0#][Vb]31

The rule in (16) derives words like fizéramos 'we did, pluperf.' [fi»zERåmuS] or

fazíamos'wedid,pastimperf.'[få»ziåmuS].

Following the observation of different stress descriptions for non‐verbs and verbs,

Mateus (1983) first elaborated a description on stress in EP, according to whom stress

dependsonthemorphologicalconstituencyoftheword.Accordingtotheauthor'sanalysis,in

nouns and adjectives, stress falls on the last syllable of the stem (exceptions are lexically

markedassuch),whereasinverbs,stressfallsonthelastsyllableofthestemorinthetheme

vowel(FutureandConditionalTensesareexceptional,andmarkedassuch).Thisdescriptive

workstillheldsomeproblems. Innouns likeviagem 'trip'orórfão 'orphan'stressdoesnot

fallon the lastsyllableof thestem,and inverbs forms likebaterei 'Iwillbeat'orbateria 'I

wouldbeat',stressdoesnotfall,neitherinthelastvowelofthestemnorthethemevowel.

Since Mateus' (1983) description, many authors have established a relationship

betweenstressassignmentandwords'morphologicalconstituency(Andrade,1992;Andrade

&Laks,1992;Lee,1995;Pereira,1999;Mateus&d’Andrade,2000;Mateusetal.,2003).

According to the description in Gramática da Língua Portuguesa (Mateus et al.

200332),stressinnon‐verbsisgenerallyonthelastvowelofthestemandinthepenultimate

vowelofthelexicalwordinverbs33.

Instances in (17), below, account for stress assignment in non‐verbs, having into

accountthemorphologicalconstituencyofthewords(followingMateusetal.,2003).

31 [Vb] stands for 'verb'. Mateus & Andrade (2000:119) later evoke this rule with slight changes in theformalization:V→ [+stress]/ [ _ ] ((+C0V)C0V)C0# ] [Vb].Bothaccount forverb forms like falas [»falåS] 'youspeak',falámos[få»lamuS]'wespoke,simp.past'andfalávamos[få»lavåmuS]'wespoke,pastimperf.'.32MostofthedescriptionsmadeinMateusetal.(2003)withrespecttowordstressfollowPereira(1999).WewillpresentPereira'smetricalanalysisahead,insection1.2.3.33Bythe'non‐verbs'category,wemeannouns,adjectivesandallmorphosyntacticcategoriesinternaldotheNounPhrase (except cliticdeterminers)asall these categorieshave similar inflection (e.g. adjectivesagree ingenderandnumberwiththenountheymodifyandtheyhavethesamemorphologicalconstituencyasnouns)andbehavesimilarlywithrespecttostressassignment(Mateus&Andrade,2000;Mateusetal.,2003:1050,fn16).

22

(17) Non‐verbsparadigm:

a. mesa'table' [»mezå] mes]Stem+a34

leite'milk' [»låjtˆ] leit]Stem+e

b. animal'animal' [åni»ma…] animal]Stem

café'coffee' [kå»fE] café]Stem

c. mesinha'tabledim.' [mˆ»zi¯å] mesinh]Stem+a

The informationaboveshows that, innon‐verbs,wordstress is in the lastvowelof

thestem(17a,b.andc.).Whenthelastvowelofthestemislexicallymarkedas'notstress',

stresswithdrawsonesyllable(17d.).

d. júbilo'exultation' [»Zubilu] júbil]Stem+o

fácil'easy' [»fasi…] fácil]Stem

Anotherimportantfactisthatnon‐verbsstressedinthelastsyllabletendnottohave

the class/gender marker: animal 'animal' [åni»ma…], amor 'love' [å»moR], papel 'paper'

[på»pE…],café'coffee'[kå»fE],jacaré'aligator'[Zåkå»RE]).

Verbshaveamoreirregularstressassignmentanditisgenerallyinthepenultimate

vowelofthelexicalword,asshownin(20).However,morphologicalconstituencyiscrucial

toguidestress,inparticularthemevowel,tenseandmoodsuffixes.

Inverbs,stressfalls:

‐ on the (vowel of the)Tense andMood suffix in Future andConditional tenses, as

presentedin(18):

(18) a.calará's/hewillshutup' [kålå»Ra] calará]TM

b.calaremos'wewillshutup' [kålå»RemuS] calare]TMmos

c.calaríamos'wewouldshutup' [kålå»RiåmuS] calarí]TMamos

34Thegendermarker inPortuguese is/‐a/,/‐o/,/‐e/or∅,surfacingas[å], [u], [ˆ]or∅, respectively,andit isneverstressed.[å]marks,ingeneral,femininewords(nouns,adjectives,determinersandpronouns),whereas[u],[ˆ] or∅ mark, in general, masculine words (nouns, adjectives, determiners and pronouns). Some nouns andadjectiveslikemapa'map'[»mapå]orpoeta'poet'[pu»Etå]areexceptional,astheyaremasculine.

23

‐onthethemevowel,ifitisfollowedbyaTenseandMoodsuffix,shownin(19):

(19) a.calei'Ishutup,simp.past' [kå»låj] cale]Ti]TM/PN

b.morreu's/hedied' [mu»{ew] morre]Tu]TM/PN

c.caláramos'weshutup,pluperf.' [kå»laRåmuS] calá]Tra]TMmos]PN

d.morrêramos'wedied,pluperf.' [mu»{eRåmuS] morrêTra]TMmos]PN

‐onthepenultimatesyllableofthelexicalword,ifanyofthepreviousconditionsdo

notapply,asdemonstratedin(20):35

(20) a.calo'Ishutuppres.' [»kalu] calo]T/TM/PN

b.comemos'weeat' [ku»memuS] come]Tmos]PN

c.dormimos'wesleep' [duR»mimuS] dormi]Tmos]PN

Conjunctionsandprepositionsareunstressedingeneral,andadverbseitherintegrate

the general stress pattern (penultimate stress) or are exceptional for historical reasons

(Mateusetal.2003:1050,fn1636).

1.2.2.AcousticdescriptionofEPwordstress

InEP, the stressedvowelexhibitshighervalues fordurationand intensity than the

unstressed vowel(s) of the word (Delgado‐Martins, 198637, 1988; Mateus et al., 2003).

Andrade&Viana (1989) suggest that: (i)durationdegrees correspond todegreesof stress

and(ii)stressedsyllablesarelongerthanunstressedsyllables.Delgado‐Martins(1986,1988)

shows that energy is the acoustic correlate for stress perception and production in

Portuguese.Theauthordefines energyas "l'intégralede l'intensitépar ladurée" (Delgado‐

Martins,1986:30).Inthissense,durationalonedoesnotsignalstress,butrathertheamount

ofintensitythatisputintoavowelinarelativeamountoftime.Thefigurespresentedin(21)

and(22)illustratetwovowelswiththesameenergybutdifferentintensityandduration.

35SeeAppendixAforasystemicrelationshipbetweenverbstensesandstressinPortuguese.36Cf.footnote33.37 The conclusions presented in Delgado‐Martins (1986:80) are taken from a former study (Delgado‐Martins,1977).

24

(21) (22) dB dB

0 1 0 1 cs cs

Despitethefactthat(21)istwiceaslongeras(22),(22)isproducedwiththedouble

intensity of (21). In this case, according to Delgado‐Martins (1986), (21) and (22) are

producedwiththesameamountofenergy.

The author analyzed intensity, fundamental frequency, duration and energy in the

vowel [i], in threewordsdiffering in thestressedvowelonly:explícito 'explicit' [S»plisitu]/

explicito 'Imakeexplicit' [Spli»situ]/explicitou 's/hemadeexplicit' [Splisi»to].Sheanalyzed

the three words in three different contexts (in isolation, in a fixed sentence and in a free

sentence)andconcludedthat(Delgado‐Martins,1986:30):

(i) durationandenergyonlycuedthestressedvowelinthelastorantepenultimate

syllableoftheword;

(ii) fundamental frequencyand intensity isnota cue forwordstress,but ratherof

thephrasecontext.

(iii) in the penultimate syllable, there are no regular acoustic cues that account for

stressedvowels.

The studies conducted by Delgado‐Martins (1986, 1988), later confirmed and

extended in Frota (2000) confirmed that fundamental frequency is mainly related to

intonationaleventsatthephraseandutterancelevel,anddoesnotdirectlycuewordstress.

Intensityandduration,computedtogetherundertheformofenergy,aretherelevant

acoustic parameters to cue word stress (though not in the default position, i.e., in the

penultimate vowel). However, it is worth noticing that if, on the one hand, duration and

intensityplayapivotalrolewithintheEPstresssystem,ontheotherhand,"thereisageneral

agreementthatdurationisaby‐productofstressing"(Mateus&Andrade,2000:118),rather

thanastress‐conditioningfactor.

1.2.3.Metricalanalyses

Inthissection,wewillfirstpresenttheanalysesforEPandsecondlyforBP,sincethe

two varieties of the language have often been studied from two different perspectives. In

general,authorsfocusingonBPmainlyproposedastressalgorithmbasedonsyllableweight

25

(Bisol, 1993, 1999;Wetzels, 2003, 200638),whereas authors onEPproposed that stress in

Portuguese relies on morphological constituency (Andrade, 1992, 1996, 1997; Andrade &

Laks,1992;Mateus&Andrade,2000;Pereira,1999).

Aswewillfurtherseeinthisreview,despitethegreatexplanatoryadequacyofboth

approaches,severalconflictingissuesarisefromthecomparisonofthetwoanalyses39:ifwe

consideramorphology‐basedalgorithm,thedomainforwordstressisthesteminnon‐verbs

and the lexicalword forverbs,whereas theweight‐basedalgorithmgenerallypredicts that

the domain for stress is the lexical word40. Additionally, the foot has not always been

mentionedasarelevantprosodicconstituentinthemetricalandrhythmicorganizationwith

respecttowordstress(Pereira,1999).

TheworkonPortuguesestresswithinthemetrical theorybegantoarise in the late

eighties.Within a non‐linear phonology approach, Andrade (1988/1992, 1996, 1997) and

Andrade&Laks(1992)proposedametricalanalysisofwordstressforPortuguese.Thework

bytheseauthorswasseminalwithrespecttotheestablishmentofthemetricalandrhythmic

principlesforwordstressinPortuguese.Mostoftheauthorsthatproposedmetricalanalyses

forword stress in Portuguese relied onmorphological constituency, though some authors

havesuggestedthatweight,andnotmorphology,playsaroleonstressassignment.

Andrade (1988/1992:109) postulated four basic principles for Portuguese stress,

withinthemetricalphonologyframework:

(i) Feetarebinaryandprominentontheleft;

(ii) Wordtreesarebuiltfromrighttoleft;

(iii) Anelementthatstandsaloneis,byitself,afoot;

(iv) Oncefoottreesarebuilt,buildbinarytreesprominenttotheright,untiltheword

levelisreached41.

Theauthorpostulatesthefollowingruleforstressassignmentinnon‐verbs:"Ifthere

arenovowels in the context ...]Stem ]N,A, then the last vowelof the stem constitutes itself a

38OneexceptionisLee(1995),whoproposedthatstressinBPreliesonmorphologicalconstituency.39 In this sectionwewill present themetrical analyses proposed by the various authors. These issueswill beaddressedingreaterdetailinsection1.2.5..40The assumption according towhich the lexicalword is thedomain for stress inweight‐based approaches towordstressisnotcategorical.Forinstance,Wetzels(2006)defendsaweight‐sensitivestressalgorithmfornon‐verbs and a morphology‐based algorithm for verbs. Lee (2006, 2007) defends that stress is the result ofmorphologicalandprosodic/rhythmicconstraints.Syllableweight isonlyrelevant fornon‐verbsstressed inthelastsyllable.Thisissuewillbeaddressedfurtherinthissection.41MytranslationfromtheoriginalinPortuguese:"a.Ospéssãobinárioseproeminentesàesquerda;b.Asárvoresconstroem‐sedadireitaparaaesqueda; c.Umelementoqueseencontra sozinhoconstitui,por si só,umpé;d.Umavezconstruídasasárvoresdepé,constroem‐seasárvoresbinárias,proeminentesàdireita,atéaoníveldapalavra."(Andrade,1992:109).

26

foot"(Andrade,1992:11042).Theprinciplespresentedaboveaccount forwords likecabelo

'hair',jacaré'aligator'andamor'love',presentedin(23a.),(23b.)and(23c.),respectively.

(23) Metrical analysis for paroxytonic and oxytonic Nouns and Adjectives (Andrade,

1992:110)

a. b. c.

* * *

**. *.* **.

cabel]o]'hair' jacaré]]'aligator' amor]e]'love'

Inthecaseofwordslikeamor'love',rapaz'boy'orhospital'hospital',withword‐final

consonant, the finalgendermarker/‐e/ isassumed43.Toderiveproparoxytonicwords like

catastrófico'catastrophic',thelastvowelofthestemislexicallymarkedasextrametrical44,as

representedin(24).

(24) MetricalanalysisforproparoxytonicNounsandAdjectives(Andrade,1992:111):

*

*.*.

catastrófic]o'catastrophic'

DespitethedescriptivepowerofAndrade's(1988/1992)proposal,anargumentation

fortheextrametricalcharacterofwordinternalsuffixesremainsabsent.

In verbs, the general rule for stress assignment is that the stressed vowel is, by

default, the theme vowel. Tensemarkers are extrametrical (except Future and Conditional

Tense markers). One principle must be applied to Present and Past tenses, referring that

"stress the lastvowelof the stem45, if after thatvowelonlyone (metricalorextrametrical)

42Idem:"Senãoexistirumavogalnocontexto...]R.D.]N,A,entãoaúltimavogaldoRadicalDerivacionalconstituiumpé,porsisó"(Andrade,1992:110).43InlinewiththePortugueseLinearPhonologyanalysisofthattime(cf.Mateus,1975;Andrade,1977).44Thisvowelispartofthesuffix‐ic‐,thatmeans'relativeto'.Othersuffixes,suchas‐ito,‐eo,‐io,‐érrimoand‐ésimo,etc. present similar behavior (Andrade, 1992: 111,fn6) and they will be considered as stress­repellent insubsequentworkbytheauthor.45Theauthorassumesthatinverbs,thestemcorrespondstothetheme:root+themevowel.

27

vowel appears" (Andrade, 1992:12946). In (25) we show Andrade's (1988/1992) metrical

representationforthePresentandPasttenses.

(25) MetricalanalysisforPresentandPasttenses(Andrade,1992:118,123,121):

a. b. c.

* * *

*.*. *.* *.*

acaba]mos]47 acaba]ste] acaba]va]

'wefinish'(pres.)'youfinished'(simp.past)'I/s/hefinished'(pastimperf.)

InFutureandConditionaltensesthestemassumedendsinaconsonant48.Thus,the

rule forFuture andConditional tenseswouldbe: if the stemends in a consonant, then the

following vowel constitutes a foot itself49. In (26),wepresent themetrical analysis for the

FutureandConditionaltenses,accordingtoAndrade(1992).

(26) MetricalanalysisforFutureandConditionaltenses(Andrade,1992:125,127):

* *

* *

.*.* .*.*

acabar]ei acabar]ia50

'Iwillfinish' 'Iwouldfinish'

46My translation from the original in Portuguese: "No sistema verbal [o acento] é tambémna última vogal doRadicalDerivacional,sedepoisdelasóhouverumavogal,métricaouextramétrica,comexcepçãodoFuturoedoCondicional(...)"(Andrade,1992:129).47 InaPresentverbformlikeacabo 'I finish', thephonologicalrepresentationassumedisacaba]+o].ThethemevowelandthePersonandNumber(PN)suffixaresubjecttocoalescenceandstresswithdrawsonesyllableback.ThesamehappenswiththePresentSubjunctive:e.g.acaba]e‐>acabe,acaba]es‐>acabes,etc.48Theargumentbroughtuptoassumeastemendinginaconsonants, inFutureandConditionaltenses,arethemesoclitic constructions like falar­te­emos 'we will talk to you', in which the PN morpheme (underlined) isdetachedfromthe'stem'(bold).49TheformalizationpresentedisC]Stem(X)V⇒C]R.D(X)*V,(X)beinganobjectcliticpronoun,and*Vbeingafoot.Forpracticalreasonsregardingtextease,wereplacedtheformalizationpresentedforthefootedvowelby*V.SeeAndrade(1992:125,129).50 In the Conditional tense the final vowel ‐a‐ in acabaria 'I would finish', acabaríamos 'we would finish' isconsideredasextrametricalaswell,since it is thetraceof theverb formhaver 'theretobe'(inthePresentandImperfecttenses,respectively),whichusedtobepartoftheFuturetenseinLatin:acabar+hia.ThesamehappensintheFuture(acabar+hei),thoughtheauthordoesnotassumeanextrametricalvowelinthiscase(*acabarei),asitwouldentailincorrectsurfaceforms(*acabárei).

28

Later,Andrade&Laks(1992)proposeanalternativeanalysisforprimarywordstress

inPortuguese. In this approach, the authors clearly state that "main stress falls on the last

vowel of the stem, or in the penultimate, if there is an extrametrical vowel" (Andrade &

Laks:1992:1751). The four principles established in Andrade (1992) were replaced by one

generalrhythmicprinciple(in(27))andthreeparameters(in(28)).

(27) RhythmicPrinciple: "Rhythmicwave,peak‐troughanchoredat theright, firstpeak

strong"(Andrade&Laks,1992:19)

(28) Parametersforstressassignment(Andrade&Laks,1992:20):

1.InL0projectthevowels;

2.InL0,perfectgridright‐left,through,firstpeakstrong;

3.InL1,Endrule,right

TwomainaspectsdistinguishAndrade&Laks'(1992)fromAndrade's(1992)paper:

one, the authors establish one single principle for nouns and verbs ((27)); second, the

authorsassumeaword‐finalemptyslotinathematicwordslikecafé_andhospital_.Thefinal

syllableisnolongerafootitself,butratherpartofatrochaicfoot,wheretheweakpositionis

notphoneticallyrealized(seetherepresentationsformodelo 'model',hospital 'hospital'and

café 'coffee' ‐ (29a.,b.,c., respectively)). In the case of café andhospital, aword‐final empty

position is assumed52. Using a 'grid‐only' formalism, the authors account for stress

assignmentinPortuguese:

(29) StressinNouns/Adjectives(Andrade&Laks,1992:20,21)

a. b. c. d.

* * * *

.*. *.*..*. *.*..

modelo hospital_café_ catastrófico

'model' 'hospital' 'coffee' 'catastrophic'

Some 'morphological specificities' (Andrade & Laks, 1992:26) interfere with the

rhythmic principle mentioned in (27): the defective gender marker and some suffixes in

51My translation from the original in Portuguese: "o acento principal incide na última vogal do radical, ou napenúltimanocasodehaverumavogalextramétrica"(Andrade&Laks,1992:17).52TheformalizationistakenfromAndrade&Laks(1992:20).

29

nouns(­ic­,­simo,­rrimo53,etc.)areexceptionalandmustbeconsideredastroughs,suchas

tensemarkersandagglutinatedPNsuffixes,inverbs.ThecaseoftheFutureandConditional

tensesisnotmentionedinthepaper.

Forverbs,theapplicationoftheprinciplein(27)andtheparametersin(28)allows

forthefollowingrepresentations:

(30) StressinVerbs(uponAndrade&Laks,1992:24)

a. b. c.

* * *

.*. .*. .*..

parti'Ibroke' bateu's/hehit' falávamos'wespoke'

Inthisproposal,verbanalysisissimplifiedcomparingtoAndrade's(1992)proposal,

since the peak‐trough alternations, and not metrical information within morphological

constituents,arethegeneralprinciplesgoverningrhythm.However,extrametricalityisstilla

problem.Thoughmoreadequatelymotivated in theoretical terms ‐ since it is considereda

lexicaltrough‐thereisstillnoempiricalmotivationforthatclaim.

It is worth noticing that all the analyses from Andrade and colleagues rely on the

assumption thatPortuguesewordstress isan interactionbetweenrhythmicprinciplesand

morphological information. This assumption will be followed in subsequent works from

otherauthors.

WithintheIdsardi's(1992)54framework,Pereira(1999)accountsforwordstressin

thelanguage.LikesomeofherPortuguesepredecessors(namely,Andrade,LaksandMateus),

theauthordefendsthatstressinthelanguageisaninteractionbetweenrhythmicprinciples

and morphological factors. Pereira (1999) defends that the stem is the domain for stress

assignment innon‐verbs (Pereira, 1999:129) and the lexicalword is thedomain for stress

assignmentinverbs(Pereira,1999:169).In(31)wewillshowthebasicparametersforword

stress and in (32) we will apply the parameters to a metric grid, as proposed by Pereira

(1999)forPortuguesenouns.

53 The authors use the term 'stress‐repellents' for the first time in this paper (p. 21), to argue for the lexicallymarkedcharacterofthesesuffixes.54Forareviewandupdateonthisframework,cf.Halle&Idsardi(1995).

30

(31) Basicparametersforwordstressinnouns‐unmarked(Pereira,1999:129‐139):

a. SyllableBoundaryProjectionparameter:projectaline0elementforeachsyllable;

b. Edge Marking parameter: place a left parenthesis to the left of the right‐most

element(Line0);

c. IterativeConstructionofConstituents:placeaparenthesiseverytwoelements,from

theright‐mostelement(Line0);

d. HeadLocationparameter:projecttheleft‐mostelementofeachconstituentontothe

nextlineofthegrid(Line1);

e. Edge Marking parameter: place a right parenthesis to the right of the right‐most

element(Line1);

f. HeadLocationparameter:projectinLine2theright‐mostelementoftheconstituent

(Line1).

In(32)wewillshowtheapplicationoftheparametersin(31),forthewordspedaço

'bit'andchafariz'fountain'.

(32) Applicationoftheparametersin(31)‐Pereira(1999:137):

Line2 x x

Line1 x) xx)

Line0 x(x (xx(x

pedaç]o'bit' chafariz]'fountain'

It is worthwhilementioning that Pereira (1999) does not refer to themetrical (or

extrametrical)statusofthewordmarker.Theauthorstatesthat"theapplicationofthestress

rules apply to the stem, ignoring the subsequent morphological constituents" (Pereira,

1999:12955).

For marked cases, the stress domain‐final syllable is lexically marked with the

information of 'do not stress' within a specific syllable (e.g. súplica 'suplication', âmbar

'amber').Inthesecases,theEdgeMarkingparameterinline0doesnotapply,asmentioned

in(33):

55My translation from the original in Portuguese: "(...) as regras de atribuição do acento [aplicam‐se] sobre oradicalderivacional,ignorandoosconstituintesmorfológicosqueseseguem"(Pereira,1999:129).

31

(33) Basicparametersforwordstressinnouns‐marked(Pereira,1999:139‐142):

a. SyllableBoundaryProjectionparameter:projectaline0elementforeachsyllable;

b. EdgeMarkingparameter:doesnotapply(Line0);

c. IterativeConstructionofConstituents:placeaparenthesiseverytwoelements,from

theright‐mostelement(Line0);

d. HeadLocationparameter:projecttheleft‐mostelementofeachconstituentontothe

nextlineofthegrid(Line1);

e. Edge Marking parameter: place a right parenthesis to the right of the right‐most

element(Line1);

f. HeadLocationparameter:projectinLine2theright‐mostelementoftheconstituent

(Line1).

(34) Applicationoftheparametersin(33)‐Pereira(1999:141):

Line2 x x

Line1 x) x)

Line0 (xx (xx

súplic]a'supplication' âmbar]'amber'

Inverbs,amorecomplexalgorithmisproposed.Inordertoderivethecorrectstress

placement, information regarding certain suffixes is required. The themevowel signals the

default stress position. We now recall the three principles that guide stress location in

Portugueseverbs,accordingtoPereira(1999).InPortugueseverbs,stressfalls:

(i) inthe(vowelofthe)TenseandMoodsuffixinFutureandConditionaltenses;

(ii) inthethemevowel,ifitisfollowedbyaTenseandMoodsuffix;

(iii) in the penultimate syllable of the morphological word, if any of the previous

conditionsapply.

In(35)wewillpresentthebasicparametersforverbstress.

32

(35) Basicparametersforwordstressinverbs(Pereira,1999:169‐174)

a. SyllableBoundaryProjectionparameter:projectaline0elementforeachsyllable;

b. MoodandTensesuffixEdgeMarkingparameter:placealeftparenthesistotheleftof

theleft‐mostelement(Line0);

c. ThemevowelEdgeMarkingparameter:placealeftparenthesistotheleftoftheleft‐

mostelement(Line0);

d. IterativeConstructionofConstituents:placeaparenthesiseverytwoelements,from

theright‐mostelement(Line0);

e. HeadLocationparameter:projecttheleft‐mostelementofeachconstituentontothe

nextlineofthegrid;

f. Edge Marking parameter: place a right parenthesis to the right of the right‐most

element(Line1);

g. HeadLocation parameter: project the right‐most element of each constituent onto

thenextlineofthegrid.

In Figure 7wewill showa summary of the order of conditions application (Line0

first),dependingontheverbtenses:

Present Past FutureandConditional

Line1 HeadLocationParameter

EdgeMarkingParameter

HeadLocationParameter

EdgeMarkingParameterHeadLocationParameter

EdgeMarkingParameter

Line0 HeadLocationParameter

Iterative Construction of

Constituents

SyllableBoundary

ProjectionParameter

HeadLocationParameter

Edge Marking Parameter

(Themevowel)

SyllableBoundary

ProjectionParameter

HeadLocationParameter

Edge Marking Parameter (Mood and

TenseSuffix)

Syllable Boundary Projection

Parameter

Figure7.OrderofapplicationofstressparametersinPortugueseverbs

Theapplicationof thestressparametersaccording to (35)andFigure7ordering is

givenin(36):

33

(36) Stressparametersapplicationinverbs(Pereira,1999:174‐189):

Present Past FutureandConditionalLine1 x

x)(xx)cant]o'Ising'

xxx(xxcant]e]i'Isang'

x(xxx(xxcant]a]re]i'Iwillsing'

Line0 x(xx)cant]o'Ising'

xx(xxcant]e]i'Isang'

xxx(xxcant]a]re]i'Iwillsing'

FollowingAndrade& Laks (1992) andPereira (1999),Mateus& d'Andrade (2000)

establishaprinciple thatallows foralternationofstrong(S)andweak(W)syllables,peaks

and troughs, respectively, assuming the need for a SW rhythmic pattern. According to the

PrincipleoftheRhythmicWave,"[t]herhythmicwaveistriggeredbytrough‐firstanchoredat

right, first peak strong." (Mateus & d'Andrade, 2000:122). Prominences are licensed in

different levels: the firstone is thesyllable level (L0), thesecondone is therhythmicpeaks

level(L1)and,thethirdoneisthelevelinwhichthestrengthenedelementinL1appears(L2).

Prominencesareassignedintoa'perfectgrid',accordingtothefollowingparameters:

(37) "In L0 project the syllables; on L0 right­to­left perfect grid, trough; on L1, End Rule,

right."(Mateus&Andrade,2000:122)

Theparameterizedformalizationoftheseprincipleswouldbe:

(38) InL0PerfectGridright­left,trough,firstpeakstrong.(Mateus&Andrade,2000:122)

In (39) we present the scheme for word stress assignment in the wordsmodelo

'model'andorganizado'organized'(thesetwowordsbelongtothenon‐verbs'paradigm).

(39) Non‐verbs(Mateus&Andrade,2000:123)

L2x x

L1 .x. .x.x.

L0xxx xxxxx

model]u'model' organizad]u'organized'

In (40)we present the scheme for stress assignment in verbs forms falei 'I spoke,

simp.past'andfalava'Ispokepastimperf.'

34

(40) Verbs(Mateus&Andrade,2000:126,127)

L2xx

L1 .x..x.

L0xxxxxx

fala]i'Ispoke'fala]va'Ispoke,pastimperf.'

Inthewordshospital]'hospital'orcafé]'coffee',theauthorsassumea'defectiveclass

marker'andpostulateanemptyVslot,whichisplacedafterthestem((41)).Inthemarked

cases, like in the words hospital 'hospital', útil] 'useful' or catastrófico] 'catastrophic', the

authorsassumeapre‐assignedrhythmictrough.

(41) Non‐verbs(Mateus&Andrade,2000:123,124):

L2x x

L1 x.x. x..

L0xxxx xxx

hospital]'hospital' útil] 'useful'

The metrical analyses presented thus far have accounted mostly for stress

assignment in EP. As for BP, Bisol (1992, 1993) presents a metrical analysis assuming a

quantity‐sensitivity approach56. Bisol's proposal for word stress in Portuguese assuming

quantity‐sensitivity was seminal, but shows some problems, namely within the verbs'

paradigm,whichstillpresentsagreatamountofcasesandexceptions.

Theauthorestablishesaprimarystressruleaccordingtowhichthedomainforword

stress istheword.TheQuantity‐Sensitive(QS)rulestatesthefollowing:"assignanasterisk

(*) to the final heavy syllable, i.e., the last branching‐rhyme syllable; otherwise, construct

non‐iteratively a binary left‐headed constituent (* .) at the right boundary of the word"

(Bisol,1993:21).Theauthoragreeswiththefactthatmorphologicalinformationisrelevant

to the stress algorithm and determines that the stress rule is cyclic in non‐verbs and

noncyclic inverbs57,andapplieswheneveraderivationalmorphemeisaddedinnon‐verbs,

whereasinverbs,theabove‐mentionedstressruleisnon‐cyclic.

56 It is worth noticing that stress placement in both varieties is generally the same, despite some differencesregardingloanwords(cf.Wetzels,2006)andwordswithepentheticvowels(e.g.adapta's/headapts'isproducedas[å»daptå]inEP,andas[a»dapita]or[ada»pitå]inBP(Andrade,1997).57 Bisol (1994:20) states that: "[T]he difference (between stress assignment in non‐verbs and in verbs) comesfromthemorphologicalstructures,whichgivesrisetothefollowing:while innon‐verbs(2)[theprimarystress

35

Extrametricality is still necessary to account for final and antepenultimate stress.

However,accordingtoBisol,extrametricalityinPortugueseonlyappliesinfinalconsonants‐

whichisthecaseforunstressedfinalheavysyllables(e.g.,útil'useful')‐,orsyllables‐inthe

caseofwordswithantepenultimatestress(e.g.,número'number').

ThreeadditionalrulesarenecessaryforstressassignmentinPortuguese:

(i) Formation of Prosodic Constituents (FPC) ‐ Adjoins a light syllable to the

precedingone,creatingaleft‐headedconstituent;

(ii) Stray Syllable Adjunction (SSA) ‐ Incorporate the stray syllable as the weak

memberofanadjacentfoot;

(iii) EndRuleFinal(ERF)‐Projecttheasteriskcreatedasthemainstressoftheword.

The proposal from Bisol (1993) allows for the following representations of non‐

verbs:

(42) Stressassignmentinnon‐verbs(Bisol,1993:21‐24):

a. b. c. d. e.

Lexicon:/kaz+a/ /koronel/ /numer+o/ /util//kafEC/

Syllabif.:kaza koronel numero util kafEC

QS:‐‐‐‐ (*) ‐‐‐‐ ‐‐‐‐ (*)

Extram.:‐‐‐‐ ‐‐‐‐ <ro> <l> ‐‐‐‐

FPC:(*.) ‐‐‐‐ (*.) (*.) ‐‐‐‐

SSA:‐‐‐‐ ‐‐‐‐ (*..) ‐‐‐‐ ‐‐‐‐

ERF:(*) (*) (*) (*) (*)

Output:káza koronél númeru útil kafé

'house' 'colonel' 'number' 'useful' 'coffee'

In (42c.) and (42d.), the final syllable and the final consonant of the words,

respectively, are peripheral and considered extrametrical. In (42e.), an abstractword‐final

consonantintheword‐finalrhymeisassumedinthephonologicalrepresentation.Evidence

supportingthisassumptioncomesfromderivedwords.Forinstance,aderivedwordofcafé

bears an internal consonant, cafeteira 'coffee kettle', cafezal 'field of coffee', robô 'robot',

robótico'robotic',etc.

rule,whosedomainofapplicationistheword]isacyclicrule,inaccordancewithHalle&Vergnaud,inverbs(2)[idem]isanoncyclicrulewhichappliesonlyonce,whentheformationofthewordisaccomplished."

36

Inverbs,adifferentruleforextrametricalityisnecessary.Forthegeneralruleapplied

tonon‐verbs,tworulesmustbeadded:(i)Markasextrametricalthefinalsyllableofthefirst

and second persons plural of the Imperfective tenses; (ii) Otherwise, mark the consonant

withstatusofinflection(Bisol,1993:27).

In the case of the Future tenses (Future and Conditional), which are, according to

Andrade (1992), problematic from the perspective of a stress assignment, the author

stipulatesanotherrule,AvoidStressShock(ASS),whichsay"Erase*"(Bisol,1993:28).

Theapplicationoftherulesestablishedforverbs,accordingtoBisol(1993)resultsin

therepresentationsin(43).

(43) Stressassignmentinverbs(Bisol,1993:28):

a. b. c. d.

Lexicon:/kaNtaS/ /kaNtar/ /kaNtasemoS/ /falar//ei/

Syllabif.:kaNtaS kaNtar kaNtasemoS falarei

Extram.:<S> <moS>

QS: (*) (*)(*)

FPC: (*.) (*.)

SSA: (*..)

ASS: (*)

ERF: (*) (*) (*) (*)

Output:kãntas kantár kantásemus falaréi

'yousing' 'tosing' 'ifwesang' 'Iwillspeak'

AccordingtoBisol(1993),thegeneralizationaccountingforweight‐sensitivityinBP

is the following: stress falls on thepenultimate syllable of theword (on the trochaic foot),

unless the final syllable isheavy (e.g.,gáto 'cat',abérto 'open',butpomár 'orchard',coronél

'colonel').Thefootofthelanguageis,therefore,themoraictrochee.Inverbs,thegeneralrule

applies (e.g., cánto 'I sing', cantár 'to sing'), but, additionally, we need to specify as

extrametricalthefinalsyllableofthefirstandsecondpersonpluraloftheImperfectivetenses

(e.g., cantávamos 'we sang'), and the consonant with status of inflection (e.g. cántas 'you

sing').

Contrary to Bisol, Lee (1995) proposes an analysis of word stress in Portuguese

wherestress ismorphology‐dependentandproposes rhythmicprincipleswithinametrical

approach to word stress in Portuguese. The author claims that Portuguese is not weight‐

sensitive and establishes two domains for stress assignment: α and β. The α‐level is the

domainforstressassignmentinnouns,β isthedomainforstressassignmentinverbs. The

37

authorassumesthatthedomainforstressassignmentinnon‐verbsisthestemandthatthe

wordstructureofnon‐verbs isorganized inbinary, right‐headedconstituents (asshown in

(44a.)and(44c.),respectively).SincewordstressinPortuguesealwaysfallsononeofthelast

three syllables of aword, it is non‐iterative (see (44b.)). The parsing direction,within the

worddomainisright‐to‐left(aspresentedin(44d.)).

(44) Stressrulesinnon‐verbs‐unmarked‐Lee(1995:153):

Domainα:stem

a.Binaryconstituent

b.Non‐iterative

c.Right‐headed

d.RighttoLeft58

Theprinciplesin(44)accountforthewordsin(45)‐Lee(1995:153):

(45) café]'coffee' almoç]o'lunch' tonel]'largebarrel'

(.*) (.*) (.*)

In (46) and (47),wewill showLee's analyses forword stress innon‐verbs, for the

marked cases. The difference between (44) and (46) is the head direction of the foot

((44,46b.)).Whereas intheunmarkedcase, thefoot isrightheaded, inthemarkedcasethe

footisleft‐headed.

(46) Stressrulesinnon‐verbs‐marked‐Lee(1995:154)

Domainα:stem

a.Binaryconstituents

b.Left‐headed

c.RighttoLeft

d.Non‐iterative

Therulesin(46)accountforthewordsin(47)‐Lee(1995:155):

(47) túnel]'tunnel' abóbor]a'pumpkin'

(*.) (*.)

58Mytranslation fromtheoriginal inPortuguese:"Domínio:radicalderivacional;a.Constituintebinário;b.Nãoiterativo;c.Cabeçaàdireita;d.Direitaparaaesquerda"(Lee,1995:153).

38

AccordingtoLee'sanalysis,thegendermarkerinnounsisextrametricalandnounsin

Portugueseare,bydefault,iambic.Theauthordefendsthegendermarkerextrametricalityby

arguing that it fulfills the peripherality condition of extrametricality, it is deleted in

derivationalprocesses (e.g.cama 'bed' ‐>camainha 'camadim.';sapato 'shoe' ‐>sapatoeiro

'shoemaker')anditisneverinstressedsyllable.

Asforstressassignmentinverbs,thatis,intheβ‐Level,theauthorproposesthatthe

word structure of verbs is organized in binary, left‐headed constituents (thus, in trochaic

feet),whicharebuiltnon‐iteratively.Theparsingdirectionwithintheworddomainisright‐

to‐left.

In (48)wepresent the formalizationof theparametersproposedbyLee (1995) for

theunmarkedcaseofverbsstress.

(48) Stressrulesinverbs‐unmarked(Lee,1995:160):

Domainβ:lexicalword(fala]'s/hespeaks',falamos]'wespeak')

a.Binaryconstituents

b.Left‐headed

c.Non‐iterative

d.RighttoLeft

Therulesin(48)wouldaccountforthewordsin(49)(Lee,1995:161):

(49) computo'Icompute' falam'theyspeak' falamos'wespeak'

(*.) (*.) (*.)

Asfornon‐verbs,themarkedandunmarkedcaseinverbsonlydiffersinthedirection

ofthefoothead.Intheunmarkedcase,feetareleft‐headed,whereasinthemarkedcase,the

feetareright‐headed.

Themarkedcounterpartfor(48)isexpressedin(50):

(50) Stressrulesinverbs‐marked(Lee,1995:162)

Domainβ:lexicalword(bati]'Ihit,simp.past)',baterá]'s/hewillhit')

a.Binaryconstituents

b.Right‐headed

c.Non‐iterative

d.RighttoLeft

39

In(51)wepresentthewordsaccountedforbytherulein(50)(Lee,1995:162):

(51) bati'Ihit,simp.past) baterá's/hewillhit' falar'tospeak'

(.*) (.*) (.*)

AsBisol(1993), theauthorconsidersthatthe2ndp.pl.of thePasttenses(falávamos

'we spoke, past imperf.' [få»lavåmuS], falássemos 'we spoke, past imp. subj.' [få»lasˆmuS],

faláramos'wespokepluperf.'[få»laRåmuS])hasafinalextrametricalsyllable(­mos).

WithinanOTapproach,Lee(2001,2006,2007) laterdefendedthatPortuguese isa

hybrid system: in non‐verbs, stress assignment is rhythmic (weight‐sensitive), whereas

stress assignment in verbs is purely morphological. Based on the evolution from Latin to

Portuguese and the differences resulting from the change (Lee, 2006), namely, the loss of

casemarkersinnouns(e.g.amo:rem(Lat.)>amor(Port.)'love')59,theauthormaintainedthe

ideaaccordingtowhichmorphologicalconstituencyisnecessarytoexplainstressassignment

inthelatter.Theauthormaintainedtheideaofaniambicdefaultfootinnon‐verbs(i.e.,main

stressisontheheadoftheright‐mostsyllableofabinaryfootwithinthestem).Also,within

an OT perspective, Lee (2001, 2006, 2007) argued for the importance of morphological

constituency in stress assignment, in verbs. The difference between Lee (1995) and Lee

(2001, 2006, 2007) is that in the former, the author argues for the strict morphological

dependenceofwordstress,whereas in the latter, an interactionbetweenphonologicaland

morphologicalpropertiesisdefended.

AlthoughLee'sproposalisabletodescribeandexplainwordstressinPortuguese,his

workposesonemainproblem:it isnottheoreticallyplausiblethat languageshavetwofeet

types (one default and onemarked). According to Goldsmith (1989), Hayes (1995), Kager

(2007),vandeVeijver(1998),amongothers,eachlanguagesetsone,andoneonly,foottype

that must be applied to syllables in order to build metrical trees. Every syllable standing

alonemustbeanalyzedasadegeneratefootorbeconsideredasextrametrical(Hayes,1995).

In summary,metrical theories, either relying on aweight‐sensitive ormorphology‐

dependent algorithm, are able to account for stress assignment in Portuguese. Metrical

theories determine the organization of the strong‐weak alternations of the rhythmicwave

within the word domain. Furthermore, by establishing, not only metrical, but also

extrametrical positions in generally unaccounted data (namely SWW words), metrical

theoriescanpredicttheconstructionofwell‐formedprosodicwordsinPortuguese.

59ForamoredetailedviewontheevolutionofwordstressfromLatintoPortuguese,seealsoPereira(1997)andPereira(2007).

40

1.2.4. Syllable types, word shapes and word stress in EP ­ frequency

information

Another important issue regarding a description of a language is frequency

information. Itprovidespresumabletendenciesandpatterns,whichcanprovidearguments

fordifferentanalysesonagivenlinguisticaspect.

Mateus&Andrade (2000:109) affirm that "[f]or themajority (over 70 per cent) of

nouns,adjectivesandadverbsendinginanoralvowel,stressfallsonthesyllablebeforethe

last"60.Mateuset al. (2003:1050,fn16) specify that 70%of the nouns and adjectives in the

languagebearstressinthelastvowelofthestem61.

InFigures8,9and10wepresentfrequencyinformationonwordshapesandstress

distribution in EP, from an oral corpus from the 1990’s62. In Figure 8, 'S' stands for a

monosyllable, 'SS' stands for a disyllable and '≥SSS' stands for words with three or more

syllables.

60Noinformationonthesourceofthelinguisticmaterialunderlyingthesenumbersisprovided.61Idem.62 Frequencies extracted from the corpus TA90PE,with 22994 orthographicwords, using©Martins, Vigário&Frota, 2004‐2007, v.1: FreP ‐ Frequency information in Portuguese(http://www.fl.ul.pt/LaboratorioFonetica/frep/).

41

Figure8.WordshapedistributioninEuropeanPortuguese(verbsandnon­verbs

together).

Figure9.Distributionofwordstressindisyllables(verbsandnon­verbstogether).

Figure10.Distributionofwordstressintri­syllables(verbsandnon­verbstogether).

Figure8showsthatPortuguesehasnearlyequalamountofmonosyllablesandwords

longerthanthreesyllables(28.6%and27%,respectively).Nevertheless,disyllablesarethe

mostcommonwordshapeinthelanguage(44.4%).InthetotalamountofwordsinEP,19%

aretrisyllabicand8%havemorethan3syllables(intotal,27%ofwordsarelongerthan3

syllables ‐Vigário,Freitas&Frota,2006:188).Also,7.4%haveamonosyllabicCVstructure

and 21.2% are monosyllables and have a non‐CV structure (CVG, CVGN, CVC, CVN, CVNC,

CVGNC‐Vigário,Freitas&Frota,2006:188).

Asforthedistributionofstresspositionpersyllablewithinthewords(Figures9and

10),weobserve that indisyllables, stress falls predominantly (73.39%) in thepenultimate

syllable. In trisyllables, 80% of the words have penultimate stress (Figure 10). The data

presented in Figures 9 and 10 lead us to assume that Portuguese has predominantly a

trochaicrhythm.

DataonBPwrittentextsshowsthat63%ofthewordsaretrochaic,12%areiambic

and25%haveotherwordshapes(Cintra,1997).Researchersonwordstress inEPandBP

acknowledge also that word prominence in Portuguese is predominantly organized into

42

trochaic feet (‐SW ‐ Bisol, 1999; Pereira, 1999: 135,170). However, final stress is also

possible, presumably due to syllable‐final consonants in non‐verbs, as suggested by Bisol

(1992, 1993, 1999) and Wetzels (2006). The data in (52) provides information on the

percentageofeachRhymestructureinstressedandunstressedpositioninEP.

(52) Distribution of Rhymes (%) per stress position (adapted from Vigário, Martins &

Frota,2006):

Rhymestructure Stressed Unstressed TOTAL

(C)V 19.1 45.26 64.36

(C)VC63 4.87 9.16 14.03

(C)VN 5.31 2.71 8.02

(C)VGN 4.36 1.26 5.62

(C)VG 3.43 0.74 4.17

(C)VGC 1.20 0.01 1.21

Other64 2.59

The distribution of syllable types between stressed and unstressed position in EP

showsthat:

(i) CVisthemostfrequentsyllabletype;

(ii) BranchingRhymes(VC)aremorefrequentinunstressedposition;

(iii) Nasalvowels(VN),oralandnasaldiphthongs(VGandVGN)aremorefrequentin

stressedposition;

(iv) SimultaneousBranchingRhymesandBranchingNuclei(VGC)aremorefrequent

instressedposition.

Bisol(1992)investigatedstresspositioninwordswithafinalconsonant65inBPand

theresultsindicatedthat78%ofthewordsendinginaconsonantareoxytonicand22%are

paroxytonic.

63The(C)VCsyllablesaccountedforheretakeintoconsideration,notonlysyllable‐final/R/and/l/,butalso,/s/.Thephoneme/s/maycorrespondtothepluralmarker,asincasas'houses[»kazåS],ortoaphonemewhichdoesnotcorrespondtoamorphologicalconstituent,asinrapaz'boy'/{å»paS/.64In'Other'weinclude(C)VGNCand(C)VNCRhymes.65Thedatawerecollectedfromwrittentext(DicionárioDeltaLarousse).

43

1.2.5.IssuesraisedinthedescriptionsonPortuguesewordstress

ThereviewmadethusfarshowsthatthedescriptionsofPortuguesewordstresshave

beenthesubjecttoseveralcontroversies.Oneofthefirststress‐relatedissuesonPortuguese

regards the pertinacity of feet in the language. The second one concerns the

weight/morphologydependenceofwordstressinPortuguese.

Inthefollowingsub‐sectionswewillsummarizetheissueconcerningthepertinacity

of feet in Portuguese (section 1.2.5.1.), and the issue regarding the weight/morphology

dependenceofwordstressinthelanguage(1.2.5.2.).

1.2.5.1.ThepertinacityoffeetinPortuguese

AfewworksonPortuguese(mainlyEP)claimthatthereisnoempiricalevidencefor

the foot in the language (Mateus et al. 2003:1060; Vigário, 2003:336). Bisol (1999) and

Wetzels(1992,1995,2006),however,mentiontwophonologicalprocessesthat involvethe

footandthusmayprovideevidenceforthisconstituentinBP66.Thesetwoprocessesarethe

Dactylic Lowering and the Spondaic Lowering andwewill describe them in the following

paragraphs.

TheDactylicLoweringisaprocessthatprohibitsmidandhighvowelsinthestressed

syllableofadactylicfoot(SWW).Forinstance:

(53) a.rótulo'label'‐[»{çtulu]butnot*[»{otulu]67

b.pétala'petal'‐[»pEtålå]butnot*[»petålå]

TheDactylicLoweringdoesnotshowexceptionsinwordswithderivationalsuffixes

(e.g., fon[»ç]logo 'phonologist', [»E]pico 'epic', etc) but has a few exceptions in non‐derived

words(e.g., f[»o]lego 'deepbreath',b[»e]bedo/b[»e]bado 'drunk',p[»e]ssego 'peach',tr[»o]pego

'lame',es[»o]fago(BP)'esophagus',est[»o]mago'stomach').

TheSpondaicLoweringisaconstraintthatdoesnotallowformidvowelswithinthe

strong syllable of the spondee foot, that is, within the stressed vowel of a word with a

'heavy'68finalunstressedsyllable.

66EventhoughitisareferenceforBP,theprocessestheauthorreferstoarevalidandsuitableforEPaswell.67Instancesin(53)and(54)weretakenfromWetzels(2006).68TheauthorassumesthatPortugueseisweight‐sensitive.

44

(54) a.móvel'mobile'‐[»mçvE…]butnot*[»movE…]

b.repórter'reporter'‐[{ˆ»pçRtER]butnot*[{ˆ»poRtER]

TheSpondaicLoweringdoesnotoccurifthepluralmarkerisadjoined(e.g.,m[»e]sas

'tablepl.'andnot*m[»E]sas).

Both the Dactylic Lowering and the Spondaic Lowering do not apply to words in

whichverbsuffixesareadjoined(inthecaseofthedactyllowering‐e.g.,perd[»e]ssemos 'we

lost, past imperf. subj.'; in the caseof the spondaic lowering ‐ e.g.,esqu[»e]sas ‐ 'you forget,

pres.subj.').

Although Wetzel's arguments have a broad descriptive adequacy, some studies

clearlyclaimthatthereisnophonologicalevidenceforthefootinthelanguage(Mateusetal.,

2003: 1060; Vigário, 2003:336). Wetzels proposal has some problems, disfavoring the

evidenceforthefoot(eitherthedactylorthespondee)inPortuguese(Vigário,p.c.):

(i) ExceptionstotheDactylicLoweringwherenolowvowelsarefoundsuggestthat

the'neutralized'vowelsmaybelowvowelsunderlyingly,anddonotresultfrom

aruleapplication.Theauthor(Wetzels,1992,1995)andBisol(1999)pointout

some of thewords towhich the Dactylic Lowering do not apply (e.g., f[»o]lego

'deep breath', b[»e]bedo/b[»e]bado 'drunk', p[»e]ssego 'peach', tr[»o]pego 'lame',

es[»o]fago(BP)'esophagus',est[»o]mago'stomach',[»e]xito);

(ii) The fact that the Spondaic Lowering does not apply in inflected forms (e.g.,

m[»e]sas 'tablepl.'),andthefactthatitisnoticeableinnon‐derived(e.g.,d[»ç]lar

'dolar',C[»E]sar'name')andderivedwords(e.g.,m[»ç]vel'mobile',d[»ç]cil'docile')

suggests that lowvowelscanberepresented in the lexiconand itmaynotbea

strict phonological process (Bisol, 1999; Wetzels, 1995, 2006), but rather a

morpho‐phonologicalprocess.

AnotherempiricalargumentdisfavoringthepertinacityoffeetinPortuguese,regards

theformationofhypochoristicsandtheprocessesofwordtruncationinthelanguage.Ifwe

assume that hypochoristics reflects some degree of prosodic organization (McCarthy &

Prince, 1995), namelyminimalwords and the predominant feet in a language,wewill see

thatthereisnotaparticularfootrequirementinPortuguese,asS,WSandSWhypochoristics

are observable in the language (Pereira, 1999:96). Nicknames like Zé (José) [»zE], Fá

45

(Mafalda) [»fa], Zezé (José) [zE»zE], Dani (Daniel) [da»ni], Chico (Francisco) [»Siku] ou Lena

(Helena)[»lenå]arefrequentinPortuguese.Additionally,wordtruncationmighthavevaried

prosodic formats (telemóvel 'cell phone' > móvel, heterossexual 'heterosexual' > hétero,

internet'internet'>net‐Vigário,2003:254).AccordingtoVigário(2003:336)thetargetofa

truncationprocess69inEPdoesnotseemtobethefoot,butrathertheProsodicWord.

1.2.5.2.Weight­sensitivityormorphology­dependence.

The debate in the literature on stress in Portuguese confirms that the distinction

betweenweight‐sensitiveandmorphology‐dependentstressisnotasstraightforwardasitis

formanyGermaniclanguages,likeDutchorEnglish(Kager,1989).Ontheonehand,authors

claim that Portuguese is weight‐sensitive (Bisol, 1992, 1993, 1999; Carvalho, 1987, 1988;

Wetzels, 1992, 2003, 2006); on the other hand, authors defend that morphological

constituency constrains the location of stress in Portuguese (Andrade, 1992, 1996, 1997;

Andrade&Laks,1992;Lee,1995;Mateus,1983;Mateus&d'Andrade,2000;Pereira,1999).

Also,theissuewasoftentakenasa'BP‐weight‐sensitive'versus'EP‐morphology‐dependent'.

Withrespecttothefrequentthetheoreticalclaimsappliedtoonesystem(EPorBP),

Pereira(1999)affirms"Itdoesnotseemnecessarytometospecifysuchdifference,asthere

arenodifferencesinstresslocation[...]betweenthetwovarietiesofthelanguage"(Pereira,

1999:83,fn79)70.

Carvalho (1987, 1988) argues that EP, but notBP, has unstressed vowel reduction,

since full unstressed vowels should be interpreted as complex segments (in this case, VV).

Mergingvowelsphenomena suchas theone found inaajuda ([å å»Zudå]), aphrase that is

commonlyproducedas[a»Zudå]'thehelp',areexamplesofevidenceforweightplayingarole

in EP phonology. The reduced character of unstressed vowels constitutes evidence for the

phonological value of theweight in the language. Based on these assumptions, the author

specifies that heavy syllables are those where no vowel reduction is found, whereas light

syllableshavereducedvowels.AccordingtoCarvalho(1988),heavysyllables inPortuguese

are:

69Theauthorusesthedesignation'clipping',forwhatwerefertoastruncation.70Mytranslationfromtheoriginal inPortuguese:"Nãomeparecenecessárioespecificartaldiferença,dadonãohaver,noquerespeitaàlocalizaçãodoacentotónico,queéoqueaquiestáemcausa,diferençassignificativaentreasduasvariantesdalíngua"(Pereira,1999:83,fn79).Andrade(1996,1997)makethesameremark.

46

(i) Thestressedsyllables,which,ingeneral,havefullvowels‐bola[»bçlå]'ball',mala

[»malå]'suitcase',meta[»mEtå]'finishline';

(ii) Thesyllableswithavowel followedby/l/oraglide ‐ e.g. selvagem [sE…»vaZå‚j‚]

'wild',faisão[faj»zå‚w‚]'grouse';

(iii) Thesyllableswithnasalvowels‐e.g.pintar[pi‚»taR]'topaint';

(iv) The unstressed syllables with low vowels (/E, a, ç/) ‐ e.g. pregar [pRE»gaR] 'to

preach',corar[kç»RaR]'toblush',ganhar[ga»¯aR]'towin'.

Intheexamplesgiveninthelinesabove,heavysyllablesappeartobeirrespectiveof

stressassignment71.Actually,themainproblemwithCarvalho'sanalysisisthatheconsiders

heavyvowelsthatarefullbutmightbeunstressed.Theauthoractuallydoesnotuseweight

toclaimforaweight‐sensitivestressinthelanguage,butratherhearguesforthecontrastive

roleofsyllableweightinthephonologyofEP.

Bisol (1992, 1993, 1999) andWetzels (1992, 1995, 2003, 2006) defend that BP is

weight‐sensitive (although thevastmajorityof arguments adducedare suitable forBPand

EP).Bisol(1993,1999)72assumesthatbothnounsandverbsaresensitivetosyllableweight,

whereasWetzels(2006)onlyconsidersweight‐sensitivityinnouns.Theformerassumesthat

stressinBPisweight‐sensitivebasedonthreetypesofevidence:

(i) Final stress is preferred when a word ends in a consonant73, i.e., whenever a

word ends in a heavy syllable, stress falls on that syllable, and not on the

penultimate,asillustratedin(55):

(55) rapaz'boy'[{å»paS],colher'spoon'[ku»¥ER],anel'ring'[å»nE…])

(ii) Whenthewordendsinavowel,penultimatestressispreferred,i.e.,whenaword

endsinalightsyllable,stressfallsonthepenultimatesyllable,asshownin(56):

(56) gato'cat'[»gatu],boneca'doll'[bu»nEkå]).

71ForacriticismonCarvalho(1987,1988),cf.Pereira(1999:101‐119).72Bisol(1999)presentsareviewonpreviousworksfromtheauthor(Bisol,1992,1993).73Weassumethatthisgeneralizationappliestowordsinwhichtheword‐final/‐s/isnotthepluralmarker,likeinrapaz 'boy' /{å»paS/,nariz 'nose' /nå»RiS/, cabaz 'basket' /kå»baS/, etc, as opposed towords like casas 'houses'/»kazåS/, gatos 'cats' /»gatuS/, etc. On this assumption, the pluralmarker /‐s/ is irrelevant for syllable weightpurposes.

47

(iii) Whenthepenultimatesyllableofawordisheavy,stressneverwithdrawstothe

antepenultimatesyllable, i.e.,wheneverthepenultimatesyllable isheavy,stress

cannotsurpassandfallsonthatsyllable,aspresentedin(57):

(57) cadastro 'criminal record' [kå»daStRu]/*[»kadåStRu], covarde 'coward' [ku»vaRdˆ]/

*[»kuvaRdˆ]).

Bothauthorsconsiderthatmostwordswithfinalstressedopen‐syllable(café'coffee'

[kå»fE], javali 'wildpig'[Zåvå»li],sofá 'couch'[su»fa])areofforeignorigin,asthemajorityof

thesewords(ifnotall)haveanAfrican,French,Englishandindigenousbasis.

In summary, four facts suggest that weight plays a role in stress assignment

(Carvalho,1987,1988;Bisol,1994,1999;Wetzels,1992,2003,2006):

(i) EP has full stressed vowels and unreduced unstressed vowels, suggesting that

full unstressed vowels might be the result of a coalescence phenomenon and

therefore have complex (heavy) segments74 ‐ e.g., s[E]lvágem 'wild', c[a]ixóte

'box',séni[ç]r'senior',c[ç]rar'toblush'.

(ii) In thenouns' system,heavysyllables, i.e., syllablesending ina consonant (/R, l,

s75/) or a glide overwhelmingly bear stress, in general (cf. table in (52), with

frequency information on syllable types in stressed and unstressed position).

SyllableweightcountsforstressassignmentdirectlyandindirectlyattheRhyme

level (i.e., syllables are heavy at the level of the Rhyme and at the level of the

Nucleus ‐ /VC, VG, VN, VNC, VGC, VGN, VGNC/) ‐ e.g., balão 'balloon', carapáu

'mackerel',rapáz'boy',amór'love',anél'ring'.

(iii) In Portuguese, antepenultimate syllables are never stressed when the

penultimate syllable is heavy ('CV.CVC.CV) ‐ e.g., cérebro 'brain' and cadastro

'criminalrecord'but*cádastro;

(iv) Stressedmid vowels in prefinal syllables followed by a finalheavy syllable are

neutralized (Spondaic Lowering) ‐ e.g., móvel 'desk'/'*m[o]vel', órfão

'orphan'/'*['o]rfão.

74Accordingtothisperspective,EPisweight‐sensitiveandBPisnot(Carvalho,1988).75Cf.footnote73.

48

The alternative approach, according to which word stress in Portuguese relies on

morphological information, namely on morphological constituency, was defended by

Andrade(1992,1996,1997),Andrade&Lack(1992),Lee(1995),Mateus(1983),Mateus&

d’Andrade (2000), Pereira (1999), Vigário (2003). From the descriptions presented (cf.

section 1.2.1. and references toMateus, 1983 andMateus et al., 2003 therein), threemain

factssuggestthatmorphologyplaysaroleinPortuguesewordstressassignment(Lee,1995,

2001,2006,2007;Mateus&Andrade,2000;Pereira,1999):

(i) Non‐verbsandverbsaresubjecttodifferentwordstressalgorithms;

(ii) Innon‐verbs,stressgenerallyfallsonthelastsyllableofthestem;

(iii) Non‐verbs'rulederivesoxytonicathematicwordswithoutfinalconsonant‐café]

'coffee',chaminé]'chimney',sofá]'couch',champô]'shampoo'76;

(iv) Inverbs,regularitiesastostressassignmentinmorphologicalconstituents,and

notas referring towordpositionor syllableweight (namely themevowelsand

tense and mood suffixes) are observable ‐ e.g., calará 's/he will shut up',

calaríamos'wewillshutup'.

Afterwehavepresentedthegeneraldescriptionofwordstressfromacross‐linguistic

perspective and the analyses proposed for Portuguese word stress, a few questions and

problemsareyettobelisted.

From the reports presented above, the claim for a strict weight‐sensitive or

morphology‐dependent stress in Portuguese is hard to defend. Some authors state that

Portuguesewordstress isweight‐sensitiveandotherauthorsdefendthat it ismorphology‐

dependent.However,mostoftheauthorsagreethatstressisarhythmicfeaturethatinteracts

withmorphology.

The assumption of a weight‐sensitive or morphology‐dependent word stress in

Portuguesemayhaveseveralimplicationsforwordstressanalysesinthelanguage.Thefirst

76Althoughit isnotthemostfrequentstresspatterninPortuguese,manywords,fromthecommonvocabulary,have word final stress in an open syllable. Here we list some of them: café 'coffee' [kå»fE], jacaré 'aligator'[Zåkå»RE], javali 'wild pig' [Zåvå»li], sofá 'couch' [su»fa], sopé 'mountain bottom' [su»pE], capilé 'sweet drink'[kåpi»lE],aloé'aloe'[ålu»E],avô'grandfather'[å»vo],avó'grandmother'[å»vç],balancé'swinger'[bålå‚»sE],champô'shampoo' [Så‚»po], canapé 'starter' [kånå»pE], chulé 'stink (feet)' [Su»lE], comité 'comitee' [kumi»tE], esquimó'esquimo'[Ski»mç],chaminé'chimney'[Såmi»nE],ilhó'grommet'[i»¥ç],oboé'oboe'[obu»E],pontapé'kick'[po‚tå»pE],trenó'bobsled'[tRˆ»nç].

49

implicationconcernsthedomainofapplicationofstressandthesecondimplicationregards

determinationofextrametricality.

Inmostoftheanalysisdefendingmorphology‐dependence(Andrade,1992;Andrade

&Laks,1992;Mateus,1983;Mateusetal.,2000,2003;Pereira,1999),thedomainforstress

application is the stem in non‐verbs and the lexical word in verbs. In the majority of the

approaches defending weight‐sensitivity, the domain for word stress is the lexical word

(Bisol,1992,1993,1999;Wetzels,2003,200677).

The determination of which constituents are extrametrical also depends on the

approach taken. Assuming that Portuguese word stress relies on the morphological

constituency, the morphological information out of the stem might be considered as

extrametrical(Andrade,1992;Lee,1995).Inweight‐sensitiveanalyses,SWWnon‐verbsmay

bear an extrametrical syllable, as well as SW non‐verbs with a final heavy syllable (Bisol,

1992,1993;Wetzels,2006).

The pertinacity of feet in the language is also a matter of debate in the literature,

someauthorsprovidingevidencefortheSpondaicandDactylicfootandothersclearlystating

thatthefootisnotrelevantinanyphonologicalprocess.Often,theauthors'analysesonBPor

EP may lead to the idea that the two varieties of the language may present different

behaviourtowardswordstress,whichdoesnotholdtrue.

Assumingthatstress inPortuguese isnot lexicallyspecified‐ the literaturepointed,

otherwise,fora'rule‐based'stress78,eithermorphologicallyconditionedorweight‐sensitive‐

,childrenacquiringPortugueseneedtolearntheruleforwordstress.Buthowdotheylearn

wordstress?Howwillachildbuildwordsinordertomarkwordprominence?

Inthisdissertationweaimatsheddinglightontheconflictingissuesthathavebeen

listed.BasedonthespeechproductionsofPortuguesechildren,weintendtocontributewith

newempiricaldatatothedebateonthenatureofPortuguesewordstress,particularlyinEP.

77Asmentionedinfootnote40,Lee(2006,2007)constitutesanexceptiontotheseapproaches,assumingweight‐sensitivityandmorphology‐dependenceinnon‐verbsandastrictmorphology‐dependenceinverbs.78DespiteearlyassumptionsofPortuguesewordstressasa'freestress'language(i.e.,stresshadnofixedlocationanditsplacementwasunpredictable‐Barbosa,1965/1983),phonologicalanalysesprovidedrobustevidenceforafixed,predictablestressassignment.

50

51

2.Theacquisitionofwordstress

In the previous chapter, we descrubed the general principles of word stress, both

cross‐linguistically and in Portuguese. We observed that word stress is not a universal

linguistic feature. It is, though, a phonological aspect of many languages, used for

phonologicalcontrast. InPortuguese, thedescriptionsonwordstresshaveposedanumber

of problems, namely concerning its domain of application (either the stem or the lexical

word),thefootshape,weight‐sensitivity,sensitivitytowordclassandnatureandtheextent

ofextrametricality79.

Inthischapter,wewillreviewthemaintopicsmentionedintheliteratureregarding

theacquisitionanddevelopmentofwordstressandwordshape,withspecialreferenceto:

(i) Therepresentationofearlywords;

(ii) Theinitialstresspatterns,aswellastheirdevelopment;

(iii) Thenatureandroleofearlyfillersandreduplicationsinprosodicdevelopment;

(iv) Wordstressacquisitionfromanacousticperspective;

(v) The meaning of the acquisition of stress patterns for a theory of language

acquisition.

Wewillfirstpresenttheacousticstudiescarriedonwordstressacquisition(section

2.1.). In section 2.2., we will describe the developmental paths for word stress across

languages. In this section,wewill give special attention to the earlywords' representation

(section2.2.1.)andtheorderofacquisitionofthedifferentstresspatterns(section2.2.2.).In

section 2.3. and 2.4., we will focus on two aspects which have often been evoked by the

literature on prosodic acquisition and the acquisition of stress patterns: fillers sounds and

reduplications,respectively.Insection2.5.,wewillpresenttheresultsfrompreviousstudies

onPortugueseword stress acquisition and, finally, in section2.6.,wewill review themain

issues expressed for the acquisition of word stress, and we will present our research

questionsandhypotheses.

79TheseproblemshavebeenlistedrecentlybyLee(2006).

52

2.1.Acousticstudiesonwordstressacquisition

Theacousticstudyofwordstress,bothinadultsandinchildren’sspeech,becamea

common resource, as more accurate instrumental methods were made available by

technology. In fact, the need for instrumental data becomes crucial when it comes to

identifyingparticularaspectsofchildren’sspeech(eithersegments,syllables,stressorother

prosodicaspects),whichmightnotbeadult‐like.

Earlyspeechisdifficulttotranscribeandwordprominencecanbehardtoidentifyin

astagewhenmeaninglessandmeaningfulspeechoverlap,whenreduplicationsarefrequent

andwhengreatvariationispresent.Bernhardt&Stemberger(1998:443)pointoutthat‘[t]he

acousticcuesforstressarecomplex,andthereareindicationsthatyoungchildrenmaynotbe

abletocontrolthemverywell.(...)Whentranscribingthespeechofveryyoungchildren,itis

oftendifficulttotranscribestressconsistently.’

In the acoustic studies, researchers measure the relevant acoustic parameters for

word stress, in order to describe how children assign prominence, and, consequently, to

argue for the presence or absence of a given default stress pattern. When studying word

stressacquisitionortheacquisitionofrhythm,researchersusuallycarryoutadescriptionon

how the acoustic parameters (mainly fundamental frequency, intensity and duration) are

usedbychildren,andbringempiricalevidencefororagainstatheoreticalclaim,frequently

relatedtotheearlyrepresentationofwordstressandfootinginagivenlanguage.

Inthefollowingparagraphs,wewillreviewavailablestudiesfocusingontheacoustic

propertiesofstressduringacquisition.Wewillfurthermoreshowthattheusechildrenmake

of theacousticparameters inorder to correctlyproducewordstressmaynotbemastered

untiltheageof2;0.

Thework by Allen&Hawkins (1980)was pivotal in the acoustic analysis ofword

stressfromalanguageacquisitionperspective.Theauthorscarriedoutanacousticstudyon

stress andaccent80 in the speechproductionsof threeEnglish‐speaking children, aged2;8‐

3;4. The authors measured duration and fundamental frequency in 100 utterances (50,

collected in two separate sessions)with twoormorewords, and reported that the speech

productionsofthreechildrenverymuchresembledtheadultspeech:finalandpost‐nuclear

stressedsyllables tendedtobe longer thanthe initialonesand fallingcontourswere found

formostof thesyllable types.Theauthors furthershowed that theobservedchildrenused

bothacousticparameterstoderivetarget‐likeaccentedphrases.

Pollock,Brammer&Hageman(1993)investigatedtheuseoffundamentalfrequency,

80By 'stress', the authorsmean 'wordprominence' (e.g. 'balloon', 'camel'); by accent theauthorsmeanphrasalstressorphrasalprominence(e.g.'thebear').Foranotionofstressandaccent,cf.Beckman(1986).

53

intensityanddurationtostresswordsinyoungchildren'sspeech.Theyrecordedproductions

ofagroupofeighteenEnglish‐speakingchildren–3groupsofsixchildrenwith2‐,3‐and4

yearsold–andanalyzed themacoustically,bymeasuring fundamental frequency, intensity

anddurationofnoveldisyllabicwords.Theirresultssuggestthatonly3and4yearoldsmade

correct use of fundamental frequency, intensity and duration to mark a target‐stressed

syllable.Also, inthethreeagegroupsthestressedsyllablewassignificantlylongerthanthe

unstressedsyllable.However,2yearoldsproducedonaveragelongerstressedsyllablesthan

3and4yearolds,anddidnotcorrectlycontrolfundamentalfrequencyandintensitytomark

the stressed syllable, both in target 'CVCV and CV'CVwords. Although perceptual analysis

(also carried out) did not detect persistent use of level stress, it showed that stress

assignmentin2yearolds'productionswasatthechancelevel81.Theexperimentscarriedout

byPollocketal.(1993)pointoutthattheuse2year‐oldsmakeoftheacousticparametersto

markword stress is not yet controlled and that the transcription ofword stress of young

children'sspeechcanbeadifficult task.Childrenover3yearsold,on thecontrary,already

controlledalltheacousticparametersinthecorrectsensetomarktheprominentvowelina

word.

Examining the acoustic correlates of stress in spontaneous productions of children

aged 1;6‐2;6, Kehoe, Stoel‐Gammon & Buder (1995) showed that young English‐speaking

childrenwereabletousestresscontrastinanadult‐likefashioninmostoftheirproductions

(by using higher fundamental frequency, greater amplitude or longer duration to contrast

syllables).However,theuseoftheacousticparameterswasnotyetcompletelycontrolled,as

tokensperceivedasincorrectlystressed(30%)haddifferentacousticpropertiesthanthose

perceived as correct. The authors argued for an insufficient control over the phonetic

parameters in marking stress, rather than the incorrect use of the lexical stress pattern

(mostly SW, for English). The authors further observed considerable variability among

childreninthesensethatsomechildrentendedtoincorrectlystresswordsmorethanother,

thussuggestingthat"stressshiftappearstobeanidiosyncraticqualityofchildren'sspeech"

(Kehoeetal.,1995:349).

Lleó & Arias (2007) provided a theoretical analysis (within the Optimality Theory

framework) of word and phrasal stress acquisition, by acoustically analyzing spontaneous

speechdataintwoSpanishmonolingualsaged1;0‐2;6.TheresultsfromLleó&Arias(2007)

indicated that the children mastered word stress since their early productions, despite

havinganeffectoffinallengtheninginsingletrochees,showingthattheuseofduration(but

not amplitude and pitch) in trochees is not yet controlled. In tri‐ and quadrisyllableswith

81 "Resultsof theperceptualanalyses indicated that interjudgereliabilitywasparticularlypoor for theyoungersubjects"(Pollocketal.,1993:198).

54

penultimate stress, both children expectedly lengthened and raised the pitch of the

penultimate vowel but the use of amplitude varied. In quadrisyllables, both children

masteredduration, aspenultimate syllableswere lengthened in general.An analysis of the

stresserrorsperformedbythetwochildrensuggestedthatfewtrocheeswerenotproduced

as such.However, some target iambic‐shapedwordswereproducedas trochees,ataround

1;8‐1;9.Levelstresswasthemostfrequentstrategyusedbythesechildrenwhenwordstress

wasnotproducedtarget‐like.Lleó&Arias'resultssuggestedthatSpanish‐speakingchildren

mirrorthetarget languagetendency,althoughtheydonotnecessarilyproducetheacoustic

correlatesofstressinanadult‐likemanner.This,infact,ledadultstoperceivewordstressin

anotyetadult‐likemanneror,astheauthorsstate,inan'immatureway'.

From a compared perspective, Vihman, DePaolis & Davis (1998) studied

stress/accent82 acquisition in English and French‐speaking children, both carrying out an

acoustic analysis and aperception‐basedone, until de25‐wordpoint (16English‐speaking

childrenand5French‐speakingchildren).Theresultsoftheacousticanalysissuggestedthat

Frenchchildrenlengthenthesecondvowelofdisyllablesinanadult‐likefashion.Asforpitch

andintensity,Frenchchildrentendedtousehigherfundamentalfrequencyandamplitudein

thesecondsyllable,butsomevariabilitywasobserved.English‐speakingchildrenshoweda

highlevelofvariabilityinduration,pitchandamplitude.

Also in English and French, Rose & Champdoizeau (2008) presented data from a

bilingual English‐French speaking child, aged from 2;0 to 4;02. The authors analyzed

fundamental frequency, intensity and duration in disyllables and observed that the child's

EnglishandFrenchproductionsweremainlytarget‐like,bothinrhythmictermsaswellasin

terms of the realization of the relevant acoustic parameters for each one of the target

languages. English disyllables were produced with higher pitch and intensity in the first

vowelwhereasFrenchdisyllableswereproducedwithlongersecondvowel.

Insum,acousticresearchontheacquisitionofwordstressandwordprominencehad

leadustobelievethatthereseemstobeconflictingresultsonhowearlythemasteryofthe

acoustic parameters to produceword stress target‐like occurs, though a general pattern is

observed: before 2;0, children may not be able to correctly produce word stress. Some

evidence of incorrectword stress production includes thenon adult‐likeproductionof the

acousticparameters, theuseof levelstressor theproductionofstresserrors (for instance,

producing iambs in target trochees or vice‐versa). Cross‐linguistic differences were also

attested,namelyinanalysescomparingdatafrombilingualchilden.After2;0,childrenseem

to produce word prominence in an adult‐like manner, either by contrasting the syllables

82Thereisnowordstressbutonlyphrase(‐final)accentinFrench(Delattre,1965;Dell,1984).

55

within aword or by correctly using the relevant acoustic parameters in order to produce

target‐likestressedwords.

2.2.Theacquisitionofwordstressacrosslanguages

In this section, wewill present a review on themajorworks on the acquisition of

wordstress.

Theresearchontheacquisitionofwordstressandstresspatternshasbeenfocused

on three aspects of children's linguistic behavior: (i) children's productions of the stress

patterns (either comparing the children's productions with target stress patterns or only

observing children's output forms), (ii) children's truncation patterns and (iii) children's

stresserrors.

In the following paragraphs we review the literature on the acquisition of stress

patterns and word shape, bearing in mind the outputs for children speaking different

languages.Inaddition,wewillpresentcross‐linguisticandlanguage‐specifictendenciesand

theinterpretationofthefindingscarriedoutbytheauthors.

Inthissection,wewillpresentsomeproposalsfortherepresentationofearlywords

(section2.2.1.)andfortheacquisitionofstresspatternsacrosslanguages(section2.2.2.).

2.2.1.Earlywords'representation

Within the field of prosodic acquisition in general, and word stress acquisition in

particular, researchers have often discussed the shape early words have at the onset of

production,bothfromarepresentationalperspectiveandaproductionpointofview.

Inthissection,wewillintroducetheproposalsfortheshapeofearlywordsproduced

by children speakingdifferent languages, bypointingoutwhat is theprosodic structureof

these earlywords, namelywith respect to amonosyllabic or disyllabic prosodic template.

Information regarding the later stages of word production – namely, in respect to the

acquisitionanddevelopmentoftargetdisyllablesandpolisyllables,wherestresspatternsare

noticeable–willbedealtwithinthefollowingsection.

In his early work on child language, Jakobson (1941/1969) observed that young

children started with a basic CV syllable (though V and VC are also possible), in order to

maximally contrast the speech sounds produced. During the early periods of speech

production, these initial syllables might appear singly or be reduplicated (e.g. papa and

mama),asdescribedbytheauthor:

56

"Pourcequiestde l'ordredecesconstituants, laséquence“consonnesuiviede

voyelle”semblequasicontraignante(...).Pendant lapériodedebabildudéveloppement

de l'enfant, unegrandepartiedes syllabesprononcées consistent en la successiond'un

sonvocaliqueetd'unearticulationconsonantique. (...)Dans le languagedupetitenfant,

lestermespapa‐maman,demêmequelespremièresmotsformantunité,necomportent

pas de consonnes différentes, et les formes disyllabiques répètent habituellement une

seuleetmêmeconsonne."(Jakobson,1941/1969:124,125)

Cross‐linguistic researchhas indeeddemonstrated that thebasicunit children start

withinproductionisamonosyllable(Ingram,1989,1992;Vihman,1992).Itdoesnotseem

clear,however,whether theproductionof thesyllable inanearlystagemight indicate that

thesylableisa'lexicalorganizingunit'(Vihman,1992:415).

Several reports in the literature have suggested that the first articulated and

meaningful utterances are the product of the contrast between a consonant and a vowel

(MacNeilage&Davis,1993,2000,2001),whichcanbeproducedinisolationorinarepetition

ofsyllables,andthatitmayreflectthechildren'sphonologicalstructure.

Studying the prosodic development of children, several authors claim that, at the

onset of word production, the children's unit of analysis is the syllable (Demuth, 1995,

1996a,b; Demuth & Fee, 1995; Fee, 1996; Fikkert; 1994; Johnson & Salidis, 1996). These

authors proposed that children speaking two Germanic languages ‐ English and Dutch ‐

mainlyproduce theCoreSyllable,eitherCVorCV:,as, in thisstage,vowel length isnotyet

mastered.

In theonsetofwordproduction,childrenmainlypayattentionto thesegmentsand

thesyllablestructureandmostlyproducecoresyllables,asillustratedin(58).

StageI:CoreSyllables‐CV(Novowellengthdistinctions)

(58) Monosyllabic productions ‐ unmastery of vowel quantity (Dutch83 and English ‐

Demuth,1995,1996b):

Orthogr. Target Output Child,Ageklaar /kla˘r/ [ka˘],[kA]dit /dIt/ [ti˘],[tI]

J.,1;4‐1;5

juice /dzus/ [du],[du˘]juice /dzus/ [gu],[gu˘]

PJ,1;8

83Allexamples(DutchandEnglish)areinDemuth(1995,1996b).TheDutchrenditionsthereinweretakenfromFikkert(1994).

57

Thedevelopmentalpathproposedbytheseauthors(Fee,1992;Demuth&Fee,1995;

Demuth 1995, 1996a,b) predicts that, in the early stages, a prosodic wordwould be a CV

monosyllable.Andthiswasindeedobservedinotherlanguages,bothGermanicandRomance

languages, namelyGerman, Catalan and French (Lleó&Demuth, 1999, for German; Prieto,

2006 for Catalan; and Rose, 2000 and Demuth & Johnson, 2003, for French), as we will

describebelow.

Comparing theprosodic acquisitionofGermanand Spanish, Lleó&Demuth (1999)

also claim that theearlywords found in the speechofGerman‐speaking childrenareoften

monosyllables,thoughwithaCVCstructure,asshownin(59)84.

(59) MonosyllabicproductionsofGerman‐speakingchildren(Lleó&Demuth,1999:2):

Orthogr. Gloss Target Output Child,Agekaputt 'broken' /ka»pUt/ [»pUth] Britta,1;7.11Fasan 'pheasant' /fa»za˘n/ [»zån] Thomas,1;8.2karton 'box' /kar»to˘n/ [»ton] Thomas,1;9.0kaputt 'broken' /ka»pUt/ [»puX] Marion,1;10.5

An earlymonosyllabic tendencywas found for Catalan in Prieto (2006). Early CVC

monosyllables were observed, though CV syllables could also be produced, both resulting

from/CVC/truncationor/CV/targets,asillustratedin(60).

(60) MonosyllabicproductionofCatalan‐speakingchildren(Prieto,2006:245):

Orthogr. Gloss Target Output Child,Agemà 'hand' /»ma/ [»ma] Ot,1;5.4.tot 'all' /»tot/ [»tHo] Lluís,1;2.20pa 'bread' /»pHa/ [»pHa] Lluís,1;3.20carn 'meat' /»karn/ [»tan] Anna,1;2./1;5llum 'light' /»¥um/ [»bum] Laura,1;7.20

InFrench,Rose(2000)observedthatchildrencouldproduceCVwordsforCVtargets.

TargetCVCwordsaremostlyrealizedasCV,aswell.Therenditionsin(61)illustratetheearly

tendencyforCVformsinFrench.

84Furtherahead, insection2.2.2.,wewillnote thatmanyauthorsargue for theearlyproductionofadisyllabictrocheeinSpanish.

58

(61) MonosyllabicproductionofFrench‐speakingchildren(Rose,2000:99,101,113):

Orthogr. Gloss Target Output Child,AgeGuy 'name' /gi/ [gi] Clara,1;03.07l'eau 'thewater' /lo/ [lç]oui 'yes' /wi/ [wi˘] Clara,1;04.07

non 'no' /nç‚/ [na] Théo,1;11.10oui 'yes' /wi/ [we] Théo,2;03.06livre 'book' /liv/ [ji] Clara,1;04.14pomme 'apple' /pçm/ [bç˘] Clara,1;06.22pique '(it)pikes' /pIk/ [pI] Théo,2;01.19voir '(to)see' /vwa“/ [va] Théo,2;02.16

Thesefindingsindicatethat,attheonsetofwordproduction,Frenchchildrenmight

beprocessingthesyllable,byproducingonlymonosyllabicforms.However,thefindingsfrom

Demuth & Johnson (2003), indicate that one Parisian French‐speaking child, aged 1;1‐1;8,

producedCVorreduplicatedCVCVwords(CV1CV1),forCVandCVCVtargets,respectively,as

shownin(62).

(62) Production of monosyllabic and reduplicated words of a French‐speaking child

(Demuth&Johnson,2003:221):

Orthogr. Gloss Target Output Agemerci 'thankyou' /mE{si/ [Si]/[si] 1;3pupée 'doll' /pupe/ [pepe] 1;3tout 'all' /tu/ [tu] 1;3pain 'bread' /pE‚/ [pa] 1;4là 'there' /la/ [na] 1;4chapeau 'hat' /Sapo/ [popo] 1;4pelle 'shovel' /pEl/ [pepe] 1;4canne 'stick' /kan/ [tata] 1;4chat 'cat' /Sa/ [a] 1;5oeuf 'egg' /øf/ [toto] 1;5

Co‐occurringmono‐ and disyllabic reduplicated forms, provide evidence against an

obligatory binary or disyllabic foot in the early speech of French children and support the

ideathat,atthebeginning,Frenchchildrenareprocessingthesyllable,eveniftheyuseitin

utteranceslargerthanamonosyllable.

The results in the literature for the early word's representation suggest the

productionofasyllable,intheearlystagesofwordproduction.Thesyllableunderchildren's

59

attentioncansurfaceasamonosyllable(CV,CV:orCVC),asinDutch,EnglishandGerman,or

bereduplicated(CV1CV1),asinFrench.

2.2.2.Theacquisitionofstresspatterns

Assoonaswordswithmore thanonesyllable start toemerge in children's speech,

the acquisition of word stress as a contrasting feature may start to take place. The

contributionsfromavastarrayoflanguagestothetopicofwordstressacquisitionhavebeen

often related to aspects suchas the initial foot structure and thedevelopmentof theword

structure.However,cross‐linguisticcomparisons,aswellasdescriptionsmadeonparticular

languages,havenotalwaysbeenconsensual.

A review on Germanic languages (mainly English and Dutch) suggests that, after

childrenstartproducingmainlymonosyllables,theygoupintheprosodichierarchy,passing

throughthefoot,totheprosodicworddomain(Demuth,1995;Fikkert,1994).Aconsiderable

amountof studieson the field further suggests that, inGermanic languages,namelyDutch,

EnglishandGermantheinitialfootisthetrochee(Allen&Hawkins,1979,1980;Bernhard&

Stemberger, 1998;Demuth, 1996a; Fikkert, 1994; Gerken, 1990, 1994, 1996; Kehoe, 1998,

2000,2001;Kehoe&Stoel‐Gammon,1997;Pater,1997).Note thatGermanic languagesare

considered to be heavily trochaic (Kager, 1989). Expectedly, children speaking Germanic

languagesproducetrochaicwordsearlierthaniambiconesandtheytendtotruncateWSW

trisyllablespreferably intoSWwords.Additionally, the initialweak(unstressed)syllableof

WSwordsisoftendeletedintheearlystagesofwordproduction,whiletheweaksyllableof

SWwordsispreserved('(ba)lloon'asopposedto'camel').

Inaseminalinvestigationonphonologicalrhythm,Allen&Hawkins(1979)establish

a few cues thatmay account for a definition of the rhythm across languages. The authors

suggest that children's early reduplicative disyllables, children's early lexicon and nursery

rhymesmayindicatethedefaultrhythmofalanguageandshowthebiaschildrenareprone

to during language acquisition. Also, the authors add that the truncation patterns in

trisyllabic words may indicate the children's preference for a specific (SW or WS) initial

rhythm or stress pattern. Looking at reduplications, early lexicon and truncation patterns,

Allen&Hawkins (1979) report the resultsof anEnglish‐speaking child, indicating that the

earlyword shapes of that childweremostly trochaic. Likewise, nursery rhymes in English

suggestaverystrongbiastowardsatrochaicfoot.

TheTrochaicBiasHypothesis, initially proposedbyAllen&Hawkins (1979), had a

great repercussion throughout the subsequent years, especially inGermanic languages like

English, Dutch and German. In the following year, Allen & Hawkins (1980) undertook a

60

productionstudywherechildrenaged3;7‐6;7werepresentedwithnon‐sensewordssimilar

insegmentalcontent,butcontrastinginthestressposition(e.g.[»taki]and[ta»ki]).Children

perceived the difference between SW and WS words but had difficulty in producing WS

(eitherusinginappropriatestressordeletingtheinitialsyllable).Theseresultsconfirmedthe

resultsfoundpreviously,indicatingthatchildren'searlyproductionswerebiasedtowardsa

trochaictemplate.

In the nineties,manyworks,mostly fromGermanic languages, confirmed the early

trochaictendencyfoundbyAllenandHawkins.Gerken(1990,1994,1996)investigatedthe

omissionpatternsinEnglish‐speakingchildrenandconcludedthatitwasmorelikelythatthe

initialweak syllableofWSwasomitted,when compared to theweak syllableof target SW

words, both in isolation and at the sentence level. English‐speaking children more easily

maintained syllableswithin a trochaic foot, thusproviding further support for the claimof

theearlytrochaictendencyinEnglish.

Fikkert(1994)alsofoundanearlytrochaictendencyinthespeechofDutchchildren.

Inacomprehensiveworkonprosodicacquisition,Fikkert(1994)proposedadevelopmental

pathforwordstressacquisitioninDutch.BasedonthecomputationalmodelfromDresher&

Kaye (1990), Fikkert accounts for children's behavior within the Principle & Parameters

model. The author focuses, not only on the earlyword shape, but also on theword stress

patternsacrossdevelopment.Fikkert's(1994)proposalfortheacquisitionofwordstressin

Dutchconstitutes,indeed,aclearexampleofalearnabilitymodelforwordstress.

After an initial stage (Stage 0) in which Dutch children mainly produce CV or CV:

monosyllables85,DutchacquisitiondataindicatedthatSWwordswereproducedtarget‐like86,

whereasWSwordswerepronetotruncation.

In(63),wepresentsomerenditionsfromNoortjeinStage1(Fikkert,1994).

(63) Stage 1 of word stress acquisition in Dutch ‐ /WS/ produced as a monosyllable

(Fikkert,1994:206):

Orthogr. Gloss Target Output Child,Agebanaan 'banana' /ba˘»na˘n/ [»na˘m] Noortje,2;3.7konijn 'rabbit' /ko˘»nEin/ [»kEin] Noortje,2;3.23

85Cf.section2.2.1..86Fikkert(1994:201)showsthepercentagesofdeletionoftheunstressedsyllablein/SW/and/WS/wordsandthe percentage of stress errors in the same words. In both tables, /SW/ words had very low percentages,suggesting that theywereproduced target‐like from thebeginning.However, inChapter6, theauthorpresentsresultsfor/S/or/‐WS/words,butnotfor/SW/.

61

InStage1,disyllablesbecomepartofthechild'ssystem.Atthispoint,thechildneeds

to know (i) how feet are parsed in his/her language (Left‐to‐Right or Right‐to‐Left), (ii)

whereistheheadofthefoot(attheRightedgeorattheLeftedge),(iii)whetherweightplays

a role in his/her language and, in particular, on stress assignment, (iv)whether there are

extrametrical syllables and, finally (v) whether feet in his/her language are binary or

unbounded.

In(64)wepresenttheparametersettingforwordstressacquisitionatthisstage.

(64) Parameters set at Stage 1 for word stress acquisition (Dutch acquisition data ‐

Fikkert,1994:280):

Directionalityparameter Feetarebuiltfromthe [Right]

Headednessparameter Feetarestrongonthe [Left]

QSparameter FeetareQS [No]

Extrametricalityparameter Thereisanextrametricalsyllable [No]

B/Uparameter Feetare [Binary]

Accordingtotheauthor,thefactthatchildrenmainlyproducetrocheesandtruncate

iambs is, in itself,anevidence foraRight‐to‐Leftparsingandaquantity‐insensitive,binary,

Left‐headedfoot.Itisaguedthat,ifchildrenwereprocessingfeetdirectionalityfromLeft‐to‐

Right(intheassumptionofbinarity),oneshouldexpectthatWSwereproducedmaintaining

theleftmostsyllablesofthetargetwordandSWwordswouldbetruncated.Furthermore,the

rightmostsyllablewouldonlybemaintainedif itwerestrong.Conversely, inaRight‐to‐Left

approach,theleftmostfootandsyllablewouldalwaysbemaintained,evenifitwereaweak

one. Since evidence for a Right‐to‐Left directionality was found, the author proposes that

RightisthedefaultvaluefordirectionalityandLeftisthedefaultvalueforheadedness.

At Stage 1, there is no evidence that Dutch children have set a marked value for

directionality, headedness and extrametricality. At this stage, the parameters for edge of

extrametricality,mainstressanddefootingparameterareirrelevant.

At Stage 2, the parameters set aremaintained equal to Stage 1, butDutch children

startprocessingdisyllables,both in targetSWandWS.However, thechildnowputsall the

targetwords–/WS/included–intoaquantity‐insensitivetrochaicfoot,asshownin(65).

62

(65) Stage 2 of word stress acquisition ‐ /WS/ words produced as [SW] (Fikkert,

1994:203):

Orthogr. Gloss Target Output Child,Ageguitar 'guitar' /Xi˘»ta˘r/ [»hi˘ta˘] Catootje,1;10.11konijn 'rabbit' /ko˘»nEin/ [»kIna˘] Catootje,1;10.11

At Stage 3, Dutch children realize that the output form does not alwaysmatch the

input (namely when producing trochees out of target iambs) and that disyllables can be

either SW or WS. At this stage, Dutch children produce iambic words with level stress,

providingevidenceforaprosodicwordlongerthanonefoot.SinceWSdonotconformafoot,

s/heproducestwofeet,thatis,awordwithtwomainstresses.

(66) Stage 3 of word stress acquisition ‐ two feet and level stress (Fikkert,

1994:214,288):

Orthogr. Gloss Target Output Agekasteel 'castle' /«kAs»te˘l/ [»tAs´»sO˘] Tirza,1;11.19konijntjes 'bunnys' /ko˘»nEintÉj´s/ [»kçnI»nEintÉj´s] Tirza,2;6.12David 'name' /»da˘vIt/ [»tA»fu˘n] Robin,2;1.26tractor 'truck' /»trEktçr/ [»tAk»tç®] Robin,2;2.37

At Stage 3, Dutch children set the parameter values for the quantity‐sensitivity,

weight, iterativityandbinarity/unboundednessparameter.Theyrealize that their language

isquantity‐sensitiveandwhatcountsasaheavysyllable.Thechildclosessyllables(VC)that

countasheavyandrealizesthattherearetwotypesoffeetinhis/herlanguage‐/SW/and

/WS/ ‐, whose prominences are assigned on the basis of syllable weight. The words like

/'CV.(C)VC/arenotproduced target‐like,butwith levelstress (['CV.'(C)VC]), since the final

syllableof thesewords isextrametricalandextrametricality isnotyetacquired.Therefore,

theyputthesameamountofprominenceinbothsyllables: inthesyllablecorrespondingto

theheadofthetrochaicfoot(thepenultimatesyllable)andinthefinalheavysyllable.

At this stage, children also realize that words can be longer than one foot, which

triggerstheiterativityparametertoitsmarkedvalue[Iterative].Likewise,ontheassumption

that Dutch has binary constituents (e.g., it has polisyllabic words with penultimate stress,

thus,itisnotunbounded),whenchildrenperceivethatstressisQS,theyrealizethatfeetare

binary, setting the [binary] value for B/U parameter. At this point, children's productions

havemaximallybinaryfeet.

63

In(67),wepresenttheparametersthataresetatstageIIIofwordstressacquisition

inDutch(Fikkert,1994).

(67) ParameterssetatStage3forwordstressacquisition(Fikkert,1994:290):

QSparameter FeetareQS [Yes]

Weightparameter Closedsyllablesareheavy [Yes]

Iterativityparameter Feetareiterative [Yes]

B/Uparameter Feetare [Binary]

Finally, atStage4, childrencorrectlyproducedisyllabicwordswith final stressand

settheword‐treedominanceparameter‐[Right].Thefactthatchildrenareabletocorrectly

producedisyllabicwordswith finalstress,butnotpolisyllabicwordswith finalmainstress

providesevidenceforthesettingofthemarkedvaluefortheObligatoryBranchingness(OB)

parameter.Atthisstage,childrenrealizethatfeetthemselvesbranchinthelanguageandthat

mainstressmustbeassignedtotherightmostbranchingfoot.

(68) Stage4ofwordstressacquisition‐ProsodicWords(Fikkert,1994:291):

Orthogr. Gloss Target Output Child,Agemicrofoon 'microphone' /«mi˘kro˘»fo˘n/ [»mi˘k´«so˘n] Tirza,2;0.5krokodil 'aligator' /«kro˘ko˘»dIl/ [»ko˘k´l«tçj] Tirza,2;1.17

Atthisstage,Dutchchildrensettheremainingparametersforwordstress:themain

stressparameterandtheOBparameter.

(69) ParameterssetatStage4forwordstressacquisition(Fikkert,1994:297):

Mainstressparameter Theword‐treeisstrongonthe [Right]

OBparameter Afootreceivingmainstressmustbebranching[Yes]

AccordingtoFikkert(1994),wordstressacquisitioninDutchmightbesummarized

inthefollowingfivestages:

64

Stage0: /S/PW thewordconsistsofonecoresyllable.

Stage1: /SW/PW children realize that the words might be

disyllabicandatrochaicfootisprocessed‐/WS/

are truncated into [S], whereas /SW/ do not

undergoanydevelopmentalpattern;/WSW/are

truncatedinto[SW]andnot[S].

Stage2: /SW/PW [SW] and /WS/ ‐> [SW]: this stage is different

from Stage 1 in the sense that /WS/ do not

become [S] anymorebut [SW] instead (through

the addition of a syllable). Metathesis and

duplications support the claim that the syllable,

andnotthefoot,isbeingprocessed

Stage3: /σΣσΣ/PW in /WS/ ‐ the 2 syllables are perceived as

belonging to twodifferent feetandbothvowels

are prominent (level stress). Children realize

thatthewordcanbelongerthanonefoot.

Stage4: /‐SW/PWand/‐

WS/PW

Target‐likeproduction.

Table1.DevelopmentalstagesforwordstressacquisitioninDutch(Fikkert,1994)

The data from Fikkert (1994) were later extended to English (Demuth, 1995,

1996a,b,c).Aspresentedinsection2.1.1.,Demuth&Fee(1995),Demuth(1995,1996a,b),Fee

(1996)andJohnson&Salidis(1996)proposethatprosodicacquisitioninEnglishstartswith

a monosyllable, later evolving to larger and more complex structure. According to the

authors,dataaspresented in(70) illustrate thisearlystage inprosodicacquisition,both in

DutchandinEnglish.

65

(70) Stage I ofprosodic acquisition:CoreSyllables ‐CV (Novowel lengthdistinctions ‐

Demuth,1995,1996b)87:

Orthogr. Target Output Child,Ageklaar /kla˘r/ [ka˘],[kA]dit /dIt/ [ti˘],[tI] J.,1;4‐1;5

juice /dzus/ [du],[du˘]juice /dzus/ [gu],[gu˘] PJ,1;8

Afteran initialstagewherechildrenmainlyproducemonosyllables,prosodicwords

would be either monosyllabic or disyllabic, but obligatorily bimoraic (withminimally two

short vowels, aVC syllable or a long vowel). In (71)wepresent instances fromStage II of

prosodicacquisition:

(71) Stage II of prosodic acquisition: Minimal Words/Binary Feet (Demuth, 1995,

1996b):

Orthogr. Gloss Target Output Child,Ageaap 'monkey' /»a˘p/ [»a˘p´] T.,1;4‐1;6CoreSyllables‐(C)VCVtuin 'garden' /toyn/ [»tøynA] N.,2;5‐2;7deze 'these' /»deze/ [teif],[de˘s]poes 'cat' /pu˘s/ [pu˘s] J.,1;6‐1;7

Closedsyllables‐(C)VC

egg ‐‐‐ /Eg/ [i/],[E/],[Eg] MH,1;7

stoel 'chair' /stu˘l/ [ty˘]daar 'there' /da˘r/ [da˘] J.,1;10‐2;0

cheese ‐‐‐ /tSiz/ [di]

Vowel length distinctions‐(C)VV

nose ‐‐‐ /noz/ [nu] AS,2;2

Inthethirdstage,childrenwouldstartbuildingupafootandassigningprominenceto

avowelintheprosodicword.Instancesin(72)accountforStageIIIofprosodicacquisition.

87Cf.footnote83,onDutchdatainDemuth(1995,1996a,b).

66

(72) Stage III of prosodic acquisition: Prosodic words larger than a binary foot (one

stressperfoot):

Orthogr. Gloss Target Output Child,Agemicrofoon 'microphone' /«mi˘kro˘»fo˘n/ [»mi˘k´»fo˘n] R.,1;10‐2;1olifant 'elephant' /»o˘li˘«fAnt/ [»o˘fA»fA˘n] J.,2;1‐2;4 tomato ­­­ /t´»mato/ [»ma˘do],[d´»ma˘do]elephant ­­­ /»El´f´nt/ [»Ebininin]motorbike ­­­ /»mot´rbaik/ [»mu˘g´ga˘baik]

AS,2;3

Finally,atstageIV,childrenwouldproduceaprosodicwordtarget‐like.

(73) StageIVofprosodicacquisition:Prosodicwords(targetform):

Orthogr. Gloss Target Output Child,Agepantoffels 'slippers' /«pAn»tçfçls/ [«pAn»tçf´ls] C.,2;4limonade 'lemonade' /«li˘mo˘»na˘d´/ [«mi˘mo˘»ma˘tA] N.,2;9

The comparative analyses of English andDutch prosodic development suggest that

theacquisitionofstresspatternsstartswithashorterunit(aCVsyllable)andevolves,inthe

courseofdevelopment,intoalonger,morecomplexstructure.Bothanalysesdefendthatthe

early foot is a trochee (i.e., awordwith a Strong‐Weakpattern ‐ e.g. ‘camel’).BothFikkert

(1994)andDemuth(1995,1996a,b)defendabottom‐upapproachofprosodicacquisition,as

formulatedin(74).

67

(74) Bottom‐upapproachtoprosodicacquisition:

U... g I... g Φ... g C ... g ω

[»ku˘»nEi]gΣ

[»ku˘lA] gσ

[»tEin]

Kehoe (1998, 2000, 2001), Kehoe & Stoel‐Gammon (1997), and Pater (1997)

confirmedtheearlytendencyforthedeletionofinitialunstressedsyllablesoftrisyllabicand

quadrisyllabic words in English‐speaking children. A higher deletion rate in the initial

stressed syllables was found. However, the results suggested that the truncation patterns

wereconditioned,notonlybymetricalaspects(Pater,1997),butalsobythelackofprosodic

salienceofinitialweaksyllables(Kehoe,2000,2001).

Analyzingthespontaneousandelicitedspeechof18English‐speakingchildrenaged

1;10‐2;10, Kehoe (1998) proposed a metrical approach to word stress acquisition and

comparedthedataobtainedwiththeonesproposedforDutchbyFikkert(1994).Theauthor

investigatedchildren'sproductionerrorsin/'SWS/,/SW'S/,/SWW/and/WSW/noveland

familiar words. The results indicated 3 main stages in prosodic development of English‐

speakingchildren.

In Stage I, denominated the 'Trochaic constraint stage', English children aged 1;10

mainlyproduceoneor two‐syllable truncations conforming a trochaic foot, though /'SWS/

and/SW'S/wordswereepisodicallyproduced.At this stage,Englishchildrenweredealing

with foot‐typeanddirectionality, truncating targetwords to [SW] from the rightmost edge

(e.g.,avocado[»kHAdo]).Theyoungerchildrenoftenusedstressshiftinthedirectionofheavy

unstressedsyllables,showingthataruleonweight‐sensitivitywasactiveatthatstage.

Stage II is an 'experimental stage'. Children aged 2;4 produced target trisyllables

68

mostlywith level and incorrect stress forms, sinceonlyone foot is beingprocessed in this

stage.InStageII,childrenarelearningextrametricality,quantity‐sensitivityandmainstress.

At Stage III, which characterized the speech of 2;10 year‐olds, few errors were

observed.Trisyllableswithinitialmainstressarecorrectlyproduced.

ThoughthereisstrongagreementfoundinGermaniclanguagesregardingtheshape

ofearlydisyllabicwords,someauthorsrejectedtheprimacyofthetrochaictendency,mainly

inEnglish(Klein,1984;Pollocketal.,1993;Vihman,DePaolis&Davis,1998).Inacasestudy,

Klein (1984) foundgreatvariability in themultisyllabicutterancesofoneEnglish‐speaking

child aged 2;0. Some words were produced correctly, others were misstressed and some

otherswere producedwith level stress,which, the author argues, provides evidence for a

neutral start and the 'lexical primacy during early stages of learning word stress', since

childrendidnotcommitmanystresserrors,providingevidencefortheabsenceofanyrule

(or regularization) taking place. In an acoustic study on the acquisition of word stress in

English‐speakingchildren,Pollocketal.(1993)argueforaneutralstartinlearningtostress,

since no indisputable evidence for the early ‘trochaic bias’ was found, although "some

evidenceofatrochaicbiaswasseeninthecompletedeletionof[initial]unstressedsyllables"

(p.198).

Using both acoustic measurements and perceptual analyses, Vihman, DePaolis &

Davis(1998)studiedthespeechofFrenchandEnglish‐speakingchildren.Theresultsofthe

perceptual analysis demonstrated that French children's productions were mostly iambic.

American children, on the contrary, showed more unbalanced results. Five (out of 9)

producedmoretrochees,whereas3ofthemproducedmainlyiambsand1ofthemproduced

as many trochees as iambs. Upon these results, the authors challenge the trochaic bias

hypothesisforEnglishlanguage,andproposethatananalysisofAmericandatashouldtake

into account the amount of disyllabic phrases, like 'the ball' or 'all gone' (rather than

disyllabic words), consisting of a function word plus a monosyllable due to the 'frequent

occurrenceintheinputofmonosyllabicwordscarryingthenuclearaccentofthephraseand

preceded by an unstressed syllable, often a function word such as an article' (Vihman,

DePaolis,&Davis,1998:942). Inbrief, in theearlystagesofwordproduction,childrenmay

not relyonwordprominence,butonphrasal stress instead.Rose&Champdoizeau(2008),

also acoustically analyzing the speech of a bilingual French and English‐speaking child,

clearlystandagainsttheuniversaltrochaicbiasandarguethatthetarget‐languageproperties

playanimportantroleinthechildren'sgrammaticalinterpretationofwordstress.

TheearlydominanceofthetrochaicbiasfoundinGermaniclanguages,however,did

not always find a match during the period of acquisition of non‐Germanic languages,

providingevidenceagainsttheuniversalorunmarkedstatusofthetrochaicfoot.

69

Authors,mainlyworking onnon‐Germanic languages also suggested that children’s

earlywordsmaybeneutralintermsofstresspatterns,meaningthatchildren’searlywords

do not display a particular stress pattern, either trochaic or iambic (Hochberg, 1988a, on

SpanishandTzakosta,2004,onGreek).Theresultsand theanalysesproposed forSpanish,

French, Catalan, Greek, Hebrew, Sesotho, K'iche' and BP indicate conflicting results with

respecttothestresspatternsdisplayedbychildrenduringacquisition.

Early reports on Spanish (Hochberg, 1988a) mention an important degree of

variabilityintheearlyspeechofchildren,regardingapreferencetowardsaparticularstress

pattern,namelythetrochee,themostfrequentfootinthelanguage.Theauthorreportedthat,

in the early speech of Spanish children (from 19 to 22months‐old) stress placementwas

random,thoughstressaccuracy(i.e.,therateofcorrectstressplacement)increasedwithage.

The author defended that children's productions are not biased towards a trochee, thus

claiming a neutral start approach to stress. Despite the great variability found, Hochberg

(1988b) further defended that stress learning is rule‐driven, based on spontaneous and

imitatedproductionsofSpanish‐speakingchildrenaged3,4and5.Theauthorobservedthat

irregularpatternstendedtoberegularized,thusarguingagainstalexically‐basedmechanism

forwordstress.

Conversely,inacomparativestudyonSpanishandGerman,bothtrochaiclanguages,

Lleó&Demuth(1999)analyzedthespontaneousspeechofthreeSpanish‐speakingchildren

andfourGerman‐speakingchildrenbetweentheagesof1;4‐2;3.Theauthorsconfirmthat,in

thebeginning,Spanishspeaking‐childrenmainlyhaveasyllabictrochee,asshownin(75a.),

whereasGermanspeaking‐childrenpreferredamoraictrochee,asillustratedin(75b.).

(75) Lleó&Demuth's representationproposal for theearlywordshape inSpanishand

German(Lleó&Demuth,1999:388):

a.Spanish b.German

PW PW

g g

Σ Σ

3 g

CVCV CVC

88 The page number referred to is relative to the manuscript version of the paper, downloaded fromwww.cog.brown.edu/~demuth/articles/1999%20Lleo&Demuth.pdf,downloadedonNovember30th2007.

70

InacomparisonbetweenSpanishandEnglishprosodicacquisition,Roark&Demuth

(2000) conclude that Spanish‐speaking children produce words longer than a disyllable

earlier thanEnglish,Dutch andGerman‐speaking children. Soon Spanish‐speaking children

gothroughastagewhereCV'CVCVareproduced,whilstEnglish‐speakingchildrenstillprefer

CVCwords.Spanishchildrenwould,accordingtoRoark&Demuth(2000),beabletoproduce

adisyllabictrochaicfootearlierthanEnglish‐speakingchildren.Thelaterfindingspresented

in Demuth (2001b) for Spanish acquisition demonstrated that, in his/her early stages, an

ArgentineanSpanish‐speakingchild, isable toproducedisyllableswithboth trochaic (as in

(76))andiambic(asin(77))stresspatterns.

(76) EarlytrochaicproductionsinSpanish(Demuth,2001:8):

Orthogr. Gloss Target Output Child,Agecaja 'box' /»kaxa/ [»kaxa] esta 'this' /»esta/ [»eta]

S.,1;8

(77) EarlyiambicproductionsinSpanish(Demuth,2001:8):

Orthogr. Gloss Target Output Child,Agepapá 'daddy' /pa»pa/ [pa»pa] acá 'here' /a»ka/ [a»ka]

S.,1;8

Atthisstage,however,iambicandtrochaicfeetshouldnotbeavailableinthechild's

system.Instead,thestressedsyllableofiambsismappedontoamonosyllabicfoot,withthe

precedingsyllablerepresentedatahigherleveloftheprosodichierarchy.Therepresentation

of thetrochees(caja 'box')and iambs(papá 'daddy')accordingtoDemuth(2001b:8) is the

following:

(78) a.Trochees b.Iambs

PW PW g 2 Σ gΣ 3 g g σs σw σ σs

g g g g

'ka xa pa 'pa

The arguments for this representation arise from theproductionofWSWwords as

SW and neverWS, as well as from the production of early determiners, before truncated

71

WSWwords.Atthisstage,trisyllablesaremostlytruncatedtodisyllables(79).

(79) EarlyWSWtruncationinSpanish‐[SW](Demuth,2001b:8):

Orthogr. Gloss Target Output Child,Agehamaca 'hammock' /a»maka/ [»maka] S.,1;8ventana 'window' /ben»tana/ [»tana] S.(1;9)

Early target‐like production in disyllables and the truncation patterns found for

wordslongerthantwosyllables(namely,/WSW/truncatedto[SW])suggestthatinSpanish,

likeinEnglishandDutch,theMinimalWordrestrictionconformstoadisyllabictrochaicfoot.

Thus,ratherthanhavingtwofeettypes(aniambicandatrochaicone),theauthorarguesthat

it ispossiblethatthechildusesaprosodicstructureliketheonein(78b.),whileproducing

[WS]forms.

Hebrew‐speaking children tend to prefer the trochaic stress pattern, which is less

frequent in the language (Adam & Bat‐El, 2008). The authors argue that, given the high

frequency of iambic nouns in the language, data from a Hebrew‐speaking child provided

evidence for a universal bias towards a trochaic pattern. The child under observation

producedtargettrochaicformsearlierthantargetiambicformsandthispatternwaskeptfor

alongperiod,asillustratedintherenditionsin(80).

(80) EarlytrochaictendencyinHebrew(Adam&Bat‐El,2009:264)89:

Orthogr. Gloss Output Agesáfta 'gradma' tátatúki 'parrot' kúkitapúax 'apple' púax

/‐SW/

banana 'banana' nánakapít 'spoon' tikipopotám 'hyppo' máta/tatá

1;03.14‐1;04.24

bakbúk 'bottle' búk/bakbúk 1;05.04‐1;05.08

/‐WS/

kivsá 'sheep' sa/kísa/kisá 1;05.15‐1;05.29

Tzakosta(2004)investigatedtheacquisitionofstressby11Greekchildrenaged1;7‐

3;5,inspontaneouscontext.InGreek,atrochaic,quantity‐insensitivelanguage(Drachman&

Malikouti‐Drachman,1999),childrendidnotprefertrochaictoiambicstresspatternsasthey

alternatebetweenbothinearlywordproductiondata(Tzakosta,2004).Earlytruncationof

both /SW/ and /WS/ forms provide evidence, both for an early monosyllabic word and

89Nophonetictranscriptionisprovided.

72

against a trochaic tendency. In (81)we present some renditions of Greek children, where

truncationof/SW/and/WS/tomonosyllableareattested.

(81) NeutraltendencyinGreek(Tzakosta,2004b:10390):

Gloss Target Output Chid,Age'fig' /»siko/ [ko] B.,1;10'those' /a»fta/ [ta] B.,1;11'toplay' /»pezi/ [pe]'carriage' /ka»rotsi/ [jo]

F.,1;11

'from' /a»po/ [po] Ma.,2;8.7

InSesothoandCatalan,bothtrochaic languages‐thoughSesothoisa languagewith

vowel lengthening in the penultimate syllable of a phrase andCatalan has an approximate

amountoftrocheesandiambs(trocheesstillbeingmorefrequent)‐childrenseemtomirror

thetendencyofthetargetlanguageandarefaithfultotrocheessincethebeginningofword

production(Demuth,1996aandPrieto,2006,respectively).

In(82)weillustratethetrochaictendencyfoundinSesotho.

(82) EarlytrochaictendencyinSesotho(Demuth,1996a:17391):

Orthogr. Gloss Outputntate 'father' tatema­simba 'chips' timpachelete 'money' teete

AccordingtoPrieto(2006),inafirststageofprosodicdevelopment,Catalanchildren

mainlyproduceCVCmonosyllables,thoughCVtargetscouldbeproducedfaithfullyaswell92.

After that stage, children's productions were constrained by a maximal disyllabic foot:

trochaicwordsareproducedtarget‐like,whereasiambicwordsaremostlytruncatedto[S],

asshownin(83).

90InstancesweretakenfromTzakosta(2004b).Noorthographictranscriptionisprovided.91 Instances in Demuth (1996a) were taken from Connelly (1984:73‐74). Though no stress mark is provided,Demuth (1996a:173) states, below the instances, that "[p]roductions in the one‐word stage in Sesotho canthereforeberepresentedbyastrong‐weaktrochaicfoot,justlikethatshownforEnglishandDutch".92Cf.instancesin(60),intheprevioussection.

73

(83) EarlytrochaictendencyinCatalanprosodicacquisition(Prieto,2006:246):

Orthogr. Gloss Target Output Child,Ageglobus balloon /»glçBus/ [»BçBu] Lluís,1;0.5papa 'daddy' /»pap´/ [»pHapHå] Lluís,1;1.10hola 'hello' /»çl´/ [»o˘wå] Lluís,1;2.15

/SW/

terra 'earth' /»tEl´/ [»tElå] Lluís,1;6.7tractor 'tractor' /tR´k»to/ [»tHo] Lluís,1;3.20Joan 'name' /Zu»an/ [»an] Lluís,1;6.7

/WS/

sisplau 'please' /sis»plaw/ [»paw] Ot,1;6.24

In iambic languages such as K’iché or French93, children display variable behavior.

French children tend tomirror the iambic tendency of the target language or they do not

favor any stress patterns at all (Demuth, 1996a; Vihman, DePaolis, & Davis, 1998; Braud,

2003;Demuth&Johnson,2003;Rose&Champdoizeau,2008).

Examining the productions of two children acquiring Quebec French, Rose (2000)

found that CV and CV'CV words are preferably produced in the early stages of word

production. As shown in section 2.1.1., Rose (2000) found that, at an initial stage, the two

children observed could produce CV words for CV targets (84a.). Target CVC words are

mostly realized as CV (84b.) and reduplication is frequent in disyllabic targets in both

children,althoughtheattemptedwordsmayalreadybereduplicatedforms(84c.).In(84),we

showaninitialtendencyformonosyllabicforms(CVandCVC)andtheco‐occurrenceofthese

formswithreduplications.

(84) EarlymonosyllabicformsinQuebecFrench(Rose,2000:99,101,113):

Orthogr. Gloss Target Output Child,AgeGuy 'name' /gi/ [gi] Clara,1;03.07l'eau 'thewater' /lo/ [lç] Clara,1;04.07

a./(C)V/

non 'no' /nç‚/ [na] Théo,1;11.10livre 'book' /liv/ [ji] Clara,1;04.14pomme 'apple' /pçm/ [bç˘] Clara,1;06.22pique '(it)pricks' /pIk/ [pI] Théo,2;01.19

b./(C)VC/

voir '(to)see' /vwa“/ [va] Théo,2;02.16coucou 'pick‐a‐boo' /kuku/ [gu»gu]pain 'bread' /bobo/ [bo»bo] Théo,1;10.27

crayon 'pencil' /kXEjç)/ [ke»jç] Clara1;07.27

c./CVCV/

poisson 'fish' /pwasç)/ [pç»sç] Théo,2;04.28

93 It isworthreminding thatFrenchdoesnothavean iambic footbutrather ithas fixedphrase‐finalaccent,asprominentsyllablesarealwaysattheendofaphonologicalphrase(Delattre,1965;Dell,1984).

74

Thedatashowedabovesuggestthatinthebeginning,bothCVandCV'CVwordsare

possible in the early speech of Quebec French‐speaking children.The co‐occurrence of CV

andCVCVpatternssuggestthat,inFrench,abinaryfoot,orthefootitself,maynotbeunder

children's attention, at the onset of prosodic acquisition. The data from French in Rose

(2000)werelaterconfirmedinBraud(2003)andDemuth&Johnson(2003).

Braud (2003) clearly stands against an iambic foot in the acquisition of French, as

monosyllables and disyllables are produced faithfully from the beginning, and trisyllables

mayhaveahighrateofsyllablepreservationaswell.Inmostcases,thedetermineriskeptin

thechildren'sproduction,likeinunescargot'asnail'/E‚nEskargo/producedas[E‚kago],orun

hélicopter 'anhelicopter' /E‚nelikçptEr/producedas [E‚nekçktEr]. The author argues for the

processingof anearlyprosodic template larger than the foot, or even theword, inFrench.

Theearlyreduplications,noticeablenotonlyindisyllablesbutalsoinpolisyllables,werenot

theresultofaninternalorganizationwithinthefoot,but, instead,thefulfillmentofa larger

prosodic template. The results found in Braud (2003) suggest that prominence was not

assignedonthebasisofthefoot,butratheronthebasisofalargerprosodicconstituent,such

asthephonologicalphrase,asfurthermoreoccursinadultFrench.

Demuth&Johnson(2003)alsofoundthat, instageI,theFrenchchildobservedalso

producedCVorreduplicatedCVCVwords(CV1CV1),forCVandCVCVtargets,respectively.

In (85), we recall the instances presented in section 2.1.1., showing the possible

occurrenceofmonosyllabic,alongwithreduplicatedformsintheacquisitionofFrench.

(85) Early tendency for monosyllabic (and reduplicated) forms in French (Demuth &

Johnson,2003:221,222):

Orthogr. Gloss Target Output Agemerci 'thankyou' /mE{si/ [Si]/[si]pupée 'doll' /pupe/ [pepe]tout 'all' /tu/ [tu]

1;3

pain 'bread' /pE‚/ [pa]là 'there' /la/ [na]chapeau 'hat' /Sapo/ [popo]pelle 'shovel' /pEl/ [pepe]canne 'stick' /kan/ [tata]

1;4

chat 'cat' /Sa/ [a]oeuf 'egg' /øf/ [toto] 1;5

75

At this stage, the child reduplicated CVC targets (/CVC/ ‐> [CV1CV1]) and truncated

trisyllablesinto[CVCV],asshownin(86).

(86) TendencyforreduplicationinFrench(Demuth&Jonhson,2003:219,221):

Orthogr. Gloss Target Output Agepoupée 'doll' /pupe/ [pepe] 1;3pelle 'shovel' /pEl/ [pepe] 1;4domino 'domino' /domino/ [çjç] 1;4balle 'ball' /bal/ [baba]porte 'door' /pç{t/ [pçpç]omnibus 'bus' /çmnibys/ [byby]

1;5

sausisson 'sausage' /sosisç‚/ [tçti] 1;6

AsRose(2000)andBraud's (2003), the findings fromDemuthand Johnson(2003),

namelytheoccurrenceof(C)V‐(C)VCValternatingforms,suggestthatbinaryfeetarenotan

obligatoryrequirementinFrenchprosodicacquisitionandthatstressmightnotbeassigned

on thebasisofa (binary) foot.Oneaspect that isdistinctiveofFrench,and isalso found in

Spanish (Demuth, 2001b) and Portuguese (Baia, 2008; Santos, 1995, 2001, 2007; Stoel‐

Gammon, 1976), is the one concerning reduplicatedwords.Authors have referred that the

early speech of Portuguese children is highly reduplicated and that it was used as an

empiricalargumenttoclaimfortheprocessingofaprosodicstructurelargerthanasyllable

orevenafoot(Santos,2001).Wewillcomebacktothis issuewhenPortugueseacquisition

dataareatstake94.

Thereviewmadethusfarenablesustoobservethatthereisagreatvariabilityinthe

earlystagesofwordproductionconcerningwordshapeandwordstress.Thepropertiesof

the target language, the children's ages that are taken into account and the type of data

considered, may in fact lead to wavering, conflicting results and may imply too many

irresolute paths. In Table 2, we will summarize the results for the languages reviewed,

concerningtheshapeofearlywordsandstresspatterns.

94Cf.section5.2.,furtherinthisdissertation.

76

Table2.Shapeofearlywords­cross­linguisticinformation

StudyLanguageRhythmintheTargetQSAgeProductionDevelopmentalProsodicTendency

Fikkert(1994)Demuth(1996)DutchTrochaicY1;0‐2;9Spontaneous

(12children)σ>SW]Σ>ω

Klein(1984)EnglishTrochaicY2;0SpontaneousExperimental(1child)

Neutral

Tzakosta(2004)GreekTrochaicN1;7‐3;5Spontaneousσ>Neutral

Hochberg(1988)1;7‐2;2Spontaneousandimitated(4children)

Neutral

Demuth(2001)

SpanishTrochaicY(nounsonly)

1;8‐2;3Spontaneous(1child)

σ>SW]Σ

Bat­El(2008)HebrewTrochaicN1;02‐1;07Spontaneous(1child)SW]Σ

Prieto(2006)CatalanTrochaicN1;1‐2;11Spontaneous(4children)

σ>SW]Σ

Demuth(1996)SesothoPhrasalstress(penultimatelengthening‐trochaic)

Y(penultimatelengthening

phrasefinally)

NotavailableNotavailableSW]Σ

Rose(2000)1;0‐4;0Spontaneous(2children)

Braud(2003)1;9‐5;0SpontaneousExperimental

Demuth(2003)1;1‐1;8Spontaneous(1child)

Nofoot:σ,CV.CV,CV.CV.CV

Vihmanetal.(1998)

FrenchPhrasalstress(final)N

>25wpSpontaneous(5children)WS]φ

Demuth(1996)K'iche'IambicYNotavailableNotavailableσ

77

InTable2,weobservethatwhenalanguageistrochaic,eitheraneutraloratrochaic

tendency is found. Inmany languages,anearlyprocessingofamonosyllable is frequent. In

French, a language with phrase‐final accent, no evidence for an iambic foot was found.

Instead,ahighvariabilitybetweenmonosyllabicanddisyllabicformswasnoticed.Inclearly

iambic systems, such as K'iche', a heavymonosyllable (CV: or CVC) was earlier produced.

From the table above,we observe that iambs are disfavored in early acquisition data. The

table showed above demonstrates that early prosodic acquisition is often an echo of

language‐specificproperties.

2.3.Thenatureandroleoffillersoundsinearlyprosodicacquisition

Thenatureandroleoffillersoundsinearlyacquisitionhasbeenlongdiscussedinthe

literature.Thequestionsmentionedare:whatisthelinguisticstatusoffillersyllables?What

canfillersoundstellusabouttheearlyprocessingofwordshapeandstresspatterns?

Most of the literature is unanimous in considering that fillersmakepart of a set of

characteristicsofthespeechofyoungchildren(e.g.,Bottari,Cipriani&Chilosi,1994;Demuth,

2001b; Peters, 1993, 1995, 1997, 2001a,b,c; Peters & Menn, 1993; Veneziano & Sinclair,

2000).Childrenacquiringmanylanguagesproducefillersoundsandsomechildrenusethem

morethanothers(Simonsen,2001).Theproductionof fillersoundsseemstobeveryoften

related to rhythmic, intonational and prosodic aspects of acquisition, as childrenmay use

themtoattainthepropertiesoftheadultspeechby,forinstance,fulfillingagivenrhythmicor

prosodictemplatethatismotivatedbyprosodicconstraintsofthetargetlanguage.Theearly

prosodic representation of both fillers and reduplicated words is, however, subject to

discussion.Inthefollowingparagraphswewillpresentareviewonthesetwoaspectsofearly

speechproduction.

Fillersoundsarecommonlyobservedintheearlyspeechofchildrenanditiscommon

to interpret them as the result of an interaction between phonology, morphosyntax and

pragmatics. The behavior of fillers varies within children acquiring the same language.

Literature reports have pointed that some children have a more 'gestaltist' approach,

following an acquisition path in production more consistent with whole phrases (thus

producingmore'filler'elements),whereasothermayhaveamore'analytic'perspectiveand

producemoreisolatedsyllables(thusproducingless'filler'elements)(Simonsen,2001).

All authors, however, agree that filler syllables result from the interaction between

phonologyandmorphosyntax,inastagewhenthedifferentmodulesofgrammararenotyet

78

completelymastered. The phonological status of these 'unglossable syllables'95 has been a

matterofdebateintheliteratureonphonologicalandmorphosyntacticacquisition(Bottari,

Cipriani&Chilosi,1994;Demuth,2001a,b,2007;Demuth&Tremblay,2008;Gerken,Landau

&Remez,1990;Gerken&McIntosh,1993;Lleó,1997,2001a,2001b;Lleó&Demuth,1999;

Peters&Menn,1993;Peters,1997,2001a,b,c;Veneziano&Sinclair,2000;Veneziano,1997,

2001).Specifically,theroleoffillersyllablesinprenominalpositionhasbeencloselyrelated

to the early awareness and production of determiners, although their co‐occurrence with

verbsandadverbsmightbeproblematicunderthisapproach96.

In an early study on perception and production of function morphemes in young

children, Gerken, Landau & Remez (1990) demonstrated that 2‐year‐olds are able to

distinguish between grammatical function morphemes and non‐function morphemes in

perception.Gerken&McIntosh(1993)complementedthepreviousanalysisshowingthat2‐

year‐olds are able to discriminate functionmorphemes such as determiners from pseudo‐

auxiliaries, thoughtheyarenotabletoproducethemtarget‐like.Theseresultssuggest that

youngchildrenprocessgrammaticalfunctionmorphemes,eveniftheyarenotabletorealize

themphonetically. Following this lineof research, awide setof studies supported the idea

thatproduction limitationsconstrainchildrenfromusinggrammatical functionmorphemes

inspeech,attheearlystages(Bernhardt&Stemberger,1998;Demuth,2001b,2007;Gerken,

1996;Gennari&Demuth,1997;Lleó&Demuth,1999).Despitethefactthatchildrenarenot

able to produce grammatical morphemes target‐like from the beginning, the use of filler

soundsisobservedinmanylanguages,andtherelationshipbetweentheproductionofthese

fillersoundsandtheirevolutionintogrammaticalmorphemeswasoftensuggested.

Oneofthemainproblemsresearchersworkingwithdatafromchildren'searlyspeech

have to deal with regards the criteria that distinguish filler from non‐filler sounds.

Frequently, the turningpointsbetween theproductionof fillersandproto‐morphemesand

between the production of proto‐morphemes and morphemes (determiners, pronouns,

modals, auxiliaries, etc.) are not easy to tackle. Veneziano (2001) and Peters (2001a,b)

established a group of phonological andmorphological features, as well as functional and

productionproperties,allowingforanidentificationofthedifferentstageschildrenmightgo

through in the production of filler syllables. According to the authors, threemoments are

distinguishable in the evolution of filler sounds in the speech of young children: a

premorphologicalstage,aprotomorphologicalstageand,finally,amorphologicalstage.

InTable3,wepresentPeter's(2001b)summaryonthestatusandfunctionoffillers

95Theexpression'unglossablesyllables'istakenfromPeters(2001a,b).96Theoccurrenceof fillersoundsbeforeverbsandadverbswasobserved inPortuguese(Costa&Freitas,2001andMiguel&Freitas,1998).Thisissuewillbediscussedinsection2.5.2..

79

incross‐linguisticacquisitiondata:

Phonological Functional Morphological Production

Premorphology

‐Fullsyllable;‐ Limited setofvowels;‐No/fewconsonants;

‐ Prosodicextender;‐ Not lexicallyselected; ‐‐‐‐‐‐‐‐‐‐‐‐‐

Phonologicalextension ofthe item it isattachedto.

Protomorphology

Somematchtosetofmorphemesinthisposition;

‐Rhythmicplaceholder;‐Lexicallyselective;‐Idiosyncratic;

‐Morphologicalplaceholder;‐Maybeamalgamated;

Fullmorphemes

Matchthetargetwithinarticulatoryability;

Approachingadult;

‐Splitintosubclasses;‐Systematic;

Becomingautomatized

Table3.Characteristicsoffillersatdifferentstagesofdevelopment(fromPeters,2001b:232)

In a premorphological stage, filler syllables produced are normally full syllables

(mostly vowels, though consonantsmay appear). In (87)we present some instances of an

English‐speakingchild,observedinPeter'sstudies(e.g.,Peters,1993,1995,2001a,b;Peters

&Menn,1993).

Inthefollowingexamples,'D'standsfor'Daddy','S'standsfor'Seth'and'^'standsfor

mainstress(agebetweenbrackets).

(87) Renditionsoffillersyllablesinthepremorphologicalstage97:

S:m^p(l)ay?m^p(l)ay?(1;6)

At this stage, fillers' function is to extend aword or phrase and they do not select

specific lexical items (theymayappearwithnouns, verbs, adverbs, etc.). Fillers soundsare

unglossableatthispoint.

Ataprotomorphologicalstage,thephonologicalformoffillersstartstoresemblethe

form of the morphemes that will occupy that position. They mark a rhythmic and

97 The examples were taken from the CHILDES website(http://childes.psy.cmu.edu/topics/fillers/characteristics.html,http://childes.psy.cmu.edu/topics/fillers/pre/pre.html andhttp://childes.psy.cmu.edu/topics/fillers/proto/proto.html – downloaded on October 31st, 2006). Peters usesCHILDES notations (for further information on the notation criteria and database information cf.http://childes.psy.cmu.edu/manuals/).

80

morphologicalpositionwithintheutteranceandstartbeinglexicallyselective.

(88) Seth'srenditionsoffillersyllablesintheprotomorphologicalstage98:

a.Proto‐wannawithproto‐article:

S:ngcwoseuh^door?(1;10)

S:wuhbrutshuh^teef?(1;11)

b.Proto‐subject:

S:uh^emptydere.(2;1)

D:it'semptythere.that'sright.

Finally, at the stage of full morphemes, a systematic match between the child's

utterances and the adult target is attained and children produce articles, modals and

auxiliariesinatarget‐likemanner.

The'identificationproblem'iscloselyrelatedtoanotherproblem,whichisthestatus

androleoffillersoundsinemergentgrammars.Asstatedabove,fillersaregenerallyassumed

tobe theprecursorsof laterphonologicalandmorphosyntactic structures suchas function

words, normally reflecting children's pathway from phonology to morphology. Fillers can

thus be interpreted as bootstrapping elements to the awareness of other grammatical

modules,i.e.,fillersoundsevolvetomorphosyntacticelements.

Many authors have suggested that the use of filler sounds by children might be a

strategytodrivetheirspeechtofeet,wordsorphrasesas,inmanycases,fillersoundsevolve

to grammatical function morphemes, such as determiners, auxiliaries and verb or noun

inflectionmarks (Demuth&Tremblay,2008;Gennari&Demuth,1997;Gerken,1996;Lleó,

2001;Lleó&Demuth,1999;Veneziano,2001).Reportsintheliteratureindicatethat,indeed,

theprosodicstructuremaylicensegrammaticalfunctionmorphemes.

AccordingtoGerken(1996),inaniambicEnglishphraselike'thepig'orinaniambic

word like 'giraffe', the initial weak syllable is omitted. However, in a sentence like 'Tom

pushedthezebra'thearticleisphoneticallypresent,asitislocatedinsideatrochaicfoot.

98Idem.

81

(89) Productionof 'thepig' and 'giraffe' (iambs)vs. '(Tom)pushed thezebra', (Gerken,

1996:687,690,694):

(the)(gi)

Gerken (1996) suggests that English‐speaking children only produce grammatical

morphemesinunmarkedprosodicstructuresoftheirlanguage,suchasthetrochaicfoot,as

illustrated in the above‐mentioned examples. Comparing the production of determiners in

GermanandSpanishmonolingualchildren,Lleó&Demuth(1999),Lleó(2001)foundsimilar

results.(Proto‐)determinersandmultisyllabicwordswereproducedearlierinSpanishthan

inGerman(andotherGermaniclanguages,likeEnglishandDutch).Accordingtotheauthors,

thisisbecausetheprosodicstructureofSpanishprovidestronger(morefrequent)evidence

formultisyllabicwordsandmultisyllabicphrases.

(90) Spanishexamplesofdeterminer+disyllabicnoun(Lleó&Demuth,1999:41399):

Orthogr. Gloss Output Child,Agepala 'shovel' [»ha»bEl´/]mamá 'mummy' [´»m´mEH]cubo 'bucket' [hU»guw´]

María,1;4.21

peine 'comb' [/a»pa˘lE˘h]globo 'balloon' [»åwewçh]papa 'porridge' [/Um»båbåH]

María,1;6.3

vaca 'cow' [haveva˘]boca 'mouth' [avava]

Miguel,1;4.5

vaca 'cow' [U˘n»vQva]agua 'water' [hE»a˘va]

Miguel,1;5.1

According to Lleó & Demuth (1999), after Spanish children are able to produce a

disyllabic trochee, they soon start producing /WSW/ utterances, where a trochaic foot

(/SW/)isprecededbyanunfootedsyllable.Theexplanationforthisdifferenceistwo‐folded:

99Noinformationaboutthetargetphoneticformisprovided.

SW

pushedthe

SW

zebra

S(W)pigraffe

82

on the one hand, determiners in Spanish are always prosodified as proclitics, and the set

proclitic+noun is equivalent to a trisyllablewhen it is followed by a disyllable; and on the

other hand, Spanish‐speaking children have in general more multisyllabic targets than

German‐speaking children, and multisyllabic words with penultimate stress are more

frequent in Spanish than in German (holding for a frequency perspective). In German, a

differentcaseoccurrs,asdeterminersareeitherfullformsandarethemselvesafoot,orthey

arereducedandencliticizedtothepreviousProsodicWord.

(91) ArticlesinSpanish:formsandprosodification(Lleó&Demuth,1999:414):

PW[elF[»pero]] PW[laF[»pala]]

(92) Articles in German: forms and prosodification (Wiese, 1996:250, apud Lleó &

Demuth,1999:415):

PW[F[der]]PW[F[Mann]] PW[F[die]]PW[F[Klappe]]

PPh[PW[F[noch]]PW[F[ein]]PW[F[Kipper]]]

PPh[PW[F[noch]n `]]PW[F[kipper]]]

Thus,inSpanishprosodicacquisition,theearlieremergenceofgrammaticalfunction

morphemessuchasdeterminersmightbedueto theadult languageprosodizationof these

morphemes. The production of determiners or proto‐determiners within prosodically

licensedpositionswasfurtherconfirmedbyGennari&Demuth(1997)fortheacquisitionof

Spanish. The authors suggested that Spanish‐speaking children map and earlier produce

determinerswithintheCliticGroupandthePhonologicalPhrase,astheyareabletoproduce

them before disyllables (e.g., la cása 'the house' ‐ /W SW/),when trisyllables (e.g., konéxo

'rabbit'‐/WSW/)werenotproducedtarget‐like100.

Theauthorsclaimthatchildrentendtoproducethesyllablesinsidethetrochaicfoot

(producingbothsyllables in/SW/andomitting the first syllable in/WSW/words),adding

thatdeterminersandothercliticwordsareproducedbecausetheyareadjoinedtotheClitic

Grouplevel,andnottotheProsodicWordlevel.Thesefindingsprovideevidencefor:

100Thesametendencywas foundforSWSWtargets(includingProsodicWords likeescaléra 'scale'andphraseslikeàlacasa'tothehouse'),onlywithhigheromissionpercentagesinweaksyllablesthanthoseobservedinWSWtargets(e.g.konéxo'rabbit'andlacasa'thehouse').

83

(i) Theearlyproductionofdeterminersandproto‐determinerswithinprosodically

licensedpositions(namelywithinlevelsoftheprosodichierarchythatareabove

theProsodicWordlevel,liketheCliticGroupandthePhonologicalPhrase);

(ii) AnearlysensitivitytotheconstituentsofprosodicstructureabovetheProsodic

Word.Also,theseresultsindicatethatproto‐determinersmaynotbeproducedto

matchafoottemplate,butratherhigherprosodicdomains.

Withrespectto(ii),specifically,Veneziano(2001)arguesthattheuseoffillersounds

might be a strategy used to match the predominant phrase‐final accent found in French,

which is a languagewhere footing does not occur. French childrenmay produce fillers to

matchaprosodictemplatethatislargerthanthefoot.Demuth&Tremblay(2008)claimthat

French determiners first appear in prosodically licensed positions, namely within a foot

before monosyllables, or outside the foot (at the prosodic word level) in larger words.

Initially,childrenproducethembeforeamonosyllablebutnotbeforedisyllables(93b.);only

later,French‐speakingchildrenstartproducingdeterminersbeforedi‐(93c.)andtrisyllabic

words(93d.).

This order of emergence is related to different stages of prosodic development, as

presentedin(93).

(93) Prosodic development of French determiners (from Demuth & Tremblay,

2008:119)101:

a.Noclitic b.Foot‐internalclitic c.PW‐internalclitic d.Freeclitic

PP PP PP PP g g g 3 PW PW PW σPW g g 2 det 2 Ft Ft σFt (σ)Ft

2 2 det 2 2 (σ)σ (σ)σ (σ)σ (σ)σ

lex detlex lex lex

101 'Det'standsfor 'determiner'; 'lex'standsfor 'lexical item'(fromDemuth&Tremblay,2008).It isworthwhilesayingthatDemuth&Tremblay'sapproachtoprosodyinFrenchacquisitioncontradictsBraud's(2003)analysis,asthelatterclaimsthatnofeetareprocessedintheearlyspeechofFrench‐speakingchildren,butlargerprosodicconstituentsinstead(mainlyphrases).

84

Thestudiesmentionedabove(Demuth,2001b;Demuth,2007;Demuth&Tremblay,

2008;Gennari&Demuth,1997;Lleó&Demuth,1999;Roark&Demuth,2000;Peters,2001,

Simonsen, 2001, Veneziano, 2001) suggest thatthe production of fillers is not necessarily

related to feet constraints, but it can also be related to the processing of larger prosodic

constituents, such as the prosodic word, the clitic group or the phonological phrase. The

studies presented also indicate that the production of filler sounds in early language

acquisitionislanguage‐dependentandlicensedbytheprosodicstructure.

2.4.Thenatureandroleofreduplicationsinearlyprosodicacquisition

Like fillers, reduplicative speech is often considered as a production strategy,

occurring between pre‐speech and speech. Reduplications without referential value have

beencharacterizedasoneofthemainfeaturesofcanonicalbabbling(Vihman,1996:110‐11).

At thisstage,reduplicationsmaybeunderstoodasameanbywhichchildrenpractice their

articulation to later speech and by which children start training the 'language game'

(Ferguson&Macken,1983;MacNeilage&Davis,1993).

Reduplicationswithreferentialvaluewereinterpretedashavingnumerouslinguistic

functions: pragmatic, semantic, phonological and morphosyntactic. Leroy & Morgenstern

(2005:475) mention that "It [reduplication] reflects the child's acquisition of pragmatic

functionsaswellasthechild’sappropriationofthelanguagesystem".Again,somechildren,

more than others, may use reduplicative productions in early phonological acquisition,

thoughmanyauthorshaveseenreduplicationasauniversalpropertyoflanguageacquisition

(Moskowitz,1973;Fee&Ingram,1982).

Reduplications can be defined as a repetition of a syllable creating a pattern that

shares consonantal and/orvocalic features ([CV1CV1]–e.g., 'another'producedas [¯e˘¯e] ‐

Waterson,1971,apudKlein,2005:71‐or'candy'producedas[¯a¯i]‐Ingram,1979:145).It

can either be the extension of a monosyllable (e.g., 'cat' produced as [kQkQ]), or it can

duplicate a syllable in adult disyllables or multisyllabic words (e.g., 'water' produced as

[wçwç]). Though reduplications are present in children's early speech, the subject has not

beenwidelydebatedintheliteratureonchildlanguageandevenlesssointheliteratureon

earlyprosodicrepresentations.

Reduplications have been studied by a few authors, mainly focusing on the

relationship between the early production of reduplications and the later production of

monosyllabicandmultisyllabicwordsorclosedsyllablesandmulti‐wordutterances(Fee&

85

Ingram, 1982; Ferguson, 1983; Schwartz, Leonard, Wilcox & Folger, 1980; Schwartz &

Leonard,1983;Veneziano,Sinclair&Berthoud,1990;Klein,2005).

Lleó(1990)relatedtheproductionofreduplicationswiththeearlyappropriationof

the phonological system of the language children are acquiring and, specifically, with the

constructionoftheirlanguage'sprosodicstructure,whichfeet,wordsorphrasesarepartof.

According to the author, reduplications are the result of children's 'ability to create

phonological rules' and the capacity to "work out certain phonological processes" (Lleó,

1990:274).Reduplicationisnotnecessarilyrelatedtoreducedsegmentalinventoriesandthe

need to replace not acquired segments, as the child observed in the study increased

reduplicationsashomonymydecreased.Instead,theauthorarguesthatthepatternsattested

inthechildobservedindicatethatthewordisthephonologicalunitunderprocessinginthe

earlystagesofphonologicalacquisition.Thechildstartstoproducetrisyllablesbyrepeating

onesyllableofthetargetword,asifthechildisattemptingtoproducethewholewordwith

the syllabic material available in his/her system. These results were later reported by

Vihman (1996) as empirical evidence for a 'whole‐word' phonology during early language

development.

In French acquisition, reduplications are frequent102 (Braud, 2003; Rose, 2000;

Wauquier‐Gravelines,2003).Untilratherlateindevelopment,Frenchspeaking‐childrenuse

both truncation and reduplication to fulfill a phonological template that is larger than a

syllableora foot.Reduplicationsmaybeused inproducingmono‐, tri‐andpolisyllables,as

theinstancesbelowindicate.

(94) Frenchreduplicationsina3year‐oldchild‐targetmonosyllablesandtrisyllables103

(Wauquier‐Gravelines,2003:10104):

Orthogr. Gloss Outputours 'bear' [E‚nunu“s]/σ/os 'bone' [E‚nonos]arrasoir 'watering‐can' [E‚{o{çzwa“]/[E‚{ea{çzwa“]coccinelle 'lady‐bird' [ynkokosinEl]aspirateur 'vacuumcleaner' [E‚{a{astatø“]

/σσσ/

hélikoptère 'helicopter' [E‚nenikçtE“]

102Seealsoexamples(84)‐(86),inthischapter.103ThisstudyisconductedusingdatafromClara,theQuebecFrench‐speakingchildfromRose(2000),andClaire,astandardFrench‐speakingchild.TherenditionspresentedherearefromClaire.104Thepagenumberreferredtointhisreferenceisrelativetothemanuscriptversionofthepaper,downloadedfromhttp://pagesperso­orange.fr/Sophie.Wauquier/download/Publications.htmonMay3rd,2009.

86

Disyllables are rarely reduplicated and reduplications tend to occur in large target

words.SinceFrenchisalanguagewithphrase‐finalaccentandnoevidenceforfooting,Braud

(2003) and Wauquier‐Gravelines (2003) suggest that morphological templates, in which

determinersandpartofalexicalwordareproducedandareunderthechildren'sattentionin

earlyphonologicalacquisition.Therefore,theproductionofreduplicationsisastrategyused

tomatch, not a prosodic constituent (like a syllable or a foot), but rather amorphological

template‐alexicalunit‐,inwhichtheearlydeterminerispartofthelexicalunit,andnotan

independentmorphologicalunit.

Insummary, fillersandreduplicationsappeartobefrequent inaperiodofprosodic

organization,whenfeetmightbe‐butarenotnecessarily‐underprocessing.Theyappearas

strategies used by the children to appropriate the system they are acquiring, bymatching

feet, word, phrasal or even morphological templates. These strategies might be used, not

becauseoffeetsizeandshapeconstraints,butratherduetowordorphrasesizeandshape

requirements.

2.5.TheacquisitionofwordstressinPortuguese

Inthissectionwewillpresentthedescriptionsandanalysesonstressacquisitionin

Portuguese, that have been purported to this date. We will review the studies where the

acquisitionofwordshapesandstresspatternshasbeeninvestigated,bothinBPandEP,with

special reference to the studies on the production of filler sounds and reduplications in

Portuguese‐speaking children. Additionally, a review of the acoustic properties of word

stressandstress‐relatedaspectsduringacquisitionwillbecarriedout.

2.5.1.Theacquisitionofwordshapesandstresspatterns

InBP,severalstudieswereconductedinordertoinvestigatetheacquisitionofword

stressandstresspatterns(Baia,2006,2008a,b,c,d;Bonilha,2005;Rapp,1994;Santos,2001,

2005, 2007; Stoel‐Gammon, 1976). In EP, a preliminary research on the acquisition of the

Prosodic Word, from a frequency perspective, was conducted, though the acquisition and

development of stress patterns has not been the focus of the analysis (Vigário, Freitas &

Frota,2006).However,thesesetsofstudieshavesomeincompatiblefindings.

Rapp (1994) firstproposed thatBrazilian childrenwoulddisplayan initial trochaic

tendency.Rapp'sconclusionsweredrawnfromacorpusofspeechdatafrom8childrenaged

from1;6 to2;0.This researchconsisted in collectionof experimentaldataandwas carried

87

outtoanalyzethetruncationpatternsinBPearlywords.

Interestingly,inmostresearchbasedonspontaneousdata,theshapeofearlywords

more frequently observed in Portuguese conforms to an iamb (Baia, 2006, 2008; Bonilha,

2005;Santos,2001,2003,2005,2007a,b;Stoel‐Gammon,1976).

In a work on the acquisition of primary word stress in BP, Santos (2001) mainly

focusedontheacquisitionof intonational,stress‐relatedaspects,basedonthespontaneous

speech productions of two Brazilian children. The author establishes four moments of

primarystressacquisition:

(i) In an early stage of word production, children are acquiring different

intonationalcontours105,appliedindifferentpragmaticcontexts.Itisnotpossible

to distinguish word and intonational prominence, as they are able to produce

utteranceslargerthanasyllableorafootinvariablemanners.Theproductionof

phonologicalconstituentslargerthanasyllableorafoot(namely,byadditionof

fillersyllables‐cf.instancesin(95))andtheobservationofdifferentintonational

contours lead the author to suggest that children are not dealing with word

stressatthisstage,butratherphrasalstress.

(95) Segmental and syllabicmaterial large thana syllableora foot,produced tomatch

severalintonationalcontours(Santos,2001:234,235):

Orthogr. Gloss Target Output Child,Agebanana 'banana' /ba»na‚na/ [m´ma»mana] R.,1;6.3aconteceucomagarrafa

'it happened withthebottle'

/kwaga»xafa/ [kotika»fafa] R.,1;6.6

ajanela 'thewindow' /aZa»nEla/ [kale»lEla] R.,1;6.6.fecha 'close' /»fESa/ [a»fEsa] R.,1;6.22aqui 'here' /a»ki/ [a»ki] T.,1;5.21passarinho 'birdie' /pasa»Ri‚¯u/ [paj»Si‚¯u] T.,1;6.3sapato 'shoe' /sa»patu/ [sa»patU] T.,1;7.23

(ii) After an initial stage (from approximately 1;8 to 2;3), children set one default

contour: (L)LH*(L)%. This contour mainly consists of a head (optionally

preceded by a syllable) + nucleus and a tail. The nucleus is the last stressed

syllable of the tone group. The head is associated to the first stressed syllable

occurring before the nucleus (and all the unstressed syllables that appear in

105SeeSantos(2001,2003,2005)forthevariedcontoursfoundinBPearlyacquisition.

88

between). The tail consists of all the post‐nuclear syllables. In this stage, the

Brazilian children observed fulfilled this contour with segmental and syllabic

material,suchastargetstressedsyllablesandfillersounds,asshownin(96).

(96) Fulfillingofthe(L)LH*(L)%contour:

(L) L H* (L) %‐>intonationalcontour

(σ) ∪ (σ) ‐>iambicstructure

[a »pej] põe'put'

[be »ke dU] brinquedo'toy'

[/´ mi »ni na] menina'girl'

At this point, word stress is not yet being computed, and phrasal accent (right‐

headed)isstillatstake.In(97)wepresentsomerenditionsofthetwochildrenobservedby

Santos(2001),whereseveralwordshapesareattempted, inordertofulfill the(L)LH*(L)%

intonationalcontour.

(97) Fulfillingthe(L)LH*(L)%intonationalcontour(Santos,2001:243,244,245):

Orthogr. Gloss Target Output Child,Agediferente 'different' /dife»Re‚j‚tSi/ [fe»Re‚tSi] T.,1;9.24pau 'stick' /»paw/ [a»paw] T.,1;11.2bicho 'animal' /»biSu/ [/m»biso] T.,1;11.4música 'song' /»muzika/ [mu»zika] T.,2;0.21machucou 'hurt' /maSu»kow/ [mas»ko] T.,2;0.26menina 'girl' /me»ni‚na/ [/´mi»nina] T.,2;2.28palhacinho 'clowndim.' /pasa»Ri‚¯u/ [pa»si‚¯U] R.,1;8.25coelhinho 'bunny' /koe»¥i‚¯u/ [ka»li‚¯U] R.,1;8.25brinquedo 'toy' /bRi‚n»kedu/ [be»kedU] R.,1;10.0adivinha 'guess' /adZi»vi‚¯a/ [dZi»vi‚¯å] R.,1;11.12gravador 'keep‐recorder' /gRava»doR/ [gra»vadoR] R.,2;0.5

In the examples presented abovewe observe that children producemostly di‐ and

trisyllabic words, both final stressed and penultimate stressed, and fulfill the (L)LH*(L)%

intonational contour. Filler sounds are possible (e.g., [a»paw]), aswell as stress shift ‐ e.g.,

[mu»zika]).

89

(iii) Ina thirdphase (from2;3onwards), childrenmainlyproducedisyllabicwords.

Atthisstage,childrenproduceboth[SW]and[WS]words,providingevidencefor

abinaryphrasalconstituentbeingprocessed.Atthispoint,Brazilianchildrenuse

both types of words ([SW] and [WS]) to fulfill one fixed intonational contour:

LH*L%.Therefore,twodomainsinteractinstressassignment:thewordandthe

phrase. In (98), we schematize the mapping of Brazilian children's disyllabic

productionontotheLH*L%intonationalcontour.

(98) Disyllabic ([SW] and [WS]) words mapped onto a LH*L% intonational contour

(Santos,2005:81):

L H* L%‐>intonationalpattern

(w s) # ‐>iambicpattern

(s w)# ‐>trochaicpattern

Therepresentationin(98)accountsfortheproductionsin(99).

(99) Binaryphrasalconstituent‐[SW]and[WS](Santos,2001:249,253):

Orthogr. Gloss Target Output Child,Ageblocos 'notebooks' /»blçkus/ [»blçkUs] T.,2;3.4agora 'now' /a»gçRa/ [a»gç] T.,2;3.4verdura 'vegetables' /ver»dura/ [ver»du] R.,2;0.20rabo 'tail' /»xabu/ [»xabU] R.,2;2.2sabão 'soap' /sa»ba‚w‚/ [sa»ba‚w‚] R.,2;2.19

The examplespresented above show theproductionof both [SW] and [WS]words.

According to the author, these twoword shapes are used to fulfill the LH*L% intonational

contourfoundinathirdmomentoftheacquisitionofBrazilianPortuguese.

(iv) Finally, children learn that stress and accent have different principles. SWW

words (with extrametrical syllables) are produced target‐like. Thesewords do

not fit the LH*L% contour anymore and that provides evidence for children

having learnt the word stress algorithm in the language. In (100) we present

some examples of the production of words with an extrametrical syllable

(/SWW/)bythetwoBrazilianchildrenobservedinSantos(2001).

90

(100) Acquisitionofthestressalgorithm:extrametricality(Santos,2001:261):

Orthogr. Gloss Target Output Child,Agemúsica 'song' /»muzika/ [»muzika] T.,2;3.4plástico 'plastic' /»plastSiku/ [»plastSiku] T.,2;3.4Malévola 'Cruella' /ma»lEvola/ [ma»lEvalå] R.,2;6.19lógico 'logical' /»lçZiku/ [»lçZikU] R.,3;2.19

Santos'(2001)workprovidesavaluablecontributiontothediscussionontheearly

stress patterns found during Portuguese acquisition. It suggests an interchangeable focus

between lexical and intonational aspects and relates phrasal stress with the early iambic

tendency found. According to the author, the algorithm for word stress is only mastered

when/SWW/words,thatis,extrametricality,areacquired.Santos(2001),however,didnot

allowforarobustdiscussionontherelevanceofweightduringtheacquisitionofwordstress

in BP, as the children's productions were not conclusive. Though the children observed

ignored final heavy syllables, producing ['CV.CVC] words, ‐ and providing evidence for

weight‐insensitivity‐,simultaneously,theycouldalsoproduceoxytonicpatterns([CV.'CV]or

[CV.'CVC]) since theearly stagesofwordproduction ‐ supportingweight‐sensitivity. Stress

errors in the direction of the heavy syllablewere not observed either. The author further

arguesthatbothphonologicalandmorphologicalaspectsinteractintheacquisitionofword

stressinBP,sinceBrazilianchildrenareabletoproducedderivedwordswithcorrectstress

placement.

Apart from the developmental path described above, one main conclusion can be

drawn from Santos' proposal: it suggests thatword stress acquisition in BP proceeds top‐

down, that is, it starts from the upper‐levels of the prosodic hierarchy (such as the

intonationalphrase).

ContrarytoSantos(2001,2003,2005),Bonilha(2005)arguesforanearlyacquisition

ofwordstressinBP.TheBrazilianchildunderobservationdidnotproduceagreatamountof

filler sounds and larger utterances, providing evidence against the role of intonational

structureduringtheacquisitionofwordstress.TheresultsfoundinBonilha(2005)further

suggest that there is a neutral start in BP stress acquisition, though the data presented

indicateaslight tendency for iambs,namely through the tendency topreserve thestressed

and initial syllables ([SW] and [WS]). In the early stages of phonological acquisition,

monosyllables and disyllables prevail, and disyllables can be both [SW] and [WS]. As for

weight, Bonilha (2005) found evidence for weight‐sensitivity in the speech of the child

observed,since:

91

(i) Words with final stressed heavy syllable (/CV.'CVG/ words) were produced

correctlyfromthebeginning,asshownin(101).

(101) Correctproductionoftargetswithfinalheavysyllables(Bonilha,2005:311)106:

Orthogr. Gloss Output Child,Agemamãe 'mommy' [ma‚»ma‚j‚] G.,1;1.22papai 'daddy' [pa»paj] G.,1;5.20baton 'lipstick' [ba»to‚w] G.,1;6.3

The non‐production of final heavy syllables with Coda consonants (e.g., tractor

'tractor'/tRa»toR/)wasassociatedtothelateracquisitionofspecificsegmentsintheCoda.

(ii) Words with penultimate stress (/'CV.CV/), words with final stressed heavy

syllable(/CV.'CVC/)andwordswithfinalstressedlightsyllable(/CV.'CV/)were

acquired earlier than words with penultimate stress but with final heavy

syllables(/'CV.CVC/)andwordswithantepenultimatestress(/'CV.CV.CV/)107.

Theauthorarguesfortheunmarkedcharacterof/‐'CV.CV/,/CV.'CVG/and/CV.'CV/

wordsandthemarkedcharacterof/'CV.CV.CV/and/'CV.CVC/words,assuggestedbyBisol

(1992).

Inbothstudies(Santos,2001andBonilha,2005),thestrategyofstressshiftinheavy

syllablestolightsyllablesandviceversawasscarce.

Inlaterwork,Santos(2007)examinedtheearlystresspatternsintheearlyspeechof

BP‐speaking children, comparing nouns and verbs, as well as reduplicated and non‐

reduplicatedwords108.Regardingnouns,theresultsinSantos(2007:40)suggestthat:

(i) /SW/ tokens were mostly produced correctly from the beginning, though

truncationto[S]wasobservable,especiallyintheearlysessions;

(ii) reduplicated /WS/ tokens were mostly produced correctly from the

beginning;

(iii) non‐reduplicated /WS/ tokens were in much lesser amount than /WS/

106Nophonetictranscriptionforthetargetisprovided.107No renditionsof the child arepresented. For comparative results on theproductionof /'CV.CV/, /CV.'CVC/,/CV'CV/and/'CV.CVC/wordsintheBrazilianchildobserved,cf.Bonilha(2005:317).108 The author designates as 'familiar lexicon' the early lexicon produced by the children. From the data, it ispossible to see thatmost of thosewordswere reduplications (andmonosyllabic forms such aspai 'father' ‐ cf.Tables 14, 15 and 16 in Santos, 2007:50‐52). For that reason we overextend the concept and generallydenominateitasreduplications.

92

reduplicatedtokens;

(iv) non‐reduplicated/WS/nounswereeitherproducedcorrectlyortruncatedto

[S], though, due to the reduced number of tokens, the percentage values

presentedmightbemisleading;

(v) /SW/tokensweremorefrequentthannon‐reduplicative/WS/tokens;

(vi) reduplicative/WS/tokensweremorefrequentthan/SW/tokens;

(vii) monosyllablesweremostlyproducedcorrectlyfromthebeginning.

Inverbs,theresultsinSantos(2007:52‐59)showedthat:

(i) /SW/ tokens were mostly produced correctly from the beginning (though

someinstancesoftruncationcouldbefound);

(ii) /WS/ tokens were mostly produced correctly from the beginning (though

some instancesof truncationcouldbe found,but insmalleramounts than in

/SW/);

(iii) monosyllablesweremostlyproducedcorrectlyfromthebeginning.

Giventhehigherpercentageratesin/WS/andthetendencyfor[WS]reduplications,

theauthorarguesforanearlyiambicfootinBP.Theearlyiambictendencyattestedmightbe

due to the fact that the unstressedword‐final wordmarker in non‐verbs is extrametrical.

ThisproposalwasalreadysuggestedinSantos(2001)andsupportedLee's(1995)analysis.

Inverbs,andsincemostearlyverbformsarenon‐finite(Infinitives),Santos(2007)suggests

thatmorphosyntacticreasonsunderlietheearlyproductionof/WS/verbforms109.

AccordingtoSantos(2007),inordertoproducetrochees,Brazilianchildrenmust,on

the one hand, learn extrametricality110 in non‐verbs and, on the other hand, master

morphologicalcontrasts(inflectionandderivation).Intheirearlystages,thewordsproduced

byBrazilianchildrenmostlyconsist inreduplicatedwordswhichdonothavewordmarker

(e.g.,mamã‘mommy’,papá‘daddy’,bebé‘baby’),asopposedtowordssuchascas]+a‘house’,

gat]+o ‘cat’, sapat]+o 'shoe'). As soon as children learn extrametricality and master

morphological contrasts (gender, tense, number contrasts) they will be able to produce

trochaicwords. Examples fromSantos (2007), from the early and the later stages ofword

109FollowingLee(1995),accordingtowhomthedefaultfootinnon‐verbsisaniambandinverbsisatrochee,theBrazilian children's early productions should favor trochaic forms. However, that was not the case. Santos(2007:109,110)arguesthat"[t]houghthedefaultvalueofstressinverbsconformstoatrochee,morphosyntacticreasons force the production of iambs, rather than the one of trochees." (My translation from the original inPortuguese:"Assim,emboraovalordefaultdeacentuaçãodosverbossejaotroqueu,razõesmorfosintáticasfazemcomqueaproduçãodeiambossupereadostroqueus.")110Cf.Chapter1,section1.2.3..

93

production in BP, are shown in (102) and (103), respectively, showing the unmastery of

morphologicalcontrastinnounsandinverbs.

(102) Early stages of word production in BP, before the morphological contrasts are

mastered(Santos,2007:45,54):

Orthogr. Gloss Target Output Child,Agesapato 'shoe' /sa»pato/ [pa»pa] R.,1;3abre 'open' /»abRe/ [»abe],[»be] R.,1;4abre 'open' /»abRe/ [»abe],[a»bu],[»bu] R.,1;5sapato 'shoe' /sa»pato/ [pa»pati],[pa»paki],[pa»pa]

abre 'open' /»abRe/ [a»pe]

R.,1;6

Intheseexamplesweobservethatthetargetfinalwordmarker(e.g.,sapato'shoe')is

notproduced. Instead, the child reduplicates the stressedsyllableorproduces itwith filler

insertionat the leftedge,creatingan iambicpattern,or,better,producesatrocheewithout

producingthewordmarker(e.g.,[pa»pati],[pa»paki]).Wordstressinverbsisvariable.

Later,thewordmarkerisproduced,inflectionismastered,andwordsareproduced

target‐like.

(103) Later stages ofword production in BP, after gender contrast ismastered (Santos,

2007:42,45,54):

Orthogr. Gloss Target Output Child,Agebola 'ball' /»bçla/ [»bçla] R.,1;8abre 'open' /»abRe/ [»abe]bola 'ball' /»bçla/ [»bçla]

R.,1;9

sapato 'shoe' /sa»pato/ [sa»patu]abre 'open' /»abRe/ [»abe]

R.,1;11

sapato 'shoe' /sa»pato/ [sa»patu]abre 'open' /»abRe/ [»abe],[»abi]

R.,2;20

BasedonthedifferencesfoundinBPandDutchdata(Fikkert,1994),i.e.,basedonthe

factthattheproductionsofBP‐speakingchildrendisplayanearlyiambictendencyandDutch

children's productions display a trochaic one, Santos (2007) defends that the parameter

valueforthefootcannotbespecifiedintheearlygrammar.Theauthorarguesthat,contrary

toDutchchildren,whichdisplayanearly[SW]tendency,theearlydisyllabicproductionsof

BP‐speakingchildrenaremostly[WS],eithertheyaretheresultofepenthesisontheleftof

94

the circumscribed syllable or they are the product of a target‐like production of a

reduplicated iamb.Thedifferent tendencies found inBPandDutchprovideevidence foran

unspecified default parameter value for the foot head (that is, no 'left' or 'right' value is

presentattheonsetofwordproductioninBP).

Given the contradicting results from experimental and spontaneous data regarding

theearlystresspatternsinPortuguese(Rapp,1994;Santos,2001,2007;Bonilha,2005),and

aiming at defining the initial prosodic template in the acquisition of BP, Baia (2008a,b,c,d)

carriedoutasetofstudies,aimingatinvestigatingtheinitialwordshapeinBP.Experimental

datafrom42Brazilianchildrenaged1;5‐3;0werecollected,withacomplementarycorpusof

spontaneousspeechfrom1childaged1;5‐3;0.Theresultsfromspontaneousdataconfirmed

an initial iambic tendency, and a later tendency for trochaic words. The results from the

experimentalstudy,however,suggestedamoreneutralapproachtotheacquisitionofstress

patterns in BP‐speaking children, as a higher production rate for trochees was observed,

thoughthedifferencebetweentrocheesandiambsproducedtarget‐likewasnotstatistically

significant.Baia(2008a)demonstratedthattherewasadifferencebetweenspontaneousand

experimental data in the prosodic acquisition of Brazilian‐speaking children: experimental

data showed that iambs were more frequently truncated than trochees in an early stage.

Trochees,however,weremorefrequentlypronetostressshift.TargetWSWwordstendedto

be truncatedto [SW],whereas/WWS/tendedtobe truncatedto [WS].TargetSWWwords

were overwhelmingly produced as [SW]. Through the comparison between spontaneous

data, in which the whole lexicon is taken into account, and experimental data, where no

reduplications were elicited, Baia (2008b) observed a neutral start approach to prosodic

acquisition in BP. Indeed, spontaneous child data in Portuguese (EP and BP) has a high

frequency of reduplicated iambic targets like bebé/bebê 'baby' or mamã/mamãe111

'mommy').Disregardingthereduplications,Baia(2008a)doesnotfindaniambictendencyin

early prosodic acquisition in BP. Therefore, the author argues for an apparent iambic

tendency in BP prosodic acquisition. Experimental data did not corroborate spontaneous

data because the author did not incorporate reduplicated target words that are iambs as

stimuli in the experimental study. These findings seem to explain the contradicting results

between experimental studies (Rapp, 1994), which tend to indicate an early trochaic

tendency, and spontaneous data (namely, Santos, 2001, 2007), which tend to indicate an

earlyiambictendency.Insum,thetendencyfoundinspontaneousdataandthecomparison

betweenspontaneousandexperimentaldataindicatedthat:

111Bebé/bebêandmamã/mamãeregardthewords'baby'and'mommy',inEPandBP,respectively.

95

(i) ReduplicationsarefrequentinPortuguese;

(ii) Reduplications heavily contribute to the early iambic tendency found in the

Brazilianchildren'sspeech.

The studies referred to thus far, regard the acquisition of word stress and stress

patterns in BP. The study from Vigário, Freitas & Frota (2006), however, focused on the

acquisitionoftheProsodicWordinEP.

Vigário, Freitas & Frota (2006) investigated the speech productions of three112

Portuguese‐speaking children, João, Inês andMarta, fromde ages of 0;10‐2;0. The authors

studiedtheshapesoftheearlywordsproducedbythesechildrenandevaluatedtheroleof

frequency of adult language and child‐directed speech within the children's outputs. The

resultsindicatedthattherearenominimalityormaximalityrequirementsonwordsizeinthe

earlyspeechofPortuguesechildren.Theyshowedthat:

(i) monosyllabicCV targetswereproduced target‐like from thebeginningofword

production,asdemonstratedin(104).

(104) CVtargetsproducedaccordingly(Vigário,Freitas&Frota,2006:192):

Orthogr. Gloss Target Output Child,Agedá 'give‐imp.' /»da/ [»da]pé 'foot' /»pE/ [»pE]

Inês,1;0.25

dá 'give‐imp.' /»da/ [»då]é '(it)is' /»E/ [»E]

João,1;02.01

é '(it)is' /»E/ [»E]pé 'foot' /»pE/ [»pe]

Marta,1;02.0

(ii) monosyllabicCVC113targetswereearlyproducedas[CV],asshownbelow.

112Thischild,Inês,isthesamethatispartofourdatabase.113AsCVCtargets,theauthorsconsideredallnon‐CVmonosyllables,thatis,allmonosyllablesnotendinginoralvowel(meaningmonosyllableswithoralandnasaldiphthongs,aswellasnasalvowelsandsyllablesclosedwith/s/,/l/or/R/).

96

(105) CVCtargetsproducedas[CV](Vigário,Freitas&Frota,2006:192):

Orthogr. Gloss Target Output Child,Agemão 'hand' /»må‚w‚/ [»må] Inês,1;01.30quer 's/hewants' /»kER/ [»kE] Marta,1;02.0cão 'dog' /»kå‚w‚/ [»kç] Marta,1;03.08cão 'dog' /»kå‚w‚/ [»kå] João,1;06.18

(iii) disyllabic targets (/SW/ or /WS/) were mainly truncated to [CV], although

target‐likeproductionswerepossibleaswell.

(106) Disyllabic targets truncated or produced target‐like (Vigário, Freitas & Frota,

2006:193):

Orthogr. Gloss Target Output Child,Ageavó 'grandmother' /å»vç/ [å»då] João,1;0.12tia 'aunt' /»tiå/ [»tiå] João,1;08.13praia 'beach' /»pRajå/ [»pa] João,2;0.19balão 'balloon' /bå»lå‚w‚/ [»law] Marta,1;03.08bolso 'pocket' /»bo…su/ [»bçtå] Inês,1;08.02

(iv) trisyllabic targets (/CV'CVCV/) were not avoided in the early stages and they

emerged early in the children's speech. Truncation of /CVCVCV/ to [CV] and

[CVCV](either[SW]and[WS])andtarget‐likeproductionswerepossibleaswell.

(107) Truncation of CVCVCV targets to [CV] and [CVCV] ([SW] and [WS]) and target‐like

productions(Vigário,Freitas&Frota,2006:195):

Orthogr. Gloss Target Output Child,Agecaracol 'snail' /kåRå»kç…/ [»toå] Marta,1;03.08coelho 'rabbit' /ku»å¥u/ [kˆ»ke¥u] Marta,1;05.17sapato 'shoe' /så»patu/ [på»på] Inês,1;04.09banana 'banana' /bå»nånå/ [å»månå] Inês,1;05.11piscina 'swimmingpool' /piS»sinå/ [pi»tinå] Inês,1;09.19sapato 'shoe' /så»patu/ [»pa] João,1;07.24

97

The authors further showed that prosodic fillers are frequent, both with

monosyllables and with longer targets, in nouns, verbs and adverbs (108)114, though no

phonologicalrepresentationorassumptiononthenatureofthesefillersoundsisproposed.

(108) Prosodic fillers in monosyllables and longer targets (Vigário, Freitas & Frota,

2006:197):

Orthogr. Gloss Target Output Child,Agemãe 'mother' /»må‚j‚/ [ˆ»må]/[»må] João,0;11.06não 'no' /»nå‚w‚/ [ˆ»nå] João,1;10.08dá 'give‐imp.' /»da/ [å»da]/[»da] Inês,1;0.25pato 'duck' /»patu/ [å»tå]/[»tå] João,0;11.06chupeta 'pacifier' /Su»petå/ [å»pi]/[»pi] Inês,1;01.30

In summary, the reviews on Portuguese word stress acquisition presented above

indicate that, at an early stage, theword shape being processedmight be amonosyllable,

though other word templates are possible. This is in line with the findings from Bonilha

(2005) and Vigário et al. (2006). At this same stage, reports on the literature state that

childrenmightbeprocessingphrasalstress,andnotwordstress,astheymayaddsegmental

material to the stressed syllable and produce words longer than a monosyllable (Frota &

Matos, 2009; Santos, 2001; Frota & Vigário, 2008). When disyllables are being produced,

conflictingresultswerefound.Someauthorsargueforaneutralapproach, i.e., fornoinitial

preferenceforatrochaicoraniambicfoot(Bonilha,2005;Vigárioetal.,2006;Baia,2008b).

One author argues for an iambic tendency (Santos, 2007) and one author claiming for a

trochaictendency(Rapp,1994).

The table presented below summarizes the findings in previous studies on

Portuguesewordstressandstress‐relatedaspectsofacquisition.

114TheoccurrenceoffillersoundsinPortugueseearlyspeechwaspreviouslydiscussedinCosta&Freitas(2001),Freitas(1996)andFreitas&Miguel(1998).Wewillreviewthisissueinthefollowingsection.

98

Study Language Target QS Age Method TendencyRapp(1994)

1;6‐2;0

Experimental(8children) SW]Σ

Santos(2001)

0;11‐3;4

Spontaneous(2children) WS]Φ

Bonilha(2005)

1;01‐3;09

Spontaneous(1child)

Neutral~Iambic

Santos(2007)

BP

1;3‐3;6

Spontaneous(2children) WS]Σ

Vigárioetal.(2006)

EP 0;10‐2;0

Spontaneous(3children) σ>Neutral

Baia(2008)

BP

Trochaicor

iambic115Undeterm.116

1;5‐3;0

Spontaneous(1child)andexperimental

cross‐sectional

(42children)

WS]Σinspontaneous

data;Neutralin

experimentaldata

Table4.Shapeofearlywords­Portuguese(BPandEP).

Table4summarizestheconflictingresultsfoundforPortugueseacquisitionofword

stress. Santos (2001) argues thatwords stress is not acquired at the early stages ofword

production. However, children's initial productions conform to an iambic phrase. Bonilha

(2005)doesnotclaimaninitialiambicfoot,thoughherresultsindicatethat[WS]wordsare

produced correctly from the beginning. On the other hand, the conflicting results found in

spontaneous and experimental studies (Baia, 2008) seem to indicate that, indeed, the high

frequency of reduplications, whose phonological status is far from clear, might bias the

resultstoan 'apparent iambictendency',wheninfact,children'sproductionsareneutral in

termsof feetproduction.Theresults fromVigário,Freitas&Frota(2006)alsopointedtoa

neutralapproach.

Thedebateddomainforwordstress,theunclearstatusandshapeoffeet,aswellas

the undetermined role of syllable weight in the target system have brought additional

problems to the possible explanations for the contradicting results found across the

literature.

115Cf.Chapter1,section1.2.3.and1.2.5.1.,andreferencesthereintoBisol(1992,1993,1999),Lee(1995,2006,2007),Pereira(1999)andWetzels(2006).116Idem.

99

2.5.2. Fillers and reduplications in the path of prosodic acquisition in

Portuguese

ForfillersandreduplicationsinPortuguese,notmanystudieshavebeencarriedout.

Thepaucityofstudies,namelyregardingfillersounds,appearstobedue,ontheonehand,to

thedifficultiesposedwhenadistinctionbetweenfillersanddeterminersisatstake.Indeed,

inEPandBPthereisaphoneticcoincidencebetweenthefillersproducedbythechildrenand

the target formofdeterminers (in general, childrenproduce [ˆ], [´] or [å] for the feminine

article [å] and for the masculine article [u]), which makes the task of identification and

distinctionoffillersanddeterminersextremelydifficult.Ontheotherhand,theproposalson

therepresentationofearlywordshavenotbeenconsensualamongauthors.Itisnotclearyet

what the early representationofwords is inPortuguese,whether it is amonosyllable or a

constituentlargerthanasyllable;whetheritisatrocheeoraniamb.

Thoughweareawarethatfillersoundsandreduplicationsaretwodistinctissuesin

phonologicaldevelopment,wewilldiscussthemtogetherhere,thoughinseparatemoments,

sincetheyhavebeenbothusedtoargueforatop‐downapproachtoprosodicacquisition.

In Portuguese, the use of fillers syllables at the onset ofwordproductionhas been

mostlyrelatedtorhythmic,intonationalandprosodicfactors(Freitas,1996;Freitas&Costa,

2001;Freitas&Miguel,1998;Santos,1995,2003,2005;Scarpa,1998,1999;Santos&Scarpa,

2005).

Santos(1995,2003,2005),Scarpa(1998,1999)andSantos&Scarpa(2005)observe

that Brazilian children use both filler sounds and reduplications to fit a phonological

constituent larger (and higher in the prosodic hierarchy) than a syllable or the prosodic

word.Theauthorssuggest that,at theonsetofwordproduction,Brazilianchildrenarenot

processingwordstressanddonotuseprominenceatthewordlevel:Brazilianchildrenrely

onhigherprosodicdomainsintheirearlyutterances.Theevidenceforaninitialprocessingof

a prosodic domain higher than the syllable or the prosodicword arises from the fact that,

initially,children'sproductionsdonotconformtoasinglesyllableandveryoftenneithera

word.Fillersounds in(Brazilian)Portugueseacquisitionare interpretedasprosodic fillers,

rather than morphological placeholders, as they are mainly used to fill an intonational

contour–a(L)LH*(L)contour(Santos,1995,2003,2005;Santos&Scarpa,2005)–andnot

necessarily tomark amorphosyntactic position. If the words selected from the target are

smaller than the intonational contour, children lengthen the utterance by means of

insertions, reduplicationsandhiatuses indiphthongs.Conversely,whenchildren's intake is

longer than the intonational contour, childrenmake their utterances shorter, bymeans of

truncation (Santos, 1995,2003).These strategies (filler insertionand reduplication, on the

100

onehand,andutterancetruncation,ontheotherhand)allowchildrentofulfill intonational

constituentsandtomarkintonationalboundaries, thusbootstrappingtheacquisitionofthe

lower levels of the prosodic hierarchy. Therefore, Santos and Scarpa argue for a top‐down

approachtoprosodicacquisition.AccordingtoSantos(1995,2001,2003,2005)andScarpa

(1998, 1999), children start building prominences from the higher levels of the prosodic

hierarchy,andgraduallyfocustheirattentioninlowerphonologicaldomains,asexemplified

in(109).

(109) Top‐downmodelofprosodicacquisition(Santos&Scarpa,2005):

U g I g Φ gC g ω g Σ gσ

[/m/u»su˘kU] 'juice'(T.,1;10.15)

[a»mE] 'onomatpoeiaforgoat'(T.,1;10.22)

[m/»biso] 'animal'(T.,1;11.04)

The use of filler sounds and reduplications in early Portuguese phonological

developmentmight be used as a strategy to fulfill an intonational contour and, thus,mark

intonationalboundariesanddrivechildrentowardsthedomainofwordstress.

Santos&Scarpa(2005)suggestedthatthechildobservedmayworkwithhigherand

lowerlevels(thephonologicalphraseandthefoot,respectively)oftheprosodichierarchy,in

order to '[maximize] a metrical template as an optimal prosodic form' (Santos & Scarpa,

2005:174),bymeansoffillersounds.

UsingPeters' (2001)proposal,Costa&Freitas (2001),Freitas (1996)andFreitas&

Miguel (1998), on EP, support Scarpa's analysis, according to which filler sounds are

producedinordertofulfillalargerrhythmictemplate.Attheonsetofwordproduction,filler

sounds in Portuguese have a prosodic, and not amorphological function. The authors use

Peters' (2001) proposal and suggest that prenominal positions have a direct path from

101

premorphology(fillers)tofullmorphology(determiners).Anintermediatestagewherefiller

soundsmay have a protomorphological status is problematic since very often there is co‐

occurrenceoffullmorphemeswithpremorphology,withfillervowelsright‐adjacenttoverbs

andadverbsuntillateinspeechdevelopment.

In(110),weprovideinstancesoffillerinsertionbeforenouns,verbsandadverbs.

(110) Fillerinsertionbeforenouns,verbsandadverbsinEP(Freitas,1996:81,82):

Orthogr. Gloss Target Output Child,Agenão 'no' /»nå‚w‚/ [å»nå] Inês,1;01dá 'give' /»da/ [å»da] Inês,1;01Bambi 'name' /»bå‚bi/ [å»bå] Inês,1;03pão 'bread' /»på‚w‚/ [å»på] Inês,1;05quer 'want' /»kER/ [´»kE] Inês,1;05

With respect to reduplications, recent literature on phonological acquisition has

mentionedthatPortugueseisindeedalanguagewhereahighrateof(iambic)reduplications

isnoticeableatthebeginningofwordproduction(Baia,2006,2008a,c,d;Santos,2007).Baia

(2008a) furthermore claims that reduplications contribute for the early iambic tendency

foundintheearlyspeechofBrazilianchildren.Baia(2008a,c)arguesthatthemethodusedin

the data collection may influence the results, as experimental studies rarely include

reduplicatedwords.

In the Portuguese‐speaking children's productions we observe reduplicated words

thatwere themselves reduplicated in theadult speech (e.g.,mamã/mamãe 'mommy',babá,

'nanny',etc117)or,alternatively,wecanfindwordsthatwerenotreduplicatedinthetarget,

but can be reduplicated by the children (e.g., [kç»kç] produced for escola 'school', [pa»pa]

produced forsapato 'shoe'). In (111),wepresent some instancesof the reduplications that

can be found in Portuguese‐speaking children, both for target‐reduplicatedwords and for

targetnon‐reduplicatedwords.

117Much of Portuguese baby‐talk is reduplicated:mamã 'mommy',papá 'daddy',bebé 'baby', cocó 'poo', chichi'pee', ó­ó 'nap', tautau (used when a child misbehave), babá 'nanny', papar 'to eat fam.', vovô 'grandpa', vovó'granny',titi'uncle/auntfam.'.

102

(111) ReduplicationsinthePortugueseearlyspeech:

Orthogr. Gloss Target Output Child,Age

Study

mãe/mamãe(BP)

'mommy' Notavailable

mamãe118 ‐‐‐‐119

pai/papai 'daddy' Notavailable

papai ‐‐‐‐

cabra/memé 'goat' Notavailable

memé ‐‐‐‐

Stoel‐Gammon(1976:23)

vovó 'grandma' /vç»vç/ [bå»bå] JoãoII,1;9.11

Freitas(1997:133)

mamã(EP) 'mommy' /må»må‚/ [må»må‚] Inês,0;11.14

bebé(EP) 'baby' /bE»bE/ [Bå»bE] Marta,1;02.0

Vigárioetal.(2006:193)

pássaro/piupiu 'bird' /»pasåRu//piw»piw/

[pi»piw] G.,1;6.17 Bonilha(2005)

cocô(BP) 'poo' /ko»ko/ [ko»ko] L.,1;8

TargetRedupl.

dodói 'woundfam.'

/do»dçj/ [do»dçj] R.,1;4Santos(2007:51)

peito/mama 'breast' Notavailable

teté ‐‐‐‐

pénis 'penis' Notavailable

pipí ‐‐‐‐

nádegas 'buttocks' Notavailable

bumbum ‐‐‐‐

Stoel‐Gammon(1976:23)

orelhas 'ears' /o»Rå¥åS/ [ge»ge] Inês,1;5.11

Freitas(1997:139)

sapato 'shoe' /så»patu/ [på»på] Inês,1;04.09

laranja 'orange' /lå»Rå‚Zå/ [lå»la˘] Inês,1;05.11

Vigárioetal.(2006:195,197)

tractor 'tractor' /tRa»toR/ [ta»ta] G.,1;5.20 Bonilha(2005)

Targetnon­redupl.

chapéu 'hat' /Sa»pEw/ [papa»paw] R.,1;8 Santos(2007:43)

Thetendencyforreduplicationcanbesostrong,bothinthechildren'sdataandinthe

adultspeech, that thechildren’snamesarealso frequentlysubjectedto thatprocess. In the

caseofthechildrenobservedinthisstudy,forinstance,Clara/»klaRå/wasoftenproducedas

[ka»ka],Inês/i»neS/wasproducedas[ne»ne]andLumaoftenreferredtoherselfas[ta»ta].

118Nophonetictranscriptionisprovided.119 Stoel‐Gammon (1976) does not provide a full description of the data underlying the research. The authorfurtherarguesthat"[t]heanalysisofBrazilianbaby‐talkpresentedhereisnotintendedasanexhaustivestudyoftheuseofbaby‐talkinBrazil"(Stoel‐Gammon,1976:22).Thepapermainlyaimsatmakingageneraldescriptiononbaby‐talkinthelanguage,fromthedifferentlinguisticperspectives:typeoflexicon,phonology,grammar,useandattitudes.

103

AlltheclaimscarriedoutonBPforanearlyiambicfoothavebeenmostlymadeupon

thesetypesofstructures(Baia,2008a;Santos,2001,2007;Stoel‐Gammon,1976).However,

literature on the topic (namely on French ‐ Braud, 2003;Wauquier‐Gravelines, 2003) has

suggestedthatreduplicationsmightnotbeusedtofulfillfootrequirementsofthelanguage.

In sum, the status and role of filler sounds and reduplications in word stress and

wordshapeacquisitionisyettobediscussedinEP.Despiteevidencethattheycancontribute

tothefulfillmentofafootinthelanguage,datahavealsodemonstratedthattheuseoffiller

sounds and reduplications might be related to the processing of larger phonological

constituents thatwill bootstrap children's attention toword prominence. The use of these

strategies during acquisition might be interpreted as evidence for a top‐down model of

prosodic acquisition, which is inconsistent with previous findings for Germanic languages

likeDutchorEnglish(Demuth,1995;Fikkert,1994).

2.5.3.AcousticstudiesonwordstressacquisitioninPortuguese

In EP, very few studies have been carried out on the acoustic properties of word

stressduringacquisition.

Frota & Vigário (2008) and Frota & Matos (2009) focus on the intonational and

durationalpatterns (respectively)ofaPortuguesechild120.Frota&Vigário (2008)describe

thatlevelstress(e.g.,bola'ball'/»bçlå/producedas[»pa»pa])andstressshiftinthedirection

of iambs (e.g., bola 'ball' /»bçlå/ produced as [pa»pa]) is frequent at the beginning. These

resultswereborneoutbothfromtheperceptivetranscriptionandtonalassociationobserved

in the initial stages. Before 1;09, L. does not produce thepitch alignment, the inventory of

pitchaccentsandboundarytonesinanadult‐likemanner,thoughthechoiceoftonalevents

ismasteredsince1;05.Inthespeechofthechildobserved,finalstressstabilizesearlier(from

1;4onwards)thanpenultimatestress(from1;10onwards).After1;09,thechildmastersthe

alignment,theinventoryofpitchaccentsandboundarytonesand,simultaneously,theword

stress patterns and accent. Based on the evidence of stress, segmental duration and pitch

accent patterns, Frota & Vigário (2008) propose an account of prosodic phrasing during

developmentinEP.Theauthorsfurthermoresuggestaninterplaybetweenwordstressand

accent during in EP prosodic acquisition, with an initial stage where the child observed

producedasyllableasaprosodicwordandaphrase(Stage1:S=PW=phrase)andasecond

120ThechildunderobservationinthisstudyisLuma.Thecorpusis,however,differentfromtheonecollectedforthis dissertation. The corpus inwhich Frota & Vigário (2008) and Frota &Matos (2009) based their analyses(LumaLiDa) consists in daily collections, whereas our corpus (Phon_EP_Mono) consisted in biweekly datacollection.ToaccessLumaLiDa,cf.http://www.fl.ul.pt/LaboratorioFonetica/LumaLiDa.htm.

104

stage, where the child distinguishes the syllable from the prosodic word (Stage 2:

S≠PW=phrase).

Frota&Matos(2009)ellaboratesadevelopmentalproposaloftheprosodicstructure

inEP,basedon thedurationpatternsofL., the samechild.Theauthors confirmedFrota&

Vigário's findings, namelywith respect to the above‐mentioned stage1 and2.Theauthors

detectedtwomomentsinthedurationpatternsofL.:thefirstmoment(i.e.,until1;04),where

astrongcorrelationbetweenthedurationofsyllablesandthenumberofsyllablesinhigher

prosodic domains (namely, the prosodic word and the intonational phrase) was found,

providedevidenceforaphasewherethechilddoesnotdistinguishthesyllable,theprosodic

word and the intonational phrase (Stage 1: S=PW=phrase). Secondly, disyllabic words

emerge (i.e., after 1;04, until 1;08‐09) and, after thismoment, no correlation between the

syllable,ontheonehand,andtheProsodicWordandtheIntonationalPhrase,ontheother

hand is found(Stage2:S≠PW=phrase).Thedataadditionallypointedtoa thirdmoment, in

whichthechilddistinguishestheprosodicwordandthephrase,namelybyproducingphrase

final lengthening (Stage 3: S≠PW≠phrase). Frota&Matos (2009) also showed that syllable

positionwithin theutterance,butnotstress,affectsyllableduration.Theseresults leadthe

authorstosuggestthat,initially(inStage1),L.hasa'harmonic'prosodicprocessing.

Acousticstudiesontheacquisitionofwordstressandstress‐relatedaspectsinEPare

scarce.However,thefewstudiescarriedoutthusfarsuggestthat,initially,differentlevelsof

the prosodic structure, namely the intonational and phonological phrase, as well as the

syllable,mightbeinteractingduringprosodicacquisition.

2.6.Summaryofthemainissues,researchquestionsandhypotheses

Thereviewsmadethusfarhavedemonstratedthat,learningtoassignstressinvolves

the processing of a set of principles that are related to syllables, feet and prosodicwords,

which,aswehaveseeninChapter1,aretheprosodicconstituentswordstressregardsto.

Playing the linguist's role, the child needs to realize, first of all, whether stress in

her/hislanguageisrule‐basedorlexical,thatis,whetherstressinherlanguageisaproperty

ofspecificmorphemesorwords. Ifstress is learnedona lexicalbasis,stresserrorsarenot

expected, as they are not the result of a generalization. In this case, children may simply

produce stress, in the target stressed syllable, and no patterns are observed. In the

assumptionofunspecifiedlexicalrepresentations,andsinceinlanguageswithlexicalstress,

stressisapropertyofthewordsandmorphemes,childrenacquiringlexicalstressshouldnot

haveproductionswithanywordprominenceinthebeginning.Assoonasstressisspecified

105

inthelexicon,wordsareproducedaccordinglyandoncemorenostresserrorsareexpected.

If, otherwise,word stress isnot learnedona lexical basis, and it is rule‐driven, errors and

productionstrategiesareexpected.Until thestressrule is learned, thechildneedstoknow

thatstressfallsonspecificconstituents(namely,onfeet,wordsormorphemes).Additionally,

it is necessary that the child realizes whether the language she is acquiring is weight‐

sensitiveandwhatthedirectionoftherhythmicwaveisforstresspurposes.Thechildneeds

toknowwhatisafoot,whetheritisrelevantinherlanguageornot,and,ifso,whereisthe

foot head. Additionally, the child needs to learnwhat to rely on in order to assign stress:

whetheronmorphologicalconstituencyoronphonologicalinformation.

In sum, according to the main topics presented in Chapter 1, in a language like

Portuguese,wherestress isapropertyofwords(thuscalled 'ProsodicWords'), the learner

needstoacquire:

(i) thedomainofwordstress;

(ii) whetherstressinhis/herlanguageisweight‐sensitive;

(iii) whetherstressinhis/herlanguageissensitivetowordclasses;

(iv) whatisthedirectionalityofmainstress;

(v) whereshoulds/hestartbuildingwordsfrom;

(vi) whereistheheadofafoot;

(vii) whichconstituentsareconsideredasextrametrical.

Fromtheperspectiveoftheadultlanguagedescription,EPfacesfourmainproblems:

thefirstoneregardstheidentificationofthedomainforstress,thesecondoneisrelatedto

theestablishmentofthefootshape,thethirdoneconcernsquantity‐sensitivityandthefourth

oneisrelatedtoextrametricality.

Given the conflicting and often contradictory results found for the acquisition and

development of stress patterns in Portuguese (Baia, 2008a; Bonilha, 2005; Rapp, 1994;

Santos, 2001, 2007; Stoel‐Gammon, 1976), these are the research questions we aim to

addressinthisdissertation:

(i) HowiswordstressacquiredinPortuguese?

(ii) What can thedevelopmentofword shapeand stresspatterns tell us about the

stressalgorithminPortuguese?

106

Cross‐linguistic information showed that languages with a trochaic rhythm (at the

word level) either show a trochaic or a neutral tendency. Inmost of these languages, the

domain forstress is thephonologicalword(Dutch,English,Spanish,Catalan)andquantity‐

sensitivity is observed. An iambic rhythm has been only claimed for languages where

prominenceisdomain‐final,suchasFrench,and,inthatparticularcase,thefootmightnotbe

thetargetofchildren'sattention.

Even though previousworks on the acquisition of word stress in Portuguese have

suggestedanearlyiambictendency,ontheabove‐mentionedassumptions‐thatPortuguese

is a languagewith a trochaic rhythm, as furthermore suggested by the literature onword

stress and by the frequency information provided in Chapter 1 ‐ we hypothesize that

Portuguesechildrenwilldisplayanearlytrochaictendency.

AssumingthatPortuguesechildrenaresensitivetothetargetlanguageproperties,we

expectthatPortuguesechildrenmirrortherhythmofthetargetlanguage.Theobservationof

atrochaictendencyshouldbefoundinthe:

(i) Earlyproductionoftargettrochees;

(ii) Truncationof/WS/wordsto[S];

(iii) Truncationof/WSW/to[SW];

(iv) Laterproductionof/WWS/and/SWW/words.

If these patterns are to be found, evidence for an early sensitivity to the trochaic

rhythm of the language will be provided. Additionally, these results might indicate that a

binary[SW]constituent,andnotthestem,isbeingprocessedasthedomainforwordstress.

Thesepatternswillfurthermoreprovideevidenceforanalgorithmproposingthatthedefault

stressposition inPortuguese is in thepenultimatesyllableof the lexicalword(Bisol,1992,

1993;Wetzels,2006). In this case,an investigationon theroleofweight‐sensitivitywillbe

required. In particular, we hypothesize that if children are sensitive to weight, then it is

expectedthatchildrenobeythegeneralweight‐basedrule forstressassignment(stressthe

penultimatesyllableofthelexicalword,orthefinalsyllableifitisheavy‐Bisol,1992,1993;

Wetzels, 2006). Evidences for stress‐attraction in heavy syllables can be found, namely

through an earlier acquisition of unmarked /CV.'CV(C)/ words and a later acquisition of

marked /'CV.CVC/ words. Conversely, if children are sensitive to a morphology‐based

algorithm, we have to assume that Portuguese children knowmorphological constituency

since the early stages of word production. Namely, we have to assume that Portuguese‐

speakingchildrenareabletoidentify,withintheinput,whatis‐andwhatisnot‐thedomain

forwordstress.

107

Atrochaictendencywillbeconsistentwiththefindingsforothertrochaiclanguages,

eitherGermanic,suchasDutch(Fikkert,1994)andEnglish(Demuth,1996),orRomance,like

Catalan(Prieto,2006)orSpanish(Demuth,2001).

AlthoughpreviousworksonBPhavedemonstratedthatiambscanbepreferredover

trochees during Portuguese acquisition, we hypothesize that this early iambic tendency is

apparent.AnearlyiambictendencyishardtoargueforPortugueseacquisition,duetoseveral

aspects,thatwehavealreadymentionedinourreview,andwhichwenowreitterate:

(i) Typologicalreasons–asshowninChapter1121,trocheesaremorefrequentthan

iambscross‐linguistically.

(ii) Frequencyinformationoftheadultlanguage–asalsoshowninChapter1122,an

initial iambic tendency is very hard to explain in a language with a heavily

trochaic rhythm likePortuguese.Evenconsidering that thealgorithm forstress

assignmententailsanunderlyingiambicfoot,wehavetoconsiderthatchildren

areprovidedwithstrongevidenceforstressinthepenultimatesyllable.

(iii) Themorphology‐based algorithm, which predicts a domain‐final default stress

position (stress the last vowel of the stem), considers thewordmarker, a very

frequent structure in the system, as extrametrical or 'invisible' for stress

assignment.

(iv) An iambic tendency was disfavored in the path of language acquisition – as

shown in Tables 2 and 4, iambs have been clearly disfavored in the path of

acquisition.

The early iambic tendency found inPortuguesephonological acquisition, aswell as

thenatureofearly reduplicationsand fillers sounds,willbediscussed,on thebasisofnew

datafromEP.Theissueofatop‐downandbottom‐upapproachtoprosodicacquisitionwill

alsobeaddressed,onthelightoftheresults.

Inthisdissertationweaimatprovidingadescriptiveandexplanatorypathforword

stress acquisition in EP, based on the spontaneous speech production of 5 monolingual

Portuguese children. We will investigate the nature of early prosodic words and their

development in thePortuguese children's system,on thebasisof the results.We intend to

121Cf.section1.1.3.,Fig.5.122Cf.section1.2.4.,Figs.9and10.

108

presentdataonwordstressacquisitioninPortuguese,byshowingarelationshipbetweenthe

children's patterns and theprosodic structure. Additionally,we aim at providing empirical

evidence for targetanalysesonwordstress inPortugueseand for theearlyprocessingofa

phonologicalaspectoflanguage.

109

3.Methodologicalaspects

In this chapter, we will present the database that demonstrates the research

performedtoinvestigatetheacquisitionofwordstressinEP.

First, we will introduce the data considered. We will describe the collection

procedureandwewillpresentthechildren'sagesacrosssessionsandthenumberofsessions

taken per child (3.1.). Secondly, we will present the database properties, notation and

transcriptioncriteria.Wewilldescribethedatabaseandprovideinformationonthecriteria

underlying the transcription of the target and the children's productions (3.2.). Finally,we

willspecifythedevelopmentalthresholdsforemergenceandacquisition,intheanalyzeddata

(3.3.).

3.1.Thedata­subjectsandsessions

For the research conducted on the acquisition ofword stress in EPwe considered

longitudinalspontaneousspeechfrom5Portuguesemonolingualchildren:Clara,Inês,Joana,

JoãoandLuma123.

Thechildren'sspeechwascollectedeveryotherweek(inthecaseofJoãoandLuma)

andonamonthlybasis(inthecaseofInês,ClaraandJoana).Threeresearchersvideotaped

thechildren124.

Thedata fromClara, JoãoandLumawererecorded inaCanonMV210digitalvideo

camcorderandvideotapedinanalogueformat(Hi8cassettes),between2004and2007.Inês

andJoanawererecordedonaSonyHandycamVideo8,AFHi‐FiStereo,duringthe1990's.In

all cases, the analogue formatwas imported todigital formatusing iMovie© software, and

compressedfor320x240(.mov)format.

Table 5 summarizes the relevant information about the data collection and the

transcribedsessions:

123Lumaisapseudoname.124 InêsandJoanawerevideotapedbyMaria JoãoFreitas;ClarawasvideotapedbyTeresadaCostaandSusanaCorreia;LumaandJoãowerevideotapedbySusanaCorreia.

110

Age(1stsession)

Numberofsessions

Timegapbetweensessions(aprox.)

Avg.durationofsessions

Totaldurationofsessions

Clara 0;11.1125 12 46,7min. 9h54mInês 0;11.14 30 42min. 21hJoana 0;11.24 33

1month29,4min. 16h17m

João 1;0.1 22 48,6min. 18h22mLuma 0;11.23126 37 2weeks 45,1min. 30h27m

Table5.Informationaboutvideotapedsessionsanddatacollection

In the tablesbelowwewillpresent thenumberof sessionsperchild, thechildren's

agesduring the sessions and thenumberof utterancesper session.By 'numberof records

accounted' we mean the number of records that contain a word, a phrase or a sentence

phonetically transcribed. Records containing unintelligible speech were not taken into

account in these tables, as they did not provide data for the purpose of this research. The

number of types and tokensmentioned in these tables is not cumulative, from session to

session.Itmeansthat,inasessionwith11tokensandinasessionwith26tokens,sometypes

andtokensmightbeoverlapping.

Clara(birthdate:2005­06­01)

Sessionnumber

Child’sage Nr.ofrecordsaccounted Types Tokens T‐Tratio127

1 0;11.1 4 2 4 0.502 1;0.13 3 3 3 1.003 1;1.3 13 6 22 0.274 1;2.22 20 5 19 0.265 1;3.6 20 7 19 0.376 1;4.19 21 8 29 0.287 1;5.16 62 16 69 0.238 1;6.6 34 12 46 0.269 1;7.11 51 14 64 0.2210 1;8.20 132 33 220 0.1211 1;9.23 259 70 464 0.1512 1;10.15 333 56 519 0.11

125TherecordingsofClarastartedat0;9,butonlyfrom0;11onwardstargetwordswereperceived.126TherecordingsofLumastartedat0;7,butonlyfrom0;11onwardstargetwordswereperceived.127T‐TrationwascalculatedusingCLAN.Lightanddarkshadingindicatetheperiodbeforeandafterthe25wordpoint,respectively.

111

Inês(birthdate:1992­11­19)

Sessionnumber

Child’sage Nr.ofrecordsaccounted Types Tokens T‐Tratio

1 0;11.14 128 7 22 0.322 1;0.25 176 11 76 0.153 1;1.30 205 26 172 0.154 1;3.6 272 40 222 0.185 1;4.9 206 35 139 0.256 1;5.11 345 65 396 0.167 1;6.11 299 59 322 0.188 1;7.2 346 71 341 0.219 1;8.2 446 111 515 0.2210 1;9.19 510 136 898 0.1511 1;10.29 573 203 1271 0.1612 2;0.11 644 232 1904 0.1213 2;1.10 602 303 1578 0.1914 2;2.1 483 243 1467 0.1715 2;3.8 348 264 1224 0.2216 2;4.19 487 290 1664 0.1717 2;5.24 308 218 983 0.2218 2;7.16 502 329 1567 0.21

Joana(birthdate:1995­06­22)

Sessionnumber

Child’sage Nr.ofrecordsaccounted Types Tokens T‐Tratio

1 0;11.24 68 2 2 1.002 1;0.25 69 2 5 0.403 1;2.7 87 5 8 0.634 1;2.29 76 7 12 0.585 1;4.6 92 3 3 1.006 1;5.5 100 2 3 0.677 1;6.24 111 12 28 0.438 1;8.4 94 13 20 0.659 1;9.25 138 39 77 0.5110 1;10.22 206 55 125 0.4411 2;0.9 229 120 233 0.5212 2;2.19 264 135 319 0.4213 2;4.1 195 143 351 0.4114 2;6.24 420 249 984 0.25

112

João(birthdate:2004­05­02)

Sessionnumber

Child’sage Nr.ofrecordsaccounted Types Tokens T‐Tratio

1 1;0.1 53 1 6 0.172 1;0.28 63 2 5 0.403 1;1.12 112 3 13 0.234 1;1.28 52 4 12 0.335 1;2.13 75 3 14 0.216 1;2.30 61 6 19 0.327 1;3.21 115 9 45 0.208 1;4.17 99 13 36 0.369 1;5.12 113 19 57 0.3310 1;5.26 124 22 87 0.2511 1;6.16 156 19 79 0.2412 1;7.0 185 28 92 0.3013 1;7.20 191 18 107 0.1714 1;8.4 178 23 66 0.3515 1;8.25 136 26 79 0.3316 1;9.25 276 73 198 0.3717 1;10.11 288 75 227 0.3318 1;10.26 286 89 276 0.3219 1;11.10 316 109 296 0.3720 1;11.19 281 102 297 0.3421 2;0.6 261 89 243 0.3722 2;0.20 288 102 302 0.34

113

Luma(birthdate:2003­10­24)

Sessionnumber

Child’sage Nr.ofrecordsaccounted Types Tokens T‐Tratio

1 0;11.23 2 2 2 12 1;0.13 2 1 2 0.53 1;0.28 15 3 16 0.194 1;1.10 38 4 39 0.15 1;2.7 5 3 5 0.66 1;2.22 21 6 22 0.277 1;3.5 46 9 54 0.178 1;3.19 64 9 68 0.139 1;4.2 23 7 23 0.310 1;4.17 5 3 5 0.611 1;5.9 9 8 9 0.8912 1;5.24 19 6 21 0.2913 1;6.6 38 8 43 0.1914 1;6.20 27 8 30 0.2715 1;7.5 77 14 100 0.1416 1;7.19 52 7 81 0.1117 1;8.2 29 6 36 0.1718 1;8.15 40 11 48 0.2319 1;9.7 28 6 41 0.1520 1;9.29 83 22 105 0.2121 1;10.18 82 19 95 0.222 1;11.1 59 14 93 0.1523 1;11.15 85 16 112 0.1424 1;11.29 78 18 94 0.1925 2;0.13 109 22 73 0.326 2;0.27 114 32 102 0.3127 2;1.10 163 24 140 0.1728 2;2.4 155 24 151 0.1629 2;2.22 232 38 221 0.1730 2;3.26 357 80 417 0.1931 2;4.11 373 114 481 0.2432 2;4.25 477 116 721 0.1633 2;5.15 421 159 625 0.2534 2;5.20 380 164 677 0.2435 2;6.6 335 155 591 0.2636 2;6.20 432 203 842 0.2437 2;6.27 260 171 575 0.3

The data were recorded at the children's homes, in spontaneous situations. The

motherand/oranothercaretakerwerepresent,togetherwiththeresearcherinthesessions.

The children's ages considered for this researchwere comprisedbetween0;11and

2;7.Thephonetic transcriptionsofClara, JoãoandLumawereonlyavailableuntil1;10,2;0

and2;6respectively.Intheremainingchildren(InêsandJoana),werandomlyestablishedan

114

initialmaximumage(2;7and2;6128).Sinceweaimedatfocusingontheearlystagesofword

production,wedidnottakeintoaccountspeechproductionsbeyondthatagelimit.Aswewill

seeinChapters4,5and6,thisagelimitallowedustoobservetheacquisitionoftheregular

stresspattern inPortuguese (namely, /SW/,/WS/and/WSW/words129).However,due to

this limitation, we observed the emergence of /SWW/ and /WWS/ trisyllables, but the

acquisitionofthedifferentverbtensesinthecorpus130wasnotcompleted.

3.2.Databaseproperties,transcriptionandnotationcriteria

Twotrainedresearchers(bothPortuguesenative‐speakers)phoneticallytranscribed

the corpus and inserted the phonetic transcriptions inPhondatabase (Rose, MacWhinney,

Byrne, Hedlund, Maddocks, O'Brien, Wareham, 2006). The transcribed sessions were,

afterwards, inter‐changed between the two transcribers for revision. Each transcriber

inserted 50% of the children's speech and revised the 50% of the other transcriber. In

dubious cases, a third researcher carried out a blind transcription and, in the case of

persistentindecision,thatutterancewasmarkedanddisregardedfromouranalysis.

3.2.1.Databasedescriptionandfunctioning

ThedataweremanuallyinsertedinPhondatabase.Figure11representsthesession

editorwindow,wherethephonetictranscriptionswerecarriedout.

1282;6was theage limit considered in thisdissertation,but Inêswascollectedat2;5andat2;7.Therefore,wedecidedtochoosetheupperbound.129 In Chapter 5,wewill focus on targetmonosyllables, trochees, iambs and trisyllables (/WSW/, /WWS/ and/SWW/). As we will see, though /WWS/ and /SWW/ trisyllables will be analyzed, they will have a reducedamountoftokens,bothinthechildren'sintakeandinthechildren'sactualproductions.130Thisissuewillberecalledintherelevantsection(Chapter5).Asfaraswordstressacquisitionisconcerned,itisworthwhilesaying,atthispoint,thatthenon‐observationoftheacquisitionof/SWW/and/WWS/trisyllables,andthe fullacquisitionofverbs' inflectionuntil2;7mightprovideevidence forthedisparatecharacterof thesestructuresinthePortuguesechildren'sspeech.

115

Figure11.Phondatabasetranscriptionwindow

Thiswindow provides information on the child and the session being analyzed, as

wellasthechild'sage(automaticallycomputed).Abuilt‐inmediawindowplaysthevideoof

thesessionbeingtranscribed.Inthiswindowitispossibletoobservethefieldsforthe:

1. Orthographictranscription(field'Orthography');

2. Phonetictranscriptionofthetarget(field'IPATarget');

3. Phonetictranscriptionofthechild'sproduction(field'IPAActual');

4. Syllabificationofthetarget(field'TargetSyllables');

5. Syllabificationofthechild'sproduction(field'ActualSyllables');

6. Alignment between the syllables of the targetword and the syllables of theword

utteredbythechild(field'SyllableAlignment').

Phondatabaseallowedforspecificsearchesregardingthephonologicalpropertiesof

the transcriptions, both in the fields regarding the target and in the fields regarding the

children'sproductions.Asfarasthetopicofthisresearchworkisconcerned(theacquisition

ofwordstress),itwaspossibletosearchfor:

1

.2

.

3

.

4.

6.

1.

2.

3.

5.

116

(i) numberofwords;

(ii) numberofsyllableswithinaword;

(iii) syllablestructureindifferentwordpositions;

(iv) stresspatterns.

Thesearcheswereundertakenusingthe‘DataTiers’commandfortheresultsofthe

children’s‘Actual’productions.Forthecorrespondencebetweenthe‘Actual’andthe‘Target‘

tiers,weusedthe‘AlignedGroups’command.

Only phonological words were considered in this dissertation. Though clitic and

onomatopoeic words were orthographically and phonetically transcribed, they were

disregardedfromthesearchesundertakeninthisdissertation.

3.2.2.Criteriaoftranscriptionandinsertioninthedatabase

The corpus considered for the acquisition of word stress took into account the

children's speech from the beginning of word production. The children's speechwas fully

transcribedandthedataconsideredfortheanalysesweretakenfromtheentiresession.Any

adultorchild‐directedspeechwasnottranscribed.

Orthographic and phonetic transcriptions were carried out for each record of the

database,inallchildrenandinallsessions.Fields1.,2.and3.(Figure11)werefulfilledbythe

transcribers, whereas syllabification (4. and 5. ‐ 'Target Syllable' and 'Actual Syllable',

respectively), and the 'Target'‐'Actual' utterance alignment (6. ‐ 'Syllable Alignment')were

automaticallypairedwiththetranscription.

3.2.2.1.Criteriaforthetranscriptionofthetargetwords

We transcribed the target words as faithfully as possible to the adult standard EP

(Mateus & d'Andrade, 2000; Mateus, Brito, Duarte, Faria, Frota, Matos, Oliveira, Vigário &

Villalva,2003).Below,wepresentthecriteriaestablishedforthetranscriptionofthetarget

wordsproducedbythefiveobservedchildren:

1. Semi­phonologicaltranscription

The phonetic transcription of the target (within the field 'IPA Target') had a semi‐

phonological character, i.e., on the one hand it does not represent oral language as it is in

spontaneousspeechinEPbut,ontheotherhand,itisnotnecessarilyinaccordancewiththe

117

phonologicalrepresentationproposedforthetargetlanguage.Itmeansthat,forinstance,the

wordtelefone 'telephone' istranscribedas[tˆlˆ»fçnˆ]andnot*[t…»fçn]131,butawordsuchas

pneu 'tire'wasnottranscribed[pˆ»new],but[p»new]instead(phonologicalrepresentationis

/p_neu/, with an empty headed initial syllable). We chose to preserve, in the target

transcription, the vowel [ˆ], every time it is the result of vowel reduction from the

phonological vowel /e,E/. This option allowedus to preserve the adult syllables andword

boundaries,whichcouldbeimportantforthesearchescarriedoutonsyllabletypes,number

ofsyllablesperword,andstressplacement.

2. Sandhiphenomena

Asfarasbetween‐wordphonologicalprocessesareconcerned,weaccountedfor:

(i) thevoicingassimilationprocess:

e.g.,asaspretas'blackwings'[»azåS] [»pRetåS]transcribedas[»azåS][»pRetåS];

asasbrancas'whitewings'[»azåS][»bRå‚kåS]transcribedas[»azåZ][»bRå‚kåS];

(ii) the re‐syllabification of lateral consonant in word final position when the

followingwordstartswithavowel:

e.g.,solamarelo'yellowsun'[»sç…][åmå»RElu]transcribedas[»sçl][åmå»RElu];

(iii) the re‐syllabificationof the fricative consonant inword finalpositionwhen the

followingwordstartswithavowel:

e.g., asas amarelas 'yellow wings' [»azåS] [åmå»RElåS] transcribed as [»azåz]

[åmå»RElåS].

Everytimetherewereadjacentvowelsindifferentwords,wetranscribedthewords

suchastheyareproducedseparately,inordertokeepthenumberofsyllablesofeachtarget

word:

e.g.,gatoamarelo'yellowcat'[»gatu][åmå»RElu]transcribedas[»gatu][åmå»RElu]

131InEPspontaneousadultspeech,unstressed/e,E/issystematicallyreducedto[ˆ]andveryfrequentlydeleted.

118

3. –sC

–sC clusters (e.g.,escola 'school' [S»kçlå],escada 'staircase' [S»kadå]) are analyzed in

the target system (Andrade&Rodrigues, 1999) as sequences of Coda of an empty headed

syllablefollowedbyanobstruent(_C.CV).Despitehavinganorthographicvowelattheleft‐

edgeof theword, there isnosuchvowel instandardEP(e.g.,escada*[ˆS»kadå]).Therefore,

thephonetictranscriptionofthetargetwordsdidnotcarryaword‐initialvowel.

4. Diphthongsandglides

Inthetarget,glidesweretranscribedinfallingdiphthongsonly,asthesearetheonly

truediphthongs inEP (Mateus&d’Andrade, 2000).Risingdiphthongswere transcribed as

sequences of two vowels, i.e., nuclei of two phonological adjacent syllables that might be

realized as a rising diphthong (e.g., piano 'piano' was transcribed as [piȌnu] and not as

[»pjånu]).

5. Stress

Primary word stress was placed before the stressed syllable and not before the

stressedvowel.

6. Cuesgivenbytheinterlocutor

Insomecases,beforethechild'sutterances,theadultinterlocutorprovidedlinguistic

cues to the child (e.g., Adult:Olha, é uma ba... 'Look, it's a po...'; Child: ...tata, batata 'tato,

potato').Inthesecases,sincenoadultspeechwastranscribed,theinterlocutorutterancewas

representedbetweenbrackets '()' in theorthography fieldof the targetwordproducedby

thechild(e.g.,Orthography:(Ba)tata.).

3.2.2.2.Criteriaforthetranscriptionofthechildren'sproductions

The phonetic transcription carried out for the children's speech was as faithful as

possible,mainly concerningplace andmannerof articulationof the consonants, stress and

the number of syllables in a word. Unclear or dubious utterances were marked with an

asterisk(*).

7. Stress

Primary word stress was placed before the stressed syllable and not before the

119

stressedvowelinthephonetictranscriptionsofthechildren'sspeech.

When non‐natural clusters (e.g., obstruent+obstruent, nasal+nasal or

obstruent+nasal) were produced, we marked stress before the first consonant of the

consonantal sequence (e.g., /bå»lå‚w‚/ produced as [»m¯å‚w‚]), as it was not possible to

determine,at themomentof the transcriptions, inwhichstageof syllabledevelopment the

childrenweregoingthrough.

8. Pauses

Pauses were marked every time transcribers perceived a silence in the children's

speech.Theyweresignaledwith<#>(cardinal) in thephonetic transcription fieldand<...>

(ellipses)or<,>(comma)intheorthographyfield.

Throughoutthisdissertation,targetwordswillbetranscribedbetweenslashes‐e.g.

bola'ball'/»bçlå/‐,asopposedtothechildren'sactualproductions,whichwillbetranscribed

betweensquarebrackets‐e.g.,[»bçlå].

3.2.2.3.Codingandanalyzingfillersoundsandreduplications

Animportantaspectconcerningphonetictranscriptioninearlylanguageacquisition

isthedefinitionofcriteriatodeterminewhetherasoundisafillersoundornot,andhowcan

weidentifyandcategorizechildren'sreduplications.

Fillersoundsarenormallyconsideredtobeattachedtoaprosodicword.Beingoften

confounded with determiners or functors ‐ all of these being unstressed words ‐ it was

necessary to establish specific criteria to determine whether a sound attached to a word

identifiedassuchhasa fillersoundstatus,oraprosodicormorphologicalone.Despite the

admittedly difficult task of discriminating filler sounds vs. determiners in Portuguese‐

speaking children's speech, in this dissertation, the criteria followedby the transcribers to

decidewhetherasoundwasafilleroradeterminerwere,concomitantly:

(i) Gender agreement with the Noun in definite determiners (e.g., o/a 'the') ‐ in a

targetphrase likeo [u]pé [»pE] 'the footmasc.', aproduction like [ˆ»pE], [´»pE]or

[å»pE]wasconsideredaproductionwithafillersound,asopposedtothetarget‐like

production [u»pE]. In a target production like a [å] pá [»pa] 'the shovel fem.', a

production like [ˆ»pa], [´»pa] or [u»pa] was considered a production with a filler

120

sound,asopposedtothetarget‐likeproduction[å»pa].

(ii) Presence of indefinite and demonstrative determiners in a specific session (e.g.,

um/uma 'a'; este/esta 'this'; esse/essa 'that'; outro/outra 'other'; aquele/aquela

'that')‐theappearanceofindefiniteanddemonstrativedeterminersindicatedthat

the prenominal position was already available (from this moment onwards, a

determiner,andnotafiller,wasnormallyconsideredinthetranscription).

(iii) Absenceoffillersoundsbeforeverbsoradverbsinaspecificsession‐since,from

beginning,childrenusedfillersoundsinallsortsofwords(mostlynouns,verbsand

adverbs), theabsenceof filler soundsbeforeverbsor adverbs in a child's speech

indicatedthatfillersoundswerenotatuseanymore.

Despite thepre‐defined criteria, the transcriptionand codingof filler sounds in the

Portuguese children's speech were problematic. At the early stages, filler syllables were

clearly identifiable as such and, at the later stages, determinerswere clearly noticeable as

well.However, the identificationanddistinctionbetweenfillersyllablesanddeterminers in

theintermediatestagesofwordproductionwereadmittedlydifficulttocarryout.

With regard to reduplications, and as far as the early stages of phonological

acquisition are concerned, they canbe amultiple repetitionof a given syllable (e.g.,mamã

'mommy' produced as [må‚»må‚måmå] or [måmåmå], Inês, 0;11.14,) when no pauses were

detected between syllables, or a simple repetition of a given syllable (e.g., sapato 'shoe'

produced as [papa], Joana, 1;9.25). For this reason we will distinguish these two types of

reduplicationsbydesignatingtheformeras'multiplereduplications'andthelatteras'simple

reduplications'. In thetablesand figuresshowed in theresults'chapters(Chapter5and6),

multiple reduplications will be abbreviated as [CVCV...], whereas simple (disyllabic)

reduplicationswillbeabbreviatedandreferred toas [CV1CV1].Thesewillbeopposednon‐

reduplicantdisyllables,whichwewillabbreviateas[CV1CV2].

3.3.Criteriaforemergenceandacquisition

The discussion on when one should consider that a structure is acquired in the

children's system has been a matter of debate in the literature on language acquisition

(Bernhardt&Stemberger, 1998:2‐17) andhasoftenbeen related to themeaningof 'Stage'

121

(Ingram,1989:32‐58).

At the beginning, many of the children's utterances, either viewed from a

phonological perspective or not, tend to be non adult‐like. Of course, at this period, some

commonpropertiesbetweenthechild'sproductionandthetargetwordsarenoticeableand

thatisthereasonwhyparents,caretakersandresearchersidentifyachild'sproductionasa

wordoraphraseinthetargetsystem(Vihman&McCune,1994).

Whenstudyingthe languageacquisitionprocess intypicallydevelopedchildren, the

expected developmental path normally displays a continual increasing rate of target‐like

production,fromtheearlymomenttothelatemomentofobservation.Fromtheperiodwhen

childrensimplydonotselectagivenstructureintheirintake132,toamomentwhentokensof

that same structure are totally produced target‐like, normally three moments are

distinguishable: themomentofemergence, themomentofacquisition(ormastery)andthe

momentofstability(Matzenauer,1990).

Inthisdissertationwewillconsiderasemergentorinacquisitionprocessastructure

thathasaconsistentproductionbelow(≤)the49%target‐like.By'consistent',wemeanone

production appearing at least one time in one session, with continuity in the following

sessions. We will consider as acquired but not stable, a structure that has a target‐like

productionratebetween50%and75%target‐like(basedonFreitas,1997133).However,we

willconsiderthistarget‐likeproductionratetowhichcumulativelycorrespondstoatleast10

tokensinthechild'sintake.Forthetimebeing,wewillconsiderthatachildacquiredagiven

structure,ass/heproducesitcorrectlymostofthetime,thoughs/hemightoftenproduceit

with some instability. The stability point will be attained when a child produces a given

structurethathasarateoftarget‐likeproductionbetween76%and100%.

132Theterm'intake'hasbeenrecentlyusedtodesignatethewordsthatchildrenselectandattemptinproduction(Fikket&Levelt,2008).Thistermcompeteswiththeexpression'targetword'.133Cf.Freitas,1997:136,fn2.

122

123

4.Apreliminaryacousticstudyontheacquisitionofwordstress

Theacousticanalysiscarriedoutinthisdissertationwasmainlymotivatedbythefact

thatwordstresswasdifficulttoidentifyintheearlyproductionsofPortuguesechildren.As

referred to inChapter2134,Bernhardt&Stemberger (1998)mention that the identification

andanalysisofwordstressinthespeechofyoungchildrenmightbeadifficulttasktocarry

out, due tomaturational aspects intrinsic to thedevelopment and, specifically, to the great

variability found, both intra‐child and between‐children, in the early stages of speech

production.Infact,duringthephonetictranscriptionscarriedoutonthespeechproductions

of the children considered in our corpus, the difficulties found in identifying word

prominenceandthefrequentperceptionoflevelstressintheearlystagesofwordproduction

ledustoconductanacousticanalysisaimingat(i)identifyingwordprominenceintheearly

disyllablesand(ii)bringingreliabilitytothephonetictranscriptions.

Apartfrombeinganabstract(phonological)propertyofsomelanguages,obeyingto

specificregularitiesandfunctioningprinciples,instresslanguages,wordstresscorresponds,

also, to variations in the realization of physical (phonetic) correlates, like fundamental

frequency,intensityandduration,withinvowels(Hayes,1995;Kager,2007).

AsmentionedinChapter1,wordstressinEPismainlyrealizedbymeansofavowel

beingproducedwithhigher energy in relationshipwith the remainder vowels of theword

(Delgado‐Martins,1986,1988).Accordingtotheauthor,energyistheresultoftheintegralof

intensitybyduration.Within thesameprosodicword,stressedvowelsarenormally longer

andproducedwithgreater intensity thanunstressedvowels.Taken together, intensityand

duration are, thus, the relevant cues for producing and perceiving word stress in EP.

Fundamental frequency, on the contrary, ismainly related to intonational events and it is

related to the accent at higher phonological domains, such as the phonological and the

intonationalphrases(Frota,2000).

As presented in Chapter 2135, several studies have been conducted on the acoustic

propertiesofwordstress,withinalanguageacquisitionperspective(Allen&Hawkins,1980;

Kehoe, Stoel‐Gammon & Buder, 1995; Lleó & Arias, 2007; Pollock, Brammer & Hageman,

1993; Rose & Champdoizeau, 2008; Vihman, DePaolis & Davis, 1998). The results of the

studies suggest that, before 2;0, childrenmay not be able to correctlymaster the relevant

acoustic parameters for word stress (fundamental frequency, intensity and duration) and,

therefore, children might not be able to contrast the vowels within a word in order to

produce stress in a target‐like fashion. The results of these studies further point to the

134Cf.section2.1..135Cf.section2.1..

124

frequent recursion to stress shift and level stress, i.e., caseswherenowordprominence is

perceivedwithinthevowelsofaword,intheearlystagesofwordproduction.

Following previous reports in the literature according towhichword stress in the

earlystagesmightnotbemastered(Kehoe,Stoel‐Gammon&Buder,1995;Lleó&Arias,2007;

Pollock, Brammer Hageman, 1993; Vihman, DePaolis & Davis, 1998), and considering the

initial difficulties in identifyingword stress in the early speech of Portuguese children,we

hypothesizethat,inthebeginningofwordproduction,Portuguesechildrenmaynotcontrast

thetwovowelsinearlydisyllables.

The present study is a preliminary research conducted on a sample of the corpus

consideredinthedissertationproject136,aimingatstudyingtheacousticpropertiesofword

stress. In this chapter, special attention will be given to the production of each acoustic

parameter,andtotheproductiontheacousticparametersinthedirectionofaspecifictarget

word shape (/SW/ or /WS/). The acoustic analysis will enable us to identify word stress

throughthemeasurementoftheacousticparameters,bringingreliabilitytotheperception‐

basedtranscriptions.Forthatpurpose,wewillconductaqualitativeanalysisoftheacoustic

parameters related to word stress cross‐linguistically (fundamental frequency137, intensity

andduration138).Wewillinvestigatethemeanvaluesofeachacousticparameterseparately,

in both vowels of /SW/ and /WS/ and across sessions. We will analyze the two vowels

producedindisyllables,acrosssessions.Ineachacousticparameter,wewillplotthemeansof

eachvowel(V1andV2) indifferentsessions,both in/SW/and/WS/.Additionally,wewill

investigatethepresenceofasignificantdifferencebetweenthemeanvaluesofthevowelsin

disyllablesinthesamesession,peracousticparameterandperchild.After,wewillconducta

quantitative analysis, by measuring the amount of prominence put into each one of the

vowelsproduced indisyllablesby thechildrenunderanalysis.Sincedurationand intensity

are the relevant acoustic parameters for word stress in EP (Delgado‐Martins, 1977, 1986,

1988;Frota,2000139),wewillstandardizeandsumupthevaluesforintensityanddurationin

each vowel, for /SW/ and /WS/. The results found allowed to investigate the amount of

prominenceputintoagivenvowelofadisyllableproduced,tounderstandwhetherchildren's

productionswerecontrastiveforwordsstressand,consequently,toidentifywordstress.

Thischapterwillbeorganizedasfollows:wewillfirstpresentthemethodcarriedout

ontheacousticanalysis(section4.1.).Secondly,wewillpresenttheresults(section4.2.).The

results'sectionwillbedividedintotwoparts:afirstpart,regardingthequalitativeanalysis

136More detailed information on the corpus and themethod used in the acoustic analysis will be provided insection4.1.,ahead.137FundamentalfrequencywillalsobereferredtoasFØ.138 Aswe observed in Chapter 2, section 2.1., these are the parameters used by all authors investigatingwordstressacquisitionwithinanacousticperspective.139ForfurtherdetailsontheacousticpropertiesofwordstressinEP,cf.Chapter1,section1.2.2..

125

on the acoustic parameters observed separately (4.2.1.), and a second part, regarding the

quantitativeanalysisonwordprominence inthetwovowelsof thedisyllablesproducedby

thetwochildrenanalyzed(4.2.2.).Finally,wewilldiscusstheresultsfoundanddrawsome

generalconclusions(section4.3.and4.4.,respectively).

4.1.Method

Given the preliminary and exploratory nature of this acoustic analysis140 and the

time‐consumingcharacterofthetaskdeveloped,wesampledthespeechproductionsoftwo

ofthechildrenparticipatinginourcorpus,JoanaandInês.Thesewerethetwochildrentobe

transcribedandthetwochildrenwherethetranscriptionproblemsfirstarose.

Weconsideredthespeechproductionsfrom11sessions(session1‐11)inJoana,and

thespeechproductionsfrom7sessions(session1‐5,session10andsession12)inInês.We

analyzedthespeechproductionsofthetwochildrenuntilstatisticallysignificantdifferences

between the two syllables were found, with respect to the production of intensity and

duration, which, as we have seen in the previous section, are the relevant acoustic

parameters forwordstress inEP. In Joana,andsincemostofherspeechproductionswere

monosyllables,we analyzed all the sessions until the stressmastery (i.e., until /WS/were

producedwithhighervaluesof intensityanddurationinthesecondvowel,and/SW/were

producedwithhighervaluesofintensityanddurationinthefirstvowel).SinceInêswasthe

most productive child in the database and given that she had a great amount of disyllabic

productions,wedecidedtoanalyzeherfiveinitialsessions.However, insession5,thechild

onlyproducedthreeanalyzable/SW/words.Therefore,weselectedsession10,analyzingthe

first15iambictargetsandthefirst15trochaictargets.Inthissection,target‐likeproductions

of the stress patterns were observed in /SW/, but not in /WS/. After this, we randomly

selected session 12, analyzing the first 15 target iambs and the first 15 target trochees. In

session12,wefoundatarget‐likeproductionofbothstresspatternsinInêsspeech.

After the transcriptiontask,weretrievedall thedisyllabicproductionsof Joanaand

Inês,usingthe'Datatiers'commandinPhonsearchmodule.Withthiscommanditispossible

tosearchfordisyllabicwordsinthechildren'sproductiontier('Actualstresspattern').Here,

we inserted trochaic, iambicand levelstresscodes, inorder togetall thepossiblerelevant

disyllabicwordshapesproducedbythechildren.

Inthetablebelowwepresentthenumberoftargetdisyllables(tokens)inJoanaand

140Itisworthnoticingthattheacousticanalysiscarriedoutinthischapterwasconductedintheearlystagesofthephonetictranscriptions.

126

Inês'productions,acrosssessions.

Sessions Joana Age Inês AgeS1 0 0;11.24 11 0;11.14

S2 2 1;0.25 41 1;0.25S3 1 1;2.7 61 1;1.30S4 0 1;2.29 112 1;3.6S5 0 1;4.6 65 1;4.9S6 1 1;5.5 S7 1 1;6.24 S8 3 1;8.4 S9 37 1;9.25 S10 25 1;10.22 305 1;9.19S11 72 2;0.9 S12 467 2;0.11TOTAL 142 1062

Table6.Numberofdisyllabicwords(tokens)producedbyJoanaandInêsacrosssessions

FortheacousticanalysisweusedPraatsoftware,measuringbothvowels(V1andV2)

ofactualdisyllablesformaximumFØandintensity,andthetotaldurationofthatsamevowel.

Since the corpus considered is based on spontaneous, unstructured speech, some

additionalremarksmustbemade,inordertomotivatethereducedamountofsessionsand

targetsconsideredforthisanalysis.

Firstly, productionswith simultaneous speech or simultaneous noise, unintelligible

productionsandmultiple reduplications (with3ormore syllables)were frequentbut they

notanalyzed.

Secondly, from session 10 onwards, Inês started to produce two‐word utterances.

Duetointonationalfactors,weonlyanalyzeddisyllabicwordsinphrasefinalposition.With

Inês,weanalyzedallthedisyllablesuntilsession5,andthefirst25recordsinsessions10and

12. Until the end of the acoustic analysis, Joana had not reached the two‐word point.

Therefore,all thedisyllables in Joanawereanalyzed,althoughonly70wereconsidered for

theresultsoftheacousticanalysis.

Table7summarizesthenumberoftokensacousticallyanalyzedineachchild,across

sessions.

127

Joana InêsSessions /‐SW/ /‐WS/ /‐SW/ /‐WS/

S1 0 0 1 5S2 0 2 1 18S3 0 0 4 23S4 0 0 11 21S5 0 0 3 26S6 0 0 S7 0 0 S8 0 0 S9 7 11 S10 4 12 15 15S11 14 20 S12 15 15TOTAL 70 171

Table7.NumberoftokensconsideredandanalyzedpersessioninJoanaandInêsdisyllabicwords

Table7showsthat thenumberof tokensacousticallyanalyzed ineachchild isvery

different.InJoana,70tokenswereconsideredfortheacousticanalysis,whereasinInês,that

number increased to 171. Specifically, we observe that, when compared to Inês', Joana's

intakeisscarceindisyllabictargets.Thischilddoesnothaveany/SW/wordsandhasonly2

analyzable/WS/words(insession2).Inaddition,itisworthnoticingthatinsessions1,2,3

and5inInês,few/SW/tokenswereanalyzed.

After registering the values of the three acoustic parameters, in the two vowels of

disyllablesproducedbyeachchild,outliersweredisregarded.Afterwards,wecarriedouta

qualitativeanalysis,byanalyzingeachacousticparameterseparately,bothinthefirstandthe

second vowel.We compared each acoustic parameter in the first and the second vowel, in

targettrocheesandintargetiambs.Sincenormaldistributionandhomogeneityofvariances

wasnotobtained,weusedanon‐parametrictest(WilcoxonSigned‐RankTest,twotailed)to

testfordifferencesbetweenthefirstandthesecondvowelofchildren'sproductions(onthe

hypothesisaccordingtowhichV1washigherthanV2,andthecontrary).Asignificancelevel

ofp<.05willbeconsidered,meaningthatastatisticallysignificantdifferenceis foundwhen

thepvalueisequalorsmallerthan.05.

Subsequently,wecarriedouta standardization task inorder to statisticallyanalyze

thedataandtocomparetherelativestrengthofeachvowel inrelationtooneanother.The

standardization task enabled us to rescale the acoustic parameters values into various

comparablevalues (Abdi,2007).Wenormalize the resultsby finding themeanofduration

and intensity (the two relevant acousticparameters forEP)by session, andby subtracting

thismeanvaluetothevalueofthatparameterforagivenword.Theresultwasdividedbythe

128

standarddeviationof that sameparameter, in that session.The standard score (z)maybe

formulatedinthefollowingterms:

z=,whereχisarawscore(inthiscase,thevalueoftheacousticparameter)tobe

standardized,µisthemeanandσisthestandarddeviation.

InTables8,9and10wepresentanexampleofthemethodusedtostandardizethe

acousticparameters.Table8showstherawvaluesforintensityanddurationinbothvowels

forthewordmamã'mommy',insession2:

Session Record TargWord IntV1(dB)

IntV2(dB)

DurV1(s)

DurV2(s)

2 42 mamã 85.72 81.61 .137 .338Table8.Instanceofacousticmeasurementsforaword­rawscores

Theaverageforintensityanddurationinthatsessionwas81.34dBand.2160sand

the standard deviationwas 3.18 dB and .1136 s. The result of the standardization for the

wordpresentedwouldthenbe:

Session Record TargWord IntV1 IntV2 DurV1 DurV22 42 mamã 1.38 .08 ‐.70 1.07

Table9.Instanceofstandardizedvaluesfortherelevantacousticparameters

Thestandardizationresultsareconsistentwiththerawscores.Ifwelookatintensity

values, we verify that the first vowel was produced with larger intensity. After the

standardizationtask,thefirstvowelremainswithahighervalue.

Theresults fromthestandardizationof intensityanddurationwere, then,summed‐

up for each vowel, in order to give us one single relative value for each vowel, as shown

below:

Session Record TargWord IntV1 IntV2 DurV1 DurV22 42 mamã 1.38 .08 ‐.70 1.08

ΣΣ

Session Record TargWord V1 V22 42 mamã .68 1.16

Table10.SumofstandardizedvaluesforV1andV2

χ - µ

σ

129

Fromtheresultsofthestandardizationtask,itispossibletoconcludethatthesecond

vowelwasthestrengthenedvowel.

After standardizing and summing‐up the duration and intensity values for each

vowel,wecarriedoutapairedsamplet‐test,inordertotestwhetherthedifferencebetween

the first and the second vowel was significant or not in the course of development (and,

specifically,whichoneofthevowelswasproducedwithhighervalues).

AlltheresultswerecomputedusingSPSS11forMacintosh.

4.2.Results

In this sectionwewillpresent the results from theacoustic analysis carriedouton

thedisyllabicproductionsoftwoEP‐speakingchildren,InêsandJoana141.

First,wewillconductaqualitativeanalysis,byaccountingforthemeanvaluesofeach

acoustic parameter (fundamental frequency, intensity and duration). Afterwards, we will

conduct a quantitative analysiswherewewill consider the amount of prominence (i.e., of

intensity + duration) in each vowel of disyllables, across sessions, in the speech of two

childrenconsidered.

4.2.1.Qualitativeanalysis

In this section, a comparison between the use of each acoustic parameter will be

carried out.We will show the results of the average for each vowel (V1 and V2), in each

acoustic parameter (fundamental frequency, duration and intensity), across sessions. It is

worth noticing that, in Joana's case, the number of sessions per target stress pattern is

different,as,insession2thischildhastwoanalyzable/WS/words,butno/SW/.Theresults

fromeachchild,for/SW/and/WS/willbeshownseparately.

4.2.1.1.Fundamentalfrequency

Fundamentalfrequencyhasakeyroleinthewordstressproductionandperception

in many languages (namely, English). Given its cross‐linguistic relevance for word stress

productionandperception,bothinadultandinchild language,ananalysisonthevaluesof

141Forfurtherdetailsonthechildren,cf.Chapter3.,section3.1..

130

fundamental frequency was, thus, necessary. However, it is worthwhile noticing that

fundamentalfrequencyisnotarelevantacousticparameterforwordstressinEP.According

toDelgado‐Martins(1986,1988)andFrota(2000),variationinthefundamentalfrequencyis

arelevantcue,bothinperceptionandproduction,forphrasalstressandintonationalevents.

Inthefiguresbelow,wepresenttheresultsoftheaverageforfundamentalfrequency

acrosssessions,inJoanaandInês'productionsof/WS/and/SW/words.Figures12and13

showtheacousticanalysisforFØinJoana'sproductions,whereasFigures14and15present

thesametypeofdataforInês'productions.142

Figure12.FØvaluespersessionforV1andV2in/WS/(Joana)143

Figure13.FØvaluespersessionforV1andV2in/SW/(Joana)

Figures12and13showthat,insession2,/WS/wordsareproducedwithhighvalues

inFØ(inV1theFØvaluesisnearly500Hz,whereasinV2,theFØiscloseto600Hz).Inthe

following sessions, the average values for FØ decreases dramatically, both for /WS/ and

/SW/,tovaluesapproximatelybetween300and400Hz.

Fromthefiguresshownabove,wealsonoticethat,insession2,thesecondvowelwas

142Infollowingfigures(12‐23),astar(*)marksadifferencestatisticallysignificant(p<.05).143Insession2,Joanaonlyhas2tokens.

*

*

131

produced in average with higher FØ values, in target /WS/. However, in the following

sessions(sessions9and10),thefirstvowelwasalwaysslightlyhigherthanthesecondone.

Onlyinthelastsession(session11),asignificantdifferencebetweenthefirstandthesecond

vowelin/WS/wasobserved(T=59,p= .02),withhighervaluesinFØforV2(Mdn144V1=

350.4;MdnV2=389.2).Intarget/SW/,V1isproducedonaveragewithlowerFØvaluethan

V2,insession9.Insession10,V1isproducedwithhigherFØvaluethanV2(MdnV1=350.3;

MdnV2=344.4),thoughthedifferenceisnotstatisticallysignificant(T=3,p=.11).Onlyin

session11,V1isproducedwithhigherFØvaluethanV2(MdnV1=413.2;MdnV2=364.3),

andthatdifferenceisstatisticallysignificant(T=26,p=.03).

Figure14.FØvaluespersessionforV1andV2in/WS/(Inês)

Figure15.FØvaluespersessionforV1andV2in/SW/(Inês)

Figures14and15showthat,inInês,theFØvaluesofthefirstandthesecondvowel

of Inês' early disyllables (/WS/ and /SW/) are very balanced, except in the last observed

session(session12).Untilsession10,eitherthereisalmostnonoticeabledifferencebetween

144Non‐parametricstatisticsarecomputedonthebasisofthemedian,andnotthemean.

*

132

thetwovowelsorthefirstvoweltendstobeproducedwithhigherFØvalues,regardlessof

thetargetform.Onlyinthelastanalyzedsession(session12),InêsproduceshigherFØvalues

in the second vowel of target /WS/ words, though no significant statistical difference is

observed(MdnV1=501.6,MdnV2=474.6,T=54,p=.37).In/SW/words,however,higher

statisticallysignificantFØvaluesinthefirstvowelarenoticeable(MdnV1=455.8,MdnV2=

380.4,T=20,p=.02).

The general results for FØ suggest that Joana only uses fundamental frequency

contrastivelyinthelastsessionandInêsdoesnotusesfundamentalfrequencycontrastively

almost during the entire period under analysis, except for /SW/, in the last session. As

referredtointheliteratureonthetargetsystem,fundamentalfrequencyisnotarelevantcue

forwordstressandtheresultsfoundseemtosupportthisconclusionaswell.

4.2.1.2.Intensity

IntensityplaysamajorroleintheproductionofwordstressinEP,asstressedvowels

are produced with greater intensity, i.e., louder, than unstressed ones (Delgado‐Martins,

1986, 1988). Given its importance in the production of word stress in EP, an analysis on

valuesofintensityintheproductionsofthetwochildrenanalyzedwasnecessary.

Inthissectionwewillpresenttheresultsoftheintensityparameterinthedisyllabic

productionsofJoanaandInês.

InFigures16 and17wepresent thedata from Joana and inFigures18 and19we

showtheresultsfromInês.

Figure16.IntensityvaluespersessionforV1andV2in/WS/(Joana)

*

133

Figure17.IntensityvaluespersessionforV1andV2in/SW/(Joana)145

Figure 16 shows that in Joana's target /WS/ there is a sudden decrease in the

intensityvaluesfromsession2tosession9.AsobservedinFØ,Joanahasexaggeratedvalues

forintensityintarget/WS/,insession2(between80and90dB).Insession9and10,target

/WS/arebalancedintermsofintensityvaluesforV1andV2.Onlyinsession11astatistically

significantdifferenceisobservedintarget/WS/,withgreaterintensityvaluesinV2(MdnV1

=68.2,MdnV2=72.2,T=15,p=.00).Intarget/SW/(Figure17),session9haslowaverage

valuesofintensity,bothinV1andV2(between66and68dB).Insession10,V2isproduced

withmuchgreaterintensity,thoughthevaluespresentedareonlyrelativetofourtokens.In

session11,V1isproducedwithasignificantlyhigherintensitythanV2(MdnV1=71.5,Mdn

V2=67.6,T=25,p=.02).

Figures18and19presentthevaluesofintensityforInês.

Figure18.IntensityvaluespersessionforV1andV2in/WS/(Inês)

145Session10withonly4tokens.

*

134

Figure19.IntensityvaluespersessionforV1andV2in/SW/(Inês)146

TheresultsforintensityinInês'productionsshowthatthevaluesareagainbalanced,

bothfortarget/WS/and/SW/.Intarget/WS/words,thereisasmalldifferencebetweenthe

first and the second vowel, along all the sessions. Until the last session, no noticeable

differences are observed between V1 and V2. In target /SW/ and until session 3, the last

vowel tends to be produced with greater intensity than the initial one. From session 4

onwards,theoppositehappens:thefirstvowelisproducedwithgreaterintensity.Again,the

differencesarenotsignificantuntil session10,mainlydue to thereducednumberof/SW/

tokens per session (cf. Table 7). In the last session, the difference between V1 and V2

becomes statistically significant,with V1 being producedwith greater intensity (MdnV1=

73.2,MdnV2=66.2,T=4,p=.00).

Insum,intensitydoesnotseemtobeanacousticparameterusedtocontrastvowels

intheonsetofwordproduction.Joanacontrastivelyusesintensity,bothin/WS/and/SW/in

the last sessions. Inês contrasts the two vowelswith intensity in the last sessions aswell,

thoughonlyin/SW/.

4.2.1.3.Duration

Likeintensity(and,especially,withintensity),durationisanacousticparameterthat

is crucial to produce (and perceive) stressed vowels, which tends to be longer than

unstressedones inEP(Delgado‐Martins,1986,1988).Oftenconsidered,cross‐linguistically,

as an indicator of phonological length contrast, duration in EP, however, seems to be, as

suggested by Mateus & Andrade (2000), longer vowels are, normally, the result of stress,

ratherthanastressattractor.

In this section, we will present the results for the duration in Joana and Inês'

146Sessions1and2withonly1tokeneach;session5with3tokensonly.

*

135

disyllabicproductions.

Figures20and21refertothedurationvaluesproducedbyJoana.

Figure20.DurationvaluespersessionforV1andV2in/WS/(Joana)147

Figure21.DurationvaluespersessionforV1andV2in/SW/(Joana)

Figures 20 and 21 show that, until session 10, the V2 is longer than the first one,

regardlessofthetargetstresspattern.

Regarding target /WS/, in session 10 and 11 a significant difference between the

durationvaluesinV1andV2isobserved(T=0,p=.00insession10,T=1,p=.00insession

11).Inthesesessions,V2islongerthanV1(MdnV1=.14,MdnV2=.39,insession10,MdnV1

=.12,MdnV2=.21).

Thesecondvowel intarget/SW/tendstobe longerthanV1,butonly insession10

there isa significant statisticaldifference. In session11,however,V2 is, inaverage, longer,

butnostatisticaldifferencewasobserved(MdnV1=.14,MdnV2=.13,T=30,p=.13).

Figures22and23showInês'valuesfordurationin/WS/and/SW/.

147Cf.footnote143.

*

*

*

136

Figure22.DurationvaluespersessionforV1andV2in/WS/(Inês)

Figure23.DurationvaluespersessionforV1andV2in/SW/(Inês)148

Figures22and23showthat,untilsession4,thesecondvowelwasingenerallonger

than the firstone, irrespectivelyof the target stresspattern.Fromsession5onwards, Inês

lengthensthesecondvoweloftarget/WS/andthefirstvoweloftarget/SW/.Thedifferences

betweenV1andV2areonlystatisticallysignificant in/WS/, insessions1to12(exceptfor

session2and5).Intarget/SW/,thedurationalvaluesforV1andV2arestatisticallydifferent

onlyinsession12(MdnV1=.19,MdnV2=.15,T=11,p=.00).

Insummary,thetwochildrenappeartousemoreandearlierdurationthantheother

acoustic parameters. Inês, in particular,mainly uses duration in /WS/ to contrast the two

vowelsinthewords.In/SW/,thechildonlyusesdurationcontrastivelyinthelastsessions.

Joanadisplaysatendencytolengthenthesecondvowel.Significantdifferenceswillbefound

inthetwolastsessionsin/WS/andinthepenultimatesessionin/SW/.

148Insession1and2Inêshasonly1token(ineachsession).

***

**

137

4.2.1.4.Summaryforthequalitativeanalysis

The results for the qualitative analysis show that much variability is found in the

productionoftheacousticparametersforwordstress,bothbetweenchildrenandwithinthe

samechild.

JoanahasnoconsistentproductioninFØintheearlysessionsand,later,sheisableto

usehigherFØinthefirstvowelin/SW/andinthesecondvowelin/WS/(thoughcontrast

between V1 and V2 is higher in /SW/). As for intensity, the child generally uses greater

intensityinthesecondvowel,irrespectiveofthetargetform.Inthelastsessionhowever,the

differencebetween the firstand thesecondvowelof/SW/and/WS/ is significantand the

child puts greater intensity in V1, in /SW/, and in V2, in /WS/. As for duration, the child

generally uses higher duration in V2, irrespective of the target form (in /WS/, the last

sessionsarecorrectlycontrastedintermsofduration,i.e.,V2issignificantlylongerthanV1).

In /SW/, however, duration does not seem to be used contrastively in the last session of

observation.

InêsdoesnotuseFØorintensitytocontrastthetwovowelsinherearlyproductions.

Only in the last session thedifferencebetweenFØand intensity values in the first and the

secondvowelsissignificant,thoughitonlyappliesto/SW/.Duration,however,isadifferent

matter.Thischildgenerallyuses longerduration in theV2of/WS/acrossdevelopment. In

/SW/,durationonlybecomescontrastiveinthelastsession.

Fromtheresultsreported in thissection, it ishardtodefendthat, takenseparately,

theacousticparameters forwordstressaremastered.Thevariability found,bothbetween‐

children and within‐child, suggests no control over the phonetic parameters relevant to

producewordprominence.However,itisnoticeablethat/WS/iambstendtoproducedwith

highervaluesofFØ,intensityandduration,inbothchildren.

4.2.2. Quantitative analysis of the acoustic parameters for EP word stress

acquisition

Inthissection,wewillpresenttheresultsfromthestandardizationtask.Inthistask,

westandardized(ornormalized) thevaluesof intensityandduration, therelevantacoustic

parameters forwordstress inEP (Delgado‐Martins,1986,1988;Frota,2000), inV1and in

V2,separately.Thepurposeofstandardizingthevaluesofthesetwoparameterswastomake

themcomparable,sincetheacousticparametershavedifferentunitsofmeasurement(dBin

intensity, and seconds induration).Therefore,we standardized the valuesof intensity and

durationaccordingtothemethoddescribedinsection4.1.,and,afterwards,wesummedup

138

thevaluesfor intensityanddurationintheV1ofeachwordproducedbythechild,andthe

sameforV2.ThisprocedureallowedustohaveavalueforV1andavalueforV2ineachone

ofthedisyllablesproduced,concerningtheamountof 'prominence'(i.e. intensity+duration)

put in each of them. The comparison carried out between these two values allowed us to

verifywhichwasthemostprominentvowelofadisyllable,and,thus,providingacousticand

statistical evidence for an early stresspattern (trochaic, iambicorneutral) in the analyzed

speechofInêsandJoana.

Note that, from a perceptive perspective and as suggested by the reports on the

acousticdescriptionsofwordstressinthetargetsystem,weassumethatEPtranscribersare

trainedtodetectwordstressbasedondurationandloudness(intensity)ofagivenvowel,in

relationwiththeothervowel(s)oftheword.

Wewill first present the results from Joana, sessionby session, and afterwardswe

willpresenttheresultsfromInês,fromsession1to12.

Table11summarizestheresultsfoundforthestandardizationtaskinJoana,for/WS/

and/SW/149.

/WS/ /SW/Session

V1 V2 V1 V2

M=‐.78,SE=.62 M=‐.002,SE=1.62 M=.53,SE=.93 M=.89,SE=1.4S9

t(10)=‐.54,p=.30 t(6)=‐.56,p=.30

M=‐.88,SE=.66 M=.84,SE=.62 M=1.79,SE=1.96 M=1.25,SE=1.82S10

t(11)=‐2.57,p=.01 t(3)=.57,p=.30

M=‐1.1,SE=.30 M=1.47,SE=.45 M=.10,SE=.47 M=‐.54,SE=.70S11

t(19)=‐4.26,p=.00 t(13)=.76,p=.01

Table11.Standardizationresults(Joana)150

Insession9,Joana'sproductionofvowelsindisyllableswasnotyetcontrastive.The

results fromthestandardizationsuggest that there isnosignificantdifferencebetween the

twovowels,both in target/WS/(t(10)= ‐.54,p= .30),and in target/SW/(t(6)= ‐.56,p=

.30).Insession10,wefounddifferentresultsfortarget/WS/andtarget/SW/.Target/WS/

wereproducedwithvowelcontrast,asthemeanofthefirstvowelwassignificantlylower(M

=‐.88,SE=.66)thanthelastvowel(M=.84,SE=.62)(t(11)=‐2.57,p=.01).Target/SW/,

however,didnotshowsignificantdifferencesbetween the twovowels (t(3)= .57,p= .30).

149Relevantresultswillbemarkedinbold.150'M'standsformean,'SE'standsforstandarderror.

139

Finally,insession11,bothtargetpatternswerecontrastive:thesecondvoweloftarget/WS/

wassignificantlyhigher than the firstone (V1 ‐M= ‐1.1,SE= .30;V2 ‐M=1.47,SE= .45;

t(19)=‐4.26,p= .00),andthe firstvowelof target/‐SW/wassignificantlyhigherthanthe

finalone(V1‐M=.10,SE=.47;V2‐M=‐.54,SE=.70,t(13)=.76,p=.01).

Table12summarizesJoana'sproductionofdisyllablesintermsofvowelcontrast.

/­WS/ /­SW/

Session9 V1=V2 V1=V2

Session10 V1<V2 V1=V2

Session11 V1<V2 V1>V2

Table12.Summaryofthestandardizationresults(Joana)

InJoana'sproduction,shefirsttendsto levelthetwovowelsofthewords.Lateron,

sheisabletoproduce/SW/and/WS/accordingly.

Inês'productionofvowelsindisyllableswascontrastivesincethebeginning,butonly

inwhatconcernsthevaluesfor/WS/151.

151Relevantresultswillbemarkedinbold.

140

/WS/ /SW/Sessions

V1 V2 V1 V2

M=‐.42,SE=.27 M=1.03,SE=.34 ‐‐ ‐‐S1

t(4)=‐4.58,p=.00 ‐‐ ‐‐

M=‐.48,SE=.33 M=.09,SE=.34 ‐‐ ‐‐S2

t(17)=‐1.64,p=.06 ‐‐ ‐‐

M=‐.36,SE=.26 M=.34,SE=.40 M=‐.41,SE=.55 M=.59,SE=.44S3

t(22)=‐2.82,p=.00 t(3)=‐2.05,p=.06

M=‐.29,SE=.31 M=.44,SE=.29 M=‐.53,SE=.48 M=.07,SE=.68S4

t(20)=‐2.52,p=.01 t(10)=‐1.08,p=.15

M=‐.41,SE=.12 M=.43,SE=.22 M=.41,SE=.15 M=‐.58,SE=.43S5

t(25)=‐3.42,p=.00 t(2)=3.1,p=.04

M=‐.06,SE=.31 M=‐.007,SE=.31 M=.03,SE=.26 M=‐.27,SE=.19S10

t(13)=‐.19,p=.42 t(14)=2.12,p=.02

M=‐.05,SE=.28 M=.20,SE=.27 M=.21,SE=.24 M=‐.67,SE=.18S12

t(14)=‐1.91,p=.03 t(13)=5.31,p=.00

Table13.Standardizationresults(Inês)

Theresults fromthestandardizationsuggest that, insession1, there isasignificant

differencebetweenthetwovowelsintarget/WS/152(V1‐M=‐.42,SE=.27;V2‐M=1.03,SE

= .34, t(4) = ‐4.58, p = .00). In session 2, target /WS/153 showed a marginal significant

differencebetweenthetwovowels,thesecondvowelbeingproduced,inaverage,withhigher

valuesthanthefirstvowel(V1‐M=‐.48,SE= .33;V2‐M= .09,SE= .34,t(17)=‐1.64,p=

.06). The data from session 3 demonstrated that the second vowel tended to be produced

with significantly higher values in target /WS/, as the second vowel was produced with

significantlyhighervaluesthanthefirstvowel(V1‐M=‐.36,SE=.26;V2‐M=.34,SE=.40,

t(22)=‐2.82,p=.00).Intarget/SW/,only4tokenswereobserve,whichdidnotallowusto

runanyrobuststatisticaltest.Insession4,nosignificantdifferencewasfoundbetweenthe

vowels in target /SW/ (t(10) = ‐1.08, p = .15) although in target /WS/ the vowels were

contrastive in the expected sense: the second vowelwas producedwith significant higher

vowelsthanthefirstone(V1‐M=‐.29,SE=.31,V2‐M=.44,SE=.29,t(20)=‐2.52,p=.01).

In session 5, again, target /WS/ have significantly different vowels, in accordance to the

152InInês'session1,onlyonetokenoftarget/SW/wasproduced.153InInês'session2,onlyonetokenoftarget/SW/wasproduced.

141

target(V1‐M=‐.41,SE=.12;V2‐M=.43,SE=.22,t(25)=‐3.42,p=.00).Intarget/SW/,the

resultspoint toastatisticallysignificantdifferencebetweenV1andV2(t(2)=3.1,p= .04),

though only 3 tokens were analyzed. In session 10, only target /SW/ were produced

contrastively(t(14)=2.12,p=.02).Finally,insession12,Inês'productionsaretarget‐like:in

target/WS/thesecondvowelisstatisticallydifferentfromthefirstone(V1‐M=‐.05,SE=

.28;V2‐M=.20,SE=.27,t(14)=‐1.91,p=.03)andthesameoccursintarget/SW/(V1‐M=

.21,SE=.24;V2‐M=‐.67,SE=.18,t(13)=5.31,p=.00).Thesecondvowelintarget/WS/is

higherthanthefirstoneandtheoppositeisvalidfortarget/SW/.

Table14summarizesInês'productionofdisyllablesintermsofvowelcontrast.

/­WS/ /­SW/

Session1 V1<V2 ‐‐

Session2 V1=V2154 ‐‐

Session3 V1<V2 V1=V2155

Session4 V1<V2 V1=V2

Session5 V1<V2 V1>V2156

Session10 V1=V2 V1>V2

Session12 V1<V2 V1>V2

Table14.Summaryofthestandardizationresults(Inês)

Inês followsanearlydevelopmentofwordstresswhere shedemonstratesanearly

masteryof iambsanda latermasteryof trochees,by theuseof level stressandbymaking

iambsoutoftrochees.Itisworthmentioning,however,thatInêsusedmainlyreduplications,

inherinitialspeech.

4.3.Discussion

Inthischapterwehaveshowedtheresultsofapreliminaryacousticanalysiscarried

out on a speech sample of twoPortuguese children, Inês and Joana. First,we analyzed the

disyllabic productions of the two children, accounting for themean values of fundamental

frequency,intensityandduration,separately.Afterwards,wecarriedoutananalysisaiming

atquantifyingtheamountofprominenceputintothetwovowelsofthedisyllablesproduced

154Inthissession,Joanahasonlytwo/WS/tokens.155Insession3,Inêshasonlyfour/SW/tokens.156Insession5,Inêshasonlythreetokens.

142

by the two children under observation, considering only the two relevant acoustic

parametersforwordstressinthetargetsystem:intensityandduration.

The investigation of the acoustic parameters used to cue word stress in cross‐

linguisticacquisitionhassuggestedthat,before2;0,theseacousticparametersmightnotyet

beenmastered.WehaveshownthatdatafromEnglish(Kehoe,Stoel‐Gammon&Buder,1995;

Pollock, Brammer & Hageman, 1993), suggesting that childrenmight not use fundamental

frequency, intensityorduration inanadult‐likemanner, inorder tocontrast thevowels in

multisyllables.SimilarresultswerefoundinVihman,DePaolis&Davis(1998),whereahigh

degree of variability in the production of the acoustic parameters in English‐speaking

childrenobserveduntil the25word‐point.Studiescarriedout inolderchildren(2,3and4

year‐olds), not only in English but also in other languages (like French and Spanish),

however,indicatethatthemasteryoftheacousticparametersand,consequently,themastery

ofwordstress,takeplacelaterindevelopment(Allen&Hawkins,1980;Lléo&Arias,2007;

Pollock,Brammer&Hageman,1993;Rose&Champdoizeau,2008).

In EP, stressed vowels are producedwith higher values of energy than unstressed

vowels,energybeingtheintegraloftheintensitybytheduration.Intensityanddurationare,

thus, the acoustic parameters relevant for word stress, whereas fundamental frequency is

used to cue intonational events and normally cues prominences at higher phonological

domains(suchasthephonologicalortheintonationalphrase).

From the analysis carried out taking into account a comparison of each acoustic

parameterand thecomputationof intensityandduration,weobserved that Inês showeda

tendencyforhighervaluesinthesecondvowelofdisyllables,butonlyin/WS/157.Intarget/‐

SW/, the contrastbetween the first and the secondvowel in the correctdirection (V1>V2)

occurslater(session5).Joanamainlyusedlevelstresssincetheearlysessions,untilsession

10.Insession11,sheisabletoproduce/SW/and/WS/target‐like.

InTable15,wesummarizetheresultsoftheacousticanalysis.Inthistable,'='stands

forlevelstress.

Joana Inês

/­SW/ /­WS/ /­SW/ /­WS/

Moment1 V1=V2 V1=V2 V1=V2 [WS]

Moment2 [SW] [WS] [SW] [WS]

Table15.Summaryofthefindingsintheacousticanalysis

157ItisworthremarkingthatmostofInês'early[WS]wordsproducedarereduplications.

143

TheresultssummarizedinTable15suggestthat:

(i) at the beginning, different children may use different strategies to deal with

prominenceinthedisyllables;

(ii) atthebeginning,wordprominencemaynotexist(asinJoana);

(iii) atthebeginning,atendencyfor[WS]productionsmayexist(asinInês);

(iv) atalatermoment,thefirstvowelin/SW/wordsisproducedwithhighervalues

oftheacousticparameters;

(v) at a later moment, the second vowel in /WS/ words is produced with higher

valuesoftheacousticparameters.

The preliminary study carried out in this chapter indicates that, in one of the two

childrenobserved(Inês),thesecondvowelof/WS/wordswasproducedwithhighervalues

of the relevant acoustic parameters. In /SW/ words, the two vowels of the disyllables

analyzed were not generally produced contrastively. In Joana, the absence of contrast

betweenthetwovowelsofearlydisyllableswasattested,thoughlater,both/SW/and/WS/

wordswereproducedwithhighervaluesinthetargetstressedvowel.Thefrequentcasesof

levelstressaccountedforinthisacousticanalysishavesupportedthedifficultiesfoundinthe

transcriptionoftheinitialspeechproductionsofthetwochildrenanalyzed.

In Joana's early sessions (until session10), significantdifferencesbetween the first

andthesecondvowelwerenotfoundintheproductionoffundamentalfrequency,intensity

and duration. In the last session (session 11), only duration in /SW/ was not produced

contrastively.ThequantitativeanalysisconfirmedthatonlyinthelastobservedsessionJoana

produced/SW/and/WS/target‐like.

InInês'data,thesecondvowelwassignificantlylongerthanthefirstvowel,intarget

iambs. In the last observed session (session 12), Inês used duration contrastively in /SW/

and/WS/(she lengthens the firstvowel in/SW/and thesecondvowel in/WS/).Sheuses

fundamentalfrequencycontrastivelyonlyin/SW/(withhigherpitchinthefirstvowel),and

incorrectlyuseshigherintensityinthesecondvowelof/SW/.Theresultsofthequantitative

analysisshowedthat, inherearlysessions, Inês tends tostrengthenthe lastvowel, inboth

/WS/and/SW/.Inthelastobservedsession,Inêscorrectlyproduces/SW/and/WS/.

If the childrenobservedweremasteringprominencewithin theworddomain, then

wewouldexpectthatthefirstandthesecondvowelwereproducedwithacousticcontrast(in

eachacousticparameterandintheoverallanalysis),inthedirectionofthetargetprominent

vowel. If childrenhadalreadymastered theacousticparameters,wewouldexpect themto

produce the relevant acoustic parameters for word stress (intensity and duration) in the

144

expecteddirection,consideringthestresspatterninthetargetword(V1>V2in/SW/,V1<V2

in/WS/).ThispatternwasobservedinInês,onlyintarget/‐WS/.Infact,atthebeginningof

word production, a great variation was observed: in Inês, /‐WS/, but not /‐SW/ were

producedaccordingly;inJoana,levelstresswasmostlycommon.

Theresultsfoundconfirmandlegitimatetheresearchers’difficultiesinthechildren's

perceptive transcriptions at the early stagesofwordproduction. For instance, in Joana, no

significantdifferenceswerefoundinthevaluesoftheacousticparametersofthetwovowels

ofdisyllablesatthebeginning(Figure12,13,16,17,20and22,andTable11).InInês,only

duration seemed to be contrastive (Figure 22 and 23, and Table 13). The overall findings

fromthisacousticanalysisconfirmBernhardt&Stemberger(1998),accordingtowhomthe

phonetictranscriptionofchildren'searlyspeechmaybedifficultandinconsistent.Also, the

data confirmed what has previously found by Goffman (1985), Pollock et al. (1993) and

Kehoeetal.(1995).Thetranscriptionsofearlyspeechwithrespecttowordstressmayonly

havearound60%ofreliability.

Somestudiesreportahighdegreeofvariabilityinthewaychildrenusewordstress,

namely,inthewaychildrenproducetheacousticparameters(namely,Kehoeetal.,1995,for

English‐speaking children, and Vihman et al., 1998, for English and French‐speaking

children).Otherauthorsclaimthatchildren'searlyproductionsarenotaccurateintheiruse

of the relevant acoustic parameters, though they may correspond to the target language

tendency (Lleó&Arias, 2007).Our results confirmed that childrendonotuse the acoustic

parameterstarget‐like,aswellastheydonotnecessarilystrengthenthecorrectvowel.

The results brought up by the acoustic analysis suggest that, in the early stages of

wordproduction,childrenmightberelyingonprosodicdomainshigherthanthefootoreven

theprosodicword(e.g.,thephonologicalortheintonationalphrase).Thisanalysishasbeen

suggested by Vihman et al. (1998) to explain the high variability found in the speech

productions forEnglish‐speaking children.Whenacoustically analyzing the early speechof

American children, the authors found that not always children's utterances had a trochaic

shape(Vihmanetal.,1998),andsuggestedthatthechildrenobservedmightnotbeproducing

wordstress,butratherphrasalstress,basedonthehighinputfrequencyofiambicphrasesin

earlychild‐directedspeech(allgone,theball,etc.).

4.4.Summary

In this chapter we analyzed the early speech of two Portuguese‐speaking children

fromanacousticperspective.

145

Wecarriedoutaqualitativeanalysisoftheaveragevaluesofeachacousticparameter

(fundamentalfrequency,intensityandduration)inthefirstandsecondvowelofboth/SW/

and /WS/ produced by two Portuguese children. Afterwards,we undertook a quantitative

analysis, where we considered intensity and duration only, since, according to the target

analysis, these are the two relevant acoustic parameters forword stress inEP. In order to

comparethestrengthofeachvowel,westandardizetheresultsfromintensityandduration

andsummedupeachofthemseparately.Statisticaltestingallowedustoobservethat,atthe

beginning,InêsandJoanaarenotproducingwordstressadult‐like,astheproductionofthe

relevantacousticparametersforwordsstressarenotmasteredintheearlymomentofword

production. Our results further confirmed reports on the literature according to which,

before 2 years‐old, children are not able to correctly differentiate stressed and unstressed

vowels in multisyllabic utterances. Our results were consistent with previous findings in

studiesfocusingontheacousticcorrelatesofwordstressduringacquisition,whichindicate

thatearlyspeechmaynotbecontrastiveinstressterms.Thelaterobservationoftarget‐like

productions, both in /WS/ and in /SW/ indicates that there might be a moment in the

children'sproductionswherewordprominencefallsintoplace.

In the following chapter we will analyze the stress patterns produced by the five

children present in our corpus and confirm that, in fact, a turning point occurs during the

acquisitionofwordshapeandstresspatterns.

146

147

5.TheacquisitionofstresspatternsinEP

InthischapterwewillpresenttheresultsoftheacquisitionofstresspatternsinEP,in

thespeechoffivePortuguesechildren.

In Chapter 2 we showed that the observation of the stress patterns during early

acquisition has been one of themeans used by researchers to investigateword stress and

wordshapeacquisitioncross‐linguistically.Thoughinmostoftheliterature,thestudyonthe

mannerhowchildrenbuildfeetandwordshavebeenmainlyusedtostudytheacquisitionof

wordshape(andnotnecessarilywordstressanditsalgorithm),theobservationoftheway

prosodicwords–withtheirstrongandweaksyllables–evolvehasbeenanefficientmethod

tostudytheacquisitionofwordstress(Fikkert,1994;Kehoe,1998,Santos,2007;Tzakosta,

2004).

In the literature review carried out in Chapter 2, we compared the acquisition of

stresspatternsacrosslanguages.Weobservedthatthedevelopmentofprosodicwordscould

start with a single syllable (like in Dutch and English ‐ Fikkert, 1994 and Demuth, 1995,

respectively) and evolve to larger prosodic units in the course of development. On the

contrary, in French, longer words are observed from the beginning of word production

(Braud,2003;Wauquier‐Gravelines,2003).Spanishhasanacquisitionpathsimilartoother

trochaic languages, likeDutchandEnglish, and shows that at thebeginning,monosyllables

prevail(Lleó&Demuth,1999).However,longerwordsmayoccurearlierthaninEnglish,due

tolanguagefrequencypatterns(SpanishhasahigherfrequencyoflongerwordsthanEnglish

‐Roark&Demuth,2000).

Though some general trends have been observed in the literature reports (for

instance, in early acquisition of trochaic languages, children tend to mirror the trochaic

tendencyof thetarget),someconflictingresultshavebeenreportedaswell. Inmanycases,

despitesomeauthorsclaimforagivenstresspattern,othersinvestigatingthesamelanguage

claimforanotsocleartrend.ThatisthecaseofEnglish(Klein,1984)andSpanish(Hochberg,

1988a)where a neutral start, and not only a trochaic approach has been proposed during

acquisition. The case of Portuguese appears as evenmore complicated, since studies have

claimedboth fora trochaic (Rapp,1994)andan iambic tendency (e.g., Santos,2007;Stoel‐

Gammon,1976).

Inthischapter,wewillexaminethedevelopmentofprosodicwordsasawindowto

study the acquisition of word stress in EP. The research questions that we pose, in this

chapter, are: what can the development of stress patterns tell us about the acquisition of

wordstressinEP?

Despite the conflicting results previously found, both cross‐linguistically and in

148

Portuguese, in Chapter 2 we hypothesized that Portuguese children will display an early

trochaictendency.Furthermore,westatedthat, ifourhypothesis isconfirmed,weexpectto

find:

(i) Anearlyproductionoftargettrochees;

(ii) Truncationof/WS/wordsto[S];

(iii) Truncationof/WSW/to[SW];

(iv) Laterproductionof/WWS/and/SWW/words.

Onthebasisoftheresults,theimplicationsonhowPortuguesechildrenacquirethe

stress algorithm of the target language will be addressed, discussing aspects such as the

presumeddefaultstresspatternintheprocessofprosodicacquisition.Theimplicationsfora

theoryofphonologicalacquisitionwillalsobediscussed,namelyconsideringatop‐downora

bottom‐upapproachtophonologicalacquisition.

Thischapterisorganizedasfollows.Insection5.1.,wewillpresenttheresultsforthe

acquisitionofstresspatterns,basedofthespeechproductionsofthefivechildrenintroduced

inChapter3.Wewillpresentthreedifferenttypesofdataanalysis:firstly,wewillinvestigate

thechildren'sproductionpatterns,irrespectiveofthetargetword(section5.1.1.);secondly,

wewill account for the children's productionswith respect to the targetword shapes and

stresspatterns(/S/,/SW/,/WS/,/WSW/,/WWS/and/SWW/‐section5.1.2.).Thirdly,we

will investigate theproduction strategiesusedby the children in the same targetwordsas

mentioned in section 5.1.2., with special attention to reduplication, epenthesis and

truncation. In section 5.2., we will discuss the results found. In this section we will first

proposeafour‐stagedevelopmentalpathfortheacquisitionofstresspatternsinEP(section

5.2.1.); secondly, we will provide empirical and theoretical arguments that disfavor the

assumptionofadefaultiambicfootintheacquisitionofEP.Thirdly,wewilldiscusstheshape

of early words in the five Portuguese children observed, hypothesizing on an interaction

betweenprominence fromhigherprosodicdomains (such as thephonological phrase) and

theprocessinglowerprosodicdomains(suchasthesyllable).Finally,insection5.2.4.,wewill

proposeanddiscussaneutralstartapproachfortheacquisitionofwordstressinEP.Section

5.3.willsummarizethemainfindingsofthischapter.

149

5.1.Results

Inthischapterwewillpresenttheresultsontheacquisitionofwordshapeandstress

patterns158 in the speech of five Portuguese children. In order to do so, we analyzed (i)

children's production patterns (5.2.1), (ii) children's faithful productions (5.2.2.) and (iii)

children'sproductionstrategies(5.2.3.)regardingwordshapeandstresspatterns.Children's

production patterns concerns the children's preferences in production, irrespective of the

targetform.Faithfulnessisrelativetothechildren'starget‐likeproduction,andthestrategies

regardtheproductionsofchildrenwhentarget‐likeformswerenotproduced.Intheresults

for the production patterns, absolute values will be showed, whereas in the results for

faithfulness and strategies the children’s productions will be accounted for in the form of

percentagesof thestructurebeinganalyzed.Eventhoughsomeoverlappingandredundant

informationcanbepresentintheresultsfortheproductionpatternsandthepercentagesof

faithfulness,weconsiderthatthecombinationofthesetwomethodsaccountsmorereliably

forthedevelopmentpathofchildrenwithrespecttostresspatterns.

5.1.1.Productionpatternsinearlywords

Inthissection,wewillpresenttheresultsfortheproductionpatternswithrespectto

[S],[SW],[WS]and[WSW]wordshapesinthespeechofthefiveobservedchildren.

Firstly, we will present the results from the early sessions in the form of adapted

Guttmanscales159(Tables16‐20),sinceproportionalmeasurementssuchaspercentagesdo

notrealisticallyaccountforthedataintheearlystages,whenveryfewtypesandtokensare

produced(5.1.1.1.).Secondly,wewillconductananalysisontheearlywordsproducedbythe

children,bycomparingtheproductionofreduplicatedandnon‐reduplicatediambs,aswellas

productions resulting from the epenthesis, and theother earlyword shapesproduced ([S],

[SW])(5.1.1.2.).

158 Sincewe considered (stressed)monosyllables in our analysis and these arenot unanimously considered as'stresspatterns',inthecourseofthischapterwewillreferto'wordshapeandstresspatterns'everytimewewanttoincludemonosyllablesintheanalyses.159 A Guttman scale (from Guttman, 1980) accounts for binary answers, ranked in a table. Here, we used thenumberoftokens,ratherthanabinarysymbol(yes/no,X/Ø,etc).

150

5.1.1.1.Wordshapepreference

ThenumbersinthecellsoftheGuttmanscalesaccountforthestresspatternsactually

produced, irrespective of the target form. The purpose of these tables is to show the

children’spreferenceandnottheacquisitionofagivenwordshape,sinceweassumethat,in

theearlystagesofspeechproduction,children’sgeneralpreferenceforaspecificwordshape

mayindicatewhichwordtemplateorprosodicconstituentisbeingprocessed.

In the Guttman scales presented below, the cells represent the number of tokens

producedbyeachchildperwordshape([S],[SW],[WS]and[WSW]160).[SWW]and[WWS]

formswere not taken into account, due the reduced number of thesewords shapes in all

children'sspeech,especiallybefore2;6.Thesetablesincludeallproductionsofagivenword

shape,eitherthatconstitutesatarget‐likeproduction,atruncatedformorit istheresultof

reduplicationor epenthesis. Light shaded cells represent thedevelopmental path, after the

consistentproductionofagivenwordshape.By ‘consistent’,wemeanthat thechildhadto

produce, in the course of aminimum of 3 subsequent sessions at least 1 token of a given

pattern.Cellswithonly1tokenofanywordshapewerenevershaded,asitcouldcorrespond

toasporadicutterance.Thedarkshadedcolumnineachtablerepresentsthemomentwhen

the number of [SW] words surpasses the number of [WS] words produced. This turning

point161– fromapredominanceof [WS]wordstoapredominanceof [SW]–appears in the

fivechildren162.

Table16represents thenumberof tokensproducedbyClaraperwordsession,per

wordshape.

S1 S2 S3 S4 S5 S6 S7 S8 S9 S10[WSW] ‐ ‐ ‐ 1 ‐ 1 4 ‐ 4 9[SW] 1 ‐ 3 ‐ 3 4 5 10 11 61[S] ‐ 1 5 7 6 7 23 14 20 46[WS] 3 2 3 7 7 10 24 9 15 34

Table16.Numberoftokenspersession,perwordshape–actualproductions(Clara)

In Clara, we observe that, at the beginning, she has a preference for [WS] words.

Monosyllables emerge after [WS]. In an early moment (until session 5), Clara produces

mostly [WS] and monosyllables. Later (after session 5), [SW] words emerge and in the

160Inthissection,[S]standsforamonosyllable.161 This turning point is actually an importantmoment that we will also observe in the results presented forfaithfulness(5.1.2.).162AlthoughintheGuttmanscaleswecanobserveapointwhentrocheessurpassiambs,wewillseethatinthefaithful productions (section 5.1.2.), this turning point does not appear, probably due to the fact that Clara’sspeechisonlyavailableuntil1;10.

151

followingsessions(6,7,8and9)theyarestilllessfrequentthan[WS]and[S].Insession10,

[SW]wordsbecomepredominant.Session10thusrepresentsapotentialturningpointinthe

child'sspeech.[WSW]wordsemergeinsession9andarescarceinClara'sspeech.

In (112),wepresentClara's emergencepath for [WS], [S], [SW]and [WSW], in this

sameorder.

(112) Clara–orderofemergenceofwordshapes:

Orthogr. Gloss Target Output Ageolá 'hello' /ç»la/ [å»la] 0;11.1(S1)dá 'giveimp.' /»da/ [å»da] 1;0.13(S2)[WS]bebé 'baby' /bE»bE/ [Bˆ»Be]/[på»pWå] 1;1.3(S3)não 'no' /»nå‚w‚/ [»nå‚w‚]dá 'giveimp.' /»da/ [»d´] 1;1.3(S3)

dá 'giveimp.' /»da/ [»ta] 1;2.22(S4)[S]

carro 'car' /»ka{u/ [»ka]água 'water' /»agWå/ [»a˘Bå] 1;3.6(S5)

água 'water' /»agWå/ [»abWå] 1;4.19(S6)[SW]mano 'brotherfam.' /»månu/ [»månu] 1;6.6(S8)Aurora 'name' /aw»RçRå/ [å˘»jçjå] 1;7.11(S9)menina 'girl' /mˆ»ninå/ [m廯inå][WSW]sapato 'shoe' /så»patu/ [tu»patu] 1;8.20(S10)

Table 17 summarizes early words' production in Inês, according to the criteria

mentionedforClara'sdata.

S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12

[WSW] ‐ ‐ 2 3 ‐ 11 3 13 12 50 83 87[SW] ‐ 2 5 4 6 2 3 27 99 198 241 290

[S] 3 20 60 63 61 148 139 127 135 219 284 508[WS] 11 28 36 74 48 137 113 60 117 116 151 139

Table17.Numberoftokenspersession,perwordshape–actualproductions(Inês)

Table17indicatesthatInês'firstwordshavea[WS]form(cf.sessions1‐9).Insession

2, monosyllables emerge and become very frequent in her speech. Monosyllables are the

predominant format in 9 out of 12 sessions). [SW]words emerge aftermonosyllables, but

only from session 8 onwards they are producedwith higher frequency. In session 10, the

number of [SW] tokens surpasses the number of [WS] tokens in Inês's speech (198words

produced as [SW], contra 116 words produced as [WS]). Thus, session 10 represents a

turning point in the child's speech, with respect to the development of word shapes.

Trisyllablesemergeinsession6.

152

In (113) we present Inês' data, concerning the emergence of word shapes. The

followingorderofemergenceisattested:[WS]>[S]>[SW]>[WSW].

(113) Inês–orderofemergenceofwordshapes:

Orthogr. Gloss Target Output AgeFernanda 'name' /fˆR»nå‚då/ [i»å‚˘]mamã 'mommy' /må»må‚/ [må»må‚]bebé 'baby' /bE»bE/ [B´»BE] Inês 'name' /i»neS/ [ne»ne]

0;11.14(S1)

mamã 'mommy' /må»må‚/ [å»må‚]mamã 'mommy' /må»må‚/ [må»må‚]dá 'giveimp.' /»da/ [a»dJa]

[WS]

não 'no' /»nå‚w‚/ [ˆ»nå‚w‚]

1;0.25(S2)

Inês 'name' /i»neS/ [»¯˘e]toma 'takeimp.' /»tçmå/ [»tç]mamã 'mommy' /må»må‚/ [»må‚]dá 'giveimp.' /»da/ [»da˘]

1;0.25(S2)

Mário 'name' /»mariu/ [»må]chupeta 'pacifier' /Su»petå/ [»pe]cão 'dog' /»kå‚w‚/ [ga]

[S]

Isabel 'name' /izå»bE…/ [»bE]

1;1.30(S3)

chupeta 'pacifier' /Su»petå/ [»pepe]Isabel 'name' /izå»bE…/ [»BEBE] 1;1.30(S3)

não 'no' /»nå‚w‚/ [»¯a¯å]Inês 'name' /i»neS/ [»¯e¯e] 1;3.6(S4)

papa 'foodfam.' /»papå/ [»papa]Fernanda 'name' /fˆR»nå‚då/ [»nånå] 1;4.9(S5)

pé 'foot' /»pE/ [pE»pEpE]

[SW]

girafa 'giraffe' /Zi»Rafå/ [ƒa»ƒaƒ˘a] 1;5.11(S6)

aqui 'here' /å»ki/ [å»kJitJ´]leitinho 'milkdim.' /låj»ti¯u/ [å»tJi¯e] 1;5.11(S6)

não 'no' /»nå‚w‚/ [å»nono][WSW]

pé 'foot' /»pE/ [pE»pEpE] 1;6.11(S7)

Table 18 summarizes João's preferences in the development of [WS], [S], [SW] and

[WSW],withthissameorderofemergence.

153

Table18.Numberoftokenspersession,perwordshapes–actualproductions(João)

Itisworthnoticingthat,inJoão’scase,[S]and[WS]wordsseemtoco‐occurfromthe

beginningoftheobservationuntilsession15.Despiteanearlyemergenceof[WS]words(cf.

sessions2and3),monosyllableshaveproductionratesveryclosetotheproductionratesof

[WS]words, fromsession4 tosession6.Fromsession7 tosession10,monosyllablesstart

beingproducedinhigherratesthan[WS].Fromsession11to14,[WS]wordsprevailover[S]

again.[SW]wordsemergeinsession7andhavelowproductionratesuntilsession15.From

session15onwards,[S]wordsgenerallysurpass[WS].Insession16,thereisanevenamount

of[SW]and[WS]but,fromsession17onwards,[SW]wordshighlysurpasstheproductionof

[WS]words.Trisyllablesonlyoccurlater,insession15.

In (114) we present instances of João's productions for the word shapes under

observation ([S], [WS], [SW] and [WSW]). These instances illustrate the following order of

emergence:[WS]>[S]>[SW]>[WSW].

(114) João–orderofemergenceofwordshapes:

Orthogr. Gloss Target Output Agedá 'giveimp.' /»da/ [å»dA]olá 'hello' /ç»la/ [ç»a] 1;0.28(S2)

água 'water' /»agWå/ [å»wa]dá 'giveimp.' /»da/ [da»da]/[å»da]

[WS]

olá 'hello' /ç»la/ [¥å»¥a]1;1.12(S3)

mamã 'mommy' /må»må‚/ [»må] 1;1.28(S4)bola 'ball' /»bçlå/ [»bå˘][S]bolacha 'cookie' /bu»laSå/ [»bå] 1;2.13(S5)

papa 'foodfam.' /»papå/ [»papå] 1;3.21(S7)bolo 'cake' /»bolu/ [»bowu] 1;4.17(S8)uva 'grape' /»uvå/ [»du˘då] 1;5.12(S9)bola 'ball' /»bçlå/ [»boa] 1;5.26(S10)panda 'pandabear' /»på‚då/ [»papa] 1;7.0(S13)

[SW]

Cidália 'name' /si»daliå/ [»dajå] 1;7.20(S13)Adriana 'name' /å»dRiånå/ [å»nånå] 1;8.25(S15)crocodilo 'aligator' /kRuku»dilu/ [di»lilu][WSW]quentinho 'warmdim.' /ke‚»ti¯u/ [ti»tinu] 1;9.25(S16)

S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 S13 S14 S15 S16 S17 S18 S19 S20

[WSW] ‐ ‐ 1 ‐ ‐ ‐ ‐ ‐ 1 1 1 2 ‐ ‐ 3 33 37 52 48 61[SW] 1 ‐ ‐ 5 ‐ ‐ 4 1 7 7 11 11 8 8 6 41 63 94 82 101[S] ‐ 1 ‐ 3 7 7 33 26 33 35 29 23 37 15 37 52 60 25 67 57[WS] ‐ 2 9 4 7 8 6 6 5 31 32 46 55 37 21 41 25 44 32 25

154

InTable19wepresenttheresultsofJoana'sdevelopmentintermsof[S],[WS],[SW]

and[WSW]words'production.

S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 S13 S14

[WSW] ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 1 5 14 5 14 74[SW] ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 3 9 20 102 90 198

[WS] ‐ 2 1 ‐ ‐ ‐ 1 3 20 15 49 16 24 57[S] 2 2 4 7 2 2 23 9 35 81 104 105 88 232

Table19.Numberoftokenspersession,perwordshape–actualproductions(Joana)

Joana’sdata (Table19) indicate that shehasanearlypreference formonosyllables,

andthatthispreferenceismaintaineduntilsession8.Thefirst[WS]and[SW]wordsoccur

laterindevelopment(bothwordshapesappearconsistentlyinsession8and9,respectively).

However,thechilddisplaysanacquisitionpathforwordshapessimilartotheoneexhibited

bytheotherchildreninthesensethatsheprefers[WS]wordsover[SW]:[S]>[WS]>[SW]>

[WSW].Itisworthnoticing,aswell,thatsession12representsaturningpointinthechild's

speech, as that is the session where a preference for [SW] over [WS] is observed. The

instancesin(115)illustratetheabove‐mentionedorderofemergenceinJoana'sspeech([S]>

[WS]>[SW]>[WSW]).

(115) Joana–orderofemergenceofwordshapes:

Orthogr. Gloss Target Output Agemamã 'mommy' /må»må‚/ [»må] meia 'sock' /»måjå/ [»mQ‚] 1;2.7(S3)

pé 'foot' /»pE/ [»pE](não)há 'thereis(not)' /»nå‚w‚»a/ [»¯Q˘] 1;2.29(S4)

não 'no' /»nå‚w‚ [»nQw] 1;5.5(S6)gato 'cat' /»gatu/ [»ka]

[S]

Carla 'name' /»kaRlå [»ta˘] 1;6.24(S7)

bola 'ball' /»bçlå [å»ç]há 'thereis' /»a/ [a»a] 1;8.4(S8)

pato 'duck' /»patu/ [pa»pa]Raquel 'name' /{å»kE…/ [kå»kE]sapato 'shoe' /så»patu/ [pa»pa]Joana 'name' /Zu»ånå/ [ˆ»¯å]

[WS]

escola 'school' /S»kçlå/ [kç»kç]

1;9.25(S9)

colo 'lap' /»kçlu/ [»kç˘u]papel 'paper' /på»pE…/ [pˆ»e˘u [SW]barco 'boat' /»baRku/ [»ma˘ku]palitos 'sticks' /på»lituS/ [pˆ»kika]

[WSW]chapéu 'hat' /Så»pEw/ [å»pHEju]

1;10.22(S10)

155

InTables20and21,wepresentLuma'sdevelopmentalpathfor[S], [WS], [SW]and

[WSW]outputforms163.

Table20.Numberoftokenspersession,perwordshape–actualproductions(Luma)

S25+26 S27+28 S29+30 S31+32 S33+34 S35+36[WSW] ‐ 6 5 45 34 101[SW] 2 9 62 233 399 382[WS] 117 146 198 292 344 328[S] 32 115 257 437 387 423

Table21.Numberoftokenspersession,perwordshape–actualproductions(Luma)­Continuation

InLuma,[S]and[WS]seemtoemergesimultaneouslyinherspeech(cf.session2‐4).

Nevertheless,itispossibletoobservethatshefavors[S]wordsintheearlymomentsofword

production (cf. sessions 5‐16). Though [WS] words emerge earlier than [SW] words (in

session2and11,respectively),fromsession11to14,thereisanalternatingpreferencefor

[SW]or[WS]words.Fromsession15,untilsession32,[S]and[WS]wordspredominate in

Luma's speech, though [WS]words are generallyproduced inhigher amounts (cf. sessions

17‐28).Insession33,theproductionratefor[SW]wordssurpassestheproductionratefor

[WS](though[S]wordsareproducedinhighratesuntiltheendoftheobservationperiod).

Trisyllablesemergeinsession27.

Luma’s early development follows the order of emergence [S] > [WS] > [SW] >

[WSW]. Despite the criterion for consistency presented165 and the observation of a

emergence path similar to the one observed in Joana (earlier emergence of and a higher

productionrate in[S]wordsthanin[WS]and[SW],attheearlystages),weareawarethat

Lumaisthelessconsistentchildofthefivechildrenoberved.

In(116)weshowLuma'semergencepathinthefollowingorder:[S]>[WS]>[SW]>

[WSW].

163 Often, Lumahad sessionswhere a considerable amount of a specificword shapewasproduced and in thefollowingsessiontheamountoftokensofthatsamewordshapedecreaseddramatically.Forthatreason, intheGuttmanscales,wedecidedtogroupsessionsandsumupthenumberoftokensofeachsession.164Session1inLumadoesnothaveanyoftheanalyzedwordshapes.165Cf.Chapter3,section3.3..

S2164 S3+4 S5+6 S7+8 S9+10 S11+12 S13+14 S15+16 S17+18 S19+20 S21+22 S23+24

[WSW] ‐ ‐ 1 4 ‐ ‐ ‐ ‐ ‐ 2 1 4[SW] ‐ 3 2 2 ‐ 3 15 3 ‐ 6 11 2[WS] 1 9 1 19 2 8 5 40 55 81 85 130[S] 1 10 13 55 18 7 41 101 28 38 60 38

156

(116) Luma–orderofemergenceofwordshapes.

Orthogr. Gloss Target Output Agedá 'giveimp.' /»da/ [»da]mamã 'mommy' /må»må‚/ [»nå‚] 1;0.13(S2)

[S]dá 'giveimp.' /»da/ [»de˘] 1;2.22(S6)dá 'giveimp.' /»da/ [a»tJa] 1;0.28(S3)Hopla 'name' /»çplå/ [pa»pa] 1;1.10(S4)mamã 'mommy' /må»må‚/ [må»må‚] 1;3.5(S7)dá 'giveimp.' /»da/ [å»da] 1;3.19(S8)

[WS]

dá 'giveimp.' /»da/ [da»da] 1;4.2(S9)banana 'banana' /bå»nånå/ [»nånå] 1;5.9(S11)mamã 'mommy' /må»må‚/ [»må‚må] 1;6.20(S14)Susana 'name' /su»zånå/ [»nånå] 1;7.5(S15)Lala 'name' /»lalå/ [»nånå] 1;9.29(S20)

[SW]

pato 'duck' /»patu/ [»atu] 1;10.18(S21)Noddy 'name' /»nçdi/ [tçR»¥çRi] 2;2.4(S28)Francisco 'name' /fRå‚»siSku/ [a»tidˆ˘] 2;3.26(S30)[WSW]papéis 'papers' /på»pEjS/ [på»pejSi] 2;4.11(S31)

In sum,our results indicated threedifferentdevelopmental trends in the children's

productionsforwordshape,whichwesummarizeinthefollowingtable:

Children Developmentalpathforthewordshapesproduced

Clara,InêsandJoão [WS]>[S]>[SW]>[WSW]

Joana [S]>[WS]>[SW]>[WSW]

Luma [S]~[WS]>[SW]>[WSW]

Table22.Individualdevelopmentalpathforthewordshapesproduced

Theresultsofthepreferenceofthefivechildrenregardingwordshapeindicatethat,

initially,childrenmayfavorboth[S]and[WS]. [SW]wordsemerge laterthan[S]and[WS].

[WSW]wordsemergelaterthanallotherwordshapesindevelopment.

5.1.1.2.Theshapeofearlywords

The literature has suggested that the early speech of Portuguese‐speaking children

displaysaheavytendencyforreduplicationandepenthesis(Baia,2006,2008;Santos,2001,

2007). Ithasalsobeensuggested thatmostof those reduplicationsnormallyhavean [WS]

pattern.

157

Ananalysisconductedtoaccountfortheamountofiambicortrochaicreduplications

observed in speech of the five children observed in this dissertation confirmed a heavy

tendencyfor[WS]reduplications,asshowninFigures24‐28,wheretheabsolutenumberof

[SW]and[WS]reduplicationsispresented,perchild:

Figure24.[SW]and[SW]reduplications(Clara)

Figure25.[SW]and[WS]reduplications(Inês)

Figure26.[SW]and[WS]reduplications(Joana)

158

Figure27.[SW]and[WS]reduplications(João)

Figure28.[SW]and[WS]reduplications(Luma)

Thefigurespresentedaboveshowatendencyfor[WS]reduplicationsinthespeechof

the five children observed. Even though some variation in the amount of reduplications is

attestedbetweenchildren,andasfurthermoredemonstratedintheinstancesforearlyword

shape(cf.examplesin(112)‐(116)),thefivechildrenunderobservationusereduplicationsin

theirspeechandtheyareoverwhelmingly[WS].

Inordertoinvestigatethenatureoftheearly[WS]words‐eitherthesewordshave

thesamedistributionastheother,non‐reduplicated,disyllables‐,wecarriedoutananalysis

where we zoomed into the early [WS] words produced by the five children under

observation. In thisanalysisweaccounted for theamountof [WS]non‐reduplicatedwords,

[WS]reduplicatedwordsandcasesofepenthesiscreatingan[WS]pattern,andcomparedit

withbothmonosyllablesandtrochees.Therefore,wedistinguished:

(i) Monosyllables([S]);

(ii) Iambs([WS]–e.g.,olá ‘hello’/ç»la/,balão ‘balloon’/bå»lå‚w‚/)‐here,weexclude

all reduplicatedwords andwords resulting from epenthesis at the left edge of

words;

159

(iii) Iambic reduplications ([S2]166 – e.g., papá ‘daddy’ /pa»pa/) ‐ here, non‐

reduplicated/WS/wordsareexcluded;

(iv) Iambs 'in disguise' resulting from the circumscription of the stressed syllable

precededbyafillersyllable([fS]–e.g.,creme'lotion'producedas[ˆ»kE]);

(v) Trochees([SW]–e.g.,casa‘house’).

Inthisanalysis,wetookintoaccounttheactualproductionsandnotthefaithfulones.

Therefore,thewordsanalyzedareirrespectiveofthetargetform,eitheritisreduplicatedor

not.

We compared data from two differentmoments: the early sessions of the Guttman

scales167(designatedas'Early'intheFiguresbelow),andalatersession,i.e.,thefirstsession

aftertheturningpointtoatrochaicpredominanceinthefaithfulproductions(designatedas

'Later'intheFiguresshownbelow).

Figure 29 shows the distribution of monosyllabic productions (S), iambic non‐

reduplicated forms(WS), trochaic forms(SW),syllablesprecededbya fillersound(fS)and

reduplications (S2), in two observational periods of Clara's speech. In Clara, the 'Early'

sessions correspond to her speech productions between sessions 1 and 10. The 'Later'

momentcorrespondstotheproductionsofsession12.

Figure29.Percentageofwordshapesproduced­Earlyvs.laterstages(Clara)

InClaraweobservethatinthefirstsessions(Figure29'Early')thereisapreference

for monosyllables (38%). Non‐reduplicated iambs are more frequent than trochees (22%

166Forspace‐savingreasonsinthefigures,inthissectionreduplicationsofthetype[CV1CV1]willbereferredtoas[S2].167Cf.section5.1.1..

160

contra17%,respectively).Epenthesisandreduplicationsresultinginiambicproductionsare

possiblein14%and9%ofthechildren'sproductions,respectively.

In the last session (Figure 29 'Later'),monosyllables are still preferred (35%), but

epenthesisisnolongerausedstrategy(0%)andtheremainingstresspatterns([WS],[SW)

and[S2])have increasedvalues(26%,22%and17%respectively). In the lastsession,non‐

reduplicatediambs,[SW]and[S2]arestillproducedinsimilaramounts,thussuggestingthe

absenceofapreferenceforanyofthem.

Examplesin(117)illustratetheproductionofreduplicationandepenthesisinClara's

speechinanearlierstage.Notethatmanyproducedreduplicationsarealreadyreduplicated

targets.

(117) Clara–earlyproductionofreduplicateddisyllablesandepenthesis:

Orthogr. Gloss Target Output Agebebé 'baby' /bE»bE/ [på»pWå] 1;1.3(S3)não 'no' /»nå‚w‚/ [nå»nå‚w‚]bebé 'baby' /bE»bE/ [be»be]está '(it)is' /S»ta/ [tå»ta]papá 'daddy' /på»pa/ [pa»pa]

1;2.22(S4)

não 'no' /»nå‚w‚/ [nå»nå‚w]bebé 'baby' /bE»bE/ [be»be˘] 1;3.6(S5)

Redupl.

bebé 'baby' /bE»bE/ [be»be] 1;7.17(S9)dá ‘give(me)’ /»da/ [å»da] 1;0.13(S2)não 'no' /»nå‚w‚/ [å»nå‚w‚]pinguim 'penguin' /pi‚»gWi‚/ [å»pe]bebé 'baby' / bE»bE/ [å»Be]

1;3.6(S5)

cão 'dog' /»kå‚w‚/ [ˆ»kå‚w‚]bebé 'baby' /bE»bE/ [å»pe] 1;4.19(S6)

pé 'foot' /»pE/ [a»pE]

Epenthesis

pai 'father' /»paj/ [Q»paj] 1;5.16(S7)

Intheexamplespresentedabove,weobservethatbothreduplicationandepenthesis

arestrategiesusedintheproductionoftargetreduplicatedwordsortargetmonosyllables.In

Clara'sspeech,targettrocheesarenotsubjecttoreduplicationand/orepenthesisstrategies.

InFigure30,wepresentInês'productionofmonosyllables,iambicnon‐reduplicated

words, trochees, iambic reduplications and epenthesis, in twomoments. In Inês' data, the

'Early'stageiscomprisedbetweensessions1and4.The'Later'momentcorrespondstoher

productionsinsession10.

161

Figure30.Percentageofwordshapedproduced­Earlyvs.laterstages(Inês)

Inês is a showcaseof the ‘disguised iambs’ thatmakePortugueseearlyword shape

acquisitionappear iambic. Figure30 'Early' shows that thepercentageofnon‐reduplicated

iambsishardlytraceable(3%).Intheearlystageofwordproduction,monosyllablesprevail

(48%), and non‐reduplicated iambs and trochees are scarce (3% and 4%, respectively). In

Figure 30 'Later', there is a change in the distribution of stress patterns in Inês' speech.

Reduplicated disyllables are only 12% and trochees predominate amongst the disyllables

(37%). Non‐reduplicated iambs are scarce (8%) and monosyllables are still predominant

(41%).

In(118),someinstancesofreduplicateddisyllablesandepenthesiswitha[WS]form

areshowninInês'earlyspeech(untilsession4).

162

(118) Inês–earlyproductionofreduplicateddisyllablesandepenthesis:

Orthogr. Gloss Target Output Agemamã 'mommy' /må»må‚/ [må»må‚]Inês 'name' /i»neS/ [ne»ne]bebé 'baby' /bE»bE/ [B´»BE]

0;11.14(S1)

mamã 'mommy' /må»må‚/ [ma»må‚] 1;0.25(S2)má 'bad' /»ma/ [må»ma]Inês 'name' /i»neS/ [ni»ni]má 'bad' /»ma/ [må»ma]

1;1.23(S3)

babete 'bib' /bå»bEtˆ/ [ba»ba˘]Teresa 'name' /tˆ»Rezå/ [ti»tJi]dá 'giveimp.' /»da/ [da»da]não 'no' /»nå‚w‚/ [nç»na]bóia 'buoy' /»bçjå/ [Ba»Bå]Bambi 'name' /bå‚bi/ [må»mE˘]

1;3.6(S4)

vestido 'dress' /vˆS»tidu/ [tJi»tJi]

Redupl.

cueca 'panties' /ku»Ekå/ [kE»kE] 1;5.11(S6)

dá 'giveimp.' /»da/ [/»da] 0;11.14(S1)mamã 'mommy' /må)»må)/ [å»må)] dá 'giveimp.' /»da/ [a»dJa]não 'no' /»nå‚w‚/ [´»¯´]

1;0.25(S2)

Mário 'name' /»maRiw/ [´»må] chupeta 'pacifier' /Su»petå/ [å»be]Isabel 'name' /izå»be…/ [E»be]mais 'more' /»majS/ [Q»ma]

1;1.30(S3)

babete 'bib' /bå»bEtˆ/ [å»BE] 1;3.6(S4)não 'no' /»nå‚w‚/ [廯å]

Epenthesis

banho 'bath' /»bå¯u/ [å»bå] 1;5.11(S6)

In Inês'speech,weobserve thatboth/S/,/WS/,/SW/andeven/WSW/wordsare

prone to reduplication and epenthesis strategies. As demonstrated in Figure 30, non‐

reduplicatediambsarescarceinthischild'searlyspeech.

In Figure 31, we show Joana's productions for monosyllables, non‐reduplicated

iambs, trochees, a monosyllable preceded by epenthesis and reduplications, in two

observation moments: the 'Early' moment corresponds to the speech productions uttered

until session 10. The 'Later' moment corresponds to the speech productions uttered in

session12.

163

Figure31.Percentageofwordshapesproduced­Earlyvs.laterstages(Joana)

In Joana's speech we observe a high predominance of monosyllables in the early

sessions (Figure 31 'Early').Monosyllables are 61%of Joana's early productions, and non‐

reduplicated iambs and trochees are distributed evenly (11% vs. 9%, respectively).

Reduplicateddisyllablescorrespondto15%ofJoana'searlyproductionsandarethesecond

most frequent word shape. Session 12 (Figure 31 'Later') marks a clear change in the

distribution of stress patterns in Joana's speech. In this session, monosyllables are still

predominant(49%),buttrocheesbecomemorepresent(44%).Non‐reduplicatediambsare

scarce(7%)andreduplicationsandepenthesesarenolongerproduced.

Joana'sproductionofreduplicateddisyllablesisexemplifiedin(119).

(119) Joana–earlyproductionofreduplicateddisyllablesandepenthesis:

Orthogr. Gloss Target Output Agemamã 'mommy' /må)»må)/ [må)»må)] 1;2.7(S3)bebé 'baby' /bE»bE/ [bE»Bi˘] 1;6.24(S7)pato 'duck' /»patu/ [pa»pa]Raquel 'name' /{å»kE…/ [kå»kE]sapato 'shoe' /så»patu/ [pa»pa]Bobby 'name' /bç»bi/ [på»pE˘]padrinho 'godfather' /på»dRi¯u/ [pa»pa]popó 'carfam.' /pç»pç/ [pç»pç]chá 'tea' /»Sa/ [ta»DJa]

1;9.25(S9)

escola 'school' /S»kçlå/ [kç»kç˘]

Redupl.

bombom 'candy' /bo‚»bo‚/ [bu»bu‚] 1;10.22(S10)

mamã 'mommy' /må)»må)/ [å»må‚] 1;0.25(S2)mamã 'mommy' /må»må‚/ [/»må‚] 1;2.7(S3)Epenthesisnão 'no' /»nå‚w‚/ [ˆ»¯å‚w‚] 1;1.22(S10)

164

The instances from Joana presented in (119) show the same behavior observed in

Inês: iambic reduplications and epenthesis may be observed in the production of target

monosyllables,targetreduplicatediambs,targettrocheesandlongerwords.

InFigure32,wepresentJoão'spercentagesforthedifferentwordshapesproduced:

monosyllables,non‐reduplicated iambs, trochees,amonosyllableprecededbya fillersound

andreduplications.Thefirstmomentofobservationwascomprisedbetweensessions1and

9,whereasthesecondmomentofobservationwassession17.

Figure32.Percentageofwordshapesproduced­Earlyvs.laterstages(João)

LikeJoana,João'searlyspeech(untilsession9)ismainlycomposedofmonosyllables

(62%).Reduplicateddisyllablesandepenthesisaccountfor10%ofhisspeech(2%and8%,

respectively),whereastrocheesandnon‐reduplicatediambsareevenlyproduced(12%and

16%, respectively). In session 17, the stress pattern distribution changed dramatically:

monosyllables are now only 40% of João's speech, and trochees increased significantly

(43%),contraalowproductionofnon‐reduplicatediambs(9%).Epenthesisandreduplicated

iambsarescarce(1%and7%,respectively).

In(120),wepresentJoão'srenditionsofreduplicationsandcasesofepenthesis.We

will first present the examples of epenthesis, as thiswas the strategymostly used by João

(reduplicationwasrare).

165

(120) João–earlyproductionofreduplicationandepenthesis:

Orthogr. Gloss Target Output Agedá 'giveimp.' /»da/ [å»da] 1;0.28(S2)dá 'giveimp.' /»da/ [a»da]água 'water' /»agWå/ [ˆ»wå] 1;1.12(S3)

bolacha 'cookie' /bu»laSå/ [å»bwa] 1;2.13(S5)mamã 'mommy' /må»må‚/ [å»må‚]bolacha 'cookie' /bu»laSå/ [å»bå] 1;2.30(S6)

Epenthesis

papa 'foodfam.' /»papå/ [a»pa] 1;3.21(S7)olá 'hello' /ç»la/ [¥å»¥a] 1;1.12(S3)melão 'mellon' /mˆ»lå‚w‚/ [lå»lå‚w‚]maçã 'apple' /må»så‚/ [må˘»må] 1;4.17(S8)

dá 'giveimp.' /»da/ [då»då˘] 1;5.12(S9)Redupl.

popó 'carfam.' /pç»pç/ [pa»pa] 1;5.26(S10)

In João's speech, the strategies of reduplication and epenthesis mainly occur with

iambictargetwords(though,occasionally,some/‐SW/wordslikebolacha'cookie'/bu»laSå/,

maybesubjecttothosestrategies).

Figure33showsthepercentageofwordshapesproducedbyLumaintwomoments.

Thefirstoneregardsherspeechproductionsuntilsession30.Thesecondonecorrespondto

Luma's speechproductions in session33.Monosyllables,non‐reduplicated iambs, trochees,

productions of monosyllables preceded by a filler sound and iambic reduplications were

analyzed.

Figure33.Percentageofwordshapesproduced­Earlyvs.laterstages(Luma)

Luma is the child who produced more monosyllables in the early stages of word

production (74%). She has a higher percentage of production for trochees (14%) than for

non‐reduplicated iambs (1%). Iambic reduplications and epenthesis account for 11% of

166

Luma'searlyproductions(6%and5%,respectively).Insession36(Figure33'Later'),Luma

hasamorebalanceddistributionofstresspatterns.Monosyllablesdecreaseto34%,trochees

and non‐reduplicated iambs increase to 34% and 19%, respectively. Though reduplicated

iambsincrease(13%),epenthesisisnolongerastrategybeingused.

In(121)wepresentLuma'sproductionofreduplicateddisyllablesandepenthesis.

(121) Luma–earlyproductionofreduplicationandepenthesis:

Orthogr. Gloss Target Output AgeHopla 'name' /»çplå/ [pa»pHa]bebé 'baby' /bE»bE/ [pa»pa] 1;1.10(S4)

dá 'giveimp.' /»da/ [da»da] 1;4.2(S9)banana 'banana' /bå»nånå/ [nå»nå] 1;5.24(S12)mamã 'mommy' /må)»må)/ [ma»må‚]dá 'giveimp.' /»da/ [da»da] 1;7.19(S16)

Redupl.

bola 'ball' /»bçlå/ [pa»pa] 1;8.2(S17)dá 'giveimp.' /»da/ [å»da] 1;0.13(S2)mamã 'mommy' /må)»må)/ [´»må)] dá 'giveimp.' /»da/ [a»da]Epenthesis

bola 'ball' /»bçlå/ [a»Bo]1;3.19(S8)

5.1.1.3.Summaryfortheproductionpatterns

The data regarding the amount ofmonosyllables, non‐reduplicated iambs, trochees

and 'disguised iambs' ([CV1'CV1] and [fS])producedby the five childrenunderobservation

indicatesthat:

(i) The emergence of word shapes in the speech of the five Portuguese children

observed variesmostly between [S] > [WS] > [SW] and [WSW] (in Joana), and

[WS]>[S]>[SW]and[WSW](inClara,InêsandJoão);Lumahasasimultaneous

emergenceof[S]and[WS],followedby[SW]and,later,[WSW];

(ii) At the early sessions,monosyllabicwords prevail, in the speech of all children

(thedatafromJoana,JoãoandLumaillustratetheoverwhelmingpreferencefor

monosyllabicwordsintheearlystagesofwordproduction);

(iii) Later,theamountofmonosyllablesissimilartotheonefoundfortrochees;

167

(iv) Iambicreduplicationsandepenthesisareverypresentatthebeginningbutthey

tendtodisappear;

(v) Childrenonly vary in thepreferential useof reduplicationor epenthesis: Joana

andInêspreferreduplicatedwords,JoãousesmostlyepenthesisandLumauses

bothinthesameamount;

(vi) Atthebeginning,thepercentageofnon‐reduplicatediambsandtrocheesisvery

similar(exceptinLuma,wherethepercentageoftrocheesishigherthattheone

ofiambs);

(vii) Later,thepercentageoftrocheessurpassesthepercentageofiambsinfourofthe

fivechildrenobserved(Claraistheonlyexception).

5.1.2.Faithfulnesstothetarget

In this section, we will present the results for the acquisition of stress patterns,

havinginmindthechildren’sfaithfulnesstothetarget.By'faithfulness'wemeanthetarget‐

likeproductionofawordtemplateandnotaspecificlexicalitem.Therefore,a[WS]wordlike

balão 'balloon' [bå»lå‚w‚], produced as [bå»wå‚w‚] or [bå»bå] will equally be considered as

faithful. Informationon the contrarywill beprovided,whennecessary.Thedata for target

monosyllables (/S/), target disyllables (/SW/ and /WS/) and target trisyllables (/WSW/,

/WWS/and/SWW/)willbepresented.

5.1.2.1.Monosyllables

Inthissection,wewillpresenttheresultsfortargetmonosyllablesproducedtarget‐

like.Wewillconsideras 'target‐like'a targetmonosyllable thatkeepsonesinglesyllable in

thechildren'sproduction.Targetmonosyllablesproducednon‐target‐likewillbeanalyzedin

section5.1.3.1..

Tables23‐27indicatetheacquisitionpathformonosyllabictargetsinallchildren,in

thesessionsunderanalysis. Instances in(121)‐(125)exemplifychildren'searly faithfulness

towardsmonosyllabictargets.

Table23showsClara'sdevelopmentalpathfortargetmonosyllables.

168

Session /S/ [S] %S1 ‐ ‐ ­S2 1 0 0S3 4 4 100.0S4 7 6 85.7S5 7 5 71.4S6 13 6 46.2S7 27 20 74.1S8 17 11 64.7S9 22 15 68.2S10 34 32 94.1S11 100 86 86.0S12 87 82 94.3

Table23.Monosyllablesproducedtarget­like(Clara)

InTable23,weobservethatClarahasa target‐likerateofmonosyllablesabovethe

50%pointsincesession3.Despiteadecreasingtarget‐likeratefromsession3tosession6to

valuesclose to50%,Clarahasahighpercentageofmonosyllablesproducedaccordingly in

almost all sessions until the end of the observation period, despite some instability is

observed.

In(122),weexemplifyClara'saccuracyintheproductionoftargetmonosyllables.

(122) Clara–monosyllablesproducedtarget‐like:

Orthogr. Gloss Target Output Agenão 'no' /»nå‚w‚/ [»nå‚w‚]dá 'giveimp.' /»da/ [»d´]

1;1.3(S3)

não 'no' /»nå‚w‚/ [»nå‚w‚]dá 'giveimp.' /»da/ [»ta] 1;2.22(S4)

não 'no' /»nå‚w‚/ [»¯ç]cão 'dog' /»kå‚w‚/ [»kå‚w‚]mãe 'mother' /»må‚j‚/ [»må‚˘j‚]

1;4.19(S6)

Table24accountsforInês'target‐likeproductionoftargetmonosyllables.

169

Session /S/ [S] %S1 1 0 0S2 22 13 59.1S3 68 25 36.8S4 112 45 40.2S5 39 27 69.2S6 129 81 62.8S7 80 54 67.5S8 89 64 71.9S9 97 86 88.7S10 136 119 87.5S11 236 200 84.7S12 410 374 91.2S13 259 239 92.3S14 316 288 91.1S15 226 211 93.4S16 454 416 91.6S17 236 218 92.4S18 428 358 83.6

Table24.Monosyllablesproducedtarget­like(Inês)

Inthistable,weobservethreemoments:afirstmoment,betweensession2and4,in

which Inêshas a slight decrease in the target‐like rate (59.1%,36.8%and40%); a second

moment,betweensession5and8, inwhich theproductionratesareabove50%,butsome

instability isstillobserved(percentagevaluesdonotgobeyond75%),andathirdmoment

(fromsession9onwards),inwhichthechildhasproductionvaluesabove75%.

The instances in(123)showthat Inês'productionsofearlymonosyllabicwordsare

generallytarget‐like.

(123) Inês–monosyllablesproducedtarget‐like:

Orthogr. Gloss Target Output Agedá 'giveimp.' /»da/ [»da˘]não 'no' /»nå‚w‚/ [»na]já 'now' /»Za/ [»za]

1;0.25(S2)

cão 'dog' /»kå‚w‚/ [»ga]há 'thereis' /»a/ [»a]pé 'foot' /»pE/ [»pE]mãe 'mother' /»må‚j‚/ [»ma]

1;1.20(S3)

InTable25weshowthedevelopmentalpathofJoanatowardsmonosyllables.

170

Session /S/ [S] %S1 2 2 100S2 1 1 100S3 1 1 100S4 8 8 100S5 3 3 100S6 3 2 66.7S7 7 7 100S8 8 7 87.5S9 21 18 85.7S10 62 52 83.9S11 57 48 84.2S12 59 51 86.4S13 55 49 89.1S14 229 181 74.7

Table25.Monosyllablesproducedtarget­like(Joana)

Weseethatthischildproduceshightarget‐likeproductionofmonosyllablesfromthe

beginningofwordproduction,untiltheendoftheobservationperiod.Insession6,thechild

hasanoccasionaldecreasingvalueintheproductionoftargetmonosyllables(66.7%),butthe

followingsessionsallhavevaluesoftarget‐likeproductionabove75%168.

Instancesin(124)exemplifyJoana'sproductionofmonosyllabictargets.

(124) Joana–monosyllablesproducedtarget‐like:

Orthogr. Gloss Target Output Agemãe 'mother' /»må‚j‚/ [»å‚˘]não 'no' /»nå‚w‚/ [»nå‚˘w‚] 0;11.26(S1)

há 'thereis' /»a/ [»a]pé 'foot' /»pE/ [»pE]pai 'father' /»paj/ [»pQ]

1;2.7(S3)

pai 'father' /»paj/ [»wåj]não 'no' /»nå‚w‚/ [»nQw] 1;5.5(S6)

é '(it)is' /»E/ [»E]há 'thereis' /»a/ [»a˘] 1;6.24(S7)

Table26showsJoão'sacquisitionpathtowardsmonosyllabictargets.

168Aswewillshowfurtherinthischapter,Joanaisachildwheremonosyllabicwords(eitherproducedtarget‐likeor as the result of truncations in multisyllabic target words) are frequently produced during the observationperiod.

171

Session /S/ [S] %S1 ‐ ‐ ­S2 4 1 25S3 1 ‐ 0S4 ‐ ‐ ­S5 ‐ ‐ ­S6 ‐ ‐ ­S7 3 2 66.7S8 1 1 100S9 26 16 61.5S10 13 10 76.9S11 26 12 46.2S12 39 13 33.3S13 18 15 83.3S14 12 10 83.3S15 29 27 93.1S16 41 35 85.4S17 72 55 76.4S18 31 25 80.6S19 68 62 91.2S20 52 48 92.3S21 33 26 78.8S22 71 63 88.7

Table26.Monosyllablesproducedtarget­like(João)

In this table we observe that from session 1 to session 6, João rarely selects

monosyllables.Insession7and8,fewmonosyllablesareselected(3and1,respectively)but

they are mostly produced accordingly. From session 9 to session 12, there is target‐like

production ratenormallyabove50%(though in sessions11and12, theproductionvalues

decrease to 46% and 33%, respectively). After session 13, monosyllables are produced

accordinglyinratesabove75%.

In(125)wepresentJoão'sproductiondatafortargetmonosyllables.

(125) João–monosyllablesproducedtarget‐like:

Orthogr. Gloss Target Output Agedá 'give(me)' /»da/ [»da] 1;0.28(S2)pau 'stick' /»paw/ [»pa˘j] 1;3.21(S7)pau 'stick' /»paw/ [»påw] 1;4.17(S8)mãe 'mother' /»må‚j‚/ [»må‚˘j‚]pá 'shovel' /»pa/ [»pHa] 1;5.12(S9)

dá ‘give(me)’ /»da/ [»da]pá 'shovel' /»pa/ [»pa] 1;5.26(S10)

Table27summarizesLuma'sproductiontowardstargetmonosyllables.

172

Session /S/ [S] %S1 1 0 0S2 2 1 50S3 ‐ ‐ ­S4 ‐ ‐ ­S5 ‐ ‐ ­S6 13 12 92.3S7 32 28 87.5S8 36 12 33.3S9 4 2 50.0S10 1 1 100.0S11 1 1 100.0S12 5 3 60.0S13 21 20 95.2S14 2 2 100.0S15 41 37 90.2S16 7 2 28.6S17 8 4 50.0S18 7 5 71.4S19 22 11 50.0S20 14 8 57.1S21 7 2 28.6S22 2 1 50.0S23 3 2 66.7S24 6 4 66.7S25 14 10 71.4S26 18 9 50.0S27 53 51 96.2S28 59 52 88.1S29 89 85 95.5S30 149 136 91.3S31 177 168 94.9S32 260 227 87.3S33 168 144 85.7S34 181 147 81.2S35 183 162 88.5S36 216 185 85.6S37 124 115 92.7

Table27.Monosyllablesproducedtarget­like(Luma)

Untilsession23,Lumahasahighrateoftarget‐likeproductionofmonosyllablesfrom

the beginning, though some inconsistencies might be found, both in percentual and in

types/tokens terms. For instance, in session8, the childhas36monosyllabicwords inher

intake,(producingonly33.3%ofthemaccordingly)andinthefollowingsession(session10),

sheonlyselectsonetokenforproduction.Fromsession25onwards,thechildstartsselecting

monosyllabicwordsinhigheramountsandsheisabletoproducetheminvaluesabove75%.

The instances presented in (126) illustrate Luma's production of target

173

monosyllables.

(126) Luma–monosyllablesproducedtarget‐like:

Orthogr. Gloss Target Output Agedá 'giveimp.' /»da/ [»da] 1;0.13(S2)dá 'giveimp.' /»da/ [»da] 1;2.22(S6)não 'no' /»nå‚w‚/ [»nå‚w‚] 1;3.5(S7)pau 'stick' /»paw/ [»pa] 1;3.19(S8)Po 'name' /»po/ [»pa] 1;4.2(S9)dá 'giveimp.' /»da/ [»da] 1;5.9(S11)

The results showed in this section indicate that children generally produce

monosyllables in high target‐like rates (above 50%) from the beginning. However, some

variation is found between the five observed children. Joana has a stable production

(generally above75%)ofmonosyllables from theonset ofwordproduction. InClara, Inês,

João and Luma's speech, some instability is normally attested in the first half of the

observation period, in which normally production rates between 50% and 75% are

noticeable.Afterthatmoment,hightarget‐likeproductionratesareobserved.

Thepreferenceformonosyllables,accountedforintheGuttmanscales(section5.1.1.)

fortheearlystages,isattestedbyahigherfaithfulnessrateinallthechildrenobserved.Aswe

will see in section 5.1.3., the strategies used by the childrenwhen they are not faithful to

monosyllables consist in producing di‐ or polisyllabicwords, bymeans of reduplication or

epenthesis.

5.1.2.2.Disyllables

Theresults in thissectionaccount for theproductionof targetdisyllabicwordsand

will be presented in terms of target trochees (/SW/) and target iambs (/WS/) produced

faithfully.

Given the heavy tendency for reduplicated productions and productions where

epenthesis was attested, creating an [WS] pattern, in this section we will account for the

acquisition of target trochees and iambs by distinguishing: (i) the general results for

faithfulness (including reduplicatedwords and productionswith epenthesis) and (ii), non‐

reduplicateddisyllablesanddisyllableswherenoepenthesisisobserved.

InTable28,weshowthegeneralresultsregardingClara'sacquisitionpathfortarget

trochesandiambs.

174

Session /SW/ [SW] % /WS/ [WS] %S1 1 1 100 3 3 100S2 ‐ ‐ ­ 1 1 100S3 4 3 75 9 3 33.3S4 ‐ ‐ ­ 10 6 60S5 4 2 50 8 3 37.5S6 4 3 75 6 2 33.3S7 5 1 20 29 20 69S8 6 6 100 10 7 70S9 10 5 50 23 10 43.5S10 80 59 73.8 35 20 57.1S11 161 103 64.0 82 62 75.6S12 146 76 52.1 142 110 77.5

Table28.Disyllablesproducedtarget­like(Clara)

InClara'sdata,weobserveanunstablebehaviorintheproductionoftargettrochees

andiambsuntilsession6and7,respectively.Insession7,/WS/wordshaveatarget‐likerate

above50%andstartbeingproducedinhigherrates,untiltheendoftheobservation./WS/

words aremore frequent in the child's intake. Until session 8, /SW/words are produced

inconsistently. From session 9 onwards, /SW/ words start having higher target‐like

production rate (above 50%), though some unstabilily is observed in the target‐like

productionvaluesuntiltheendoftheobservation.

The instances in (127) and (128) illustrate Clara's production of target iambs and

targettrochees,respectively.

(127) Clara–iambsproducedtarget‐like:

Orthogr. Gloss Target Output Agebebé 'baby' /bE»bE/ [be»be] 1;2.22(S4)pinguim 'penguin' /pi‚»gWi‚/ [å»pe] 1;3.6(S5)bebé 'baby' /bE»bE/ [å»pe] 1;4.19(S6)bebé 'baby' /bE»bE/ [bE»bE] 1;5.26(S7)João 'name' /Zu»å‚w‚/ [zu»å‚w‚] 1;7.11(S9)cocó 'poo' /kç»kç/ [ko»ko] 1;8.20(S10)memé 'sheepfam.' /mE»mE/ [nˆ»nE] 1;9.23(S11)aí 'there' /å»i/ [å»i] 1;9.23(S11)

175

(128) Clara–trocheesproducedtarget‐like:

Orthogr. Gloss Target Output Ageágua 'water' /»agWå/ [»a˘Bå] 1;3.6(S5)água 'water' /»agWå/ [»ak´] 1;4.19(S6)mano 'brotherfam.' /»månu/ [»månu] 1;5.16(S7)mana 'sisterfam.' /»månå/ [»månå] 1;7.11(S9)pato 'duck' /»patu/ [»patJu] 1;8.20(S10)chucha 'pacifierfam.' /»SuSå/ [»su_sa] 1;9.23(S11)esta 'thisfem.' /»EStå/ [»EtHå] 1;10.15(S12)

In the renditions presented above,we observe that Clara normally produces /WS/

wordsas reduplicationsor she inserts a filler syllableat the left edgeof the circumscribed

stressedsyllable.Also,weobservethatearly/WS/wordsareoftenreduplicatedinthetarget

(e.g.,mamã 'mommy',bebé 'baby',cocó 'poo').Conversely,when/SW/startbeingproduced

as[SW]theynormallycorrespondtononreduplicatedwords([CV1CV2]).

Table29showsInês'acquisitionpathfortargettrocheesandiambs.

Session /SW/ [SW] % /WS/ [WS] %S1 ‐ ‐ ­ 19 8 42.1S2 2 0 0 41 15 36.6S3 4 0 0 64 21 32.8S4 27 1 3.7 67 39 58.2S5 27 1 3.7 56 30 53.6S6 90 1 1.1 110 61 55.5S7 73 3 4.1 93 55 59.1S8 39 5 12.8 80 35 43.8S9 140 82 58.6 113 81 71.7S10 221 166 75.1 96 89 92.7S11 238 182 76.5 201 124 61.7S12 275 229 83.3 254 159 62.6S13 297 256 86.2 189 128 67.7S14 249 198 79.5 211 147 69.7S15 232 178 76.7 133 89 66.9S16 283 215 76.0 168 108 64.3S17 154 87 56.5 96 45 46.9S18 275 193 70.2 150 90 60.0

Table29.Disyllablesproducedtarget­like(Inês)

Inêsdisplaysadifferentbehaviortowardstargettrocheesandtargetiambs.Thefirst

observationthatwecandrawfromherdataisthatshedisplaysanacquisitionpathfortarget

trochees(thereareclear increasingproductionvalues fromtheearly to the latersessions),

whereastargetiambshaveaslowincreasingproductionrate,butnogreatvariation,fromthe

early to the later sessions. In Inês' data, trocheespresent a low target‐likeproduction rate

176

(below50%)until session8 and, from session9onwards, theproduction values generally

risetovaluesabove75%.Despitetargetiambsseemtobeacquiredearlierthantrochees(in

session4,a58.2%productionrateisattested),thetarget‐likeproductionvaluesdonotvary

muchacrossearlierandlatersessions.

In(129)and(130)wepresentInês'renditionsfortargetiambsandtargettrochees,

respectively.

(129) Inês–iambsproducedtarget‐like:

Orthogr. Gloss Target Output Agemamã 'mommy' /må»må‚/ [må»må‚] 0;11.14(S1)Inês 'name' /i»neS/ [ne»ne] 1;1.30(S3)chapéu 'hat' /Så»pEw/ [pE»pQ˘] 1;3.6(S4)aqui 'here' /å»ki/ [å»tJi] 1;5.11(S6)balão 'balloon' /bå»lå‚w‚/ [bå»bå] 1;6.11(S7)popó 'carfam.' /pç»pç/ [på»pç] bacio 'basin' /bå»siw [Bi»bi] 1;8.2(S9)

João 'name' /Zu»å‚w‚/ [du»å‚w‚] 1;10.29(S11)

(130) Inês–trocheesproducedtarget‐like:

Orthogr. Gloss Target Output Agetampa 'lid' /»tå‚på/ [»patJå] 1;8.2(S9)casa 'house' /»kazå/ [»kata] 1;9.19(S10)boa 'good' /»boå/ [»boå]sala 'livingroom' /»salå/ [»tJalå] 1;10.29(S11)

sete 'seven' /»sEtˆ/ [»dEtå] branca 'white' /»bRå‚kå/ [»bå‚kå]Bambi 'name' /»bå‚bi/ [»babi]

2;0.11(S12)

banho 'bath' /bå¯u/ [»båju]lobo 'wolf' /»lobu/ [»lubu]livro 'book' /»livRu/ [»idi]

2;1.10(S13)

nada 'nothing' /»nadå/ [»nadå]preta 'black' /»pRetå/ [»petJå] 2;2.1(S14)

AsforClara,intheexamplespresentedofInês'sspeech,weobservethatearlyiambs

tend tobe reduplicatedwords (either fromreduplicated targetwordsornot),whereas the

trochaicwordsproducedbyInêsarenormally[CV1CV2]words.

InTable30,wepresentJoana'sacquisitionpathfortargettrocheesandtargetiambs.

177

Session /SW/ [SW] % /WS/ [WS] %S1 ‐ ­ ‐ ‐ ­S2 ‐ ‐ ­ 4 0 0.0S3 2 0 0.0 3 1 33.3S4 ‐ ‐ ­ 3 0 0.0S5 ‐ ‐ ­ ‐ ‐ ­S6 ‐ ‐ ­ ‐ ‐ ­S7 16 0 0.0 3 1 33.3S8 4 0 0.0 3 0 0.0S9 13 0 0.0 29 9 31.0S10 16 2 12.5 28 10 35.7S11 45 13 28.9 54 32 59.3S12 91 57 62.6 34 10 29.4S13 74 44 59.5 42 16 38.1S14 163 108 66.3 56 28 50.0

Table30.Disyllablesproducedtarget­like(Joana)

Table30showsthatJoanaselectsandproducestargetiambsearlierthantrochees.In

the early sessions (sessions2‐9), target iambshave an early unstable production,whereas

target‐trochees are not produced. In session 10 and 11, iambs have a higher target‐like

productionratethantrochees(35.7%vs.12.5%insession10,and59.3%vs.28.9%insession

11),thoughtheamountoftrocheesinthechild'sintakeincreasesinsession10.Insession12,

weobserveaturningpointinthechild'sspeech,astheamountoftrocheesproducedtarget‐

likesurpassestheiambs.Bothtarget‐trocheesandtargetiambsremainbelowthe75%rate

oftarget‐likeproductionuntiltheendoftheobservationperiod,inJoana'sspeech.

Instances in (131) and (132) illustrate Joana's early production of target iambs

producedaccordingly,fromsession2onwards.Targettrocheesaremostlyproducedtarget‐

likefromsession12onwards.

(131) Joana–iambsproducedtarget‐like:

Orthogr. Gloss Target Output Agemamã 'mommy' /må»må‚/ [å»må‚] 1;0.25(S2)mamã 'mommy' /må»må‚/ [må»må‚] 1;2.7(S3)bebé 'baby' /bE»bE/ [be»Bi˘] 1;6.24(S7)café 'coffee' /kå»fE/ [ki»kE] 1;9.25(S9)Raquel 'name' /{å»kE…/ [kE»kE˘]avó 'grandmother' /å»vç/ [å»bç˘] 1;10.22(S10)

pastor 'shepard' /påS»toR/ [pˆ»tJo] 2;0.9(S11)colher 'spoon' /ku»¥ER/ [du»¯Ej]avó 'grandmother' /å»vç/ [å»vç] 2;2.19(S12)

178

(132) Joana–trocheesproducedtarget‐like:

Orthogr. Gloss Target Output AgeNando 'name' /»nå‚du/ [»na¯u]seco 'dry' /»seku/ [»tjEku]outra 'other' /»otRå/ [»otJå]

2;2.19(S12)

mala 'purse' /»malå/ [»mawå]pedra 'stone' /»pEdRå/ [»pEtHå]mota 'motorbike' /»mçtå/ [»mç˘tJå]

2;4.1(S13)

Paula 'name' /»pawlå/ [»pawå]casa 'house' /»kazå/ [»kajZå]prima 'cousin' /»pRimå/ [»pimå]quadro 'painting' /»kWadRu/ [»kajdJu]

2;6.24(S14)

As observed in Inês and Clara, Joana's early iambs tend to be the result of a

reduplicationstrategy(eitherfromreduplicatedtargetsornot)ortheresultofepenthesisat

the left edge of the circumscribed stressed syllable of the targetword.When trochees are

producedcorrectly,theyarenormallyproducedas[CV1CV2].

InTable31,wepresentJoão'sdevelopmentfortargettrocheesandtargetiambs.

Session /SW/ [SW] % /WS/ [WS] %S1 6 1 16.7 ‐ ‐ ­S2 ‐ ‐ ­ 1 0 0.0S3 8 3 37.5 4 3 75.0S4 6 5 83.3 6 4 66.7S5 4 0 0.0 ‐ ‐ ­S6 ‐ ‐ ­ 8 2 25.0S7 15 2 13.3 22 3 13.6S8 18 1 5.6 12 4 33.3S9 13 2 15.4 16 2 12.5S10 39 5 12.8 30 18 60.0S11 17 3 17.6 17 13 76.5S12 12 4 33.3 28 17 60.7S13 23 5 21.7 54 40 74.1S14 17 8 47.1 33 29 87.9S15 7 3 42.9 32 17 53.1S16 34 21 61.8 44 22 50.0S17 58 49 84.5 25 16 64.0S18 67 60 89.6 56 37 66.1S19 77 62 80.5 37 21 56.8S20 90 71 78.9 41 22 53.7S21 86 78 90.7 37 24 64.9S22 59 48 81.4 42 17 40.5

Table31.Disyllablesproducedtarget­like(João)

179

In Table 31, we observe that, in the early sessions (sessions 1‐6), the number of

trocheesinthechild'sintakeinhigherthanthenumberofiambs.Atthisperiod,thechildhas

areducednumberofbothstructures,andavariabletarget‐likeproductionrateisobservedin

/SW/and /WS/.Between sessions7 and16, thenumber of /SW/and /WS/ in the child's

intake is very even. At this point, iambs are generally produced with higher rates than

trochees.Iambsareproducedabovethe50%rateinsession10,whiletrocheesremainbelow

the50%until session15.However, the target‐likeproductionrate in iambsremainsbelow

75% until the end of the observation period169. After session 16, a turning point in João's

speechisobservedandtheamountoftrocheesproducedtarget‐likeoverridesiambs(andthe

number of trochaic tokens in the child's intake is higher than the number of iambs). In

session17,trocheesreachstableproductionvaluesabovethe75%.

Intheinstancesbelow(133)and(134),weshowJoão'sproductionsfortargetiambs

andtrochees.

(133) João–iambsproducedtarget‐like:

Orthogr. Gloss Target Output Ageolá 'hello' /ç»la/ [¥å»¥a] 1;1.12(S3)mamã 'mommy' /må»må‚/ [å»må] 1;2.30(S6)maçã 'apple' /må»så‚/ [må˘»må] 1;4.17(S8)papá 'daddy' /på»pa/ [på»pa˘]popó 'carfam.' /pç»pç/ [på»på˘w] 1;5.12(S9)

mamã 'mommy' /må»må‚/ [må»må‚] 1;5.26(S10)avô 'grandfather' /å»vo/ [pu»bu]memé 'sheepfam.' /mE»mE/ [må»må˘] 1;7.0(S12)

João 'name' /Zu»å‚w‚/ [nu‚»å‚w‚] 1;8.4(S14)mamã 'mommy' /må»må‚/ [må»må‚] 1;8.25(S15)café 'coffee' /kå»fE/ [tå»tE]Jesus 'name' /Zˆ»zuS/ [ju»ju]arroz 'rice' /å»{oS/ [a»jo˘]

1;9.25(S16)

avó 'grandmother' /å»vç/ [å»bç]papá 'daddy' /på»pa/ [på»pa˘] 1;10.11(S17)

169Session14seemstorepresentanexception.

180

(134) João–trocheesproducedtarget‐like:

Orthogr. Gloss Target Output Agebolo 'cake' /»bolu/ [»bo¥u]pombo 'pigeon' /»po‚bu/ [»må˘mu] 1;9.25(S16)

zebra 'zebra' /»zebRå/ [»bibå] cinco 'five' /»si‚ku/ [»titu] 1;10.11(S17)

urso 'bear' /»uRsu/ [»tutu]porta 'door' /»pçRtå/ [»bçtå] 1;10.26(S18)

gato 'cat' /»gatu/ [»tåtu]galo 'rooster' /»galu/ [»aju]trinta 'thirty' /»tRi‚tå/ [»ti‚tå]

1;11.10(S19)

mano 'brotherfam.' /»månu/ [»månu]fino 'thin' /»finu/ [»tinu]doce 'sweet' /»dosˆ/ [»toti]

1;11.19(S20)

parque 'park' /»paRkˆ/ [»paki]porta 'door' /»pçRtå/ [»pçtå] 2;0.6(S21)

Joãoalsouses/WS/reduplicationsandepenthesisintargetiambs.Earlytargetiambs

in João's speech are often reduplicated words which the child produces as such (cf. the

productions for papá 'daddy' /på»pa/, popó 'car fam.' /pç»pç/,mamã 'mommy' /må»må‚/,

memé'sheepfam.'/mE»mE/).However,reduplicationinnon‐redupliactedtargetarepossible

aswell(e.g.olá 'hello'/ç»la/,producedas[¥å»¥a],Jesus 'name'/Zˆ»zuS/producedas[ju»ju]).

On the contraray, when trochees are produced target‐like, they are produced as non‐

redupliactedwords([CV1CV2]).

InTable32,wepresentLuma'sacquisitionpathfortargettrocheesandtargetiambs.

181

Session /SW/ [SW] % /WS/ [WS] %S1 1 0 0 ‐ ‐ ­S2 ‐ ‐ ­ ‐ ‐ ­S3 9 0 0.0 ‐ ‐ S4 31 3 9.7 7 1 14.3S5 5 0 0.0 ‐ ‐ ­S6 7 2 28.6 2 1 50.0S7 22 5 22.7 ‐ ‐ ­S8 26 1 3.8 2 0 0.0S9 13 0 0.0 ‐ ‐ ­S10 2 0 0.0 2 0 0.0S11 2 0 0.0 2 1 50.0S12 2 1 50.0 6 2 33.3S13 11 1 9.1 3 0 0.0S14 8 0 0.0 20 3 15.0S15 16 5 31.3 32 5 15.6S16 1 0 0.0 69 29 42.0S17 31 0 0.0 17 14 82.4S18 14 0 0.0 21 17 81.0S19 4 0 0.0 15 12 80.0S20 31 2 6.5 51 39 76.5S21 21 0 0.0 57 46 80.7S22 33 6 18.2 30 13 43.3S23 41 0 0.0 52 42 80.8S24 34 0 0.0 48 46 95.8S25 30 1 3.3 21 21 100.0S26 24 0 0.0 41 40 97.6S27 18 0 0.0 65 54 83.1S28 30 5 16.7 61 51 83.6S29 38 8 21.1 76 60 78.9S30 122 30 24.6 103 67 65.0S31 118 42 35.6 89 65 73.0S32 208 97 46.6 110 83 75.5S33 215 130 60.5 130 87 66.9S34 217 138 63.6 142 122 85.9S35 146 101 69.2 100 74 74.0S36 231 183 79.2 216 157 72.7S37 145 90 62.1 123 90 73.2

Table32.Disyllablesproducedtarget­like(Luma)

In Table 32, we observe that Luma selects trochees earlier than iambs. However,

thereisageneralhigherproductionofiambstarget‐likesincesessionfromthebeginningof

word production, until the last session. In the early sessions (until session 27), target

trocheesarerarelyproducedaccordingly.Fromsession28onwards(2;2)anacquisitionpath

fortrocheesisobserved,i.e.,trocheesstarthavinganincreasingtarget‐likeproductionrate.

In (135) and (136) we exemplify Luma's earlier production of iambs and a later

productionoftrocheestarget‐like.

182

(135) Luma–earlyiambsproducedasiambs:

Orthogr. Gloss Target Output Agebebé 'baby' /bE»bE/ [pa»pa] 1;1.10(S4)mamã 'mommy' /må»må‚/ [må»må‚] 1;2.22(S6)mamã 'mommy' /må»må‚/ [´»må‚] 1;3.19(S8)balão 'balloon' /bå»lå‚w‚/ [Bˆ»wå] 1;5.9(S11)mamã 'mommy' /må»må‚/ [må»ma] 1;5.24(S12)Chuchu 'name' /Su»Su/ [a»Bu] 1;8.15(S18)Totô 'name' /to»to/ [to»to] 1;9.7(S19)Pati 'name' /pa»ti/ [ti»ti]aqui 'here' /å»ki/ [å»ti] 1;9.29(S20)

Bibi 'name'170 /bi»bi/ [Bi»Bi] vovô 'grandfatherfam.' /vo»vo/ [to»to] 1;10.18(S21)

Miguel 'name' /mi»gE…/ [ni»E] 1;11.15(S23)bebé 'baby' /bE»bE/ [Bå»BE] 2;0.13(S25)chichi 'pee' /Si»Si/ [si»si˘] 2;0.27(S26)

(136) Luma–trocheesproducedtarget‐likelater:

Orthogr. Gloss Target Output AgeTito 'name' /»titu/ [»titç]pedra 'stone' /»pEdRå/ [»pEdå] 2;5.15(S34)

sujo 'dirty' /»suZu/ [»CuZu]quatro 'four' /»kWatRu/ [»kWatu]alto 'tall' /»a…tu/ [»atu]

2;6.6(S35)

arco 'arch' /»aRku/ [»aku]cima 'above' /»simå/ [»fimå]baixo 'below' /»bajSu/ [»ba˘Su]brincos 'earings' /»bRi‚kuS/ [»bi‚ku‚]

2;6.20(S36)

roxo 'purple' /»{oSu/ [»{oCu]fresco 'fresh' /»freSku/ [»veku]circo 'circus' /»siRku/ [»Ciku]

2;6.27(S37)

Luma's renditions clearly confirmwhat has been previously observed in the other

children:earlyiambsproducedtarget‐likearenormallyreduplicatedwords(inLuma'scase,

mostly resulting from reduplicated target iambs). When target trochees are produced as

[SW],theyareproducedas[CV1CV2]words.

Thedatadisplayedaboveshowthat4ofthe5children(Inês,Joana,JoãoandLuma)

observedhavenolearningcurveforiambsbutdemonstrateanacquisitionpathfortrochees.

Indeed,contrarytoiambs,introchaictargetwordsweobserveagradual, increasingtarget‐

170Familiarnicknameforbisavó'grand‐grandmother'.

183

like production. The other child, Clara, selects both /SW/ and /WS/ and generally has

productionratesabovethe50%forbothstructures,sincetheonsetofwordproduction. In

Clara,itisnotpossibletodrawanyconclusiontotheearlieracquisitionof/SW/or/WS/,as

shetendstobeaccurateinherproductionsfromthebeginning171andherspeechafter1;10is

notavailable.

Insum,thedatapresentedaboveshowsthat:

(i) Initially,thepercentageoftrocheesproducedtarget‐likeisingenerallow;

(ii) Initially, thepercentageof iambsproduced target‐like ishigher than theoneof

trochees, and does not seem not to be a change in the course of prosodic

developmentforiambicwords;

(iii) The percentage of trochees produced target‐like increases in the course of

languagedevelopment;

(iv) InInês,JoanaandJoão'sspeech,aturningpointwheretrocheesoverrideiambsis

observedinagiventimeofdevelopment.

InClara, the childwhoseobservationsonly occurreduntil 1;10, no turningpoint is

noticedandboth/SW/and/WS/areproducedtarget‐likeatthelatersessions.InLuma,the

target‐like production rate is in general higher in iambs, until the end of the observation

period.

In theother children, the turningpoint takesplace in the sessionspresent inTable

33:

Inês S8>S9(1;7‐1;8)

Joana S11>S12(2;0‐2;2)

João S15>S16(1;8‐1;9)

Table33.Turningpoint­trocheesoverridingiambs

The data presented thus far seem to indicate that in their early stages, Portuguese

childrentendtobemorefaithfultoiambsthantotrochees.However,reduplicatedwordsand

epenthesisareveryfrequentinthesestages.AsshowninFigures29‐33,wherethedifferent

wordshapes,inanearlyandlaterpoint,areaccountedfor,thefrequencyoftheseproduction

patternstendstodecrease,inthecourseofdevelopment.

171 In fact,Clara isagoodexampleofa child thatusesselectionstrategies inherproduction,as sheappears toselect what she is able to produce. As we will observe further in this chapter, this behavior in Clara will befrequentinothertargetstructures.

184

In order to further show the acquisition path for non‐reduplicated iambs and

trochees,wepresentthedataconcerningthe faithfulproductionsofbothstresspatterns in

thespeechofthefivechildrenobserved.

The tables presented below concern the acquisition path for trochees and non‐

reduplicated iambs172. In thisanalysis,weaccounted fornon‐reduplicated targetsandnon‐

reduplicated actual productions only. Therefore, we did not take into account target

reduplicated forms (mamã, papá, etc.) or target non‐reduplicated forms that children

reduplicated, even though those forms could turn out being iambic (e.g., Raquel 'name'

/{å»kE…/ produced as [kE»kE]). The children's names, which in many cases were also

produced in reduplicated forms (Clara 'name' /»klaRå/ produced as [ka»ka] or Inês 'name'

/i»neS/producedas[ne»ne]),werealsonotconsidered.

Table 34 shows the acquisition path for trochees and non‐reduplicated iambs in

Clara'sspeech.

Session /SW/ [SW] % /WS/ [WS] %S1 1 1 100 3 3 100S2 ‐ ‐ ­ 1 1 100S3 4 3 75 ‐ ‐ ­S4 ‐ ‐ ­ ‐ ‐ S5 4 2 50 3 2 66.7S6 4 3 75 2 2 100S7 5 1 20 23 20 87S8 6 6 100 2 1 50S9 10 5 50 16 5 31.3S10 80 5973.8 25 20 80S11 161 103 64 67 51 76.1S12 146 7652.1 106 80 75.5

Table34.Trocheesandnon­reduplicatediambsproducedtarget­like(Clara)

Inthetablepresentedabove,weobservethreemainphases:(i)first(fromsessions1

to4),thechildrarelyselectsboth/SW/and/WS/;(ii) fromsessions5to9,weobservean

unstableproductionofbothstructures,thoughgenerallyhightarget‐likeproductionratesare

attestedinbothtargettrocheesandtargetiambs(however,areducedamountoftokensare

observedatthispoint173);(iii)fromsessions10to12trocheesarestillproducedwithsome

instability, whilst iambs display a stable behavior. In general, both /SW/ and /WS/ have

productionvalues above the50%.However, thenumberof trochaic tokens in the target is

172Non‐reduplicatediambswillbedesignatedas'iambs'or'targetiambs'hereafter.173InSession7,thenumberoftokenspresentedaccountsfor2types‐olá‘hello’andaqui‘here’‐,whichwereveryrepetedinthatsession.Thenumberis,thus,notveryindicativeofatendency.

185

higher in trochees than in iambs. In Clara’s early productions, a clear preference for a

particularstresspatternisnotobserved.However,thischildconsistently174reachesthe50%

oftarget‐likeproductioninsession9,for/SW/andinsession11,for/WS/.Despiteasimilar

acquisitionpathfor/SW/and/WS/,thischildacquires/SW/earlierthan/WS/.

Theinstancesin(137)illustrateClara'sproductionofbothtrocheesandiambs:

(137) Clara–trocheesandiambsproducedtarget‐like:

Orthogr. Gloss Target Output Ageágua 'water' /»agWå/ [»a˘Bå] 1;3.6(S5)água 'water' /»agWå/ [»ak´] 1;4.19(S6)Trocheesmano 'brotherfam.' /»månu/ [»månu] 1;5.16(S7)olá 'hello' /ç»la/ [碻ja] 1;3.6(S5)

Iambsaqui 'here' /å»ki/ [a»ki˘] 1;5.16(S7)

InTable35,wepresentthevaluesfortheacquisitionoftrocheesandiambsinInês'

speech.

Session /SW/ [SW] % /WS/ [WS] %S1 ‐ ‐ ­ ‐ ‐ ­S2 2 0 0 ‐ ‐ ­S3 4 0 0 8 0 0S4 27 1 3.7 ‐ ‐ ­S5 27 1 3.7 4 0 0S6 90 1 1.1 14 3 21.4S7 73 3 4.1 29 0 0S8 39 5 12.8 39 11 28.2S9 140 82 58.6 44 17 38.6S10 221 166 75.1 45 21 46.7S11 238 182 76.5 77 35 45.5S12 275 229 83.3 155 68 43.9S13 297 256 86.2 110 64 58.2S14 249 198 79.5 137 84 61.3S15 232 178 76.7 98 60 61.2S16 283 215 76.0 126 83 65.9S17 154 87 56.5 74 43 58.1S18 275 193 70.2 115 74 64.3

Table35.Trocheesandnon­reduplicatediambsproducedtarget­like(Inês)

InInês'sspeech,theonsetofdisyllables(fromsessions1to4)ischaracterizedbythe

scarce selection of both /SW/ and /WS/ (though /SW/ are more often selected). From

sessions 5 to 8, /SW/words start being producedmore consistently,whereas /WS/were

174Cf.criteriaofacquisitionreferredtoinsection3.3.

186

emerging.Untilsession8,areducednumberoftokens,both inthe intakeandinthechild's

productions,isattested.Fromsession9onwards,/SW/performedbetterthan/WS/andare

acquired (they surpass 50% target‐like production rate). After session 9, the number of

trocheesinthechild'sintakeincreasesdramatically,whencomparedtothenumberofiambs.

/WS/ are acquired in session 13 and never reach stabilization (≥ 75%). In Inês's speech,

trocheesareacquiredearlierthaniambs.

In(138),weillustratetheemergenceofbothtrocheesandnon‐reduplicatediambsin

Inês'speech.

(138) Inês–trocheesandiambsproducedtarget‐like:

Orthogr. Gloss Target Output Ageminha 'mine' /»mi¯å/ [»min´] 1;5.11(S6)pato 'duck' /»patu/ [»patJu] 1;6.11(S7)copo 'glass' /»kçpu/ [»kotJç]Trochees

garfo 'fork' /»gaRfu/ [»gau] 1;7.2(S8)

aqui 'here' /å»ki/ [å»tJi] 1;5.11(S6)aqui 'here' /å»ki/ [å»ki] 1;6.11(S7)Iambsavô 'grandfather' /å»vo/ [å»dJç] 1;7.2(S8)

InTable36,wepresentthevaluesoftarget‐likeproductionforiambsandtrochees,in

Joana'sspeech.

Session /SW/ [SW] % /WS/ [WS] %S1 ‐ ‐ ­ ‐ ‐ ­S2 ‐ ‐ ­ ‐ ‐ ­S3 2 0 0 ‐ ‐ ­S4 ‐ ‐ ­ ‐ ‐ ­S5 ‐ ‐ ­ ‐ ‐ ­S6 ‐ ‐ ­ ‐ ‐ ­S7 16 0 0 2 0 0S8 4 0 0 1 0 0S9 13 0 0 14 4 28.6S10 16 2 12.5 14 5 35.7S11 45 13 28.9 30 11 36.7S12 91 57 62.6 30 6 20S13 74 44 59.5 27 12 44.4S14 163 108 66.3 49 25 51.0

Table36.Trocheesandnon­reduplicatediambsproducedtarget­like(Joana)

Joana’s data also illustrate a neutral emergence of /SW/ and /WS/ and an earlier

acquisitionof/SW/.Fromsessions1tosession8,weobservethattherearesomeattempts

for both stress patterns (SW and WS), although in reduced amounts (trochees are

187

nevertheless earlier selected). Fromsessions9 to11,both/SW/and/WS/emergeand, at

thispoint, /WS/wordshavea slightlybetterperformance than/SW/.However, in session

12,/SW/surpassthe50%oftarget‐likeproductionandareacquired,whereas/WS/words

areonlyacquiredinsession14.AsinInês'data,fromagivenpointonwards(inJoana'scase,

thispointissession12),thenumberoftrochaictokensincreasesdramaticallyandovercomes

theamountofiambictokens.Joanaacquirestrocheesearlierthaniambs.

Theinstancespresentedin(139)exemplifytheacquisitionoftrocheesandiambsin

Joana'sspeech.

(139) Joana–trocheesandiambsproducedtarget‐like:

Orthogr. Gloss Target Output Agepato 'duck' /»patu/ [»paku] 1;9.25(S9)colo 'lap' /»kçlu/ [»kç˘u] 1;10.22(S10)linda 'beautiful' /»li‚då/ [»¯i‚¯å]Trochees

barco 'boat' /»baRku/ [»ma˘ku] 2;0.9(S11)

João 'name' /Zu»å‚w‚/ [u»aw]café 'coffee' /kå»fE/ [ki»kE] 1;9.25(S9)

papel 'paper' /»på»pE…/ [pˆ»e˘u] 1;10.22(S10)Iambs

balão 'balloon' /bå»lå‚w‚/ [m´»¯å‚w‚] 2;0.9(S11)

Figure37representsthenumberoftarget iambsandtrocheesproducedtarget‐like,

inJoão'sspeech.

188

Session /SW/ [SW] % /WS/ [WS] %S1 6 1 16.7 ‐ ‐ ­S2 ‐ ‐ ­ 1 1 100S3 8 3 37.5 4 4 100S4 6 5 83.3 4 4 100S5 4 0 0 ‐ ‐ ­S6 ‐ ‐ ­ 3 3 100S7 15 2 13.3 4 3 75S8 18 1 5.6 12 2 16.7S9 13 2 15.4 12 0 0S10 39 5 12.8 11 6 54.5S11 17 3 17.6 4 1 25.0S12 12 4 33.3 3 0 0S13 23 5 21.7 10 0 0S14 17 8 47.1 2 1 50S15 7 3 42.9 4 2 50S16 34 21 61.8 16 9 56.3S17 58 49 84.5 10 7 70S18 67 60 89.6 30 20 66.7S19 77 62 80.5 11 9 81.8S20 90 71 78.9 22 10 45.5S21 86 78 90.7 12 7 58.3S22 59 48 81.4 13 3 23.1

Table37.Trocheesandnon­reduplicatediambsproducedtarget­like(João)

InJoão,weobservethat,fromsessions1to6,anoccasionalselectionandproduction

ofboth/SW/and/WS/, though/WS/havehigher target‐likeproduction rates. It isworth

noticing,however,thatthetokensaccountedfor,fromsessions1to6,correspondtoasingle

type,thewordolá 'hello'/ç»la/.Fromsessions7to15,/SW/wordshaveanincreasingand

consistent target‐like production rate. /WS/ words have an inconsistent production from

sessions 7 to 13. In sessions 14 and 15 are produced with a 50% rate, but these rates

correspondto theproductionofareducednumberof target tokens(2and4,respectively),

andto3types(olá'hello'/ç»la/,Natal'Christmas'/nå»ta…/andJesus/Zˆ»zuS/).Fromsession

16onwards,both/SW/and/WS/areacquired(≥50%),though/SW/soonreachthestability

point(≥75%),whereas/WS/arestillpronetounstableproductionvalues(cf.session20and

22).

Thefollowingexamples(140),illustrateJoão'sproductionoftrocheesandiambs.

189

(140) João–trocheesandiambsproducedtarget‐like:

Orthogr. Gloss Target Output Agebolo 'cake' /»bolu/ [»bowu] 1;4.17(S8)uva 'grape' /»uvå/ [»dud˘å] 1;5.12(S9)bola 'ball' /»bçlå/ [»bawå] 1;6.16(S11)bola 'ball' /»bçlå/ [»bçja] 1;7.0(S12)uva 'grape' /»uvå/ [»u˘å] 1;8.4(S14)

Trochees

bola 'ball' /»bçlå/ [»Båjå] 1;8.25(S15)melão 'mellon' /mˆ»lå)w)/ [m´»wå)w)] 1;4.17(S8)olá 'hello' /ç»la/ [ç»wa] 1;5.12(S9)João 'name' /Zu»å)w)/ [nu)»å)w)] 1;8.4(S14)Iambs

Jesus 'Jesus' /Zˆ»zuS/ [do»ju˘] 1;8.25(S15)

Table38showstherateoftrocheesandiambsproducedtarget‐likebyLuma.

Session /SW/ [SW] % /WS/ [WS] %S1 1 0 0 ‐ ‐ ­S2 ‐ ‐ ­ ‐ ‐ ­S3 9 0 0 ‐ ‐ ­S4 31 3 9.7 ‐ ‐ ­S5 5 0 0 ‐ ‐ ­S6 7 2 28.6 ‐ ‐ ­S7 22 5 22.7 ‐ ‐ ­S8 26 1 3.8 ‐ ‐ ­S9 13 0 0 ‐ ‐ ­S10 2 0 0 ‐ ‐ ­S11 2 0 0 1 1 100S12 2 1 50 ‐ ‐ ­S13 11 3 27.3 ‐ ‐ ­S14 8 0 0 6 0 0S15 16 5 31.3 2 0 0S16 1 0 0 36 0 0S17 31 0 0 ‐ ‐ ­S18 14 0 0 ‐ ‐ ­S19 4 0 0 ‐ ‐ ­S20 31 2 6.5 16 15 93.8S21 21 0 0 7 7 100S22 33 7 21.2 ‐ ‐ ­S23 41 0 0 10 4 40S24 34 0 0 1 1 100S25 30 1 3.3 7 7 100S26 24 2 8.3 2 1 50S27 18 0 0 1 1 100S28 30 8 26.7 13 8 61.5S29 38 9 23.7 12 3 25S30 122 30 24.6 19 7 36.8S31 118 42 35.6 27 16 59.3S32 208 97 46.6 50 30 60S33 215 130 60.5 46 21 45.7

190

S34 217 138 63.6 55 39 70.9S35 146 101 69.2 35 19 54.3S36 231 183 79.2 100 66 66S37 145 90 62.1 49 26 53.1

Table38.Trocheesandnon­reduplicatediambsproducedtarget­like(Luma)

Table 38 shows a disparate emergence time for /SW/ and /WS/ in Luma's speech.

From sessions 2 to 19, trochees are selected in all sessions and occasionally produced,

whereasiambsareonlyselectedinsessions11,14,15and16(insession11,theonlytoken

attemptedisthewordbalão'balloon'/bå»lå‚w‚/).Fromsessions20to27,iambsareproduced

withhighertarget‐likeproductionratesthan/SW/,thoughtheonlythreetypescorrespond

tothetokensaccountedforinthesesessions(aqui'here'/å»ki/,avô'grandfather'/å»vo/and

Miguel 'name' /mi»gE…/). After session 28, /SW/ start having an increasing and consistent

productionisattested,butstabilityisneverreached./SW/acquisitionisattainedinsession

33.Fromsession28onwards,/WS/wordshaveahightarget‐likeproductionrateingeneral,

buttheyhaveanunstablebehavioruntiltheendoftheobservationalperiod(cf.sessions30

and34).Theacquisitionof/WS/ inLuma'sspeech isreached insession31.Lumaacquires

/WS/earlierthan/SW/.

In(141)wepresentLuma'searlyproductionsfortargettrocheesandiambs.

(141) Luma–trocheesandiambsproducedtarget‐like:

Orthogr. Gloss Target Output AgeNoddy 'name' /»nçdi/ [»nç¥i] 1;11.1(S22)bola 'ball' /»bçlå/ [»bowå˘] 2;0.13(S25)lobo 'wolf' /»lobu/ [»wobu] 2;2.4(S28)gato 'cat' /»gatu/ [»tatu] 2;2.22(S29)parvo 'silly' /»paRvu/ [»pavu 2;3.6(S30)tinta 'ink' /»ti‚tå/ [»ti‚tå]

Trochees

onze 'eleven' /»o‚zˆ/ [»o‚Zu] 2;4.11(S31)

Pati 'name' /pa»ti/ [te»ti]aqui 'here' /å»ki/ [å»ti] 1;9.29(S20)

avô 'grandfather' /å»vo/ [tç»to˘]Miguel 'name' /mi»gE…/ [ni»å‚] 1;11.15(S23)

Miguel 'name' /mi»gE…/ [ni»E] 1;11.29(S24)aqui 'here' /å»ki/ [å»ki] 2;2.4(S28)sujou 'itgotdirt' /su»Zo/ [Cu»Zo] 2;2.22(S29)aqui 'here' /å»ki/ [a»ki] 2;3.26(S30)

Iambs

saiu 's/heleft' /så»iw/ [Så»iw] 2;4.11(S31)

191

In summary, the analysis conducted on target disyllables in the speech of the five

Portuguesechildrenobserved,showedthat:

(i) Including reduplicated target words and words produced with epenthesis, an

iambictendencyisobserved;

(ii) However,non‐reduplicatedwords(/SW/and/WS/)haveadifferentacquisition

path frommost of the earlywordsobserved (which are, in their vastmajority,

reduplicated words or word produced with epenthesis creating an iambic

pattern).

(iii) Accounting for non‐reduplicated words, nor words where filler syllables are

inserted,indicatesthat/SW/and/WS/formsemergeapproximatelyatthesame

timeinthespeechofPortuguesechildren.

(iv) From a certain point in development (which we designated 'turning point'),

/SW/generallyoverride/WS/.

InTable39,wesummarizetheacquisitionof/SW/and/WS/inthespeechofthefive

childrenobserved:

Children Acquisitionpathindisyllables

Clara,Inês,Joana /SW/>/WS/

João /SW/~/WS/

Luma /WS/>/SW/

Table39.Acquisitionofdisyllables–between­childrencomparison

Inthistable,weobservethatthreechildren(Clara,InêsandJoana)acquiretrochees

earlierthaniambs,onechild(Luma)acquiresiambsearlierthantrocheesandonechildhasa

simultaneousacquisitionofbothtrocheesandiambs.

The analysis on /SW/ and /WS/ stress patterns suggests that the 'early iambic

tendency' of Portuguese early productions is biased by the high frequency of reduplicated

words([CV1'CV1])andfillersounds([fS])andmightbeonlyapparent.

192

5.1.2.3.Trisyllables

Inthissection,wewillpresentthedataforthefaithfulproductionsoftrisyllablesin

the five childrenobserved.The faithfulproductionsof /WSW/, /WWS/and/SWW/words

willbeanalyzed.

InTable40,wepresentClara'sdataforfaithful/WSW/and/WWS/targets.InClara's

productions,target/SWW/wordswerenotattested.

Session /WSW/ [WSW] % /WWS/ [WWS] %S1 ‐ ‐ ­ ‐ ‐ ­S2 ‐ ‐ ­ ‐ ‐ ­S3 ‐ ‐ ­ ‐ ‐ ­S4 ‐ ‐ ­ ‐ ‐ ­S5 ‐ ‐ ­ ‐ ‐ ­S6 ‐ ‐ ­ ‐ ‐ ­S7 3 ‐ 0 ‐ ‐ ­S8 ‐ ‐ ­ ‐ ‐ ­S9 4 2 50.0 ‐ ‐ ­S10 11 4 36.4 ‐ ‐ ­S11 21 12 57.1 1 0 0S12 22 8 36.4 5 0 0

Table40.Trisyllablesproducedtarget­like(Clara)

Clara does not select any trisyllables until session 7. The first trisyllabic forms

attempted are /WSW/ words, in session 7. Only in session 9 does she produce /WSW/

trisyllables target‐like (50%). Target /WWS/ words are attempted in session 11, but are

never produced target‐like. The production of trisyllables target‐like is maintained in low

rates(below60%)untilthelastsessionofobservation.

Thefollowinginstances((142))regardClara'sacquisitionpathfor/WSW/words.In

theearlysessions(untilsession7)no/WSW/wordsareattempted.Onlyinsession9,afew

/WSW/wordsareproducedaccordingly.

(142) Clara–/WSW/producedtarget‐like:

Orthogr. Gloss Target Output Agemenina 'girl' /mˆ»ninå/ [mˆ¯i»na]sapato 'shoe' /så»patu/ [tu»patu] 1;8.20(S10)

menina 'girl' /mˆ»ninå/ [mˆ»¯i¯å]menino 'boy' /mˆ»ninu/ [må»ninu] 1;9.23(S11)

menina 'girl' /mˆ»ninå/ [må»ninå]batata 'potato' /bå»tatå/ [båtJa»ta] 1;10.15(S12)

193

Table 41 shows the acquisition path for target trisyllables (/WSW/, /SWW/ and

/WWS/)inInês'speech.

Session /WSW/ [WSW] % /WWS/ [WWS] % /SWW/ [SWW] %S1 1 0 0 ‐ ‐ ­ ‐ ‐ ­S2 1 0 0 ‐ ‐ ­ ‐ ‐ ­S3 12 0 0 16 0 0 7 0 0S4 11 0 0 ‐ ‐ ­ 1 0 0S5 13 0 0 ‐ ‐ ­ 2 0 0S6 33 1 3 1 0 0 ‐ ‐ ­S7 35 1 2.9 ‐ ‐ ­ 21 0 0S8 33 7 21.2 2 0 0 3 0 0S9 28 7 25 4 1 25 1 0 0S10 77 33 42.9 2 0 0 ‐ ‐ ­S11 85 49 57.6 11 2 18.2 ‐ ‐ ­S12 112 74 66.1 6 1 16.7 7 2 28.5S13 125 78 62.4 22 9 40.9 5 1 20S14 75 53 70.7 14 7 50 3 0 0S15 112 77 68.8 17 8 47.1 9 6 66.7S16 106 54 50.9 12 6 50 5 0 0S17 69 37 53.6 12 1 8.3 4 0 0S18 79 45 57 14 5 35.7 3 1 33.3

Table41.Trisyllablesproducedtarget­like(Inês)

Despite trisyllabic targets beingpresent since the first session, only from session8

onwards is there a consistent emergence of trisyllables (/WSW/). Target /WSW/ are the

trisyllabic forms earlier acquired by Inês (in session 11), though their production remains

unstableuntiltheendoftheobservationperiod./WWS/emergeconsistentlyaftersession11

andthe50%pointissurpassedinsession14fortarget/WWS/.However,/WWS/wordsare

prone to instability until the last observed session. Target /SWW/ words are selected in

session3,but theyare rarelyproduced target‐like,until the last session.Though there isa

clear difference between the acquisition of /WSW/, on the one hand, and /WWS/ and

/SWW/,andon theotherhand, thestabilizationof trisyllables isnotattaineduntil the last

session./SWW/wordsaresubjecttoalateacquisition.

Theinstancesin(143)exemplifyInês'acquisitionpathfortrisyllables.

194

(143) Inês–/WSW/producedtarget‐like:

Orthogr. Gloss Target Output Agemacaco 'monkey' /må»kaku/ [ka»kaku]cabelo 'hair' /kå»belu/ [ƒeƒe»ƒe] 1;7.2(S8)

sapato 'shoe' /så»patu/ [ba»pato] 1;8.2(S9)boneca 'doll' /bu»nEkå/ [m´»¯Ekå]janelas 'windows' /Zå»nElåS/ [nå»nålåS] 1;10.29(S11)

Table42,showsthedevelopmentalpathfortrisyllablesinJoana'sspeech.

Session /WSW/ [WSW] % /WWS/ [WWS] % /SWW/ [SWW] %S1 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S2 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S3 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S4 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S5 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S6 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S7 2 0 0 ‐ ‐ ­ ‐ ‐ ­S8 1 0 0 ‐ ‐ ­ ‐ ‐ ­S9 11 0 0 1 0 0 ‐ ‐ ­S10 12 1 8.3 3 0 0 3 0 0S11 31 5 16.1 2 0 0 5 0 0S12 42 2 4.8 1 0 0 2 0 0S13 44 5 11.4 1 0 0 1 0 0S14 84 39 46.4 3 0 0 12 1 8.3

Table42.Trisyllablesproducedtarget­like(Joana)

In Joana's speech, the acquisition of trisyllables is not reached, until the end of the

observationperiod.Target/WSW/areattemptedinsession7,butonlyinsession13thereis

a increasing rateof target‐likeproduction.Thehighest score (46.4%) is reached in session

14. /WWS/ and /SWW/words are rarely attempted andproduced target‐like. /WWS/ are

onlyattempted insession9./SWW/wordsareattempted insession10,but the target‐like

ratesdonotgobeyondthe8.3%untiltheendoftheobservationperiod.

Theinstancesin(144)illustrateJoana'sproductionsof/WSW/words.

(144) Joana–/WSW/producedtarget‐like:

Orthogr. Gloss Target Output Agebigode 'moustache' /bi»gçdˆ/ [»bi˘»biBi] 1;6.24(S7)leitinho 'milkdim.' /låj»ti¯u/ [å»kJi¯å] 2;2.19(S12)amiga 'friend' /å»migå/ [å»migå] 2;4.1(S13)Joana 'name' /Zu»ånå/ [Zu»ånå]aposta 'bet' /å»pçStå/ [å»pHçtJå] 2;6.24(S14)

195

Table43summarizesJoão'sacquisitionpathfortrisyllables.

Session /WSW/ [WSW] % /WWS/ [WWS] % /SWW/ [SWW] %S1 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S2 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S3 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S4 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S5 7 1 14.3 ‐ ‐ ­ ‐ ‐ ­S6 10 0 0.0 ‐ ‐ ­ ‐ ‐ ­S7 ‐ ‐ ­ ‐ ‐ ­ 5 0 0S8 5 0 0.0 ‐ ‐ ­ ‐ ‐ ­S9 1 0 0.0 2 0 0 ‐ ‐ ­S10 1 0 0.0 ‐ ‐ ­ ‐ ‐ ­S11 8 1 12.5 ‐ ‐ ­ 3 0 0S12 7 1 14.3 1 0 0 ‐ ‐ ­S13 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S14 ‐ ‐ ­ 1 1 100 ‐ ‐ ­S15 2 0 0.0 3 0 0 ‐ ‐ ­S16 48 27 56.3 8 1 12.5 6 0 0S17 38 26 68.4 4 1 25 ‐ ‐ ­S18 62 39 62.9 1 0 0 2 0 0S19 34 23 67.6 9 3 33.3 15 3 20S20 49 28 57.1 2 1 50 8 1 12.5S21 42 15 35.7 4 2 50 3 1 33.3S22 62 48 77.4 12 5 41.7 3 0 0

Table43.Trisyllablesproducedtarget­like(João)

AsInês,Joãodisplaysaclearacquisitionpathfor/WSW/words,onthronehand,and

/WWS/and/SWW/words,ontheotherhand.Theformerareearlierselected,producedand

acquired, whereas the latter have an unstable behavior until the end of the observation

period.

ThefirsttrisyllablesattemptedbyJoãoare/WSW/(insession5),althoughhekeepsa

lowproductionrateuntilsession15.Insession16,thechildstartsproducing/WSW/words

moreconsistently(thetarget‐likeproductionvaluessurpassthe50%insesison16)andthe

target‐like production of /WSW/ is kept constant around 60% until the end of the

observationperiod.

/WWS/ and /SWW/ trisyllables are produced target‐like in very low rates, though

/WWS/areearlierandmoreselectedand/WWS/performbetterthan/SWW/.

In(145)weshowJoão'sproductionoftarget/WSW/.

196

(145) João–/WSW/producedtarget‐like:

Orthogr. Gloss Target Output Agequentinho 'warm' /ke‚»ti¯u/ [ti»tinu] 1;9.25(S16)cadeira 'chair' /kå»dåjRå/ [då»dåjRå]girafa 'giraffe' /Zi»Rafå/ [tå»tatå] cavalo 'horse' /kå»valu/ [å»jaju]

1;10.26(S18)

InTable44,wepresentLuma'starget‐likeproductionoftrisyllables.

Session /WSW/ [WSW] % /WWS/ [WWS] % /SWW/ [SWW] %S1 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S2 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S3 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S4 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S5 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S6 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S7 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S8 3 0 0 ‐ ‐ ­ ‐ ‐ ­S9 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S10 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S11 3 0 0 ‐ ‐ ­ 1 0 0S12 7 0 0 ‐ ‐ ­ 1 0 0S13 1 0 0 3 0 0 7 0 0S14 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S15 10 0 0 ‐ ‐ ­ 1 0 0S16 ‐ ‐ ­ ‐ ‐ ­ 2 0 0S17 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S18 1 0 0 ‐ ‐ ­ 4 0 0S19 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S20 2 0 0 ‐ ‐ ­ ‐ ‐ ­S21 5 0 0 ‐ ‐ ­ 2 0 0S22 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S23 ‐ ‐ ­ ‐ ‐ ­ 2 0 0S24 1 0 0 ‐ ‐ ­ 1 0 0S25 1 0 0 ‐ ‐ ­ 2 0 0S26 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S27 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S28 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S29 4 0 0 ‐ ‐ ­ ‐ ‐ ­S30 7 1 14.3 1 1 100 1 0 0S31 22 6 27.3 2 0 0 2 0 0S32 63 6 9.5 ‐ ‐ ­ ‐ ‐ ­S33 33 5 15.2 ‐ ‐ ­ 2 2 100S34 55 9 16.4 ‐ ‐ ­ 3 0 0S35 84 43 51.2 3 0 0 3 0 0S36 55 28 50.9 11 2 18.2 ‐ ‐ ­S37 64 49 76.6 11 2 18.2 ‐ ‐ ­

Table44.Trisyllablesproducedtarget­like(Luma)

197

InLuma'sspeech,weobservethat,fromsession1tosession28,trsyllablesarerarely

selected.However,duringthisperiod,/WSW/and/SWW/aremoreselectedthan/WWS/,

though no cases of target‐like production is attested in any of the target forms. The child

starts producing /WSW/ words more consistently in session 30 and in session 36 she

surpasses the 50% and acquires /WSW/ trisyllables. /SWW/ and /WWS/ words are

producedinconsistentlyuntiltheendoftheobservationperiod.

Theinstancesin(146)mostlyshowLuma'sproductionof/WSW/trisyllables.

(146) Luma–productionof/WSW/targets:

Orthogr. Gloss Target Output AgeSusana 'name' /su»zånå/ [Zu»Zånå]sequinho 'drydim.' /sˆ»ki¯u/ [si»ki¯u]antigo 'old' /å‚»tigu/ [å»tigu]

2;6.6(S35)

chuchinha 'pacifierdim.' /Su»Si¯å/ [Su»Si¯å]agora 'now' /å»gçRå/ [å»gça] Francisco 'name' /fRå‚»siSku/ [få‚»fiXku]

2;6.20(S36)

Asfortrisyllables,themoststraightforwardassumptionthatcanbedrawnfromthe

datapresentedaboveisthatEP‐speakingchildrenstartproducingthemlateindevelopment.

/WSW/areearlierselectedbythePortugeusechildrenunderobservation,whereas/WWS/

and/SWW/areproducedinconsistentlybyalmostallthechildrenatthebeginning.

Furthermore,ourresultsindicatethattheacquisitionpathpursuedbythePortuguese

childrentowardstrisyllablesis/WSW/>/WWS/>/SWW/.

5.1.2.4.Summaryforthefaithfulnesstothetarget

TheresultsfromthefaithfulproductionsindicatethatPortuguesechildrenobserved

tend toproducemonosyllables faithfully first.Faithfuldisyllablesareproduced ina second

stageandwithoutapreference for [SW]or [WS]. Ina thirdstage, threeof the fivechildren

(Clara,InêsandJoana)showedhigherandearlierfaithfulnessratesfor/SW/thanfor/WS/.

Onechild(João)hadasimultaneousacquisitionpathfor/SW/and/WS/andtheotherchild,

Luma,acquires/WS/earlierthan/SW/.Inafourthstage,/WSW/areproducedtarget‐like.

Othertrisyllables(/WWS/and/SWW/)haveingeneralanunstablebehavioruntiltheendof

theobservationalperiod.

These data suggest that a categorical iambic or trochaic tendency is hard to be

claimed. Instead, these results indicate a neutral start, eventually with a slight trochaic

tendency.

198

In the following section,wewill present the strategies from the observed children

whentheydidnotproducemonosyllables,disyllablesandtrisyllablestarget‐like.

5.1.3.Productionstrategies

Since no clear tendency was observed in the data in the target‐like productions, a

specialattention to thestrategiesusedby theobservedchildrenmustbegiven, inorder to

betteraccountforthepreferentialandtendencialwordshapesandstresspatternsproduced

bythechildren.

In this section we will focus on the strategies undertaken by the children under

observation when dealing with word shape development. We will look at monosyllables,

disyllables and trisyllables. In monosyllables, we will particularly analyze the strategies

implying epenthesis, reduplications and production of [SW] (through ungliding175). In

disyllablesand trisyllables,wewill investigate truncationonly. In this section,wewillonly

showtheresultsfortruncationandnotshowtheresultsforreduplicationandepenthesisfor

tworeasons:

(i) Thestrategyoftruncationisastrategyfoundintheliterature,bothinPortuguese

andinotherlanguages,toinvestigatetheacquisitionofstresspatterns(e.g.,Baia,

2008; Fikkert, 1994; Kehoe, 1998; Kehoe & Stoel‐Gammon, 1997; Rapp, 1996;

Santos,2007;Tzakosta,2004176).Inordertocompareourresultswiththeresults

previouslyfoundforotherlanguages,thisanalysisremainedasnecessary;

(ii) Wehave shown that disyllables (/SW/, and especially /WS/)might initially be

pronetoreduplicationandepenthesis,andthatbothofthesestrategiestendedto

disappearinthecourseofdevelopment177.

Resultsforthestrategiesusedinmonosyllables,disyllablesandtrisyllables(5.1.3.1.,

5.1.3.2. and 5.1.3.3., respectively) will be shown in marked line charts, in order to better

account for the use of the different strategies across development178. Mention to absolute

valueswillbedone,whennecessary.

Inafinalsectionwewillpresenttheresultsforstressshift.

175Aswewill observe in the five children, this strategymayoccur through theproductionof adiphthongas ahiatus:e.g.pai'father'/»paj/,producedas[»pai].176Cf.section2.2.2.,forareview.177Cf.section5.1.1.2.,Figures29‐30andinstancesin(117)‐(121).178Cf.AppendixB,forabsolutevaluesandpercentages.

199

5.1.3.1.Monosyllables

Inthissectionwewillpresentthestrategieschildrenusewhentheydonotproduce

monosyllables target‐like. Despite monosyllables being the least relevant word shape for

studying stress patterns acquisition, the strategies used by the childrenwhen they do not

producemonosyllablestarget‐likemayindicatewhatkindofwordshapeisbeingprocessed.

It is important to observe whether children extend the target word shape (namely, via

reduplication,epenthesisorungliding179)and,iftheydoso,whatstructuresdotheyselect.

In target monosyllables, we accounted for the production of [WS] reduplications

([S2]180),insertionofafillersoundattheleftedgeofthecircumscribedsyllableofthetarget

word ([fS]), multiple reduplications ([CVCV...]) and, finally, the production of a trochee,

throughungliding([SW]).

InFigure34,weshowClara'sstrategiestowardsmonosyllables.

Figure34.Strategiesinearlymonosyllables(Clara)

Weobserve that this childdoesnotusemanystrategies in theproductionof target

monosyllables. In session2, theonlymonosyllabic tokenwasproducedwithepenthesis. In

sessions 4 and 5, reduplication was used (14.3% in both sessions). In sessions 5 and 6,

epentheses were frequent (14.3% and 38.5%, respectively). In sessions 6, 7, 8 and 9,

ungliding was also observed, though in reduced amounts (7.7%, 11.1%, 29.4% and 9.1%

respectively),creatinga trochaicpattern.Multiplereduplicationswerenotvery frequent in

Clara'sspeech(cf.session6,7,8and9).Fromsession9onwards,weobservethatthereisa

reduceduseofallthestrategies.

In(147)wewillshowClara'srenditionsofepenthesisandreduplications.

179Cf.footnote175.180Cf.footnote166.

200

(147) Clara–epenthesis,reduplicationsandungliding:

Orthogr. Gloss Target Output Agedá 'giveimp.' /»da/ [å»da] 1;0.13(S2)não 'no' /»nå‚w‚/ [å»nå‚w‚] 1;3.6(S5)cão 'dog' /»kå‚w‚/ [ˆ»kå‚w‚] 1;4.9(S6)pai 'father' /»paj/ [Q»paj]

Epenthesis

pé 'foot' /»pE/ [a»pe] 1;5.16(S7)

não 'no' /»nå‚w‚/ [nå»nå‚w‚] 1;2.22(S4)Redupl.

não 'no' /»nå‚w‚/ [nå»nå‚w‚] 1;3.6(S5)cão 'dog' /»kå‚w‚/ [»kåu‚]não 'no' /»nå‚w‚/ [»nå‚u‚]mãe 'mother' /»må‚j‚/ [»må‚i‚]

1;5.16(S7)Unglid.[SW]

cão 'dog' /»kå‚w‚/ [»kå‚˘u‚˘] 1;6.6(S7)

InFigure35,weshowInês'strategiesregardingmonosyllables.

Figure35.Strategiesinmonosyllables(Inês)

Weobservedthat in the four initialsessions, longerutterancesare frequent(100%,

13.6%, 50% and 42.9%). A few cases of reduplications are also found, but not in great

amount(13.4%,insession4).Insessions2,5,6and7,epenthesisiscommon(27.3%,20.5%,

19.4%and10%,respectively).Fromsession9onwards,monosyllablesaremainlyproduced

target‐like.

In (148) we present some renditions of Inês displaying target monosyllables

producedwiththedifferentstrategies.

201

(148) Inês–reduplicationsandepenthesis:

Orthogr. Gloss Target Output Agedá 'giveimp.' /»da/ [»da»da»dQ]dá 'giveimp.' /»da/ [»da»da»da˘] 1;1.30(S3)

não 'no' /»nå‚w‚/ [nå»nå˘nånånå]Multipl.redupl.

põe 'put' /»po‚j‚/ [p´»p´p´j] 1;3.6(S4)

dá 'giveimp.' /»da/ [/»da] 0;11.14(S1)dá 'giveimp.' /»da/ [´»dA] 1;0.25(S2)dá 'giveimp.' /»da/ [å»då˘] 1;4.9(S5)

Epenthesis

pé 'foot' /»pE/ [å»pE] 1;5.11(S6)dá 'giveimp.' /»da/ [»da»da]má 'bad' /»ma/ [må»ma] 1;1.30(S3)

dá 'giveimp.' /»da/ [da»da] 1;3.6(S4)pé 'foot' /»pE/ [pE»pE] 1;3.6(S4)

Redupl.

dá 'giveimp.' /»da/ [dQ»dQ] 1;4.9(S5)

Figure36showsthestrategiesusedbyJoanaintarget‐monosyllables.

Figure36.Strategiesinmonosyllables(Joana)

In Joana, we observed that, in the initial sessions (1‐6), very few strategies are

produced (until session 5, monosyllables are produced target‐like and, in session 6, one

monosyllable in three is produced in a multiple reduplication). In session 8, [WS]

reduplications are attested (12.5%).After session9, Joanausesmostlyungliding ([SW]) to

dealwithtargetmonosyllables,thoughinverylowrates(thehigherrateis9.7%of[SW],in

sessions9and10).

In(149)wepresentsomeinstancesofJoana'sstrategiestowardsmonosyllables.

202

(149) Joana–epenthesis,unglidingandreduplication:

Orthogr. Gloss Target Output Agehá 'thereis' /»a/ [a»a˘] 1;8.4(S8)

Redupl.chá 'tea' /»Sa/ [ta»DJa] 1;9.25(S9)pai 'father' /»paj/ [»pa˘i˘] 1;9.25(S9)pai 'father' /»paj/ [»pai]Unglid.

[SW]chão 'floor' /»Så‚w‚/ [»tHA‚˘u‚] 1;10.22(S10)

quer '(s/he)wants' /»kER/ [ˆ»kE]Epenthesis

não 'no' /»nå‚w‚/ [ˆ»nå‚˘u‚] 2;0.9(S11)

Figure37showsJoão'sstrategiesfortarget‐monosyllables.

Figure37.Strategiesinmonosyllables(João)

João mainly uses epenthesis and reduplications when he does not produce

monosyllables target‐like. In sessions 2 and 3, very few tokens are selected (4 and 1,

respectively)andtheyaremainlyproducedwithepenthesis(25%and100%insession2and

3,respectively)orreduplication(50%insession2).Insessions7and9,monosyllableswith

diphthongs are produced with ungliding, creating a [SW] pattern (33.3% and 19.2%). In

sessions 10, 11, 12 and 13, monosyllables are mostly produced as reduplications (15.4%,

46.2%, 53.8% and 11.1%, respectively, in each session). From sessions 16 to 19, reduced

percentages of the different strategies are attested, though ungliding ([SW]) has higher

values (9.7% in sessions 17 and 18, and 15.2 in session 21) than the other analyzed

strategies.

In(150)weexemplifysomeofJoão'sstrategiesintargetmonosylables.Werecallthat

inJoão,thepercentageofeachstrategywashighlyreduced.

203

(150) João–epenthesisandreduplicationinmonosyllables:

Orthogr. Gloss Target Output Agedá 'giveimp.' /»da/ [å»då] 1;0.28(S2)

Epenthesisdá 'giveimp.' /»da/ [å»då] 1;1.12(S3)pau 'stick' /»paw/ [»påu] 1;5.12(S9)mãe 'mother' /»må‚j‚/ [»må‚i‚]não 'no' /»nå‚w‚/ [»nåu] 1;9.25(S15)

cão 'dog' /»kå‚w‚/ [»kå‚u‚]

Unglid.[SW]

meu 'mine' /»mew/ [»me˘u‚] 1;10.11(S16)

dá 'giveimp.' /»da/ [då»då˘] 1;5.12(S9)dá 'giveimp.' /»da/ [da»da] 1;5.26(S10)Redupl.dá 'giveimp.' /»da/ [da»d´˘] 1;6.16(S11)

Figure38showsthestrategiesusedbyLumatodealwithtargetmonosyllables.

Figure38.Strategiesinmonosyllables(Luma)

Luma is a showcase of the strategies that children may use to deal with target

monosyllables. Until session 10, the child uses multiple reduplications and epenthesis (in

session8,forinstance,thechildproduces30.6%ofeachstrategy).Fromsession10tosession

27,sheusesreduplicationinhigheramounts(insession16,thechildreduplicates57.1%of

themonosyllables selected). Between sessions 19 and 21, epenthesis is a used strategy as

well(withproductionratesof9.1%,7.1%and28.6%insessions19,20and21,respectively).

Fromsession27onwards,theamountofstrategiesusedbyLumadecreasessignificantly.

In(151),thestrategiesfortargetmonosyllablesinLuma'sspeecharegiven:

204

(151) Luma–reduplicationinmonosyllables(untilsession25):

Orthogr. Gloss Target Output Agenão 'no' /»nå‚w‚/ [nå»nå] 1;3.6(S4)não 'no' /»nå‚w‚/ [¯o»¯o]pão 'bread' /»på‚w‚/ [på»på] 1;5.11(S6)

dá 'giveimp.' /»da/ [da»da] 1;8.2(S17)Redupl.

chá 'tea' /»Sa/ [»sJa»sJa] 1;10.8(S21)não 'no' /»nå‚w‚/ [´»nå] 1;4.9(S5)dá 'giveimp.' /»da/ [a»da] 1;3.19(S8)não 'no' /»nå‚w‚/ [廯å] 1;5.11(S6)

Epenthesis

dá 'giveimp.' /»da/ [»/da] 1;7.5(S15)

The data presented indicate that the five children use epenthesis, reduplications

([CV1CV1])andlongerutterances‐[CVCV...]‐intargetmonosyllables.

However, it isworthwhile noticing that some children use production strategies in

monosyllables more than others. For instance, Inês and Clara have a great amount of

production strategies in monosyllables between sessions 1 and 8, and Luma uses these

duringtheentireobservationperiod.Joãodisplaysthesamebehaviorfromsession1to13.

Onthecontrary,Joanahasareducednumberofreduplications,epenthesisandunglidingin

targetmonosyllables.

Despite the frequent use of production strategies in target monosyllables, and as

furthermoreshownin5.1.2.1.,theseformsarenormallyproducedtarget‐likeinthespeechof

thefivechildrenobserved.

5.1.3.2.Truncationindisyllables

Tables 45‐49 show the truncation patterns displayed by Portuguese childrenwhen

dealingwithtarget/SW/andtarget/WS/.

In Table 45, we present the percentage of truncation (to [S]) in target disyllables

(/SW/and/WS/)producedbyClara.

205

Session /SW/ [S] % /WS/ [S] %S1 1 0 0 3 0 0S2 ‐ ‐ ­ 1 0 0S3 4 0 0 ‐ ‐ ­S4 ‐ ‐ ­ ‐ ‐ ­S5 4 1 25 3 0 0S6 4 0 0 2 0 0S7 5 0 0 23 3 13.0S8 6 0 0 2 1 50S9 9 1 11.1 16 6 37.5S10 80 4 5 25 7 28S11 161 3 1.9 67 11 16.4S12 146 7 4.8 106 23 21.7

Table45.Truncationof/SW/and/WS/(Clara)

InClara'sdata,weobservethattargetiambsaremorepronetotruncationthantarget

trochees. Until session 6, there is one single case of truncation (in /SW/). From session 7

onwards, target trochees have residual values for truncations, whereas trochees maintain

truncationrateshigherrates(insessions9and10,forinstance,weobserveatruncationrate

of37.5%and28%,respectively).

Sincetrocheesaremostlyproducedtarget‐likebyClara,in(152)weexemplifysome

instances of truncations for target iambs, from session 7 onwards (1;5). In the examples

presented, whenever a truncation is produced, the stressed syllable is, in general, the

circumscribedsyllable.

(152) Clara–truncationintargetiambs:

Orthogr. Gloss Target Output Ageaqui 'here' /å»ki/ [»ki] 1;5.16(S7)chapéu 'hat' /Så»pEw/ [»pEj] 1;6.6(S8)aqui 'here' /å»ki/ [»ki˘]João 'name' /Zu»å‚w‚/ [»jå‚w‚] 1;7.11(S9)

Table46presentsthepercentageoftruncationinInês'productionof/SW/and/WS/.

206

Session /SW/ [S] % /WS/ [S] %S1 ‐ ‐ ­ ‐ ‐ ­S2 2 1 50 ‐ ‐ ­S3 4 4 100 8 4 50S4 27 9 33.3 ‐ ‐ ­S5 27 17 63.0 4 3 75S6 90 45 50 14 7 50S7 73 43 58.9 29 14 48.3S8 39 18 46.2 39 19 48.7S9 140 20 14.3 44 15 34.1S10 221 25 11.3 45 18 40S11 238 10 4.2 77 18 23.4S12 275 20 7.3 155 66 42.6S13 297 14 4.7 110 13 11.8S14 249 18 7.3 137 23 16.8S15 232 11 4.7 98 15 15.3S16 283 25 8.8 126 28 22.2S17 154 17 11.0 74 10 13.5S18 275 27 9.8 115 20 17.4

Table46.Truncationof/SW/and/WS/(Inês)

InInês'datatwophasesareobserved:afirstone,untilsession8,inwhich/SW/and

/WS/areequallypronetotruncation.Fromsession9onwards,weobservethat/SW/tendto

be preserved (the truncation rates are below 15%), whereas /WS/ are still prone to

truncation, though, from session 12 onwards, the truncation rates start a decreasing path.

Fromsession13until theendoftheobservationperiod, iambsandtrocheesbothdisplaya

lowrateoftruncation(below20%).

The instances in (153) and (154) show Inês tendency for [S] truncation, in target

trochees (until session 9) and in target iambs (until session 13), respectively. It is worth

noticingthattruncationiniambsisastrategykeptforlongertimebyInês,whencomparedto

trochees.

(153) Inês–truncationintargettrochees:

Orthogr. Gloss Target Output Agemanta 'blancket' /»må‚tå/ [»m˘å‚]banho 'bath' /»bå¯u [»på] 1;3.6(S4)

pêlo 'hair' /»pelu/ [»pe] 1;4.9(S5)fralda 'diaper' /»fRa…då/ [»ka˘] 1;5.11(S6)cartas 'cards' /»kaRtåS/ [»ka]garfo 'fork' /»gaRfu/ [»ƒa] 1;6.11(S7)

porta 'door' /»pçRtå/ [»pç] 1;7.2(S8)tampa 'lid' /»tå‚på/ [»pa]perna 'leg' /»pERnå/ [»bE] 1;8.2(S9)

207

(154) Inês–truncationintargetiambs:

Orthogr. Gloss Target Output Ageaqui 'here' /å»ki/ [»ki] 1;4.9(S5)champô 'shampoo' /Så‚»po/ [»po]João 'name' /Zu»å‚w‚/ [»wå‚] 1;5.11(S6)

balão 'balloon' /bå»lå‚w‚/ [»b´]colher 'spoon' /ku»¥ER/ [»kE] 1;6.11(S7)

papel 'paper' /på»pE…/ [»pE˘] 1;7.2(S8)chapéu 'hat' /Så»pEw/ [»pEw] aqui 'here' /å»ki/ [»ki] 1;8.2(S9)

abrir 'toopen' /å»bRiR/ [»biR]chapéu 'hat' /Så»pEw/ [»Ew] 1;10.29(S11)

avó 'grandmother' /å»vç/ [»dJç] 2;0.11(S12)anões 'dwarfs' /å»no‚j‚S/ [»no‚j‚S] 2;1.10(S13)nariz 'nose' /nå»RiS/ [»di˘S] 2;1.10(S13)

Table 47 shows the percentage rate of truncation for target WS and target SW in

Joana'sspeech.

Session /SW/ [S] % /WS/ [S] %S1 ‐ ‐ ­ ‐ ‐ ­S2 ‐ ‐ ­ ‐ ‐ ­S3 2 1 50 ‐ ‐ ­S4 ‐ ‐ ­ ‐ ‐ ­S5 ‐ ‐ ­ ‐ ‐ ­S6 ‐ ‐ ­ ‐ ‐ ­S7 16 16 100 2 1 50S8 4 1 25 1 1 100S9 13 6 46.2 14 3 21.4S10 16 11 68.8 14 6 42.9S11 45 24 53.3 30 9 30S12 91 20 22 30 22 73.3S13 74 11 14.9 27 11 40.7S14 163 13 8 49 12 24.5

Table47.Truncationof/SW/and/WS/tomonosyllables(Joana)

In Joana, the truncation rate, forboth /SW/and/WS/ is in general high, though, as

expected,itdecreasesinthecourseofdevelopment.InJoana'sdata,wefoundtwotrends,for

truncation in disyllables. A first time, in which /SW/ and /WS/ are equally prone to

truncation(untilsession11),andasecondtime,inwhichtargettrocheesshowadecreasing

pathandtargetiambshaveahighertruncationrate(fromsession12to14).

208

Intheexamplesbelow((155)‐(156)),weshowJoana'srenditions,whereweobserve

the truncation patterns for target trochees (until session 11) and for target iambs (from

session7untiltheendoftheobservationperiod).

(155) Joana–truncationintargettrochees:

Orthogr. Gloss Target Output AgeCarla 'name' /»kaRlå/ [»ta˘] 1;6.24(S7)Carla 'name' /»kaRlå/ [»ka] 1;8.4(S8)gosto 'Ilike' /»gçStu/ [»gç] 1;9.25(S9)Paula 'name' /»pawlå/ [»pa] 1;10.22(S9)Nando 'name' /»nå‚du/ [»¯å‚] leite 'milk' /»låjtˆ/ [»¯åj] 2;0.9(S11)

(156) Joana–truncationintargetiambs:

Orthogr. Gloss Target Output Ageavó 'grandmother' /å»vç/ [»pç] 1;9.25(S9)cristal 'cristal' /kRiSta…/ [»taw]atrás 'behind' /å»tRaS/ [»tJaS˘] 2;2.19(S12)

chapéu 'hat' /Så»pEw/ [»pEw]bacio 'basin' /bå»siw/ [»Siw]chover 'torain' /Su»veR/ [»vej]

2;4.1(S13)

azul 'blue' /å»zu…/ [»dZow]avô 'grandfather' /å»vo/ [»vo] 2;6.24(S14)

InTable48,wepresent thepercentages for truncation in/SW/and/WS/ in João's

speech.

209

Session /SW/ [S] % /WS/ [S] %S1 6 0 0 ‐ ‐ ­S2 ‐ ‐ ­ 1 0 0S3 8 0 0 4 0 0S4 6 1 16.7 4 0 0S5 4 2 50.0 ‐ ‐ ­S6 ‐ ‐ ­ 3 0 0S7 15 10 66.7 4 1 25S8 18 15 83.3 12 5 41.7S9 13 7 53.8 12 7 58.3S10 39 18 46.2 11 5 45.5S11 17 9 52.9 4 1 25.0S12 12 5 41.7 3 2 66.7S13 23 6 26.1 10 7 70.0S14 17 2 11.8 2 1 50.0S15 7 1 14.3 4 2 50.0S16 34 5 14.7 16 3 18.8S17 58 2 3.4 10 3 30.0S18 67 0 0.0 30 0 0.0S19 77 1 1.3 11 0 0.0S20 90 5 5.6 22 2 9.1S21 86 0 0.0 12 1 8.3S22 59 1 1.7 13 1 7.7

Table48.Truncationof/SW/and/WS/tomonosyllables(João)

In João, from session 1 to 6, both structures are produced in reduced amounts

(however,theselectediambsareneverproducedwithtruncation181,whereastrocheesmay

besubjecttothisstrategy,asshowninsessions4and5).Fromsession7tosession11,the

truncationratesfortrocheestendtobehigherthanforiambs.Insession12,thistendencyis

invertedandtrocheeshaveabetterperformancethaniambs.Fromsession12tosession17,

iambsaremorepronetotruncationthantrochees.Fromsession18onwards,bothstructures

display reduced amounts of truncation. In João, truncation in trochees has a general

decreasing path (from sessions 14/15 onwards, it becomes scarce),whereas truncation in

iambsismaintainedforalongerperiodoftime(onlyaftersession18areductionintheuseof

thisstrategyisobserved).

In (157) and (158), we illustrate João's truncation patterns in /SW/ and /WS/,

respectively.

181Itisworthsayingthattheiambictokensattemptedinsession1,3,4and5allcorrespondtoasingletype,thewordolá'hello'/ç»la/.

210

(157) João–truncationintargettrochees:

Orthogr. Gloss Target Output Agebola 'ball' /»bçlå/ [»ba˘]panda 'pandabear' /»på‚då/ [»på] 1;3.21(S7)

bola 'ball' /»bçlå/ [»bo] 1;4.17(S8)mota 'motorbike' /»mçtå/ [»ta]pato 'duck' /»patu/ [»på] 1;5.12(S9)

mano 'brotherfam.' /»månu/ [»må]bola 'ball' /»bçlå/ [»bç˘] 1;5.26(S10)

mota 'motorbike' /»mçtå/ [»ta] 1;6.16(S11)

(158) João–truncationintargetiambs:

Orthogr. Gloss Target Output Agelimão 'lemon' /li»må‚w‚/ [»mo‚]avô 'grandfather' /å»vo/ [»bu] 1;5.12(S9)

limão 'lemon' /li»må‚w‚/ [»å‚˘w‚˘] 1;6.16(S11)sofá 'couch' /su»fa/ [»på˘] 1;7.20(S11)Natal 'Christmas' /nå»ta…/ [»ta] 1;8.4(S14)olá 'hello' /ç»la/ [»ja]João 'name' /Zu»å‚w‚/ [»jåw] 1;8.25(S15)

Table49 shows thepercentages for truncation in target /SW/and/WS/ inLuma's

speech.

Session /SW/ [S] % /WS/ [S] %S1 1 0 0 ‐ ‐ ­S2 ‐ ‐ ­ ‐ ‐ ­S3 9 6 66.7 ‐ ‐ ­S4 31 0 0 ‐ ‐ ­S5 5 0 0 ‐ ‐ ­S6 7 1 14.3 ‐ ‐ ­S7 22 9 40.9 ‐ ‐ ­S8 26 5 19.2 ‐ ‐ ­S9 13 9 69.2 ‐ ‐ ­S10 2 1 50 ‐ ‐ ­S11 2 1 50 1 0S12 2 1 50 ‐ ‐ ­S13 11 5 45.5 ‐ ‐ ­S14 8 1 12.5 6 6 100S15 16 4 25 2 2 100S16 1 0 0 36 36 100S17 31 6 19.4 ‐ ‐ ­S18 14 2 14.3 ‐ ‐ ­S19 4 0 0 ‐ ‐ ­S20 31 12 38.7 16 3 18.8S21 21 9 42.9 7 0 0

211

S22 33 17 51.5 ‐ ‐ ­S23 41 12 29.3 10 0 0S24 34 12 35.3 1 0 0S25 30 7 23.3 7 0 0S26 24 1 4.2 2 0 0S27 18 1 5.6 1 0 0S28 30 1 3.3 13 2 15.4S29 38 1 2.6 12 7 58.3S30 122 2 1.6 19 9 47.4S31 118 1 0.8 27 5 18.5S32 208 3 1.4 50 19 38S33 215 3 1.4 46 22 47.8S34 217 3 1.4 55 12 21.8S35 146 2 1.4 35 13 37.1S36 231 1 0.4 100 21 21S37 145 9 6.2 49 8 16.3

Table49.TruncationofWSandSWtomonosyllables(Luma)

Lumahasaninterestingpathfortruncationindisyllables.Inherdata,weobservean

inversion of the truncation patterns.We observe a higher truncation rate for /SW/ in the

beginning(until session25)anda laterhigher truncationrate for iambs(aftersession28).

The apparent tendency for higher truncation rate in target trochees at the beginning,

however, is due to the highly reduced number of target iambs in Luma's speech. The

truncationratefortargettrocheesisnothigherthanfortargetiambs.Itisonlythestrategy

thechildusestodealwiththedisyllablesproduced.

The instances in (159) and (160) illustrate Luma's truncation patterns in target

trocheesandintargetiambs,respectively:

(159) Luma–truncationintargettrochees(session20‐25):

Orthogr. Gloss Target Output Ageolha 'lookimp.' /»ç¥å/ [»a] 1;9.29(S20)gata 'cat' /»gatå/ [»ta˘] 1;11.1(S22)milho 'corn' /»mi¥u/ [»jç]gato 'cat' /»gatu/ [»ta] 1;11.29(S24)

meias 'socks' /»måjåS/ [»bE˘] 2;0.13(S25)polvo 'octopus' /»po…vu/ [»bo˘] 2;0.13(S25)

212

(160) Luma–truncationintargetiambs(session28‐37):

Orthogr. Gloss Target Output AgeInês 'name' /i»neS/ [»ne˘]sofá 'couch' /su»fa/ [»fa˘] 2;2.4(S28)

chinês 'chinese' /Si»neS/ [»ne˘] 2;2.22(S29)aqui 'here' /å»ki/ [»ki˘]ali 'there' /å»li/ [»¥i˘] 2;3.26(S30)

azul 'blue' /å»zu…/ [»nu˘] 2;4.11(S31)azul 'blue' /å»zu…/ [»nu‚] favor 'favor' /få»voR/ [»vo]ali 'there' /å»li/ [»ni‚]

2;5.20(S34)

azul 'blue' /å»zu…/ [»nu‚]adeus 'bye' /å»dewS/ [»dew] 2;6.6(S35)

buscar 'toget' /buS»kaR/ [»ka]João 'name' /Zu»å‚w‚/ [»Zå‚w‚] 2;6.20(S36)

avó 'grandmother' /å»vç/ [»vç] 2;6.27(S37)

The first observation thatwe can drawwith respect to the strategies used to deal

with target disyllables is that, when disyllables are truncated, the circumscribed and

preserved syllable is overwhelmingly the stressed syllable. In addition, the tables shown

aboveshowthat,intheearlysessions:

(i) 3children(Joana,JoãoandLuma)havehighertruncationratesintrochees;

(ii) Inêstruncatesmore/WS/;

(iii) Claradisplaysnotendencytopreservemore/SW/or/WS/.

In the last sessions, iambs are more prone to truncation than trochees in the

productions of the five children observed. Trocheeswill progressively be less truncated to

monosyllablesandbeproducedmoretarget‐like,whiletargetiambsarestilltruncated.

The truncation patterns displayed by the Portuguese children observed suggest

mixed‐interpretations: on the one hand, it is noticeable that three of the children display

highertruncationin/SW/,suggestinganiambictendency.However,ontheotherhand,this

tendencywillbeinvertedand,inthefinalsessions,iambsaremorepronetotruncationthan

/SW/, suggesting that, either therewas no iambic tendency at the beginning, or the early

tendencyfor/WS/changedto/SW/.

213

5.1.3.3.Truncationintrisyllables

Thefollowingchartsrefertothetruncationpatternsfor/WSW/,/WWS/and/SWW/

targets. Ineachchild,wewillpresentthetruncationpatternsfor/WSW/,then/WWS/and

finally/SWW/targets.Thefirstsessiondisplayedinthechartspresentedbelowrepresents

thefirstsessioninwhichatrisyllableisattemptedinachild'sspeech182.

Figures 39 and 40 show the truncation patterns in /WSW/ and /WWS/ for Clara.

Claradidnotselectany/SWW/targetduringtheobservationperiod.

182Inordertoshowtheresultsforthethreetrisyllabicstructuresandthetruncationpatternsproducedineachoneofthem,wepresent,sidebyside,thedataincharts,withpercentagevalues.Forinformationontheabsolutevalues,cf.AppendixC.

214

Figure39.Truncationpatternsfor/WSW/(Clara)

Figure40.Truncationpatternfor/WWS/(Clara)

Claraselectstarget/WSW/onlyinsession7.Inthesetargetforms,thechildclearly

showsapreferencefor[SW]truncation,asshownbyFigure39.

Target/WWS/areattemptedlater,insession11(Figure40).Insession11and12a

very reduced number of target /WWS/ were selected and produced (6 attempted and 5

produced).However,whenshefirstproducestheseforms,sheproducesthemas[WS].

Theinstancesin(161)and(162),illustrateClara'struncationsfortarget/WSW/and

/WWS/,respectively.

(161) Clara–truncationintarget/WSW/([SW]):

Orthogr. Gloss Target Output AgeAurora 'name' /aw»Rçrå/ [»lalå] 1;7.11(S9)Aurora 'name' /aw»Rçrå/ [»lçlå] 1;8.20(S10)menina 'girl' /mˆ»ninå/ [»¯i¯å]menino 'boy' /mˆ»ninu/ [»ninu] 1;9.23(S11)

macaco 'monkey' /må»kaku/ [»kaku]sapato 'shoe' /så»patu/ [»patu] Aurora 'name' /aw»Rçrå/ [»ç˘lå˘]

1;10.15(S12)

(162) Clara–truncationof/WWS/to[WS]:

Orthogr. Gloss Target Output Ageacabou 'itfinished' /åkå»bo/ [kå»bo] 1;9.23(S11)acabou 'itfinished' /åkå»bo/ [ka»bo] 1;10.15(S12)

Figures41, 42 and43 show thepercentages for truncation in /WSW/, /WWS/and

/SWW/,respectively,inInês'speech.

215

Figure41.Truncationpatternsfor/WSW/(Inês)

Figure42.Truncationpatternsfor/WWS/(Inês)

Figure43.Truncationpatternsfor/SWW/(Inês)

In Inês' truncation patterns for /WSW/, we observe an initial neutral production

(untilsession5).Fromsession6tosession9,apreferencefor[WS]isattested.[SW]and[S]

arelessproduced,buttheyareneverthelessfrequentaswell,untilsession9.Fromsession10

to15,thepercentageof[SW]and[WS]truncationissimilarandthetruncationto[S]isless

frequent,thoughthethreepatternshaveageneraldecreasingpath.Both[SW]and[WS]are

possibleand[S]becomesrare.Insession16,thenumberof[WS]truncationsincreasesagain,

morethan[SW].Aswewillshowin(162),below,the[WS]truncationpatternmostlyregards

reduplicatedforms.

In /WWS/ (Figure 42), Inês also displays a preference for [WS] truncation in all

observed sessions. Before session 8, occasional target /WWS/ were produced as

monosyllables.Thetruncationfor[S]isveryscarce,duringtheobservationperiod.Between

sessions8and12,truncationto[WS]remainshigh(above50%).Fromsession13onwards,

216

the percentage values for [WS] truncation decreases to values below 50%, and, after this

session,thechildmostlyproduces/WSW/target‐like.

In Inês'/SWW/words(Figure43), it ispossible twoobservetwodistinctphases:a

firstphase,wherethetruncationto[S]ismorefrequent(untilsession9),andasecondphase,

wherethetruncationto[SW]becomespredominant(fromsession12onwards).

In (163)wepresent someexamplesof Inês' truncatedproductions for /WSW/.We

observe that most of the [WS] truncations are, in fact, the result of epenthesis and

reduplication.

(163) Inês–truncationin/WSW/([WS],[SW]and[S]):

Orthogr. Gloss Target Output Agechupeta 'pacifier' /Su»petå/ [å»pe] 1;1.30(S3)babete 'bib' /bå»bEtˆ/ [ba»BQ] 1;3.6(S4)sapato 'shoe' /så»patu/ [på»på] 1;4.9(S5)vestido 'dress' /vˆS»tidu/ [tJi»tJi] 1;5.11(S6)chupeta 'pacifier' /Su»petå/ [be»be] 1;6.11(S7)cadeira 'chair' /kå»dåjRå/ [ƒE»ƒE] 1;7.2(S8)

[WS]

boneca 'doll' /bu»nEkå/ [me»¯E] 1;9.19(S10)chupeta 'pacifier' /Su»petå/ [»pe] 1;1.30(S3)cabelo 'hair' /kå»belu/ [»pE˘] 1;3.6(S4)boneca 'doll' /bu»nEkå/ [»¯å] 1;5.11(S6)girafa 'giraffe' /Zi»Rafå/ [»ƒa] 1;6.11(S7)

[S]

cadeira 'chair' /kå»dåjRå/ [»gE] 1;7.2(S8)vestida 'dressed' /vˆS»tidå/ [»bitå] umbigo 'bellybutton' /u‚»bigu/ [»bidu] [SW]barulho 'noise' /bå»Ru¥u/ [»buju]

1;8.2(S9)

In (164)we show instances of Inês producing target /WWS/words. Asmentioned

before, the child produces [WS] but, in their vast majority, these productions are

reduplications.

217

(164) Inês–truncationof/WWS/as[WS]:

Orthogr. Gloss Target Output Agecanguru 'kangaroo' /kå‚gu»Ru/ [ka»kE˘] 1;7.2(S8)Leonor 'name' /liu»noR/ [¯o»no] 1;8.2(S9)parabéns 'happybirthday' /påRå»bå‚j‚S/ [på»påjS] 1;9.19(S10)parabéns 'happybirthday' /påRå»bå‚j‚S/ [Bå»Båjs] 1;10.29(S11)estragou '(s/he)spoil' /StRå»go/ [ti»go] 2;0.11(S12)acabou 'itfinished' /åkå»bo/ [kå»bo] 2;1.10(S13)cachecol 'scarf' /kaSˆ»kç…/ [ka»kçj] 2;1.10(S13)

In(165),below,we illustrate twophases:an initialphasewhere/SWW/wordsare

mostly truncated to [S], and a second phase, when [SW] are mainly produced. It is

worthwhile recalling that the production of early SWW in Inês is the product of a type

repetition,thename'Mário'.

(165) Inês–truncationin/SWW/(early[S]andlater[SW]):

Orthogr. Gloss Target Output AgeMário 'name' /»maRiu/ [»må] 1;1.30(S3)

[S]Mário 'name' /»maRiu/ [»ma] 1;4.9(S5)árvore 'tree' /»aRvuRˆ/ [»abi]Bárbara 'name' /»baRbåRå/ [»babå] 2;0.11(S12)

números 'numbers' /»numˆRuS/ [»nuwoS]médico 'doctor' /»mEdiku/ [»mEku] 2;1.10(S13)

árvore 'tree' /»aRvuRˆ/ [»abi] 2;2.1(S14)Cláudia 'name' /»klawdiå/ [»ladjå] 2;3.8(S15)

[SW]

óculos 'glasses' /»çkuluS/ [»çklu] 2;5.24(S16)

InFigures44,45and46,wepresentthepercentageoftruncationin/WSW/,/WWS/

and/SWW/inJoana'sspeech.

218

Figure44.Truncationpatternsfor/WSW/(Joana)

Figure45.Truncationpatternsfor/WWS/(Joana)

Figure46.Truncationpatternsfor/SWW/(Joana)

In Joana,we observe that there is a preference for [S] truncation in target /WSW/

until session 11. Truncation to [WS] is possible from session 8 to session 11 in low rates,

though these rates arehigher than theones attested in [SW] truncation. In session12, the

childfavorstruncationto[SW]andboth[S]and[WS]becomelessfrequent.

Theproductionof target/WWS/ is rare. Joanaonlyhas9 tokens for/WWS/word,

from session 9, until the end of the observation period. When she attempts /WWS/, she

mainlytruncatesthemto[S].

/SWW/wordsarenotattempteduntillateinJoana'sspeech.Whensheselectsthese

words,sheusestruncationto[S],thoughthispatternhasadecreasingpath,contraryto[SW],

whichbecomesanavailablestrategyusedby Joanaat theendof theobservationperiod, in

/SWW/words.

In(166)weshowJoana'struncationsfor[S],[SW]and[WS],inthisorder,alongthe

observationperiod.

219

(166) Joana–truncationin/WSW/([S],[WS]and[SW]):

Orthogr. Gloss Target Output Ageboneca 'doll' /bu»nEkå/ [»bWå] 1;6.24(S7)Elvira 'name' /E…»viRå/ [»i˘] 1;9.25(S9)coelho 'rabbit' /ku»å¥u/ [»E]sapato 'shoe' /så»patu/ [»pa˘] 1;10.22(S10)

cabeça 'head' /kå»beså/ [»me]camisa 'blouse' /kå»mizå/ [»m˘i] 2;0.9(S11)

castelo 'castle' /kåS»tElu/ [»tEw]

[S]

gelados 'icecreams' /Zˆ»laduS/ [»Z˘a] 2;2.19(S12)

Joana 'name' /Zu»ånå/ [ˆ»¯å]sapato 'shoe' /så»patu/ [pa»pa]padrinho 'godfather' /på»dRi¯u/ [pa»pa]

1;9.25(S9)

gelado 'icecream' /Zˆ»ladu/ [pi»ja˘] [WS]

macaco 'monkey' /må»kaku/ [pˆ»kaw] 2;0.9(S11)

leitinho 'milkdim.' /låj»ti¯u/ [»ti¯u]menina 'girl' /mˆ»ninå/ [»minå] 2;2.19(S12)

comida 'food' /ku»midå/ [»mi¯å]bananas 'bananas' /bå»nånåS/ [»¯å¯åS] 2;4;1(S13)

vermelho 'red' /vˆR»må¥u/ [»mEju]

[SW]

baloiço 'swing' /bå»lojsu/ [»bo˘Su] 2;6.24(S14)

Instancesin(167)illustrateJoana'struncationto[S]intarget/WWS/.

(167) Joana–truncationof/WWS/to[S]:

Orthogr. Gloss Target Output Agecarapau 'mackerel' /kåRå»paw/ [»pç] 1;10.22(S10)Carabás 'name' /kåRå»baS/ [»baS˘] 2;4.1(S13)

Theinstancesbelow((168))illustrateJoana'stendencyfor[S]and[SW]truncationin

thefinalsessions.

(168) Joana–earlytruncationof/SWW/to[S]andlatertruncationto[SW]:

Orthogr. Gloss Target Output Agemáquina 'machine' /»makinå/ [»≤a]óculos 'glasses' /»çkuluS/ [»tJç] 1;10.22(S10)

óculos 'glasses' /»çkuluS/ [»dçS][S]

árvore 'tree' /»aRvuRˆ/ [»fa˘] 2;0.9(S11)

sítio 'place' /»sitiu/ [»tJiku] 2;2.19(S12)óculos 'glasses' /»çkuluS/ [»çkHu] [SW]Bárbara 'name' /»baRbåRå/ [»babå] 2;6.24(S14)

220

Figures 47, 48 and 49 show the truncation preferences for /WSW/, /WWS/ and

/SWW/words,inJoão'sspeech.

Figure47.Truncationpatternsfor/WSW/(João)

Figure48.Truncationpatternsfor/WWS/(João)

Figure49.Truncationpatternsfor/SWW/(João)

Joãodoesnotselect/WWS/until session5.However, fromsession5 tosession14,

thechilddoesnotfavoranytruncationpatternandproduces/WSW/as[S],[SW]and[WS].

From session 15 onwards, a clear preference for [SW] truncation is observable, though, in

thesesessions,/WSW/aremostlyproducedtarget‐like.

As in the previous children, target /WWS/ are selected late by João. During their

emergence,theyareproducedfirstlyas[S](untilsession15)andlateras[WS](fromsession

16onwards).

/SWW/ words are very scarce in João's speech. When they are selected they are

truncatedto[S](untilsession16)andto[SW](fromsession16onwards).

In(169)weshowsomeinstancesoftheproductionof/WSW/wordsinJoão:

221

(169) João–truncationintarget/WSW/:

Orthogr. Gloss Target Output Agebolacha 'cookie' /bu»laSå/ [»bwa] 1;2.13(S5)bolacha 'cookie' /bu»laSå/ [»bu] 1;7.0(S12)[S]comboio 'train' /ko‚»bçjo/ [»bç] 1;4.17(S7)bolacha 'cookie' /bu»laSå/ [å»bwa] 1;2.13(S5)chupeta 'pacifier' /Su»petå/ [a»pa] 1;2.30(S6)[WS]bolacha 'cookie' /bu»laSå/ [Bå»wå] 1;7.0(S12)banana 'banana' /bå»nånå/ [»nånå] 1;6.16(S11)bolacha 'cookie' /bu»laSå/ [»balå] girafa 'giraffe' /Zi»Rafå/ [»nånå] 1;7.0(S12)

banana 'banana' /bå»nånå/ [»månå] 1;8.25(S15)atende 'pickup' /å»te‚dˆ/ [»tå‚dˆ]menina 'girl' /mˆ»ninå/ [»¯inå] 1;9.25(S16)

artista 'artist' /åR»tiStå/ [»tita]alface 'lettuce' /a…»fasˆ/ [»kati] 1;10.11(S17)

ajuda 'help' /å»Zudå/ [»dudå]querido 'dear' /kˆ»Ridu/ [»tidu] 1;10.26(S18)

[SW]

barriga 'belly' /bå»{igå/ [»bidå] 1;11.10(S20)

In(170)weshowsomeofthefewtokensproducesbyJoãointarget/WWS/.

(170) João–truncationof/WWS/to[S],[WS]:

Orthogr. Gloss Target Output Ageavestruz 'ostrich' /åvˆS»tRuS/ [»tu]javali 'wildpig' /Zåvå»li/ [»di‚] 1;8.25(S15)

[S]acabou 'itfinished' /åkå»bo/ [»bo] 1;9.25(S16)avestruz 'ostrich' /åvˆS»tRuS/ [tu»tu] 1;9.25(S16)acabou 'itfinished' /åkå»bo/ [å»bo] 1;10.11(S17)[WS]biberon 'milkbottle' /bibˆ»Ro‚/ [bå»bo‚] 1;10.26(S18)

(171) João–truncationof/SWW/to[SW]:

Orthogr. Gloss Target Output AgeFátima 'name' /»fatimå/ [»patå] Júlia 'name' /»Zuliå/ [»ujå] 1;11.10(S20)

nêspera 'medlar' /»neSpˆRå/ [»pebå] 2;0.6(S21)

Figures50,51and52showthetruncationpatternsdisplayedbyLumainthecourse

oftheobservationperiod.

222

Figure50.Truncationpatternsfor/WSW/(Luma)

Figure51.Truncationpatternsfor/WWS/(Luma)

Figure52.Truncationpatternsfor/SWW/(Luma)

Luma also displays a late emergence of trisyllables. Target /WSW/ emerge late in

development. They are not selected until session 8 and until session 25, they are mostly

223

producedas[S]and,later(fromsession27onwards),truncatedto[WS].

Figure 71 shows that the percentage of /WWS/ in Luma is very scarce. In the last

sessions, the child produces /WWS/ targets as [WS]. The data from these words are not

sufficienttoallowforanyrobustgeneralization.

Target/SWW/wordsalsodisplayunstablebehaviorinthespeechofLuma,untilthe

endoftheobservationperiod.However,twomomentsareobserved:first,/SWW/wordsare

truncatedtomonosyllables(untilsession23);second,/SWW/wordsareproducedas[SW]

(sessions25,34and35).

Asfor/WWS/,wereinforcethattheresultsfor/SWW/truncationinLuma'sspeech

havetobeinterpretedcarefully,duetothereducednumberoftokensselected.

In(172),instancesoftruncationin/WSW/aregiven:

(172) Luma–truncationintarget/WSW/([SW]):

Orthogr. Gloss Target Output Agemenina 'girl' /mˆ»ninå/ [»¯inå] 2;2.22(S29)Franscisco 'name' /fRå‚»siSku/ [»tiku˘] 2;3.26(S30)menina 'girl' /mˆ»ninå/ [»ninå] 2;3.26(S30)magoa 'ithurts' /må»goå/ [»goå] 2;4.11(S31)festinha 'caress' /fˆS»ti¯å/ [»ti¯å] 2;4.25(S32)comando 'remote' /ku»må‚du/ [»må‚du] 2;5.15(S33)

In(173)wepresentsomeinstancesofLuma'struncationintarget/SWW/.

(173) Luma–truncationof/SWW/to[S]and[SW]:

Orthogr. Gloss Target Output Agepássaro 'bird' /»pasåRu/ [»bwa] 1;11.15(S23)lóbulo 'lobe' /»lçbulu/ [»nou] 1;11.29(S24)pássaro 'bird' /»pasåRu/ [»bWa] 2;0.13(S25)pássaro 'bird' /»pasåRu/ [»ba˘] 2;0.13(S25)óculos 'glasses' /»çkuluS/ [»çku]máquina 'machine' /»makinå/ [»makjå] 2;5.20(S34)

Theresults fromthetruncationpatternsobserved inthespeechof the fivechildren

indicatesadistinctdevelopmentalpathfor/WSW/,/WWS/and/SWW/trisyllables.

Table50summarizesthedevelopmentalpathoftruncationsfor/WSW/,/WWS/and

/SWW/words,inthespeechofthefivechildrenunderobservation.

224

/WSW/ >> /WWS/>> /SWW/Clara ∅>(S7‐12)[SW] ∅>(≥S11)[WS]

∅

Inês Neutral>(S6‐9)[WS] ∅>(S8‐12)[WS] ∅ > (S3‐9) [S] > (≥S13)[SW]

Joana ∅>(S7‐11)[S]>(S12)[SW] ∅>(S10‐14)[S] ∅> (S10‐11) [S]> (S14)[SW]

João ∅>(S5‐12)Neutral>(≥S12)[SW] ∅>(S9‐16)[S] ∅ > (S7‐16) [S] > (≥S17)[SW]

Luma ∅>(S8‐29)[S]>(≥S29)[SW]

∅>(≥S36)[WS] ∅>(S11‐23)[S]>(≥S34)[SW]

Table50.Developmentalpathoftruncationsintrisyllables(/WSW/,/WWS/and/SWW/)

Fromthetablepresentedabove,weobservethat,intarget/WSW/,onlyInêsfavorsa

[WS]truncationpattern,andwhen[WS]truncationsareproducedbythischild,reduplication

isobserved.Intheearlysessions,childrentendtoavoid/WWS/targets.Inasecondphaseof

theproductionof/WSW/,twochildrenusetruncationto[S](JoanaandLuma).Joãodisplays

a neutral truncation preference and Clara favors [SW]. Before /WSW/words are acquired,

truncationto[SW]isobserved(Joana,JoãoandLuma'sspeech.

The results for the truncation pattern /WSW/ do not suggest an iambic tendency.

Except for Inês, children do not favor an iambic foot. If anything, before the acquisition of

/WSW/,thePortuguesechildrenobserveddisplayapreferencefor[SW].

As /WSW/ trisyllables, /WWS/ words are not earlier selected by the children.

However, these forms emerge later than /WSW/ (attested in the speech of the five

Portuguese childrenobserved).When theyare selected, /WWS/are truncated to [S] (as in

JoanaandJoão)or[WS](Clara,InêsandLuma).

/SWW/wordsare selected laterby fourof theobservedchildren (Inês, Joana, João

andLuma).Claradoesnotselectthesewordformsatall.When/SWW/wordsareselected,

theyaretruncatedto[S]andonlylateraretheyproducedas[SW].

Target/WWS/and/SWW/areacquiredlaterthan/WSW/and,beforethat,theyare

subjectedtotruncationaswell.Theformertendtobetruncatedto[S]or[WS]andthelatter

to[S]or[SW].

5.1.3.4.Stressshiftindisyllables

In this section, we considered stress shift the strategy where children use both

syllablesoftheword,butmisplacewordstress.Forinstance,awordlikecasa'house'/»kazå/,

producedas[ka»za]wouldbeconsideredasstressshift,aswellasawordlikebalão'balloon'

225

/bå»lå‚w‚/ produces as [»bålå‚w‚]. In this account, only non‐reduplicated disyllables were

considered.

InTables51and52,weshowClara'sresultsforstressshiftin/SW/and/WS/.

Session %/SW/­[WS]1 0%(0/1)2 0%(0/0)3 0%(0/4)4 0%(0/0)5 0%(0/4)6 25%(1/4)7 0%(0/5)8 0%(0/6)9 11.11%(1/9)10 6.25%(5/80)11 1.9%(3/158)12 4.23%(6/142)

Table51.Percentageofstressshiftin/SW/(Clara)

Session %/WS/­>[SW]1 0%(0/3)2 0%(0/1)3 0%(0/0)4 0%(0/0)5 0%(0/3)6 0%(0/2)7 0%(0/23)8 0%(0/2)9 0%(0/16)10 8%(2/25)11 5.9%(4/67)12 0%(0/106)

Table52.Percentageofstressshiftin/WS/(Clara)

Intargettrochees,Clarahadamaximumof16tokenswherestressshiftoccurredina

totalof413targets. Intarget iambs,stressshiftoccurred6timesin248targets.Claradoes

notgobeyondthe25%,wherethisvaluecorrespondsto1mis‐stressedwordinfourtargets

(insession6).

InthetablesbelowweshowInês'dataforstressshift,intargetSWandWS.

226

Session %/Sw/­[wS]1 0%(0/0)2 0%(0/2)3 0%(0/4)4 3.70%(1/27)5 3.70%(1/27)6 0%(0/90)7 0%(0/74)8 5.13%(2/39)9 12.23%(17/139)10 3.64%(8/220)11 2.53%(6/237)12 2.18%(6/275)13 0.34%(1/297)14 0.40%(1/249)15 0.86%(2/232)16 0%(0/283)17 1.30%(2/154)18 0.36%(1/275)

Table53.Percentageofstressshiftin/SW/(Inês)

Session %/WS/­>[SW]1 0%(0/0)2 0%(0/0)3 12.5%(1/8)4 0%(0/0)5 25%(1/4)6 0%(0/14)7 0%(0/29)8 0%(0/39)9 0%(0/44)10 6.66%(3/45)11 5.19%(4/77)12 3.87%(6/155)13 1.81%(2/110)14 1.45%(2/137)15 3.06%(3/98)16 1.58%(2/126)17 0%(0/74)18 0.86%(1/115)

Table54.Percentageofstressshiftin/WS/(Inês)

LikeClara,Inêshasareducedpercentageofstressshift,bothintargetSWandWS.In

atotalof2624targettrocheesacrosssessions,Inêsproduced48tokenswithstressshift.In

iambs,inatotalnumberof1075targetwords,Inêsproduced25tokenswithstressshift.

Inthefollowingtables,weshowJoana'sdataforstressshiftin/SW/and/WS/.

Session %/SW/­[WS]1 0%(0/0)2 0%(0/0)3 0%(0/2)4 0%(0/0)5 0%(0/0)6 0%(0/0)7 0%(0/16)8 25%(1/4)9 7.69%(1/13)10 0%(0/16)11 6.67%(3/45)12 2.2%(2/91)13 0%(0/74)14 3.68%(6/163)

Table55.Percentageofstressshiftin/SW/(Joana)

Session %/WS/­>[SW]1 0%(0/0)2 0%(0/0)3 0%(0/0)4 0%(0/0)5 0%(0/0)6 0%(0/0)7 0%(0/2)8 0%(0/1)9 0%(0/14)10 7.14%(1/14)11 3.33%(1/30)12 0%(0/30)13 3.7%(1/27)14 2.04%(1/49)

Table56.Percentageofstressshiftin/WS/(Joana)

227

Joanaalsodisplaysa reduced rateof stress shift,both in target trocheesand target

iambs. Insession8, Joanahas25%ofstressshiftbut it corresponds to theproductionof1

tokenwithstressshift,outof4targets.Intargetiambs,thepercentageofstressshiftiseven

morereduced,neversurpassingthe8%.

InTables57and58,weshowJoão'sdataforstressshiftin/SWand/WS/.

Session %/SW/­[WS]1 0%(0/6)2 0%(0/0)3 50%(4/8)4 0%(0/6)5 50%(2/4)6 0%(0/0)7 13.33%(2/15)8 11.11%(2/18)9 15.38%(2/13)10 17.99%(7/39)11 0%(0/17)12 16.67%(2/12)13 43.5%(10/23)14 23.53%(4/17)15 15.29%(1/7)16 8.82%(3/34)17 0%(0/58)18 0%(0/67)19 0%(0/77)20 0%(0/90)21 1.16%(1/86)22 5.08%(3/59)

Table57.Percentageofstressshiftin/SW/(João)

Session %/WS/­>[SW]1 0%(0/0)2 0%(0/1)3 0%(0/4)4 0%(0/4)5 0%(0/0)6 0%(0/3)7 0%(0/4)8 0%(0/12)9 0%(0/12)10 0%(0/11)11 25%(1/4)12 0%(0/3)13 10%(1/10)14 0%(0/2)15 0%(0/4)16 18.75%(3/16)17 10%(1/10)18 10%(3/30)19 1/1120 2/2221 3/1222 2/13

Table58.Percentageofstressshiftin/WS/(João)

Joãoistheonlychildwherethepercentageofstressshiftmaysuggestthatthereisan

iambic tendency. Ifwecompare thepercentageof stress shift in target trocheesand target

iambs,weobservethatitishigherintargettrochees(cf.sessions3‐10and12‐15).Moreover,

thenumberscanreachthe50%intheearlystages(untilsession15,whenthereisareduced

numberoftargettrochees).Betweensession7and15stressshiftintargettrochees,butnot

intargetiambs,isanavailablestrategyforJoão,aswell.However,fromsession16tosession

18,highervaluesareobservedintarget/WS/to[SW].

In(174)weshowsomeinstancesofstressshift(/SW/‐>[WS])inJoão'sspeech.

228

(174) João–stressshift/SW/‐>[WS]:

Orthogr. Gloss Target Output Ageágua 'water' /»agWå/ [a»Ba] 1;3.21(S7)bolo 'cake' /»bolu/ [bo»u] 1;4.17(S8)uva 'grape' /»uvå/ [då»du˘] 1;5.12(S9)bola 'ball' /»bçlå/ [bo»a] 1;5.26(S10)bola 'ball' /»bçlå/ [bå»wa] 1;7.0(S12)bolo 'cake' /»bolu/ [bo»jo] 1;8.4(S14)

LikeJoão,Lumaalsousesstressshift.However,contrarytoJoão,Lumapreferstouse

stressshift in target iambs, thereforeproducing trochees.Thisstrategy ismorecommonat

theendoftheobservationperiod(fromsession28tosession31).

Tables59and60showLuma'sdataforstressshiftin/SW/and/WS/.

229

Session %/SW/­[WS]1 0%(0/1)2 0%(0/0)3 0%(0/9)4 0%(0/1)5 0%(0/5)6 14.29%(1/7)7 4.54%(1/22)8 0%(0/26)9 0%(0/13)10 0%(0/2)11 0%(0/2)12 0%(0/2)13 0%(0/11)14 0%(0/8)15 0%(0/16)16 0%(0/1)17 0%(0/31)18 0%(0/14)19 0%(0/4)20 0%(0/31)21 0%(0/21)22 0%(0/33)23 2.44%(1/41)24 0%(0/34)25 0%(0/30)26 8.33%(2/24)27 38.89%(7/18)28 10%(3/30)29 2.63%(1/38)30 7.38%(9/122)31 9.32%(11/118)32 3.37%(7/208)33 0.93%(2/215)34 0%(0/217)35 0%(0/146)36 0.87%(2/229)37 2.06%(3/145)

Table59.Percentageofstressshiftin/SW/(Luma)

Sessions %/WS/­>[SW]1 0%(0/0)2 0%(0/0)3 0%(0/0)4 0%(0/0)5 0%(0/0)6 0%(0/0)7 0%(0/0)8 0%(0/0)9 0%(0/0)10 0%(0/0)11 0%(0/1)12 0%(0/0)13 0%(0/0)14 0%(0/6)15 0%(0/2)16 0%(0/36)17 0%(0/0)18 0%(0/0)19 0%(0/0)20 12.5%(2/16)21 0%(0/7)22 0%(0/0)23 0%(0/10)24 0%(0/1)25 0%(0/7)26 0%(0/2)27 0%(0/1)28 23.07%(3/13)29 25%(3/12)30 31.57%(6/19)31 25.92%(7/27)32 2%(1/50)33 8.69%(4/46)34 5.45%(3/55)35 11.42%(4/35)36 5%(5/100)37 6.12%(3/49)

Table60.Percentageofstressshiftin/WS/(Luma)

230

In(175)weshowLuma'srenditionsofstressshifttowardstrochees.

(175) Luma–stressshift/WS/‐>[SW]:

Orthogr. Gloss Target Output AgePati 'name' /pa»ti/ [»pati] 2;2.4(S28)Miguel 'name' /mi»gE…/ [»niå‚] 2;2.22(S29)Mami 'name' /ma»mi/ [»mami] 2;3.26(S30)

Thefirstobservationthatwecandrawisthatstressshiftwasaninfrequentstrategy

in all children. However, two children (João and Luma) displayed stress shift in the

production of /SW/ and /WS/. Whereas João tends to misstress target /SW/ words,

producing them as [WS], Luma displays the opposite behavior, producing [SW] for target

/WS/words. In both children, stress shift is a strategy used before the acquisition of the

predominantdisyllabictemplate(/SW/or/WS/).InJoão,stressshiftto[WS]occursbefore

theacqusitionof/SW/wordsand,inLuma,stressshiftoccursbeforetheacquisitionof/WS/

words.Intheremainderchildren(Clara,InêsandJoana)stresserrorsareproducedinhighly

reducedpercentages.

5.1.3.5.Summaryfortheproductionstrategies

InthissectionweanalyzedthestratagiesusedbythefivePortuguesechildrenunder

analysis in target monosyllables, disyllables (/SW/ and /WS/) and trisyllables (/WSW/,

/WWS/and/SWW/).

In targetmonosyllables,we observed that,whenmonosyllableswerenot produced

target‐like, theyweresubjecttoreduplicationandepenthesis.Unglidingwasalsoobserved,

thoughinsmalleramounts.Thoughallchildrenusedthesestrategiesintargetmonosyllables,

Clara,JoãoandLumausedthemacrosstheentireobservationperiod(thoughinadecreasing

fashion), Inês used themmainly in the early sessions (until sessions 8‐9) and Joana used

theminverysmallamount.

As for targetdisyllables,ourdata showed that in thevastmajorityof thecases, the

syllable preserved as the result of truncation is the stressed syllable. In the early sessions,

both /SW/ and /WS/ are subject to high rates of truncation. Inês has a preference for

truncation in /WS/at thebeginning. Joana, João andLumahavehigher truncation rates in

trochees,initially,thoughmuchvariationisfoundintheearlysessions.Claradoesnotdisplay

any tendency. In the last sessions iambsaremoreprone to truncation than trochees in the

productionsofthefivechildrenobserved.

231

The results found for the strategies indisyllables indicate that, at thebeginning,no

generalpreferenceisattestedand, later,allchildrentendtohavehighertruncationratesin

/WS/.

Trisyllables were not selected in the early sessions in four of the five observed

children(Inêswastheonlychildselecting/WSW/intheonsetofwordproduction).

The first trisyllables toemerge in the children's speechwere/WSW/words.At the

beginning of /WSW/ selection, children chose variable truncation paths: Clara truncated

theseword forms to [SW]. Inês and João showed a neutral tendency (both [SW] and [WS]

werenoticeable)andJoanaandLumahadapreferencefor[S],initially.before/WSW/were

acquired,allchildren(exceptforInês)hadapreferencefor[SW]truncation.

Target/WWS/werenotearlierselectedbythechildren.Whentheywereattempted,

theyweremainlytruncatedto[S](inJoanaandJoão)or[WS](ClaraandInês).Lumausedthe

truncationto[SW],aswell.

Target /SWW/ were the trisyllabic forms later acquired. In the early moments of

selection, theywere produced as [S] by all children (except Clara, who did not select any

/SWW/word).Later,theywereproducedas[SW].

Stressshiftwasnotastrategyrecursivelyusedbythreeinthefivechildrenobserved.

Joãousedstressshiftinhigheramountsintheearlysessions,favoringtheproductionof[WS]

(/SW/wordswereproducedas[WS]),whereasLumadisplayedtheoppositebehavior(she

produced[SW]in/WS/).

The results from the strategies observed in the speech of the five children under

observationdonotsuggestaniambictendency.

5.1.4.Summaryofresults

Inthischapter,weanalyzedthestresspatternsintheproductionsoffivePortuguese

children.Weinvestigated(i)theproductionpatterns,(ii)thefaithfulnesstotargetwordsand

(iii)theproductionstrategiesusedindealingwithwordstressandwordshape.

Asfarastheproductionpatternsareconcerned,thedatapresentedshowedthat:

(i) At the early stages of word production, there is a general preference for

monosyllabic forms(cf.Tables16‐20,Figures29‐33).Thesemonosyllablesmay

resultfromtargetmonosyllables,butalsofromthetruncationoftargettrochees

andtarget(cf.instancesfor[S]in(112)‐(116));

(ii) [WS]patternsareproducedearlierthan[SW](cf.Tables16‐20);

232

(iii) Atthebeginning,[WS]areingreateramountsthan[SW](Tables16‐20);

(iv) From a certain point in development, the amount of [SW] words widely

surpassestheamountof[WS](cf.Tables16‐20);

(v) Most early words are reduplications and words produced with epenthesis,

creating[WS]patterns(cf.Figures29‐33);

(vi) Theamountofreduplicationsandepenthesisdecreasesimportantlyinthecourse

ofdevelopment(cf.Figures29‐33).

With respect to the faithfulness to the target, our results enable us to draw the

followinggeneralizations:

(vii) If results are analyzed including reduplicated targets (/CV1CV1/) and

reduplicatedproductions ([CV1CV1]), aswellasproductionswith filler syllables

([fS]),EPwordstressacquisitionhasancleariambictendency;

(viii) If non‐reduplicated words are considered, i.e., if we consider /SW/ and /WS/

(/CV1'CV2/)wordsandifwedonotconsiderintermsoffaithful,theproductions

where childrenproduce iambic reduplications ([CV1'CV1]) for /CV1'CV2/words,

no clear tendency is observed, or a slight tendency for trochees is attested (cf.

Tables34‐38);

(ix) Threechildren(Clara, Inêsand Joana)haveanearlieracquisitionof/SW/than

/WS/; one child (João) has a simultaneous acquisition of both structures; one

child (Luma)acquires/WS/earlier than/SW/ (cf.Tables34‐38andTable39,

forasummary).

The results regarding the production strategies observed in the speech of the five

childrenindicatedthat:

(x) Disyllables are prone to truncation and the same strategies as monosyllables:

reduplicationandepenthesis(cf.Table45‐49andinstances(117)‐(121));

233

(xi) When truncationswere carriedoutby the children, the stressed syllableof the

target words was overwhelmingly the preserved syllable (cf. instances (152)‐

(160)).

(xii) Truncation in/SW/and/WS/showedanearlyvariablepath,bothwithin‐child

andbetween‐children:intheearlysessions,Inêshasapreferencefortruncation

in /WS/ (cf. Table 46). Initially, Joana, João and Luma have higher truncation

rates in trochees, though some variation is found (cf. Tables 47, 48 and 49,

respectively).InClaranotendencyisobserved(cf.Table45).

(xiii) Theresultsfoundforthestrategiesindisyllablesindicatethat,atthebeginning,

no general preference is attested and, later, all children tend to have higher

truncationratesin/WS/(cf.Tables44‐49).

(xiv) Close to the moment or in the moment in which children acquire /SW/, all

childrendisplayapreferencefor[SW]truncationin/WSW/(cf.Tables34‐38and

Figures39,41,44,47,50);

(xv) OnlyInêsfavors[WS]truncatedformsintarget/WSW/(cf.Figure41);

(xvi) Ingeneral,target/WWS/areproducedeitheras[S](earlier)or[WS](later)(cf.

Figures,40,42,45,48and51);

(xvii) Ingeneral,target/SWW/areearlierproducedas[S]andlaterproducedas[SW]

(cf.Figure43,46,49and52).

(xviii) Stress shift is a strategy used in reduced percentages. Only two children

display higher percentages of this strategy: one child favors [WS] (João), the

otherchildfavors[SW](Luma)(cf.Tables51‐60).

It is worth noticing, also, that we found variable behavior with respect to stress

patterns and word shape in the early sessions (both within‐child and between‐children).

Later,thisvariabilitywasneutralizedandchildrenappeartohaveamorecommongroundin

theproductionofstresspatterns.

234

In sum, the acquisition of stress patterns by the five Portuguese children observed

mightbesketchedasfollows:/SW/(~/WS/183)>(/WS/)>/WSW/>/WWS/>/SWW/.

Inthefollowingsection,wewilldiscusstheresultsfoundandproposeaphonological

analysisfortheacquisitionofstresspatternsinEP.

5.2.Discussion

Inthissection,wewilldiscusstheresultsfoundintheacquisitionofstresspatternsin

EP,basedontheempiricalevidencefromthespeechproductionsoffivePortuguesechildren.

This discussionwill be divided into two parts: in the first part (5.2.1.), wewill propose a

developmental scale for theacquisitionof stresspatterns (/SW/,/WS/and/WSW/) inEP.

Wewilltacklethedevelopmentalpathproposed,onthelightofourresults,andcomparethe

acquisition path undertaken by Portuguese children with the descriptions previously

conducted on other languages, namely Dutch. In the second part (5.2.2.), wewill focus on

some aspects related to theproductiondata in the early stages of the acquisition of stress

patternsandwordshape.Wewilldiscusstheapparentearlyiambicfootandthehypothesis

according to which the early acquisition of iambs might be the result of a higher level

prominence(specifically,phrasalstress).

5.2.1.Developmentalpathfortheacquisitionofwordshapeandstresspatterns

inEP

In this section, wewill propose a developmental scale for the acquisition of stress

patternsandwordshapeinEP.Thisscalewillbebasedonthespeechproductionsofthefive

Portuguesechildrenobservedinthisproject.

As summarized in Table 61, overall results suggest that, despite some language‐

specific aspects, EP word stress acquisition, proceeds much like many other trochaic

languages.

183Themark'~'indicatesasimultaneousemergenceof/SW/and/WS/attested,forinstance,inJoão.

235

1;01;11;21;31;41;51;61;71;81;91;101;112;02;12;22;32;42;52;6Claraσ

[CV1CV2]≥S4

/SW/(>50%)≥S9

/WS/(>50%)≥S10

/WSW/(>50%)≥S11

Inêsσ

[CV1CV2]≥S4

/SW/(>50%)≥S9

/WSW/(>50%)≥S11

/WS/(>50%)≥S12

Joanaσ

[CV1CV2]≥S8

/SW/(>50%)≥S12

/WS/(>50%)≥S13

/WSW/?

Joãoσ

[CV1CV2]≥S8

/SW/(>50%)/WSW/(>50%)/WS/(>50%)≥S16

Lumaσ

[CV1CV2]≥S18

/WS/(>50%)≥S31

/SW/(>50%)≥S33

/WSW/(>50%)≥S36

Table61.Individualdevelopmentalpathforwordshapesandstresspatterns

236

Table 61 depicts the developmental path pursued by each of the five children

observed,withrespect towordshapesandstresspatterns(/σ/,/SW/,/WS/and/WSW/).

The grey grading indicates the different moments of production and acquisition of the

distinct word shapes (/σ/, /SW/, /WS/ and /WSW/). We observe up to five moments of

productionforwordshapesandstresspatterns,inthespeechofthefivechildren.Noticethat

notallchildrengothroughallmomentsand,insomechildren,theproductionofthedifferent

stresspatternsmayoverlap.Therefore,inthisdiscussionwewillaccountforthegeneralpath

undertaken by the five children observed, based on the maximum number of moments

pursued. The fivemoments in the production of the stress patternsmight be described as

follows:

1. In thebeginning,monosyllablesare thepredominantwordshape inproduction

(cf. section 5.1.1.), therefore, stress contrasts are hard to define. At this point,

longerutterancesaretheproductofreduplicationofthestressedsyllableorthe

result of epenthesis before the circumscribed stressed syllable. Some children

use more reduplication (namely, Inês) and other children prefer vowel

epenthesis (namely, Clara), but all children use these strategies at the onset of

wordproduction(cf.section5.1.3.);

2. In a second moment, non‐reduplicated disyllables emerge, though in reduced

amounts and in an unstable fashion (they can be produced target‐like or

truncatedbythesamechild‐cf.sections5.1.2.2.and5.1.3.2.);

3. Inathirdmoment,variablepathsamongchildrenareattested:inClara,Inêsand

Joana,themajorityof/SW/wordsareproducedtarget‐like.Joãoacquires/SW/,

/WS/and/WSW/stresspatterns simultaneously andLumaacquires /WS/ (cf.

resultsinsection5.1.2.2.);

4. ThefourthmomentisattestedinClara,Inês,JoanaandLuma.InClaraandJoana,

the majority of /WS/ words are produced target‐like. Inês and Luma acquire

/WSW/and/WS/,respectively;

5. ThefifthmomentisattestedinClara,InêsandLumaandischaracterizedbythe

acquisitionof/WS/inInês,and/WSW/inClaraandJoana.

237

Thetwoinitialmomentswerecommontoallchildren,thoughsomevariationisfound

intheamountoftimespentonit,ineachchild.Asobservedinsection5.1.4.,Lumaistheonly

childacquiring/WS/earlierthan/SW/.

In the following paragraphs, we will put forward the acquisition stages of stress

patternsandwordshape,basedontheobservationoftheproductiondatainanalyzedinthis

dissertation.

StageI

Our findings suggest that, despite the amount of productions larger than a

monosyllable,Portuguesechildrenusethesyllabletostartbuildingwords.

The assumption of an early processing of the syllable is borne out from the data.

Indeed, three factsmotivate the identification and description of this first stage as a stage

wherethesyllableisunderthechildren'sattention:

(i) Asfarasfaithfulnessproductionsareconcerned,weobservedthatmonosyllables

are selected earlier, and they have higher rates of target‐like production, than

non‐reduplicateddisyllables(cf.Tables23‐27and34‐38).

(ii) Asfarastheproductionpatternsareconcerned,monosyllablesarethepreferred

wordshape,irrespectiveoftheshapeofthetargetword(cf.Figures29‐33).Even

the children who produce [WS] patterns earlier can show greater amounts of

[S]184.

(iii) As far as truncations are concerned, trochees and non‐reduplicated iambs can

equallybetruncatedtomonosyllables(cf.Tables45‐49).

InStageIoftheacquisitionofstresspatterns,theobservedchildrencircumscribethe

stressedsyllableofthetargetwordandproduceitinvariousmanners:eithertarget‐like(/S/

‐>[S]),withafillerinsertion([fσ])orwithreduplication([CV1CV1]).Theseproductionswere

possible both for target monosyllables, and for target disyllables (/SW/ and /WS/), as

exemplifiedin(176).

184See,forinstance,theabsolutevaluesin[S]and[WS]wordshapesinInês(Table17).

238

(176) Productionof[S],[fσ]and[CV1CV1]in/S/,/SW/and/WS/:

Orthogr. Gloss Target Output Child,Agedá 'giveimp.' /»da/ [»da˘]já 'now' /»Za/ [»za]dá 'giveimp.' /»da/ [a»dJa]

Inês,1;0.25(S2)

cão 'dog' /»kå‚w‚/ [»ga]há 'thereis' /»a/ [»a]má 'bad' /»ma/ [må»ma]

Inês,1;1.20(S3)

mãe 'mother' /»må‚j‚/ [»å‚˘]não 'no' /»nå‚w‚/ [»nå‚˘w‚]

Joana,0;11.24(S1)

pé 'foot' /»pE/ [»pE] Joana,1;2.29(S4)pai 'father' /»paj/ [»pa] Joana,1;6.24(S7)dá 'giveimp' /»da/ [a»tJa] Luma,1;0.28(S3)dá 'giveimp' /»da/ [»dQ] Luma,1;2.22(S6)

/S/

dá 'giveimp' /»da/ [da»da] Luma,1;4.2(S9)toma 'takeimp.' /»tçmå/ [»tç] Inês,1;0.25(S2)água 'water' /»agWå/ [»a˘]papa 'foodfam.' /»papå/ [»ba»ba] Inês,1;1.30(S3)

barco 'boat' /»baRku/ [»bQ˘]bóia 'buoy' /»bçjå/ [Ba»Bå] Inês,1;3.6(S4)

gato 'cat' /»gatu/ [»ka]Carla 'name' /»kaRlå/ [»ta˘]papa 'foodfam.' /»papå/ [»pa˘]

Joana,1;6.24(S7)

Carla 'name' /»kaRlå/ [»ka]gato 'cat' /»gatu/ [´»ga] Joana,1;8.4(S8)

gato 'cat' /»gatu/ [»ta]lua 'moon' /»luå/ [»nå]pato 'duck' /»patu/ [»pa]

Luma,1;3.5(S7)

bola 'ball' /»bçlå/ [»pa] Luma,1;4.2(S9)

/SW/

bola 'ball' /»bçlå/ [pa»pa] Luma,1;8.2(S17)chapéu 'hat' /Så»pEw/ [pE»pQ˘] Inês,1;3.6(S4)mamã 'mommy' /må»må‚/ [å»må‚] Joana,1;0.25(S2)mamã 'mommy' /må»må‚/ [må»må‚] Joana,1;2.7(S3)

/WS/

avó 'grandmother' /å»vç/ [»a] Joana,1;8.4(S8)

The instances showed above indicate that children correctly produce target

monosyllablesandthattheytruncatedisyllables(/SW/and/WS/)intomonosyllables.Inthe

caseoftruncation,childrenoverwhelminglycircumscribeandpreservethestressedsyllable.

239

The instances presented in (176) also show that children use epenthesis and

reduplication both in targetmonosyllables and in target disyllables185. Indeed,we observe

thatchildrenproducebothmonosyllablesanddisyllablesinthefirststageoftheacquisition

of stress patterns and word shape. The monosyllables produced might be the product of

targetmonosyllablesproducedtarget‐like(/σ/‐>[σ]),ortheresultoftruncation(/‐SW/‐>

[σ] or /‐WS/ ‐> [σ]). The disyllables produced might also be the product of target

monosyllables being reduplicated (/σ/ ‐> [CV1CV1]) or produced with epenthesis (/σ/ ‐>

[fσ]) or, finally, be the result of target disyllables being reduplicated and produced with

epenthesisaswell(/CVCV/‐>[CV1CV1],[fσ]).

Since both target monosyllables and disyllables are produced in the samemanner

(theycanbeproducedasamonosyllable,be reduplicated,orbe subject to filler insertion),

and since the frequency information on the distribution of word shapes in the target (cf.

Figure8)pointedtoamajorityofdisyllables,onemightaskwhyisthattheearlywordshape

isamonosyllable,andnotadisyllable.Also,sincethedisyllabicproductionshaveaniambic

shape,onemightaskwhytheearlywordisnotadisyllabiciamb.Onceagain,fourfactscanbe

brought up in order to argue against a disyllabic representation on the one hand, and an

iambicrepresentationontheotherhand,forearlywords:

(i) Non‐reduplicated/WS/wordsarenotearlier selectedor theyare selectedand

producedinsmallamounts.Onthecontrary,targetmonosyllablesareproduced

inhighrates.Tables62and63illustratethisbehaviorinthefirst10sessionsof

datacollection186.

185Asshowninprevioussections(cf.,inparticular,Figures29‐33,insection5.1.1.2.),productionsoflargerwords(where reduplication and epenthesis are found) mainly have a [WS] shape. The apparent iambic form ofPortuguesechildren'searlywordswillbediscussedinthefollowingsection(section5.2.2.).186Thevaluesmarkedwithastar, inClaraandJoão,correspondtotheproductionofonesingletype(olá 'hello'/ç»la/).Insession5,Claraalsoselectsonetimethewordpinguim'penguin'/pi‚»gWi‚/.

240

%/WS/SessionClara Inês Joana João Luma

S1 100* ‐ ‐ ‐ ‐S2 100* ‐ ‐ 100* ‐S3 ‐ 0 ‐ 100* ‐S4 ‐ ‐ ‐ 100* ‐S5 66.7* 0 ‐ ‐ ‐S6 100* 21.4 ‐ 100* ‐S7 87 0 0 75 ‐S8 50 28.2 0 16.7 ‐S9 31.3 38.6 28.6 0 ‐S10 80 46.7 35.7 54.5 ‐

Table62.Percentageoftarget­likenon­reduplicated/WS/words

%/S/SessionClara Inês Joana João Luma

S1 ‐ 0 100 ‐ 0S2 0 59.1 100 25 50S3 100.0 36.8 100 0 ‐S4 85.7 40.2 100 ‐ ‐S5 71.4 69.2 100 ‐ ‐S6 46.2 62.8 66.7 ‐ 92.3S7 74.1 67.5 100 66.7 87.5S8 64.7 71.9 87.5 100 33.3S9 68.2 88.7 85.7 61.5 50.0S10 94.1 87.5 83.9 76.9 100.0

Table63.Percentageofmonosyllablesproducedtarget­like

The comparison of the two tables indicates that, indeed, target monosyllables are

produced inhigherpercentages than/WS/. Ifweassumeadisyllabic /WS/ representation

forearlywords,thereisnoreasonforthepaucityofnon‐reduplicated/WS/,especiallywhen

comparedtothehighfrequencyofreduplicatediambs.

(ii) Non‐reduplicated/WS/aregenerallysubjecttotruncationforalongerperiodof

time (though notmuch longer) than /SW/, as demonstrated in section 5.1.3.2.

and illustrated in the instances in(177). In these instancesweobservethat the

samechildtruncates/WS/foralongerperiodoftimethan/SW/.

241

(177) Latertruncationin/WS/thanin/SW/:

Orthogr. Gloss Target Output Child,Ageporta 'door' /»pçRtå/ [»pç] Inês,1;7.2(S8)tampa 'lid' /»tå‚på/ [»pa]perna 'leg' /»pERnå/ [»bE] Inês,1;8.2(S9)

Paula 'name' /»pawlå/ [»pa] Joana,1;10.22(S9)Nando 'name' /»nå‚du/ [»¯å‚] leite 'milk' /»låjtˆ/ [»¯åj] Joana,2;0.9(S11)

mano 'brotherfam.' /»månu/ [»må]bola 'ball' /»bçlå/ [»bç˘]

João,1;5.26(S10)

mota 'motorbike' /»mçtå/ [»ta] João,1;6.16(S11)gato 'cat' /»gatu/ [»ta] Luma,1;11.29(S24)meias 'socks' /»måjåS/ [»bE˘] Luma,2;0.13(S25)

/SW/

polvo 'octopus' /»po…vu/ [»bo˘] Luma,2;0.13(S25)papel 'paper' /på»pE…/ [»pE˘] Inês,1;7.2(S8)chapéu 'hat' /Så»pEw/ [»pEw] aqui 'here' /å»ki/ [»ki] Inês,1;8.2(S9)

abrir 'toopen' /å»bRiR/ [»biR]chapéu 'hat' /Så»pEw/ [»Ew] Inês,1;10.29(S11)

avó 'grandmother' /å»vç/ [»dJç] Inês,2;0.11(S12)cristal 'crystal' /kRiSta…/ [»taw]atrás 'behind' /å»tRaS/ [»tJaS˘]

Joana,2;2.19(S12)

chapéu 'hat' /Så»pEw/ [»pEw]bacio 'basin' /bå»siw/ [»Siw]chover 'torain' /Su»veR/ [»vej]

Joana,2;4.1(S13)

limão 'lemon' /li»må‚w‚/ [»å‚˘w‚˘] João,1;6.16(S11)sofá 'couch' /su»fa/ [»på˘] João,1;7.20(S11)Natal 'Christmas' /nå»ta…/ [»ta] João,1;8.4(S14)Inês 'name' /i»neS/ [»ne˘]sofá 'couch' /su»fa/ [»fa˘]

Luma,2;2.4(S28)

chinês 'Chinese' /Si»neS/ [»ne˘] Luma,2;2.22(S29)aqui 'here' /å»ki/ [»ki˘]

/WS/

ali 'there' /å»li/ [»¥i˘] Luma,2;3.26(S30)

(iii) Infourofthefiveobservedchildren,stressshift in/SW/to[WS] isveryscarce

andonechildshowhighervaluesinstressshiftfor[SW](cf.Tables51‐60).

Accordingtothedata,weclaimthat,intheearlystagesofwordproductioninEP,the

children'sprosodicrepresentationconsistsinasyllableandthereisnocomplexorganization

withintheworddomain.Thecircumscribedsyllableisthetargetstressedsyllableanditcan

beusedinareduplicatedproductionorbesubjecttoepenthesisattheleft‐edge.

In (178), we represent the child's output form in Stage I. We will follow the

242

representationproposedbyDemuth(2001b:8)187, forearlyreduplicated iambs188produced

by Spanish children. Demuth'smotivation for the representation presented in (178) is the

alternatingproductionoftrisyllabic/WSW/wordsas[SW]and[WSW].However,inStageI,

wedonotfindthisalternationinPortuguese‐speakingchildren.Trisyllablesareselectedand

producedlaterand,whentheyareattemptedinthebeginning,theytendtobetruncated189.

Therefore, theassumptionof apositionat the left‐edgeof theprosodicwordmotivatedby

alternatingearlyproductionsof[SW]and[WSW]doesnotholdforEP.Instead,wemotivate

thestructurerepresentingearlywordsinPortuguesechildren(in(178))onthreeempirical

facts:

(i) Early words consist mostly in monosyllables, reduplications of the syllable

circumscribedfromthetargetwordorproductionsofthecircumscribedsyllable

precededbyafillersound,thelatter(reduplicationsandepenthesis)resultingin

iambicforms;

(ii) Theunstressedsyllablefoundintheseearlystructures(i.e.,theinitialsyllableof

a /WS/ reduplication or a filler syllable) is not mapable onto the unstressed

syllable of the target prosodic word (cf., for instance, the instances in (116)‐

(120));

(iii) EPisaprocliticlanguage190‐thereisamuchhighertendencyinthelanguageto

placeunstressedwordsattheleftofaProsodicWordthantoplaceitattheright

ofaProsodicWord.

Sincetheinitialunstressedsyllable(eitherafillersoundortheinitialsyllablewithina

/WS/ reduplication) does not correspond to the unstressed syllable of the target lexical

item191 and given the language's tendency to place clitics at the left edge of the Prosodic

Word,weassumethatunstressedsyllablesareoccupyingapositionatthe leftofthetarget

stressedsyllable inorder tomatchtheProsodicWordtemplate.Sinceweareassumingthe

syllable as the target of children's attention, in this stage, any elements preceding the

circumscribedsyllableareoptional.

187Cf.Chapter2,section2.2.2..188Infact,theearlyreduplicatedwordswitha[WS]shapearenotiambicfeet,astherepresentationproposedbytheauthorsuggest.189Cf.section5.1.3.3.,andStageIVfurtherinthissection.190 Frota, Vigário & Martins (2006:2227) state that in EP "[o]f all clitics, 97% are proclitics and only 3% areenclitics."191Aswewillobserve further inStage III,expectedly,disyllableswillbeproducedwith the twosyllablesof thetargetword.

243

Therepresentationpresentedin(178)forthechild’soutputforminStageIpredicts

thatPortuguesechildrenpickthestressedsyllablefromthetargetform(e.g.,[»kE]and[»på]

from thewordscreme 'lotion' and sofá 'sofá', respectively).This circumscribedsyllable can

surfaceasareduplicatedproduction(e.g.,[kE»kE],[må»må])orbeprecededbyafillersound

(e.g., [ˆ»kE], [å»på]). Multiple reduplications (e.g., [»kE»kE»kE]) are also accounted for by this

representation, as any syllable might be optionally preceded by any other (stressed or

unstressed)syllable.

The preference for monosyllabic words, the high production of target‐like

monosyllablesandthehighfrequencyofreduplicationandepenthesisseemtoindicatethat

Portuguese children are circumscribing the stressed syllable, interpreting it as a Prosodic

Word.

StageII

In the second stage of word stress acquisition children realize that words can be

longerthanamonosyllableandtheystartproducingnon‐reduplicateddisyllables([CV1CV2]‐

/SW/–e.g.,pato‘duck’[»patJu],Inês,1;6,‐and/WS/–e.g.avô‘grandfather’[å»dJç],Inês,1;7).

In this stage, very few types and tokens of eachword type are targeted and produced, in

general.

In (179) and (180), respectively,we illustrate the stagewhere /SW/and/WS/are

produced target‐like, though truncation tomonosyllables is still a frequent strategy, in the

samesessionofthechild.

(178) Child’soutputform–StageIOptionalsyllablesaremarkedwith()

e.g.[»pa]/[pa»pa]bola‘ball’[å»må‚]mamã'mommy'

ωg Σ g σ g

[( ºpa)pa]

[(å)må‚]

244

(179) Productionsof/SW/(truncatedandtarget‐like):

Orthogr. Gloss Target Output Child,Agecarro 'car' /»ka{u/ [»ka] Clara,1;3.6(S5)pêlo 'hair' /»pelu/ [»pe] Inês,1;4.9(S5)fralda 'diaper' /»fRa…då/ [»ka˘] Inês,1;5.11(S6)cartas 'cards' /»kaRtåS/ [»ka]garfo 'fork' /»gaRfu/ [»ƒa] Inês,1;6.11(S7)

Carla 'name' /»kaRlå/ [»ka] Joana,1;8.4(S8)luva 'glove' /»luvå/ [»bu˘] Joana,1;10.22(S10)grande 'big' /»gRå‚dˆ/ [»gA‚] Joana,2;0.9(S11)bola 'ball' /»bçlå/ [»bo] João,1;4.17(S8)mano 'brotherfam.' /»månu/ [»må] João,1;5.26(S10)

Truncated

pato 'duck' /»patu/ [»på] João,1;6.16(S11)água 'water' /»agWå/ [»a˘Bå] Clara,1;3.6(S5)água 'water' /»agWå/ [»ak´] Clara,1;4.19(S6)mano 'brotherfam.' /»månu/ [»månu] Clara,1;5.16(S7)bóia 'buoy' /»bçjå/ [»bˆ´] Inês,1;4.9(S5)minha 'mine' /»mi¯å/ [»min´] Inês,1;5.11(S6)pato 'duck' /»patu/ [»pa˘tJu] Inês,1;6.11(S7)pato 'duck' /»patu/ [»paku] Joana,1;9.25(S9)colo 'lap' /»kçlu/ [»kç˘u] Joana,1;10.22(S10)linda 'beautiful' /»li‚då/ [»¯i‚¯å] Joana,2;0.9(S11)bolo 'cake' /»bolu/ [»bowu] João,1;4.17(S8)bola 'ball' /»bçlå/ [»boå] João,1;5.26(S10)

Target­like

mota 'motorbike' /»mçtå/ [»tadå] João,1;6.16(S11)

245

(180) Productionof/WS/(truncatedandtarget‐like):

Orthogr. Gloss Target Output Child,Ageaqui 'here' /å»ki/ [»ki] Clara,1;5.16(S7)chapéu 'hat' /Så»pEw/ [»pEj] Clara,1;6.6(S8)aqui 'here' /å»ki/ [»i]João 'name' /Zu»å‚w‚/ [»dJa˘w˘]

Inês,1;4.9(S5)

champô 'shampoo' /Så‚»po/ [»po] Inês,1;5.11(S6)colher 'spoon' /ku»¥ER/ [»kE] Inês,1;6.11(S7)Raquel 'name' /{å»kE…/ [»kE] Joana,1;9.25(S9)avó 'grandmother' /å»vç/ [»fç] Joana,1;10.22(S10)Jesus 'name' /Zˆ»zuS/ [»DJu] Joana,2;0.9(S11)limão 'lemon' /li»må‚w‚/ [»mo‚] João,1;5.12(S9)Natal 'Christmas' /»nå»ta…/ [»tå˘] João,1;7.20(S13)

Truncated

olá 'hello' /ç»la/ [»ja] João,1;8.25(S15)olá 'hello' /ç»la/ [碻ja] Clara,1;3.6(S5)aqui 'here' /å»ki/ [å»ki] Clara,1;5.16(S7)aqui 'here' /å»ki/ [å»tJi] Inês,1;5.11(S6)avô 'grandfather' /å»vo/ [å»dJç] Inês,1;7.2(S8)café 'coffee' /kå»fE/ [ki»kE] Joana,1;9.25(S9)avó 'grandmother' /å»vç/ [å»bç˘] Joana,1;10.22(S10)balão 'balloon' /bå»lå‚w‚/ [m´»¯å‚w‚] Joana,2;0.9(S11)olá 'hello' /ç»la/ [ç»wa] João,1;5.26(S10)

Target­like

João 'name' /Zu»å‚w‚/ [nu‚»å‚w‚] João,1;8.4(S14)

InStageII, reduplicationsarestillpresentbutarebecoming less frequent,andnon‐

reduplicated disyllables, on the contrary, are increasingly produced. The onset of non‐

reduplicated disyllables is characterized by the simultaneous emergence of both target

trocheesandtargetiambs.

Thebeginningofdisyllabicproductions([CV1CV2])is,presumably,atransitionalstage

that some childrengo through from the finalphaseof Stage I to the activationof Stage III,

where children will show a preference for /SW/. Since no hierarchy is stabilized in the

children'sprosodictemplateinthisperiod,nofixedwordprominenceisyetassigned.

246

Therepresentationin(181)showsthat,inStageII,both/SW/and/WS/wordsmay

be attempted, as children are still decidingwhat the language rhythmat theword level is.

Thisrepresentationpredictsthatchildrentrydisyllabicwords([CV1CV2]),whichcansurface

as[SW],non‐reduplicated[WS]orasyllablepredecedbyafillersound([fS]),andcanalsobe

truncatedtoamonosyllable(bothtarget/SW/and/WS/).Duetothefactthatdisyllablesare

onlyemergingandthatitisnotpossibletopredictanypreferenceforanyfootshapeatthis

stage(bothabsolutevaluesandpercentagesshowsimilarvaluesand theseareparticularly

lowforbothtemplates),weclaimthatnodefaultfootissetatthisstageofwordproduction

inEP.

StageIII

In Stage III of word stress acquisition, Portuguese children realize that EP has a

trochaicrhythmandthatsyllablesareorganizedintoadisyllabic/SW/foot.Inthisstage:

(i) theamountof [SW]productionsoverwhelminglysurpassestheamountof [WS]

(cf.Tables16‐20);

(ii) ingeneral,/SW/havehighertarget‐likeratesthan/WS/(cf.Tables34‐38);

(iii) thetruncationrate innon‐reduplicated/WS/ishigherthanin/SW/(cf.Tables

45‐49);

(iv) reduplicationsandepenthesisarealmostinexistent(cf.Figures29‐33);

(v) thepreferentialtruncationpatternin/WSW/is[S]or[SW],ratherthan[WS](cf.

Figures39,41,44,47,50).

(181) Child’soutputform–StageIIOptionalsyllablesaremarkedwith()

e.g.[»patJu]pato‘duck’,[å»dJç]avó'grandmother'

ωg Σ g g

(σW) σS (σW)

[»pa tJu]

[å »dJç]

247

In Stage III, prominence is assigned and the trochaic foot is being processed.

Examplesin(182)illustratethechildren'sproductionofthetrochaicfoot,as/SW/wordsare

produced target‐like and /WSW/ are truncated to [SW]. In (183), we present renditions

where/WS/wordsaretruncatedto[S],inthesamesessions,whereweobservethatchildren

simultaneouslyproduce/WSW/and/SW/as[SW],buttruncate/WS/into[S].

(182) Productionof[SW]intarget/SW/and/WSW/:

Orthogr. Gloss Target Output Child,Agemana 'sisterfam' /»månå/ [»månå] Clara,1;7.11(S9)Noddy 'name' /»nçdi/ [»nçtHi] Clara,1;8.20(S10)pato 'duck' /»patu/ [»patJu] Clara,1;923(S11)isto 'this' /»iStu/ [»itJu]tampa 'lid' /»tå‚på/ [»patJå]

Inês,1;8.2(S9)

sete 'seven' /»sEtˆ/ [»dEtå]flores 'flowers' /»floRˆS/ [»tol´S] Inês,2;0.11(S12)

Nando 'name' /»nå‚du/ [»na¯u] Joana,2;2.19(S12)mota 'motorbike' /»mçtå/ [»mç˘tJå]mala 'purse' /»malå/ [»mawå] Joana,2;4.1(S13)

brinca 'play' /»bRi‚kå/ [»mi‚kå] Luma,2;4.25(S32)Noddy 'name' /»nçdi/ [»nçdi]

/SW/

queijo 'cheese' /»kåjZu/ [»keCu] Luma,2;5.20(S34)

Aurora 'name' /aw»RçRå/ [»lalå] Clara,1;7.11(S9)menina 'girl' /mˆ»ninå/ [»¯i¯å] Clara,1;9.23(S11)vestida 'dressed' /vˆS»tidå/ [»bitå]umbigo 'bellybutton' /u‚»bigu/ [»bidu] barulho 'noise' /bå»Ru¥u/ [»buju]

Inês,1;8.2(S9)

leitinho 'milkdim.' /låj»ti¯u/ [»tJi¯u]bonito 'pretty' /bu»nitu/ [»nitHu]

Joana,2;2.19(S12)

menina 'girl' /mˆ»ninå/ [»¯inå] Luma,2;4.25(S32)Susana 'name' /su»zånå/ [»dånå] Luma,2;5.15(S33)

/WSW/

comando 'remote' /ku»må‚du/ [»må‚du] Luma,2;5.20(S34)

248

(183) Truncationof/WS/:

Orthogr. Gloss Target Output Child,AgeJoão 'name' /Zu»å‚w‚/ [»wå‚w‚] Clara,1;7.11(S9)avô 'grandfather' /å»vo/ [»v˘u] Clara,1;8.20(S10)aqui 'here' /å»ki/ [»ki] Claraabrir 'toopen' /å»bRiR/ [»biR]chapéu 'hat' /Så»pEw/ [»Ew]

Inês,1;10.29(S11)

avó 'grandmother' /å»vç/ [»dJç] Inês,2;0.11(S12)anões 'dwarfs' /å»no‚j‚S/ [»no‚j‚S] nariz 'nose' /nå»RiS/ [»di˘S] Inês,2;1.10(S13)

anão 'dwarf' /å»nå‚w‚/ [»¯å‚˘w‚]maçã 'apple' /må»så‚/ [»Så‚]chapéu 'hat' /Så»pEw/ [»pEw]Manel 'name' /må»nE…/ [»¯Ew]

Joana,2;2.19(S12)

In the third stage, Portuguese children realize that their language has a trochaic

rhythm and that feet are /SW/. Therefore, in Stage III of EP word stress acquisition,

children’swordrepresentationisadisyllabictrochee.

The template presented in (184) accounts for the production of both /SW/ and

/WWS/as[SW]inStageIII.Target/WS/wordsaremappedintoatrochaictemplateandthe

initialunstressedsyllableisdeleted.Inthisstagechildrenrealizethatthedomainforstressis

adisyllabicfootandthat it is left‐headed.Also,sincechildrenoverwhelminglycircumscribe

theright‐mostfootofthetargetwords(cf.children'sproductionsfor/WSW/in(161),(163),

(184) Child’soutputform–StageIII

e.g.,[»patJå]tampa'lid',[»bidu]umbigo'bellybutton',[»Ew]chapéu'hat'(WS)

ω

g

Σ

3

σS σW

[»pa tJå]

[»bi du]

[»Ew]

249

(166),(169)and(172)),thatisanevidenceforchildrentoassumethatfeetarebuiltfromthe

right,withinwords.

StageIV

In the fourth stage for theacquisitionof stress inEP, children realize thatprosodic

words in the language can be larger than a disyllabic foot. In the fourth stage of the

acquisitionofstresspatterns,both/WS/and/WSW/areacquired.

Note, however, that Clara, Joana and Luma acquire /WS/ before /WSW/ but Inês

acquires /WSW/before /WS/ (cf. Table 61). This alternating developmental path suggests

thattheinitialsyllablesin/WS/and/WSW/wordsareprosodicallyequivalent,i.e.,theyare

notpartofthetrochaicfoot.However,assoonasaprosodicpositionoutsidethetrochaicfoot

islicensed,both/WS/and/WSW/maybeacquired.Somechildren(Clara,JoanaandLuma)

produce the smaller form (the disyllable) earlier and later the larger one (the trisyllable).

Inêsisabletoproducelargertargetwords(/WSW/),thoughdisyllableswherenodisyllabic

trochaicfootisfound(/WS/)mightstillbetruncated.

In Stage IV, /SW/, /WS/, /WSW/, /WWS/ and /SWW/ words are mapped into a

[W[SW]Σ]PW template.The target‐likeproductionof/WS/,and/WSW/ isnownoticeable in

highpercentages,though/WWS/and/SWW/arenotyetacquired.

In (185) and (186), we illustrate the target‐like production of /WS/ and /WSW/

wordsandthetruncatedformsin/WWS/and/SWW/,respectively,inthesamechildandin

thesamesessions.

250

(185) Target‐likeproductionof/WS/and/WWS/

It isworthnoticingthat,contrarytomainwordstress,secondarystressinEPisnot

assignedatthelexicalcomponent(Pereira,1999;Vigário,2003).Therefore,weconsiderthat,

atthelexicallevel,feetinEParenotiterativeand,intherepresentationin(185),allpre‐tonic

syllablesarenotpartofanyfootstructure.

From the data presented in section 5.1.2.3., we might conclude that trisyllables

acquisition in EP occurs late. At the early stages (Stage I, II and III), trisyllables are not

generally selected. In Stage IV of the acquisition of stress patterns in EP, only /WSW/ are

produced adult‐like, but not in all children (Joana is the exception). This child does not

produce/WSW/target‐likeandtendsto truncate themto[S]or [SW](i.e.,herproductions

for/WSW/areconsistentwiththeprosodictemplatepresentedinStageIII).

Inallchildren,/WWS/and/SWW/areproducedwithgreatinstabilityandwithlow

rates of faithfulness until the end of the observational period. /WWS/ tend to be earlier

truncatedto[S]andlaterto[WS]./SWW/aremainlyproducedas[SW](cf.section5.1.3.3.),

asshownbelow.

Orthogr. Gloss Target Output Child,AgeJoão 'name' /Zu»å‚w‚/ [du»å‚w‚]avô 'grandfather' /å»vo/ [å»vo] Clara,1;9.23(S11)

aqui 'here' /å»ki/ [å»ki] Clara,1;10.15(S12)anão 'dwarf' /å»nå‚w‚/ [å»nå‚w‚]bater 'tohit' /bå»teR/ [bå»teR]

Inês,2;0.11(S12)

café 'coffee' /kå»fE/ [kå»pE]

/WS/

chapéu 'hat' /Så»pEw/ [tå»pEw] Inês,2;1.10(S13)

menina 'girl' /mˆ»ninå/ [mˆ»ninå]menino 'boy' /mˆ»ninu/ [må»ninu]

Clara,1;9.23(S11)

vermelho 'red' /vˆR»må¥u/ [di»mQju]menino 'boy' /mˆ»ninu/ [mi»ninu] Inês,2;0.11(S12)

bonecos 'toys' /bu»nEkuS/ [bu»nEkuS]

/WSW/

pantufa 'slipper' /på‚»tufå/ [på»dotå]Inês,2;1.10(S13)

acabou 'itfinished' /åkå»bo/ [kå»bo˘] Clara,1;9.23(S11)acabou 'itfinished' /åkå»bo/ [ka»bo] Clara,1;10.15(S12)Isabel 'name' /izå»bE…/ [bå»bE] Inês,2;0.11(S12)cachecol 'scarf' /kaSˆkç…/ [ka»kçj]

/WWS/

biberon 'milkbottle' /bibˆ»Ro‚/ [bi»bo‚] Inês,2;1.10(S13)

árvore 'tree' /»aRvuRˆ/ [»abi]Bárbara 'name' /»baRbåRå/ [»babå] Inês,2;0.11(S12)

/SWW/números 'numbers' /»numˆRuS/ [»nuwoS] Inês,2;1.10(S13)

251

In Stage IV of the acquisition of stress patterns, Portuguese children map their

productions into a prosodic word larger than a foot ([W[SW] Σ]PW), as illustrated in the

representation above. This representation shows that target iambs, target trochees and

/WSW/mightbeproducedaccordingly.Inordertomatchthe[W[SW]Σ]PWprosodictemplate,

/WWS/ words are still truncated to [S] or [WS]192. The initial unstressed syllable is not

mappedontothetemplateandis,therefore,erased.Thepreferredstresspatternatthisstage

is still /SW/, as suggested by the production of /SWW/ targets as [SW] (e.g., números

'numbers'[»numˆRuS]‐>[»nuwoS],Inês,2;1.10).

The developmental path presented for the stress patterns found in EP is partly

comparabletotheonesfoundinothertrochaiclanguages,likeCatalan(Prieto,2006),Dutch

(Fikkert,1994),English(Demuth,1995,1996a,b,c,d;Gerken,1994,1996;Kehoe,1998)and

Spanish (Demuth, 2001; Lleó & Demuth, 1999). Indeed, as early as disyllables emerge, a

trochaic foot is the most frequently observed stress pattern in languages with a trochaic

rhythm193.However,itisworthmentioningthat,despitegeneralregularitiesaccountedforin

a cross‐linguistic analysis on the acquisition of word shape and stress patterns in some

192Itisworthnoting,atthispoint,thatonechild,Luma,tendedtoaddafinalvoweltothecircumscribedstressedof/WWS/words, thuscreatinga[SW]pattern.However, thisprocesswasonlynoticeable inwordsending inasonorant consonant (/l,R/).Thisprocesshas alreadybeenobserved inotherdataonEP (cf. Freitas, 1997).Webelieve, nevertheless, that this process is motivated by syllable structure requirements (namely, word‐finalsonorantsoccupyingtheonsetofanempty‐headedsyllable),ratherthanbythefootstructure.193Cf.Chapter2,section2.2.2..

(186) Child’soutputform–StageIVOptionalsyllablesaremarkedwith()

e.g.,[kå»pE]café'coffee',[»patu]pato'duck',[mi»ninu]menino'boy',[bå»bE]Isabel

'name'and[»abi]árvore'tree'ω

3 g Σ g 3 σ W σS σW kå »pE

»pa tu

mi »ni nu

(i) (bå) »bE

»a bi

252

languages, it is true that the acquisition of stress patterns in non‐Germanic languages has

beenthesubjectofamuchgreaterdebate(cf.Hochberg,1988a,b,onSpanish,andTzakosta,

2004, on Greek, who claim for a neutral start, and Adam & Bat‐El, 2009, for a trochaic

tendencyinHebrew,aniambiclanguage).BPdatahavecontributedtowardsthedebate,for

itsunclear‐andsometimesintriguing‐developmentalpath.

Analyzing the truncation patterns of BP‐speaking children, Rapp (1994) found that

trocheeswerefavored,whereasBaia(2006),Santos(2001,2007)andStoel‐Gammon(1976)

found an iambic tendency in spontaneous speech productions, also during BP acquisition.

Bonilha(2005)didnotclaimforaniambictendency,thoughtheresultsdemonstrateahigher

tendency of preservation of initial and stressed syllables (/SW/ and /WS/). Baia (2008)

concludedthat, indeed,reduplicationsverymuchcontributedfortheearly iambictendency

found in BP. The conflict between an early trochaic or iambic tendency thus seems to be

highlydependentontheassumptionofreduplicationsasiambsornot.

From the comparison between the results now found in EP and Dutch, a typical

Germanic language, we observe that some differences are noticeable. As referred to in

Chapter 2, Dutch is one of the languages extensively described from the word stress

acquisitionpointofview (Fikkert,1994).Therefore,wewill takeDutchasa showcaseof a

trochaic language being acquired. We recall, below, the description for the acquisition of

wordstressconductedbyFikkert(1994)forDutch,summarizedinChapter2:

253

Stage0: /S/PW Thewordconsistsofonecoresyllable.

E.g.,klaar/kla˘r/‐>[ka˘],[kA](J.,1;4‐1;5)

dit/dIt/‐>[ti˘],[tI](J.,1;4‐1;5)

Stage1: /SW/PW Childrenrealizethatthewordsmightbedisyllabic

and a trochaic foot is processed ‐ /WS/ are

truncated into [S],whereas /SW/donot undergo

any developmental pattern194; /WSW/ are

truncatedinto[SW]andnot[S].

E.g.,banan/ba˘»na˘n/‐>[»na˘m](N.,2;3.7)

konijn/ko˘»nEin/‐>[»kEin](N.,2;3.23)

Stage2: /SW/PW /SW/ ‐> [SW] and /WS/ ‐> [SW]: This stage is

different from stage 1 in the sense that /WS/ do

notbecome[S]anymorebut[SW]instead(through

the addition of a syllable). Metathesis and

reduplications support the claim that the syllable,

andnotthefoot,isbeingprocessed.

E.g.,konijn/ko˘»nEin/‐>[»kIna˘](C.,1;10.11)

guitar/Xi˘»ta˘r/‐>[»hi˘ta˘](C.,1;10.11)

Stage3: /σΣσΣ/PW In /WS/ ‐ the 2 syllables are perceived as

belongingtotwodifferentfeetandbothvowelsare

prominent (level stress). Children realize that the

wordcanbelongerthanonefoot.

E.g.,David/»da˘vIt/‐>[»tA»fu˘n](R.,2;1.26)

tractor/»trEktçr/‐>[»tAk»tç®](R.,2;2.37)

Stage4: /‐SW/PWand/‐WS/PW Target‐likeproduction.

E.g.,microfoon /«mi˘kro˘»fo˘n/ ‐> [»mi˘k´«so˘n] (T.,

2;0.5)

Figure53.DevelopmentalstagesforwordstressacquisitioninDutch(Fikkert,1994)

194Cf.footnote86,ontheabsenceofexamplesfor/SW/inFikkert(1994).

254

From the tablepresentedabove, it ispossible toobservean initial stagewhere the

syllable is under the Dutch children's attention, as observed in EP. In this stage, Dutch

speaking‐childrenmainly produceCVmonosyllables, ormonosyllableswhere vowel length

contrast is not yet mastered. In EP, the majority of children's productions in Stage I also

consist in amonosyllable. At this stage, both [CV] and other syllable types are possible195.

Also,bothinEPandDutch,theearlyfootformconformstoatrochee(thoughithappensat

Stage I in Dutch, and at Stage III in EP). Despite important similarities, three striking

differencesdistinguishDutchandEPacquisitionofstress:

(i) Firstly,accordingtoFikkert(1994),reduplicationsarerareinDutchandmainly

occur in Stage II, when children are still circumscribing the syllable from the

targetform.InEP,reduplicationsareveryfrequent,especiallyintheearlystages

ofwordproduction;

(ii) Secondly, in Dutch, epenthesis is observable at the right edge of the

circumscribed syllable of a target prosodic word (/WS/), favoring [SW];

Portuguesechildren,onthecontrary, inserta fillersoundatthe leftedgeofthe

circumscribedsyllable,favoring[WS];

(iii) Thirdly,whereasinDutchacleartruncationpreferencein/WS/isnoticeable,in

EPtruncationofboth/SW/and/WS/isobservable,attheearlystagesofword

production;

(iv) Finally, Dutch /WS/ words are acquired in the last stage of word stress

acquisition. In the Portuguese children observed, themoment of acquisition of

iambs was, in general, very close in time with the moment of acquisition of

trochees.

Basedonthereportsfromothertrochaiclanguages,atthebeginningofthischapter

wehypothesizedthat,giventhetrochaic tendencyof theadult language,a trochaic tendency

willbefound.Inlinewiththishypothesis,wewouldexpectthattrocheesappearearlierthan

iambs.Also,truncationsandatendencyforstresserrorswereexpectedtofavortrochees.

Given the results found, our hypothesis was partly confirmed. From a general

developmentalperspective,wefoundthatwordstressacquisitioninEPhadasimilarpathas

195Inthisproject,nosystematicanalysisonthesyllablestructurewascarriedout.ForacompletedescriptionontheacquisitionofthesyllablestructureinEP,cf.Freitas(1997).

255

theoneobservedforothertrochaiclanguages.Wefoundanearlyprocessingofthesyllable,

followedbyatrochaictendencyand,finally,theproductionoflongerwords.

As showed in the developmental scale presented in this section (Table 61),

summarizing the findings for the acquisition of stress patterns, the developmental path

proposed in this dissertation was observed in other trochaic languages, namely English,

Dutch, Spanish and Catalan (Demuth, 1996a,b; Fikkert, 1994; Demuth, 2001; Prieto, 2006,

respectively).

As additionally observed in other trochaic languages (e.g. Fikkert, 1994, on Dutch;

Lléo&Demuth,1999,onGermanandSpanish), fromacertainpointonwards, target iambs

weremorepronetotruncationthantargettrocheesinEPwordstressacquisition(cf.Tables

45‐49),and/WSW/werepreferably truncated to [SW](Cf.Figures39,41,44,47,50).The

truncation patterns favoring trochees have supported the cross‐linguistic claimof an early

preference for the trochaic foot during prosodic acquisition. The data found in this

dissertationalsoconfirmedthistendency.

However,ourresultswerenotascategoricalastheresultsfoundinotherlanguages,

whereacleartrochaictendencywasfound(namelyDutchandEnglish). Infact,attheearly

stagesofwordstressacquisitioninEP:

(i) Reduplicationsfavoredaniambicwordshape;

(ii) Epenthesisoccurredattheleftedgeofprosodicwords,favoringaniambicword

shape;

(iii) Neithertrocheesnoriambswerecategoricallyfavoredintruncations.

Given the results found,we defend that the early disyllables in EP have a trochaic

form, though, at the beginning, no clear tendency is observed. Therefore, and following

previousanalysisonSpanish(Hochberg,1988a)andGreek(Tzakosta,2004),weclaimfora

neutral start in the acquisition of stress patterns in EP, with an early processing of the

syllable, a subsequent processing of a disyllable without fixed prominence and a later

processingofthetrochaicfootandtheproductionoflongerwords.

5.2.2.Ontheearlystages­theapparentiambictendency

InthissectionwewilldiscusstheearlywordshapeinthespeechofthePortuguese

children analyzed.Wewill provide empirical and theoretical arguments aiming at showing

that there is no 'early iambic tendency' in EP word stress acquisition. We will also

demonstrate that the early iambic tendency postulated for BP is biased by the heavy

256

tendency for reduplications and filler insertion, which do not need to be analyzed as a

productoffootshaperequirements.

The literature review carried out in Chapter 2 on Portuguese word stress

acquisition196 pointed to a number ofworks on BP indicating that, at the early stages, the

early speech of Brazilian children displayed an unexpected [WS] pattern (Bonilha, 2005;

Santos, 1995; 2001; Stoel‐Gammon, 1976). Santos (2007) clearly suggests that the early

words produced by the Brazilian children observed conform an iambic foot. The author

assumesthatBrazilianchildrenapplythemorphology‐basedstressalgorithmforthetarget

language,accordingtowhichstressgenerallyfallsonthelastsyllableofthestem(Lee,1995).

Following Lee's (1995) analysis forword stress in adult language, Santos (2007) provides

evidencefortheanalysisofwordstressinthetargetlanguageaccordingtowhichthestemis

thedomainforstress,andthebasicfoot isaniamb(innon‐verbs)197.Sincetheinformation

outsidethestressdomainisconsideredextrametrical(Lee,1995),Brazilianchildrenproduce

[SW] nouns only when morphological contrasts are mastered and the parameter for

extrametricality, accounting for the production of syllables that are outside the stress

domain,issetto[Yes].

Itisworthmentioning,however,thattheauthorswhoobservedaniambictendency

in Portuguese (Baia, 2006; Baia, 2008b; Santos, 2007; Stoel‐Gammon, 1976) always took

reduplications into account and considered children's reduplications (either resulting from

targetreduplicationsornot),aswellasproductionswithfillersyllablesatthesamelevelas

non‐reduplicatedwords.However,sincethoseproductionstendtodisappearinthecourseof

development,oftenbeingreplacedbyproductionswherebothsyllablesaremapableintothe

targetsyllablesofthetargetword,onemightaskwhethertheseproductionshavethesame

prosodic and lexical status as typical 'prosodic words' attempted by the children198. We

believetheydonot,andwewillexplainwhyinthefollowingparagraphs.

ThedatafoundinEPweresimilartotheonesfoundinBP,namelywithrespecttothe

early (apparent) iambic tendency. Aswe observed in this and in previous sections of this

dissertation, one of the more intriguing aspects of the acquisition of word stress in

Portuguese(bothEPandBP)wasthehighfrequencyofreduplicatedwordsandepenthesis,

favoring [WS] surface forms. Indeed, if we consider early reduplications produced by the

children and thewordsproducedwith filler insertion,EPhas aheavy iambic tendency (cf.

Tables28‐32).

196Cf.Chapter2,section2.5.197Lee(1995)claimsthatthedefaultfootinnon‐verbsisaniambandthedefaultfootinverbsisatrochee.Inthisparagraphwe focusonlyon thenon‐verbssystem. In the followingchapterwewillcompare theverb'sandthenon‐verb'ssystemwithrespecttowordstress.198Theiambicformoftheseearlyproductionswillbediscussedinsection5.2.2.2..

257

In fact, the presence of reduplicated forms in child language is documented in the

literature (Fee & Ingram, 1982; Ferguson, 1983; Ferguson & Macken, 1983; Klein, 2005;

Leroy & Morgenstern; 2005; Lleó, 1990; Moskowitz, 1973; Schwartz, Leonard, Wilcox &

Folger, 1980; Schwartz& Leonard, 1983; Veneziano, Sinclair&Berthoud, 1990;Waterson,

1971; Wauquier‐Gravelines, 2003), though in general these forms do not favor a stress

patternthatiscontrarytothestresspatternobservedinthetargetlanguage.Frenchisoneof

thelanguageswherereduplications‐namelywitha[WS]format‐areattested(Braud,2003;

Rose,2000;Wauquier‐Gravelines,2003).In(187)weshowsomeexamplesofreduplication

inFrenchdata.

(187) ReduplicationsinearlyFrenchdata:

Orthogr. Gloss Target Output Child,Age Studycoucou 'peeck‐a‐boo' /kuku/ [gu»gu]pain 'bread' /bobo/ [bo»bo] Théo,1;10.27 Rose(2000)

pupée 'doll' /pupe/ [pepe] 1;3chapeau 'hat' /Sapo/ [popo]pelle 'shovel' /pEl/ [pepe]canne 'stick' /kan/ [tata]

1;4

oeuf 'egg' /øf/ [toto] 1;5

Demuth&Johnson(2003)

os 'bone' ‐‐‐‐‐ [E‚nonos] 3;0 Wauquier‐Gravelines(2003)

Despite the similarity between EP and French data, conclusions comparing

PortuguesewithFrenchclaimingthatbothareiambicmightbepremature,sinceFrenchisa

languagewherefeetarenotrelevantforstresspurposes,andwherestressis,instead,phrase‐

final.

Asshowninthisdiscussion,wehavereasonstobelievethatearlywordsarenot, in

fact, disyllabic iambic feet, as non‐reduplicated /WS/ are scarce or inexistent at the

beginning, contrary to reduplicated /WS/. A cross‐linguistic comparison conducted in

Chapter 2 of this dissertation further suggested that, if a target language is not iambic, an

iambictendencyinthechild'ssystemwasneverattested.

AlthoughpreviousworksonBPhavedemonstratedthatiambscanbepreferredover

trochees during Portuguese acquisition (Baia, 2006; Santos, 2001, 2007; Stoel‐Gammon,

1976),anearlyiambictendencyishardtoargueforintheEPacquisitiondata,duetoseveral

aspectsmentionedinourreviewandwhichwenowrecall:

258

(i) An initial iambic tendency is very hard to explain in a languagewith a heavily

trochaic rhythm, like Portuguese199. Even considering that the algorithm for

stress assignment entails an underlying iambic foot, we have to consider that

childrenareprovidedwithstrongevidenceforstressinthepenultimatesyllable

and,thus,foratrochaicrhythm.

(ii) The morphology‐based algorithm, which predicts, by default, a domain‐final

stress position (i.e., stress the last vowel of the stem)200, considers the word

markerasextrametrical,markedor'invisible'forstressassignment.Asobserved

inthedevelopmentalpathproposedinsection5.2.1.,assoonasnon‐reduplicated

disyllables emerge, thewordmarkerwasproduced.Anearlyproductionof the

wordmarkerisinconsistentwithitsextrametricalandmarkedcharacter.

(iii) Target iambshavebeenclearlydisfavoredinthepathofacquisition,asshowed

inTables2and4.

Inamoredetailedanalysis,whereproductionswithreduplicationandepenthesisare

notaccountedforasiambs,theheavyiambictendencytendstodisappear,asshowninTables

64,65and66,below.Inthesetables,weplottogetherthevaluesfor/SW/,thevaluesforthe

disguisediambs–heresimplifiedto/CV1'CV1/–andthevaluesfortargetnon‐reduplicated

iambs (/CV1'CV2/), in Inês, Joana and João's speeches. We show the number of targets

selected by each child, as well as the number of tokens produced and the corresponding

percentages.

199Cf.Chapter1,section1.2.4.,Figures9and10.200 Cf. Chapter 1, section1.2.3. and references therein toAndrade (1992),Andrade&Laks (1992), Lee (1995),Mateus&Andrade(2000)andPereira(1999).

259

Session /SW/ [SW] % /CV1'CV1/ [CV1'CV1] % /CV1'CV2/ [CV1'CV2] %S1 ‐ ‐ ­ 19 8 42.1 ‐ ‐ ­S2 2 0 0 41 15 36.6 ‐ ‐ ­S3 4 0 0 64 21 32.8 8 0 0S4 27 1 3.7 67 39 58.2 ‐ ‐ ­S5 27 1 3.7 56 30 53.6 4 0 0S6 90 1 1.1 110 61 55.5 14 3 21.4S7 73 3 4.1 93 55 59.1 29 0 0S8 39 5 12.8 80 35 43.8 39 11 28.2S9 140 82 58.6 113 81 71.7 44 17 38.6S10 221 166 75.1 96 89 92.7 45 21 46.7S11 238 182 76.5 201 124 61.7 77 35 45.5S12 275 229 83.3 254 159 62.6 155 68 43.9S13 297 256 86.2 189 128 67.7 110 64 58.2S14 249 198 79.5 211 147 69.7 137 84 61.3S15 232 178 76.7 133 89 66.9 98 60 61.2S16 283 215 76.0 168 108 64.3 126 83 65.9S17 154 87 56.5 96 45 46.9 74 43 58.1S18 275 193 70.2 150 90 60.0 115 74 64.3

Table64.Percentageoftarget­like/SW/and/WS/words,consideringreduplicationsandepenthesis(/CV1'CV1)ornot(/CV1'CV2/)­Inês

Session /SW/ [SW] % /CV1'CV1/ [CV1'CV1] % /CV1'CV2/ [CV1'CV2] %S1 ‐ ­ ‐ ‐ ­ ‐ ‐ ­S2 ‐ ‐ ­ 4 0 0.0 ‐ ‐ ­S3 2 0 0.0 3 1 33.3 ‐ ‐ ­S4 ‐ ‐ ­ 3 0 0.0 ‐ ‐ ­S5 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S6 ‐ ‐ ­ ‐ ‐ ­ ‐ ‐ ­S7 16 0 0.0 3 1 33.3 2 0 0S8 4 0 0.0 3 0 0.0 1 0 0S9 13 0 0.0 29 9 31.0 14 4 28.6S10 16 2 12.5 28 10 35.7 14 5 35.7S11 45 13 28.9 54 32 59.3 30 11 36.7S12 91 57 62.6 34 10 29.4 30 6 20S13 74 44 59.5 42 16 38.1 27 12 44.4S14 163 108 66.3 56 28 50.0 49 25 51.0

Table65.Percentageoftarget­like/SW/and/WS/words,consideringreduplicationsandepenthesis(/CV1'CV1)ornot(/CV1'CV2/)­Joana

260

Session /SW/ [SW] % /CV1'CV1/ [CV1'CV1] % /CV1'CV2/ [CV1'CV2] %S1 6 1 16.7 ‐ ‐ ­ ‐ ‐ ­S2 ‐ ‐ ­ 1 1 100.0 1 1 100201S3 8 3 37.5 4 4 100.0 4 4 100S4 6 5 83.3 6 4 66.7 4 4 100S5 4 0 0.0 ‐ ‐ ­ ‐ ‐ ­S6 ‐ ‐ ­ 8 3 37.5 3 3 100S7 15 2 13.3 22 3 13.6 4 3 75S8 18 1 5.6 12 4 33.3 12 2 16.7S9 13 2 15.4 16 2 12.5 12 0 0S10 39 5 12.8 30 18 60.0 11 6 54.5S11 17 3 17.6 17 13 76.5 4 1 25.0S12 12 4 33.3 28 17 60.7 3 0 0S13 23 5 21.7 54 40 74.1 10 0 0S14 17 8 47.1 33 29 87.9 2 1 50S15 7 3 42.9 32 17 53.1 4 2 50S16 34 21 61.8 44 22 50.0 16 9 56.3S17 58 49 84.5 25 16 64.0 10 7 70S18 67 60 89.6 56 37 66.1 30 20 66.7S19 77 62 80.5 37 21 56.8 11 9 81.8S20 90 71 78.9 41 22 53.7 22 10 45.5S21 86 78 90.7 37 24 64.9 12 7 58.3S22 59 48 81.4 42 17 40.5 13 3 23.1

Table66.Percentageoftarget­like/SW/and/WS/words,consideringreduplicationsandepenthesis(/CV1'CV1)ornot(/CV1'CV2/)­João

The tables presented above show that, disregarding reduplications, only João has

highertarget‐likeratesin/WS/,butthatseemstobetheresultofatyperepetition(theword

olá'hello').Ifwedonottakereduplicationsintoaccount,Joãohasasimultaneousemergence

ofboth/SW/and/WS/(/CV1'CV2/).Also,disregardingreduplications, theapparent iambic

tendency in Inês disappears and, in Joana, we observe that /SW/ are selected and are

acquiredearlierthan/WS/.Inaddition,weobservethatthenumberof/SW/tokensselected

ishigherthanthenumberof/CV1'CV2/tokensinthesechildren,but, intheearlystagesthe

numberof/CV1'CV1/tendstobehigherthanthenumberof/SW/.

From the data analyzed, it was also noticeable that the production of both

reduplications and fillers tends to decrease and, sometimes, disappear in the course of

development(cf.Figures29‐33).Additionally,productionswithreduplicationandepenthesis

haveadistinctemergencetimewhencomparedtotargetiambsthatwerenotreduplicatedor

subject to filler insertion. In fact, if only target iambs are taken into account, no clear

tendencyforiambsortrocheesisobserved(thoughtrocheesare,ingeneral,earlieracquired

‐ cf. Tables 34‐38).Our resultswhere no reduplications and filler soundswere considered

201Itisworthnoticingthat,inJoão,theonlynon‐reduplicated/WS/typeproducedfromsession1to6isthewordolá'hello'.

261

indicatedthat(i)threeoftheobservedchildrenacquiredtrocheesearlierthaniambs(Clara,

InêsandJoana)and(ii)attheearlystages,truncationintargetdisyllablesdidnotfavorany

stresspattern,butlatertrocheestendedtobelesspronetotruncationthaniambs(cf.Tables

45‐49).Moreover,stressshiftofthetype/SW/‐>[WS]wasonlyobservedinonechild(João),

beingveryscarceinthespeechofthreechildren(Clara,InêsandJoana).Lumahadahigher

stressshiftrateofthetype/WS/‐>[SW](cf.Tables51‐60).

Thefindingssummarizedabovesuggestthatthereisnocategoricaliambictendency

in our data. Indeed, the distinct behavior demonstrated by all childrenwith respect to the

early words (where reduplication and epenthesis is mostly found, but which tend to

disappear in the course of development) and laterwords (trochees and iambs), lead us to

considerthat,eventually,earlywordsdonothavethesameprosodicstatusastargettrochees

andiambsandthattheearlyiambictendencyseemstobeonlyapparent.Theassumptionof

anearlyiambicfootwouldnotaccountforourdatainthreedifferentmanners:

(i) On the one hand, the assumption of an early iambic foot in early word's

representation in EP would have to account both for the production of the

'disguisediambs'(reduplicationsandproductofepenthesis)andfortheproduction

of non‐reduplicated target iambs. However, we observed that, at the beginning,

iambsthatwerenottheproductofreduplicationandepenthesiswerenotselected

(byJoanaandLuma)ortheywereveryscarce(inClara,InêsandJoão).

(ii) On the other hand, the results from the faithful productions and the production

strategies(truncationandstressshift)usedintarget/SW/andinnon‐reduplicated

/WS/wordsindicatedthattarget/SW/and/WS/tendtohaveasimilarbehavior.

The emergence of both words is simultaneous, iambs and trochees are equally

prone to truncation in the early stages and stress shift does not favor either

trocheesor iambs.ThisbehaviorwasalsoobservedinGreekchildreninTzakosta

(2004), leading the author to claim for a neutral start in Greek word shape

acquisition.In(188),wepresentsomeinstancesshowingthattruncation,inGreek

earlyprosodicacquisition,doesnotfavoreither/SW/or/WS/.

262

(188) NeutraltendencyinGreek(Tzakosta,2004b:103202):

Gloss Target Output Chid,Age'fig' /»siko/ [ko] B.,1;10'those' /a»fta/ [ta] B.,1;11'toplay' /»pezi/ [pe]'carriage' /ka»rotsi/ [jo]

F.,1;11

'from' /a»po/ [po] Ma.,2;8.7

(iii) Finally,ontheassumptionofaniambicfootwhichwouldbeearlieracquiredsince

nowordmarker (which is extrametrical, according to amorphology‐based stress

algorithm203)ispresentinthesewords,twopointsareyettobeclarified:

a. First, the assumption that children produce iambs earlier because they are

stemsanddonotbearmorphologicalinformation,ontheonehand,anddonot

produce trochees because they bear morphological information which is

extrametrical is circular. In fact, the absence of the word marker might be

explainedintermsofthecircumscriptionofthetargetstressedsyllable.

b. Secondly,asobservedinsection5.2.1.andfurthermentionedinthisdiscussion,

as soonasnon‐reduplicateddisyllablesemerge, thewordmarker isproduced.

An early production of thewordmarker is inconsistentwith its extrametrical

andmarkedcharacter.

In fact, theearlierproductionof /SW/words (bearingmorphological contrastsand

extrametricality) isnotconsistentwith themuch laterproductionof/SWW/words,which,

accordingtotheliteratureonthetargetsystem(Lee,1995;Mateus&Andrade,2000;Pereira,

1999;amongothers),alsobearwordmarkerandextrametricalsyllables.

On theassumptionofanearly iambic footmotivatedbyamorphology‐basedstress

algorithm,thelaterproductionof[SW]stresspatterns(atStageIII)wouldmeanthatchildren

hadmasteredmorphologicalcontrastsandacquiredextrametricality(assuggestedbySantos,

2007, for BP). However, the explanation according towhich Portuguese‐speaking children

haveanearlieracquisitionforiambsthanfortrocheesdueto(i)anearlieracquisitionofan

202InstancesweretakenfromTzakosta(2004b).Noorthographictranscriptionisprovided.203Cf.Chapter1,section1.2.3.,andthereferencestoLee(1995)andPereira(1999)therein.Accordingtotheseauthors, stress fallson the lastsyllableof thestem.Therefore, thestresswouldbedomain‐final ([WS])andthethemevowelin/SW/wordswouldbeextrametrical.

263

unmarked stress rule in non‐verbs (stem‐final stress) and due to (ii) the unmastery of

extrametricalitydoesnotseemtoholdforEPdata.

Ourfindingsontheacquisitionof/SW/and/SWW/non‐verbsseemtoindicatethat

/SWW/ are acquired much later than /SW/. According to our data, word markers are

producedassoonasnon‐reduplicatedtrochaicwordsemerge204,andextrametricalelements

(in/SWW/non‐verbs)arenotmastereduntiltheendoftheobservationperiod205.Infact,the

datafromtheacquisitionof/SWW/wordsinEPconfirmedthat/SWW/,butnot/SW/,are

markedstructuresandareacquired late.Hence, it isnotplausible thatchildrenareable to

masterextrametricality,producingtrochees,butarenotabletoproduce/SWW/non‐verbs

target‐like,especiallywhenothertrisyllabicwordsarealreadyacquired.

Assuming that nouns constitute the vast majority of Portuguese children's early

words206andmost/SW/nounsinPortuguesehaveafinalextrametricalelement(Lee,1995)

that childrenacquire later than iambs (Santos,2007), it remainsunexplainedwhychildren

produce/SW/withthewordmarkerassoonastheyacquirenon‐reduplicateddisyllablesbut

donotproducetheextrametricalsyllableinSWWwords.

Given the three reasons presented above,wehave to take seriously the hypothesis

accordingtowhichearlyiambsarenot,infact,iambicfeet.Thesefindingsareconsistentwith

cross‐linguisticinformationontheacquisitionofstresspatternsandwiththefactthatiambs

aredisfavoredinthegeneralpathoflanguageacquisition.OnlyinFrench,alanguagewhere

feetorganization is irrelevant forprominence(as, in fact, the languagedoesnothaveword

stress), the production of long utterances, with final prominence was also noticed in the

children'sspeech(Braud,2003;Wauquier‐Gravelines,2003).Reduplicationsarefrequentin

Frenchand theyarenotmotivatedby feet constraints,but requiredby the largerprosodic

structure instead. French‐speaking children are sensitive to the rhythmic and prosodic

propertiesof the languageandattemptto fulfilla largerprosodicunit, therelevantunit for

prominence purposes in French, the Phonological Phrase. The fact that French, a language

without word stress, but with phrase‐final accent, has an alternation betweenmono‐ and

disyllabicproductionsandagreatamountof reduplicationsandmultisyllabicutterances in

theearlystagesofacquisitionallowustohypothesizethatearlyreduplicationsandfillersin

EP might not be part of the foot structure and might be due to the processing of higher

prosodicdomains.

204EmpiricalevidenceforthisobservationwillbeprovidedwithfurtherdetailinChapter6,section6.1.1.1..205Cf.section5.1.2.3.206Inthefollowingchapter,wewillconfirmthisinformation,byshowingthedistributionofthemorphosyntacticcategoriesinthedata.

264

Indeed,theearly[WS]productionsinthePortuguese‐speakingchildrenmightbedue

to the production of phrasal prominence, with higher levels of the prosodic hierarchy

(namely,thePhonologicalPhrase)interactingwiththesyllablelevelinproduction.

ThedatafoundinthisdissertationconcerningtheacquisitionofstresspatternsinEP

suggestthatchildrenstartbuildingwordsfromasyllable,thestressedsyllable.Asobserved

inother languages, likeEnglishandDutch (Demuth,1995;Fikkert, 1994, respectively), the

frequentmonosyllabicproductionsintheearlystagesofwordstressacquisition,alongwitha

subsequent complexification of the prosodic structure (Syllable > Foot > Prosodic Word),

haveprovidedevidenceforabottom‐upmodelofprosodicacquisition.Assumingthismodel,

fromthebeginning,childrenwouldprocessandproducethesyllableofthetargetform.After

that,childrenwouldprocessandproducefeet,andsoforth,inapathfromsimpler(lower)to

more complex (higher) prosodic constituents, in a gradual fashion. However, children

acquiring other languages did not necessarily follow this path and, apparently, the results

foundinEParecontradicting.Ontheonehand,weobservedagreatamountofmonosyllabic

forms in EP early speech, suggesting that the syllable is being the target of children's

attention.On theotherhand,monosyllabicutteranceswere frequently extended todi‐ and

multisyllabic utterances, suggesting that higher prosodic constituents could be under

processing by the children. Additionally, we observed that, at the early stages of word

productionaheavytendencyfor[WS]consistinginreduplicationsandsequenceswithfillers

werenoticed.

Based on these facts (tendency for monosyllabicity, with recursion to di‐ and

multisyllabicproductions,normallywitha[WS]format),wehypothesizethat:

(i) intheearlystages,theprominencebeingtargetedbythePortuguesechildrenis

notwordprominence,but,instead,phrasalprominenceand;

(ii) in the early stages, lower levels (the syllable) of the prosodic hierarchy are

interactingwithhigherlevels(namely,thePhonologicalPhrase).

Specifically,wedefendthatthereasonsthatmaycontributeforanearlyright‐headed

prominenceinPortuguesemayberelatedtonuclearaccent.

Peters (2001a,b) and Veneziano (2001) suggest that filler sounds are initially

prosodic extenders and prosodic placeholders. Filler sounds do not necessarily extend

children's productions to the foot level but, depending on the prosodic properties of the

language, theymaymatchaphrasetemplate(asproposedbyGennari&Demuth,1997and

Lleó,1990,forSpanish).

SinceEP,contrarytoEnglishandGerman, isaproclitic language, it is likelythatthe

265

initialsyllableofanearlyiamb‐aswellastheinitialsyllableofareduplication‐isprocessed

asanunstressedelementbeingplacedattheleftoftheProsodicWord(Vigário,2003).

Our results partly confirmed Fikkert (1994) and Demuth's (1995) findings in the

sense that prosodic acquisition and, in particular, word stress acquisition in trochaic

languagesstartsatthesyllable levelandgoesupintheprosodichierarchy, fromsyllableto

feet, and from feet to prosodic words. From this perspective, the acquisition of word

prominence,thus,proceedsbottom‐upinEP.Thisdirectionality,however,isnotclear‐cut,as

there appears to be an interaction between lower and higher prosodic levels, in the early

stagesofwordproduction,namely, itappeartobeaninteractionbetweenthesyllablelevel

andthephonologicalphraselevel.Thisinteractionisnoticeablein:

(i) the majority of monosyllabic productions (target‐like and as a result of

truncation) in the speech of Portuguese children, along with the observed

strategiesofreduplicationandepenthesis;

(ii) thetendencytostressthefinalsyllableinearlyutterances.

Additionally,inChapter4wehaveshownthat,attheonsetofwordproduction,Joana

and Inês did not correctly assign word prominence. Our results suggested that, at the

beginning, children might not be producing word prominence and confirmed previous

research showing that higher levels of the prosodic structure, like the phonological or the

intonationalphrase,mightbeinfluencingchildren'sproductionofrhythmintheearlystages

(Kehoeetal.,1995;Pollocketal.,1993;Vihmanetal.,1998).

Though Portuguese children were not able to produce longer target words (for

instance,trisyllableshadalateemergenceandacquisition),itwasobservedthat,indeed,they

were able to produce utterances larger than a syllable, contrary to Dutch and English‐

speakingchildren.UtteranceslongerthanasyllableintheearlyspeechofBrazilianchildren

were also observed (Santos, 1995; Scarpa, 1998, Scarpa & Santos, 2005), suggesting that

structuralpropertiescommontobothvarieties, suchasphrasalstressdirectionality (right‐

to‐left),mightbeatstakeas farasearly [WS]wordsareconcerned.Theauthorssuggested

that reduplications and epenthesis occurring at the left edge of the words at the onset of

wordaredue torhythmicconstraintsathigherprosodic levels,namelyat thePhonological

Phrase level, and claim for a top‐down processing in BP prosodic acquisition. Therefore,

before childrenareawareofword stress, other (higher)prosodicdomainsmightbeunder

children'sattention.

Frota & Vigário (2008), studying the acquisition of intonational and durational

266

patternsofoneEP‐speakingchild,L.207,observedmanycasesofstressshiftintheearlystages

and thatearlyprominencewasmainly final.Theauthorsargue thatphrasal stress is right‐

headed in Portuguese and, thus, the child was not processing word stress, but rather

prominence in higher prosodic domains (like the phonological and intonational phrase).

Frota&Matos(2009)supportedthefindingsfromFrota&Vigário(2008)andfoundthat,at

thebeginning,L.'sspeechshowedastrongcorrelationbetweenthedurationofsyllablesand

the number of syllables in higher prosodic domains (namely, at the prosodic word and

intonationalphraselevel).ThesefindingssuggestthatearlyprosodicacquisitioninEPdoes

not start exclusively bottom‐up and that lower andhigher levels of theprosodic hierarchy

mayprovidecuestotheacquisitionofwordstressalgorithm.

Interestingly, English, Dutch and German also have left‐to‐right phrasal stress and

children acquiring those languages undoubtedly startwith trochees, as soon as disyllables

emerge (Fikkert, 1994; Demuth, 1995, 1996a,b,c). The important difference between

GermaniclanguageslikeDutch,EnglishorGerman(Goedemans,vanderHulst&Visch,1996;

Kager,1989;Roca,1999),andalanguagelikeEPisthat,inGermaniclanguagestheevidence

for trochees isoverwhelming,namely, through theapplicationof themajorgeneralizations

underlyingwordstressassignment.

OneofthemaingeneralizationsofwordstressinDutchpredictsthat"primarystress

isdirectlybeforeaschwallable‐iftheschwaisprecededbyaconsonant"(Kager,1989:227).

Theauthor further states that "disyllabicwordswith schwallables arequite frequent. Such

wordswillhavestressontheirinitial(orpenultimate)syllable".Moreover,theauthoradds,

"The force of the schwallable restriction is clear from rightward stress shifts tomatch the

pattern, both in mispronunciations and in imported words [e.g., katálogus ‐> katalógus,

nótulen‐>notúlen]".

AlsocontrarytoEP,Dutchfeetarebuiltiteratively,withina/SW/alternation.Infact,

both primary and secondary stress inDutch are assigned on the basis of the SW rhythmic

repetition(e.g.,macaroni/«ma˘ka˘»ro˘ni/,limonade/«li˘mo˘»na˘d´/,televisie/«te˘l´»vi˘si˘/).

In EP, apart from stress assignment itself, there is weak evidence for the foot in

general208,andforthetrochaicfootinparticular,whichmaycausechildrentorelyonother

prosodicdomainssuchasthesyllable, theprosodicwordorthePhonologicalPhrase,while

processingprominence.

The stress algorithm in those languages is very clear (a weight‐based one) to

children: contrary toEP,weight isundoubtedly relevant for stresspurposes (e.g., inDutch,

207L.isLuma.Cf.footnote120.208BoththeSpondaicandtheDactylicLoweringrefertootherfeettypes,respectivelythespondeeandthedactyl(Wetzels,1992,1995,2006).

267

superheavysyllables‐VVC‐attractstressalmostwithoutexception209),andbothDutchand

Englishhavevowellengthcontrastwhichcountsforsyllableweight(Kager,1989).Finally,in

English and Dutch underived words, morphological constituency does not play a role in

primarystressassignment(Kager,1989).

Contrary towhatoccurrs inmanyRomance languages (Roca,1999),word stress in

Germanic languages is not under debate. In fact, clear evidence, both for the word stress

algorithmandthetrochaicfoot,mayoverruleright‐to‐leftphrasalstress.

Basedontheargumentspresented insection5.2.2.1.and in thissection,webelieve

that EP word stress acquisition does not start with an iambic foot. It starts with a

monosyllable.For the reasonspresentedabove,weclaim thatearly [WS]words frequently

found in the speech of Portuguese‐speaking children are not iambs and that there is an

apparentiambictendencyintheacquisitionofstresspatternsinEP,causedbyreduplication

andepenthesisstrategies.

5.3.Summary

Inthischapter,weanalyzed:

(i) thechildren’sproductionpatternsforwordshape;

(ii) thechildren’sfaithfulnesstowardsstresspatterns;

(iii) thestrategiesusedbythechildrentodealwithstresspatternsinagivenmoment

ofphonologicalacquisition.

Wepresentedtheresultsformonosyllables,trochees,iambsandtrisyllables(/WSW/,

/WWS/and/SWW/)inEP,bothinthetargetandinthechildren'sactualproductions.

Theresults fromthechildren’sproductionpatternsand faithfulness indicate that in

theearlystages,Portuguesechildrenseemtofavormonosyllablesand[WS]words.Onlylater

Portuguese children come up with [SW] words and later they produce [WSW] words.

Monosyllablesareproducedandmastered from thebeginningofproduction inPortuguese

children's linguistic development, though they might be subject to the same strategies as

disyllables and trisyllables, i.e., simple or multiple reduplications, and filler insertion, and

/SW/areproducedtarget‐likelaterthan/WS/.Trisyllablesareacquiredlaterthandisyllabic

209ThoughonlysuperheavysyllablesareconsideredasheavyinadultDutch,Dutchchildrenseemtobesensitivetosyllableweightinanyclosedsyllables:"(...)childrenseemtoregardanyclosedsyllableasheavy,independentlyofthevowelandindependentlyofthenatureofthefinalconsonant,sinceallclosedsyllablesarenowstressedinthechild'soutputforms."(Fikkert,1994:286).

268

words.Thefirsttrisyllableacquiredis/WSW/andboth/WWS/and/SWW/arenotacquired

untiltheendoftheobservationalperiodinthefivechildren.Thelateracquisitionof/WWS/

and,especially,/SWW/providedevidenceforthemarkercharacterofthesestructures,with

respecttotheacquisitionofstresspatterns.

In this chapter we discussed the acquisition of word stress in EP. Our results

demonstrated that EP acquisition follows an acquisition path with four stages for the

acquisitionofwordstress:

Stage I: /σ/PW ‐ childrenmainly produce a syllable (they circumscribe the stressed

syllableforatargetwordandmightproduceitinvariablemanner:asamonosyllable,usinga

reduplicationstrategyorepenthesis);

Stage II: /CV1CV2/PW ‐ non‐reduplicated disyllabic words emerge (both /SW/ and

/WS/arepossible,thoughinsmallamounts);

StageIII:/SW/Σ‐Atrochaicfootisprocessed;

StageIV: [W[SW]Σ]PW‐Words largerthanadisyllabic footareacquired(both/WS/

and/WSW/areproduced).

Thisdevelopmentalpathissimilartothepatternfoundforother languages,namely

with respect to the processing of directionality during prosodic acquisition. Portuguese

childrenstartwithasyllableanduse it to fulfillhigherprosodic templates.Themajorityof

children's productions is monosyllabic, monosyllables are produced adult‐like from the

beginning and, otherwise, they are subject to reduplication and filler insertion. These

strategies are, in fact, frequent in the production of /S/, but also /SW/ and /WS/, and

contribute for an 'apparent' iambic tendency. After this stage, disyllables emerge and both

[SW] and [WS] are possible. At this point, no word prominence is yet assigned, as many

disyllablesarepronetotruncation.Afterwards,childrenprefera[SW]pattern:/WS/arestill

truncatedto[S]but/SW/areproducedadult‐like./WSW/wordsaretruncatedto[SW].At

thisstepofthedevelopment,itappearsthatchildrenhavealreadyrealizedthatfeetare[SW]

in their language and that word trees are built from Right‐to‐Left. Trisyllables are not

selecteduntillateindevelopmentbut,whentheyemerge,childrenselectmostly/WSW/.The

target‐like production of trisyllables occurs in the fourth stage, when children realize that

prosodicwordscanbelargerthanadisyllabicfoot.

Ourproposalpartlyconfirmsabottom‐upapproachtoprosodicacquisition(Fikkert,

269

1994 andDemuth, 1995), in the sense that Portuguese children start producing a syllable,

thenadisyllabictrocheeandlaterawordlongerthanadisyllabicfoot.However,thedataalso

provided evidence supporting previous studies arguing for an early processing of higher

prosodic domains in children's early productions (Scarpa, 1998; Scarpa & Santos, 2005),

giventhatearlydi‐andmultisyllabicutteranceswerefound,mostlyhavinga[WS]shape,in

theearlyspeechofPortuguesechildren.

Before the turning point towards a trochaic tendency, Portuguese children are not

processingwordprominenceandmainlyprocessa syllable,withwhich they fulfill a larger

prosodictemplate,bymeansofreduplicationandfillerinsertion.Aninterplaybetweenlower

(syllableandfoot)andhigher(phonologicalphrase)wassuggestedbythedata.

Accordingtoourproposal,earlyiambscanbeinterpretedasmonosyllablesor'iambs

indisguise' (reduplicationsandsyllablesprecededbya filler,witha [WS] form),whichare

produced, not to fulfill feet requirements, but rather requirements from higher prosodic

domains.Underthisperspective,EPacquisitiondatadonotconfirmtheexistenceofanearly

iambic tendency. Our results further suggest that the domain for stress rules in EP is the

lexical word. If the lexical word is the domain for stress assignment, the assumption of a

morphology‐based stress algorithm remains arguable. That is, if acquisition data provide

evidenceforthe lexicalword,andnotmorphemes,asthedomainforstress, the interaction

between word stress and morphology, both in phonological acquisition and in the adult

speech,beginstofallapart.Inthefollowingchapter,wewillinvestigatetheresultsfromour

dataanalysisontheinteractionofmorphologyandweightonwordstressacquisition.

In summary, the general path pursued by all the children analyzed, taking into

accountrealiambs,andnotthe'disguisediambs',indicatethatthereisaneutralstartinthe

acquisitionofstresspatternsinEP.

270

271

6.Testingtargetanalysesonwordstresswithacquisitiondata

In this chapter we will discuss the target analyses on word stress, based on the

observationofPortuguesechildren'sdevelopmentalpatternsforwordstress,inrelationship

withmorphological informationandsyllableweight.Weaimattestingweightsensitivityor

morphologydependenceofwordstressinPortugueseand,additionally,weintendtodiscuss

the analyses for word stress in EP on the basis of new empirical data on child language

acquisition.

In Chapter 1210, we observed that word stress in Romance languages – and in

Portugueseinparticular–hasbeenwidelydiscussed(e.g.,Bisol,1993,1999;Carvalho,1987,

1988;Lee,1995,2001,2006,2007;Mateus,1983,2000,Pereira,1999;Roca,1999;Wetzels,

2002, 2006). In Portuguese (EP and BP), several analyses have been purported aiming to

explainhowword stressworks in the language, all of themagreeing in the fact that stress

assignmentinthissystemistheresultofaninteractionbetweenrhythmicandmorphological

properties. The two disputed approaches accounting for word stress in Portuguese have

defended,ontheonehand,thatstressisassignedonthebasisofsyllableweight(Bisol,1993,

1999;Carvalho,1987,1988;Wetzels,2002,2006)and,ontheotherhand,thatmorphological

constituency governs stress assignment (Andrade 1988/1992; Andrade& Laks, 1992; Lee,

1995;Mateus,1983;Mateus&Andrade,2000;Pereira,1999).

In Chapter 1, we listed the facts indicating that weight, on the one hand, and

morphology,on theother,playarole inPortuguesewordstressassignment.Wewill recall

thembelow:

A. Evidences for weight in Portuguese word stress assignment (Bisol, 1993, 1999;

Carvalho,1987,1988;Wetzels,2002,2006;Lee,2001,2006,2007):

(i) EP has full stressed vowels and full unstressed vowels, suggesting that full

unstressed vowels might be the result of a coalescence phenomenon and

therefore be complex (heavy) segments ‐ e.g., s[E]lvágem 'wild', c[a]ixóte 'box',

séni[ç]r'senior',c[ç]rar'toblush'.

210Cf.Chapter1,section1.2..

272

(ii) In thenouns' system,heavysyllables, i.e., syllablesending ina consonant (/R, l,

s211/) or a glide overwhelmingly bear stress. Syllable weight counts for stress

assignmentdirectlyandindirectlyattheRhymelevel(i.e.,syllablesareheavyat

the levelof theRhymeandat the levelof theNucleus ‐/VC,VG,VN,VNC,VGC,

VGN,VGNC/) ‐ e.g.,balão 'balloon',carapáu 'mackerel',rapáz 'boy',amór 'love',

anél'ring'.

(iii) In Portuguese, antepenultimate syllables are never stressed when the

penultimate syllable is heavy ('CV.CVC.CV) ‐ e.g., cérebro 'brain' and cadastro

'criminalrecord'but*cádastro.

(iv) Stressedmid vowels in prefinal syllables followed by a finalheavy syllable are

neutralized (Spondaic Lowering) ‐ e.g., m[»ç]vel/*m[»o]vel 'desk',

[»ç]rfão/'*['o]rfão'orphan'.

B. Evidence for morphology playing a role in Portuguese word stress assignment

(Andrade,1988/1992;Andrade&Laks,1992;Lee,1995,2001,2006,2007;Mateus,

1983;Mateus&Andrade,2000;Pereira,1999):

(i) Non‐verbsandverbsaresubjecttodifferentwordstressalgorithms.

(ii) Innon‐verbs,stressgenerallyfallsonthelastsyllableofthestem‐e.g.,gát]o'cat',

rapáz]'boy'.

(iii) Non‐verbs' rule derives oxytonic athematic words without final consonant or

glide‐e.g.,café]'coffee',chaminé]'chimney',sofá]'couch',champô]'shampoo'.

(iv) Inverbs,regularitiesastostressassignmentinmorphologicalconstituents,and

notas referring towordpositionor syllableweight (namely themevowelsand

tense and mood suffixes) are observable ‐ e.g., calará 's/he will shut up',

calaríamos'wewillshutup'.

211Cf.footnote73.

273

Insum,bothanalysessupportingweight‐sensitivityandmorphology‐dependenceare

able to describe and account for word stress assignment in the language212 and,

simultaneously, they both raise a number of problems. The approach defending weight‐

sensitivitydoesnotexplainstressedlightsyllablesinword‐finalposition(e.g.,café 'coffee'),

or, instead,postulatesthepresenceofa finalextrametricalconsonant,whichisnotrealized

phonetically (e.g., café(C) 'coffee' ‐ Bisol, 1993, 1999), to explain stressed light syllables in

word‐finalposition.Theapproachsupportingmorphology‐dependenceneglectsthefactthat

finalheavysyllablescruciallyattractstress,asshownin(189):

(189) a.Word‐final/CVs/‐e.g.,rapaz'boy'[{å»paS]

b.Word‐final/CVR/‐e.g.,amor'love'[å»moR]

c.Word‐final/CVl/‐e.g.,hotel'hotel'[ç»tE…]

d.Word‐final/CVN/‐e.g.,patim'skate'[på»ti‚]

e.Word‐final/CVGN/‐e.g.,sabão'soap'[så»bå‚w‚]

f.Word‐final/CVG/‐e.g.,carapau'mackerel'[kåRå»paw]

Thecases shown in (185)opposed to infrequentandmarkedcases like theones in

(190),whereafinalclosedsyllableisunstressed:

(190) a.lápis'pencil'[»lapiS]

b.túnel'tunel'[»tunE…]

Additionally,theapproachdefendingamorphology‐basedwordstressdoesnottake

into account that whenever a penultimate syllable is heavy (CV.'CVC.CV), stress never

withdrawsonesyllableback(e.g.,Andrade,1988/1992,1996,1997;Andrade&Laks,1992;

Pereira,1999).

Dataonwordstressacquisition,bothinPortugueseandinotherlanguages,hasbeen

mostlyfocusingontheacquisitionofstresspatterns213andrarelyfocusedontherelationship

between word stress assignment and morphological information, especially in languages

wheremorphological constituencymight be relevant for stress purposes (like Spanish and

Italian).

The acquisition of stress patterns and the observation of stress patterns across

developmentmayindicatewhatistherhythmiccharacterofthelanguage(eithertrochaicor

212Cf.Chapter2,section1.2.3..213Cf.Chapter2,section2.2..

274

iambic),whatisthedomainforstress,aswellasitmayprovideevidenceforbothwordstress

and foot directionality in a given language. It might furthermore indicate how and when

childrenacquireweight.

Fikkert(1994)proposedthatDutchchildrenstartwithaCVmonosyllable(Stage0)

andsoonstartproducing[SW]words(StageIandII)214.UntilStageII,Dutchchildrenmainly

learnthatstressintheirlanguageisattherightmostedgeofwordsandthatfeetarebinary

andleft‐headed.Quantity‐sensitivity,weightanddirectionalityareacquiredlater,inStageIII.

Thechildthenrealizesthattherearetwotypesofwordsinherlanguage‐/SW/and/WS/‐,

whoseprominencesareassignedon thebasisofsyllableweight.At thisstage,/SW/words

are produced correctly but /CV.'CVC/ and /'CV.CVC/words are producedwith level stress

(['CV.'CVC]). The child knows that closed syllables (VC) count as heavy and that feet in

his/her language are quantity‐sensitive. Since feet are left‐headed in the target language,

childrenrealizethatwordsmighthavetwofeet(inWSwords),andproduce/CV.'CVC/words

withlevelstress,asshownin(191).

(191) Representationforlevelstressin/WS/wordsinstageIIIofwordstressacquisition

inDutch(Fikkert,1994:215):

Wd3

F F 2 2

σs (σw) σs(σw)

4 4

ballon/bA»lçn/‐>[»bAn »dçn]

The words like /'CV.CVC/ are not produced target‐like, but with level stress

(['CV.'CVC]),sincethefinalsyllableofthesewordsisextrametricalandextrametricalityisnot

yet acquired. Therefore, children do not produce stress shift in target /CV.'CVC/ words

([CV.'(C)VC]),but,instead,putthesameamountofprominenceinbothsyllables.

However, as referred inChapter5215,Dutchprovides stronger evidence for syllable

weightasarelevantcuetowordstress(Kager,1989)thanEP,astheformerhassuperheavy

214ForadetaileddescriptiononDutchwordstressacquisition(Fikkert,1994),cf.Chapter2,section2.2..215Cf.Chapter5,section5.2.2..

275

syllables‐VVC‐thatattractstressalmostwithoutexception216.Also,Dutchhasvowellength

contrastwhichcountsforsyllableweight(Kager,1989).

As far as Portuguese (BP) is concerned, thework from Bonilha (2005) and Santos

(2001,2007)haveshedsomelightonthe(still)controversialissueofwordstressandword

stressacquisitioninPortuguese,asbothworksdescribedtheacquisitionpathundertakenby

Brazilianchildrenwithrespecttowordstress,suggestinganiambictendencyina language

withatrochaicrhythm(BP).

AccordingtoSantos(2001)217,atanearlystage,Brazilianchildrendonotdistinguish

betweenaccentandwordstress,astheyareabletoproduceutteranceslargerthanasyllable

ora foot.At this stage, childrenaredealingwithseveral intonationalcontours.Afterwards,

Brazilian children establish one single intonational contour (LH*L%) and fulfill it with

segmentalandsyllabicmaterial. Ina thirdstage,Brazilianchildrenrealizedisyllabicwords

andproduceboth[SW]and[WS]words.Atthisstage, theystilldonotdistinguishbetween

word and phrasal stress (lexical stress and accent, respectively), since they still use both

typesofwordstofulfillthebasicintonationalcontour(LH*L%),asshownin(192).

(192) Disyllabic ([SW] and [WS]) words mapped onto a LH*L% intonational contour

(Santos,2005:81):

L H* L%‐>intonationalpattern

(ws) #‐>iambicpattern

(sw)#‐>trochaicpattern

When SWW words are acquired, at a fourth stage, Brazilian children focus their

attentiononwordstressand learn thestressalgorithmof the language.Theauthorargues

that the acquisitionofphonological aspects suchas extrametricality, and the acquisitionof

morphologicalaspectssuchasthewordmarker,interactintheacquisitionofwordstressin

BP, since the speech productions of the twoBrazilian children observed showed that they

wereable todecomposewords (namely,byproducingderivedwords)andcorrectlyassign

stress. Furthermore, the author found no evidence for weight sensitivity in the Brazilian

childrenobserved.Below,welistthefactsthatleadtheauthortoconsiderthatweightdoes

notplayaroleinwordstressacquisitioninBP:

216Cf.Chapter5,section5.2.2..217Foramoredetaileddescriptionon thestudiesmentioned in the followingparagraphs,cf.Chapter2, section2.5..

276

(i) Wordswith finalheavystressedandunstressedsyllables (/'CV.CVC/ ‐e.g. lápis

'pencil'/»lapiS/‐and/CV.'CV(C)/‐e.g.,café'coffee'/kå»fE/,calor'heat'/kå»loR/)

wereproducedcorrectlysincethebeginningofstressacquisition,suggestingthat

childrendidnotdistinguishbetweenheavyandlightsyllables;

(ii) Stress shift was not found in /'CV.CVC/ words (e.g., lápis 'pencil' /»lapiS/),

showingthatheavysyllablesdonotnecessarilyattractstress.

Supporting Lee's (1995) analysis for Portugueseword stress, Santos (2007) argues

thatchildrenaresensitivetothefactthatwordstressreliesonmorphologicalconstituency

and that the domain for stress in non‐verbs is the stem (Lee, 1995). Initially, Brazilian

childrenproduce[WS]wordswherenowordmarkerisrealized,providingevidenceforthe

earlysensitivity to thedomain forwordstress (stems innon‐verbsand the lexicalword in

verbs),ontheonehand,andfortheextrametricalstatusofthewordmarkerin/‐SW/non‐

verbs.Thefinalsyllablein[+nouns]withwordmarkerisextrametricalandthattheearlyfoot

inBPacquisitionisaniamb(sapát]o'shoe').Inverbs,Santos(2007)foundasimilartendency

as the one found in nouns, i.e., Brazilian children tended to produce /WS/ forms (mainly

Infinitives) earlier than /SW/. As far as verb forms are concerned, Santos (2007) did not

supportLee(1995),whodefendsthat thestressdomain is the lexicalwordandthedefault

footisatrochee,sinceinmostverbformsstressfallsonthepenultimatesyllableofthelexical

word.

AsdiscussedinChapter5,bothfromacross‐linguisticperspectiveandfromatarget

frequency perspective, the assumption of an iambic default foot for Portuguese [+nouns],

which would motivate an early iambic tendency, is hard to explain. As demonstrated in

Chapter1(section1.2.4.),Portuguesehasanoverwhelmingtrochaicrhythm.Nearly75%of

thedisyllabicwordsbearstressinthepenultimatesyllable.Intrisyllables,thepercentageof

wordswithstressinthepenultimatesyllableincreasesto80%.Evenifstructuralproperties

mayprovidethechildrenwithdifferentcues(assumingLee's,1995,2006,2007proposals),

the rhythmic structure heard by Portuguese children is overwhelmingly trochaic.

Additionally,asshowninTable2218,atendencyforiambicfeetintheearlyspeechofchildren

wasnotobservedinanytrochaiclanguage.

ContrarytoSantos(2007),Bonilha(2005)defendsthatwordstressisacquiredsince

the beginning of word production. The author found a reduced number of filler sounds,

especially in verbs, which seemed to indicate that the child observed was not paying

218Cf.Chapter2,section2.2.2..

277

attention to the intonational patterns of the language. Both iambs and trochees were

produced in the early stages, though stressed and initial syllables (/SW/ and /WS/)were

normally maintained, indicating a slight tendency for iambs. Contrary to Santos (2001),

Bonilha (2005) found evidence for weight‐sensitivity in the speech of the Brazilian child

observed. The author supported her claim forweight‐sensitivity in the following empirical

evidences:

(i) Wordswith final stressedheavy syllable (/CV.'CVG/words ‐ e.g.,papai 'daddy'

/pa»paj/)wereproducedcorrectlyfromthebeginning;

(ii) Words with penultimate stress (/'CV.CV/), words with a final stressed heavy

syllable (/CV.'CVG/) andwordswith a final stressed light syllable (/CV.'CV/) ‐

e.g., pato 'duck /»patu/, papai 'daddy' /pa»paj/ and café 'coffee' /ka»fE/,

respectively‐wereacquiredearlierthanwordswithpenultimatestressbutwith

final heavy syllables (/'CV.CVC/ ‐ e.g., lápis 'pencil' /»lapis/) and before words

withantepenultimatestress(/'CV.CV.CV/‐e.g.,música'song'/»muzika/).

The author empirically supports the analysis of Bisol (1992) for word stress in

Portuguese, according to whom Portuguese is weight‐sensitive. Bisol (1992) defends the

unmarked character of penultimate stress and final stress in heavy syllable. Final stress in

lightsyllable isduetoa finalabstractconsonantfulfillingasyllable‐finalposition.Basedon

Gordon (2002) and Kenstowicz (1994, 1996), Bonilha (2005) proposes that the sonority

prominence of Coda consonants, as well as mid low vowels219 should activate weight‐

sensitivity in Portuguese. On this assumption, children would earlier acquire unmarked

stress positions (penultimate stress ‐ /'CV.CV/ ‐ and final stress in heavy syllable ‐

/CV.'CV(G/C)/) and would later acquire marked stress positions (penultimate stress with

finalheavysyllable‐/'CV.CVC/‐andantepenultimatestress‐/'CV.CV.CV/).

DespitethevaluablecontributionofbothSantos(2001,2007)andBonilha(2005)to

theissueoftheacquisitionofwordstressinPortuguese(particularlyBP),manyquestionare

stillpending,namelythoseregarding:

219Notice,however,thatthisproposaldoesnotaccountforfinalstressinwordswithaword‐finalhighvowel:aqui'here'[å»ki],rubi'ruby'[{u»bi],peru'turkey'[pˆ»Ru].

278

(i) theroleofsyllableweightinstressassignment‐arechildrensensitivetoheavy

syllableswhileacquiringwordstress?

(ii) the role of morphology in stress assignment ‐ do children display different

behaviortowardsstressassignment innon‐verbsand inverbsand,withineach

morphosyntactic category, do children display a different behavior towards

wordswithdifferentmorphologicalconstituency?

(iii) thedomain forwordstress in the language ‐do children'sproductions suggest

thatthelexicalwordormorphologicalconstituents,suchasstems,mightbethe

domainforwordstress?

Specifically,theresearchquestionthatweaimtoaddressinthischapteris:giventhe

controversial status of word stress in Portuguese, what can the children's data say about

wordstressinthissystem?

Thehypothesesthatweposeare:

A. Given the results found in previous analyses on the adult language (Andrade,

1988/1992; Andrade & Laks, 1992; Mateus, 1983; Mateus & Andrade, 2000;

Pereira,1999)andonwordstressacquisition inPortuguese(Santos,2007),we

hypothesizethatmorphologicalacquisitioninteractswithwordstressandword

shapeacquisitioninEP.

B. Giventheresultsfoundinthepreviouschapterandonpreviousanalysesforthe

child data and for adult Portuguese (e.g., Bonilha, 2005, Bisol, 1993, 1999;

Wetzels,2002,2006),wehypothesizethatwordstressinEPacquisitionrelieson

aweight‐basedalgorithm.

IfPortuguesewordstress ismorphology‐dependent(e.g.,Lee,1995;Pereira,1999),

and assuming that the word‐final information in /SW/ [+nouns], but not in [+verbs], is

extrametrical(asproposedbyLee,1995andsupportedbySantos,2007withBPacquisition

data),thenwewouldexpectthatchildrenstartwith[WS]in[+nouns]and[SW]in[+verbs].

If, otherwise, Portuguese is weight sensitive, it is expected that children obey the

generalweight‐based rule for stress assignment (Bisol, 1992, 1994;Wetzels, 2006): stress

thepenultimatesyllableof the lexicalword,or the finalsyllable if it isheavy).Evidence for

stress‐attraction in heavy syllables would be found, through earlier acquisition of the

279

unmarkedstresspattern(penultimatestress,orfinalstressifthelastsyllableisheavy)anda

lateracquisitionofthemarkedstresspattern(penultimatestressifthefinalsyllableisheavy

andantepenultimatestress).

Inthe followingsection(section6.1.),wewillpresenttheresultsof theanalyseson

the morphology‐dependence and weight‐sensitivity in the speech of the five Portuguese

childrenobserved.

In section 6.1.1., we will show the results for the interaction of morphological

information (both in the noun and in the verbs paradigm). We will compare the intake

distribution (in terms of types and tokens) in the speech of the five children under

observation, with special attention to the distribution of word classes ([+nouns] and

[+verbs]) per stress pattern (monosyllabic, /‐SW/ and /‐WS/), across sessions (section

6.1.1.1.). In section 6.1.1.2., we will look to the morphology‐dependence hypothesis in

[+nouns],bycomparingtheacquisitionoftrocheeswiththeproductionandacquisitionofthe

wordmarker.Insection6.1.1.3.,wewillshowtheresultsforverbs,lookingattheacquisition

path undertaken by the children towards verbs tenses and the strategies carried by the

childrenin/‐SW/and/‐WS/verbs.

In section 6.1.2., we will show the results for weight‐sensitivity in the data of EP‐

speaking children. We will look at the deletion of unstressed light and heavy syllables

(section 6.1.2.1.), to the developmental path for words with heavy stressed and heavy

unstressed syllables (/CV.'CV/, /CV.'CVC/ and /'CV.CVC/ ‐ section 6.1.2.2.) and, finally, we

willlookatstressshiftinthesesamewords(section6.1.2.3.).Insection6.1.3.wewillpresent

thesummaryofthemainfindings.

In section 6.2., we will discuss the results found, both for the interaction of

morphologyandtheroleofsyllableweightduringwordstressacquisitioninEP.

Finally,insection6.3,wewillsummarizethemainfindingsofthischapter.

6.1.Results

In this section, we will show the results for morphology‐dependence and weight‐

sensitivityintheobservedchildren'sspeech.

This section will be divided into two parts: firstly, we will analyze the role of

morphological interaction in the acquisition of words stress (6.1.1.); secondly, we will

investigatetheroleofsyllableweightintheacquisitionofwordstress.(6.1.2.).

280

6.1.1.Ontheinteractionofmorphologyintheacquisitionofwordstress

Inthissectionweaimatshowingwhethermorphologicalinformationisrelevantfor

stress assignment in Portuguese, from the language acquisition point of view. We will

examinethedistributionofwordshapesandstresspatterns(targetmonosyllables,/SW/and

/WS) per word class (non‐verbs and verbs) in the children's intake (6.1.1.1.)., we will

investigatetherelationshipbetweenwordstressandmorphologicalconstituentssuchasthe

wordmarkerinnon‐verbs(6.1.1.2.)andtense/moodandpersonsuffixesinverbs(6.1.1.3.).

6.1.1.1.Thedistributionofwordclassesperwordshape

In the following tableswewill showthenumberof targets selectedby thechildren

(i.e.,word in the intake)with respect toword classes (mainly [+nouns] and [+verbs]), and

word shapes (monosyllables, /‐WS/ and /‐SW/). With this data, we intend to investigate

children'ssensitivityforwordclassduringwordstressacquisition.

Whencomparedto[+nouns],multisyllabicverbsingeneralarerareatthebeginning

and trisyllabiconesarevery scarce, aswewill observe, further in this section. Sinceverbs

were very few in some children's speech, we plotted together ‐SW and ‐WS, that is, all

multisyllabicwordsendinginSWorWS(e.g.,deixa'leaveit'[»dåjSå]andaperta's/hetiesup'

[å»pERtå]belongtothe‐SWparadigm,whereasdeixar'towash'[dåj»SaR]andapertar'totieup'

[åpˆR»taR]belongtothe‐WSparadigm).

Tables67‐71depictthewordshapeoftarget[+verbs]and[+nouns]selectedbyeach

of the children analyzed with respect to word shapes (monosyllabic, trochaic and iambic

verbs).Ablackstripeindicatestheturningpointtowardsatrochaictendency.Thenumberof

tokensandthenumberoftypeswillbeprovidedinthesetables.Typeswillappearbetween

brackets.Table67presentsthedatafromthewordshapesselectedbyClara.

281

Monosyllables ­SW ­WS

+Verbs +Nouns +Verbs +Nouns +Verbs +Nouns

S1 1(1)

S2 2(2)

S3 1(1) 1(1) 3(1) 9(1)

S4 3(2) 10(2)

S5

4(2) 6(3)

S6

9(2) 4(1) 4(1)

S7 4(2) 18(4) 8(4) 6(2)

S8 6(2) 13(2) 6(1) 10(3)

S9 2(1) 18(3) 13(4) 17(2)

S10 6(2) 18(4)

90(14)

23(3)

S11 31(7) 50(7) 10(3) 173(29) 2(2) 35(7)

S12 34(4) 43(5) 21(2) 143(22) 5(1) 50(11)

Table67.Distributionofwordclassesperstresspattern(Clara)

In Table 67 we observe that [+nouns] are the majority in Clara's speech at the

beginningofwordproduction(untilsession10).Onlymonosyllabicverbsareproduceduntil

session 10 (mainly é/está '(it) is', dá 'give (imp.)' tem 's/he has' and há 'there is'). After

session11,verbsareproducedconsistently,especially/‐SW/.

InTable68,weshowthedistributionofthewordclassesinInês'speech.

282

Monosyllables ­SW ­WS

+Verbs +Nouns +Verbs +Nouns +Verbs +Nouns

S1 1(1) 1(1) 19(4)

S2 22(2)

2(1) 1(1) 41(3)

S3 31(3) 16(6) 16(3) 60(8)

S4 31(6) 17(4) 1(1) 38(16) 67(7)

S5 23(2) 5(4) 1(1) 39(14)

53(6)

S6 18(3) 52(5) 2(1) 124(31) 1(1) 99(11)

S7 20(4) 18(6) 108(28) 1(1) 81(10)

S8 8(4) 28(7) 4(4) 69(24) 1(1) 73(14)

S9 31(6) 36(11) 8(5) 159(50) 8(6) 87(16)

S10 61(9) 19(11) 29(8) 274(68) 7(6) 89(12)

S11 112(10) 46(16) 43(18) 297(94) 15(8) 136(21)

S12 176(14) 53(15) 29(11) 338(104) 32(17) 128(21)

S13 120(21) 54(17) 75(32) 360(127) 53(25) 106(25)

S14 171(14) 43(16) 60(25) 271(105) 34(18) 89(17)

S15 136(17) 46(13) 53(24) 287(101) 41(25) 50(16)

S16 201(24) 99(15) 80(30) 293(102) 49(20) 60(19)

S17 103(15) 43(16) 51(20) 196(81) 42(16) 25(15)

S18 175(24) 123(29) 82(53) 261(110) 36(26) 56(25)

Table68.Distributionofwordclassesperstresspattern(Inês)

In Table 68, we observe that, like Clara, the majority of the child's speech at the

beginning (until session 7) consists in [+nouns].Monosyllabic, but notmultisyllabic, verbs

are present from the beginning. Trochaic verbs are selected before (in session 2) iambic

verbs(insession6)andinitiallythenumberoftypesishighlyreducedinbothstructures.

Table69presentsthedistributionofstresspatternsperwordclassinJoana'sspeech.

283

Monosyllables ­SW ­WS

+Verbs +Nouns +Verbs +Nouns +Verbs +Nouns

S1 1(1)

S2

4(1)

S3 2(2)

2(1) 3(1)

S4 4(1) 2(2) 3(2)

S5 2(2)

S6 2(1)

S7 4(2) 1(1) 18(5) 2(2)

S8 4(1) 2(2)

6(4) 2(2)

S9 2(2) 17(5) 1(1) 23(12)

27(13)

S10 8(3) 43(12) 29(19) 1(1) 27(14)

S11 16(4) 24(15) 1(1) 87(55) 1(1) 50(21)

S12 14(5) 23(8) 4(1) 145(79) 2(2) 32(19)

S13 16(5) 24(9) 7(6) 112(79) 8(5) 27(15)

S14 147(11) 31(17) 43(13) 242(129) 10(9) 31(22)

Table69.Distributionofwordclassesperstresspattern(Joana)

In Joana's speech a monosyllabic tendency is noticeable, both in [+nouns] and

[+verbs].Words longer than one syllable are scarce until session 7 and onlymultisyllabic

[+nouns] are selected until that period. Multisyllabic verbs are selected in session 9

(trochees), but only from session 10 onwardsmultisyllabic verbs (both /‐SW/ and /‐WS/)

areselected.

Table70 shows thedistributionof [+nouns] and [+verbs]perword shape in João's

speech.

284

Monosyllables ­SW ­WS

+Verbs +Nouns +Verbs +Nouns +Verbs +Nouns

S1 6(1)

S2 4(1)

S3 1(1) 8(1)

S4 6(2) 2(1)

S5 14(3)

S6

10(2) 5(2)

S7 3(1) 15(4) 18(2)

S8

1(1) 23(6)

11(5)

S9 5(2) 20(3) 14(6) 2[1) 14(6)

S10 3(1) 10(3) 39(6) 1(1) 23(8)

S11 15(1) 7(1) 35(6) 17(6)

S12 20(2) 18(4) 20(7) 1(1) 26(7)

S13 8(2) 11(1) 23(3) 54(7)

S14 2(1) 4(3)

17(6) 1(1) 31(6)

S15 4(2) 12(3) 1(1) 10(7) 30(5)

S16 9(4) 17(7) 9(3) 87(31) 6(1) 34(17)

S17 3(1) 56(10) 1(1) 109(45) 3(3) 22(8)

S18 5(2) 21(3) 22(5) 104(46) 3(1) 50(14)

S19 19(5) 22(9) 9(3) 117(54) 5(4) 28(14)

S20 6(1) 28(8) 9(4) 139(52) 1(1) 35(15)

S21 1(1) 16(8) 3(3) 135(53) 35(15)

S22 20(5) 30(5) 10(4) 117(46) 2(2) 38(19)

Table70.Distributionofwordclassesperstresspattern(João)

In Table 70, we observe that, contrary to Joana, João has a small amount of

monosyllabicwords, being either [+nouns] or [+verbs].Multisyllabic /‐SW/and /‐WS/ are

selected from the beginning, though /‐SW/ [+nouns] aremore frequent than /‐WS/.Verbs

areveryscarceuntilsession15.Insession9,Joãoselectshisfirstmultisyllabicverbform(‐

285

WS/) but he keeps the same type (acabou 'it finished') until late. Multisyllabic/‐SW/ are

producedconsistentlyandmorefrequentlyfromsession15onwards.

Table71showsthedistributionofstresspatternsperwordclassinLuma'sspeech.

Monosyllables ­SW ­WS

+Verbs +Nouns +Verbs +Nouns +Verbs +Nouns

S1 1(1) 1(1)

S2 2(1)

S3 1(1) 9(1)

S4 31(2) 7(2)

S5

5(3)

S6 13(1)

7(3) 2(1)

S7 30(1) 1(1) 22(6)

S8 35(1) 1(1) 29(5) 2(1)

S9 8(2) 1(1) 13(3)

S10 1(1) 2(1) 2(1)

S11 1(1) 5(4) 2(2)

S12 5(1) 9(2) 6(1)

S13 21(1) 12(5)

3(1)

S14 8(4) 6(1) 14(2)

S15 41(1) 28(8) 3(1) 29(3)

S16 8(2) 1(1) 36(1) 33(2)

S17 8(1) 11(3) 17(2)

S18 7(2) 15(4) 21(3)

S19 22(1)

4(2) 15(3)

S20 12(2)

12(1) 22(7) 49(5)

S21 3(2) 5(3) 7(1) 19(6) 57(5)

S22 2(1) 14(1) 26(7) 30(2)

S23 2(1) 1(1) 13(1) 30(3)

52(4)

286

S24 2(1) 3(1) 29(7) 48(3)

S25 5(1) 9(2) 31(10) 21(7)

S26 1(1) 14(14) 2(1) 23(6) 41(8)

S27 2(1) 10(5) 1(1) 19(8) 65(5)

S28 1(1) 14(4) 30(6)

60(9)

S29 13(2) 17(7)

46(10) 1(1) 75(10)

S30 6(3) 18(11) 8(3) 119(33) 86(14)

S31 10(7) 22(12) 11(7) 134(48) 3(2) 71(13)

S32 32(10) 36(11) 96(29) 166(33) 59(5)

S33 48(9) 22(13) 41(20) 203(63) 4(3) 94(14)

S34 48(8) 41(14) 50(17) 220(69) 27(11) 108(10)

S35 60(9) 29(10) 52(19) 184(73) 17(8) 68(8)

S36 122(12) 23(9) 54(17) 224(76) 29(24) 116(19)

S37 70(12) 12(8) 26(17) 207(69) 23(15) 81(15)

Table71.Distributionofwordclassesperstresspattern(Luma)

InLumaweobservedthesametrendalreadyobserved in theotherchildren. In the

beginning, she mainly produces monosyllabic words ([+nouns] and [+verbs]) and

multisyllabic[+nouns].Untilsession30,thechilddoesnotselectverbsandwhenshedoesso

(session 14), she repeats the same type (the /WS/ verbmarchar 'to march') in the three

followingsessions.Insession20,Lumaproducesthe/‐SW/verbformolha'look!'andthatis

theonly/‐SW/verb formproduceduntil session27.After session30, sheproduces/‐SW/

verbformsconsistentlyandfromsession33onwards,thesameappliesto/‐WS/verbforms.

The distribution of [+nouns] and [+verbs] across sessions in all children indicates

that,attheearlystages,acomparisonbetweenthetwotypesofwordclasses([+nouns]and

[+verbs])perstresspatternisnotpossible,sincemultisyllabicverbsareveryscarce.

The distributional information on word classes and stress patterns presented

indicatesthatmultisyllabicverbsareselectedlaterthanmonosyllabicverbsinthespeechof

thefivechildrenobserved.Monosyllabicverbsareproducedfromtheearlystages,aswellas

multisyllabicnouns.

InTable72,werepresent thedevelopmentalpath formono‐andmultisyllabicnon‐

verbsandverbsinthechildren'sintake:

287

Monosyllables Multisyllables

[+Nouns] [+Verbs] [+Nouns] [+Verbs]

MomentI

MomentII

Table72.Developmentforthedistributionofwordclassesperwordshapeandstresspatterninthechildren'sintake.

Also,theresultsofthissectionshowedthatthreechildrenmoreandearlierselect/‐

WS/[+Nouns]andtwochildrenearlierselected/‐SW/[+Nouns].Note,however,thatinthese

tablestheentireintake(includingreduplicatedtargets)wastakenintoaccount.

Inthefollowingsectionswewillshowresultsfromtheanalysesontwostress‐related

aspect:(i)therelationshipbetweentheacquisitionof/SW/[+nouns]andtheproductionof

thewordmarker,(ii)verbsproductionandstrategiesinthechildrenobserved,lookingcloser

attheverbtensesproduced.

6.1.1.2.Lookingcloseratthenounparadigm

In this section we will investigate the children's production of /SW/ and the

productionofthewordmarkerin/SW/.

In Chapter 5 (section 5.1.1.2.), we showed that the early production of Portuguese

childrenmostlyhada [WS]shape.However,wealsoshowedthat these iambsweremainly

produced as reduplications or resulted from epenthesis on the left of the circumscribed

syllablesfromthetargetword.

Asalreadymentionedinthisdissertation220,Santos(2007)defendsthatearlywords

inBPconformtoaniambicfoot.Theauthorsupportsherclaiminthefactthatchildrenare

obeyingthemorphology‐basedstressalgorithminthetargetlanguage,i.e.,stressfallsonthe

lastsyllableofthestem(Lee,1995).Sincetrochees,butnotiambs,bearafinalwordmarker

that,accordingtothetargetdescriptionfollowedbytheauthor(Lee,1995),isextrametrical,

Brazilianchildrenwerenotproducingearlytrocheesduetotheunmasteryofmorphological

contrasts (mainly, gender contrasts), on the one hand, and due to the unmastery of

extrametricality.

FollowingSantos(2007),thepredictionforthissectionisthat,sincethewordmarker

bears morphological and extrametrical information, morphological contrasts, in this case,

220Cf.reviewontheacquisitionofwordstressinBP, inChapter2,section2.5.,andthediscussioninChapter5,section5.2..

288

gendercontrast(masculineandfeminine),willnotbenoticedbeforetrocheesareacquired.

Trocheeswithoutthewordmarker,whicharescarceinthetargetlanguage,however,willnot

beproducedastrocheeseither,andwillpreferablybeproducedasiambs(e.g.,Noddy'name'

/»nçdi/‐>[nç»di],lápis'pencil'/»lapiS/‐>[la»piS]).

InTables73,75,77,79and81,wewillconsiderthecorrectproductionoftheword

markerin[+nouns](di‐,tri‐andpolisyllables).InTables74,76,78,80and82,weaccounted

for the production of trochees, irrespective of the morphosyntactic category, in order to

describe the acquisition of trochees in general and not only trochaic nouns.We took into

account the correct production of thewordmarker, the production of trochees target‐like

(/»kazå/‐>[»kazå]),casesofstressshift(/»kazå/‐>[ka»za]),orcasesinwhichthetruncated

syllable was the final one ([»za]). We consider that the word marker was not produced

whenever children used truncation or reduplication (/»kazå/ ‐> [»ka] or /»kazå/ ‐>

[»kaka]/[ka»ka]).Agreystripeindicatesthesessioninwhichthetrochaicfootpredominated

(>50%)ineachchild'sspeech.

InTable73and74,weshowClara's results for theproductionof thewordmarker

andtrocheesrespectively221.

221Ahighernumberoftargetsintheproductionofthewordmarkerthanintheproductionof/SW/isduetothefact that we are considering, not only disyllabic trochees, but also all the words with word marker. A highernumberoftargetsinthetableregardingtheproductionof/SW/thaninthewordmarkerproductionisduetothefactthatweareconsidering/SW/ingeneral(verbsincluded),andnotonly/SW/nouns.

289

WMProduction %

S3 3/3 100

S4 ‐ ‐

S5 2/4 50

S6 4/4 100

S7 6/8 75

S8 6/6 100

S9 12/13 92.30

S10 72/90 80

S11 103/173 59.53

S12 80/143 55.94

Table73.Wordmarkerproduction(Clara)

Productionof/SW/ %

S3 3/4 75

S4 ‐ 0

S5 2/4 50

S6 3/4 75

S7 1/5 20

S8 6/6 100

S9 5/9 55.55

S10 59/80 73.75

S11 103/161 63.97

S12 76/146 52.05

Table74.Productionof/SW/(Clara)

Claraisaccuratefromthebeginningbothintheproductionofwordmarkersandin

the production of trochees. Though she has few tokens from both trochaic and disyllabic

wordswiththewordmarkerintheearlysessions(untilSession8),ahighrateofproduction

in both structures is observable in Clara. The tables above indicate that Claramasters the

wordmarker andgender contrasts and she is able to correctlyproduce trochees, from the

beginningofwordproduction.

In (193) we present instances of Clara's production of trochees where the word

markerisobserved.

(193) Clara'saccuracyinearlytrocheesandthewordmarker

Orthogr. Gloss Target Output Agepapa 'foodfam.' /»papå / [»tatå‚] 0;11.1(S1)papa 'foodfam.' /»papå/ [»pQpa] 1;1.3(S3)água 'water' /»agWå/ [»a˘Bå] 1;3.6(S5)água 'water' /»agWå/ [»ak´] 1;4.19(S6)mano 'brotherfam.' /»månu/ [»manu] 1;5.16(S7)

In Tables 75 and 76, we show thewordmarker production and the production of

disyllabictrochees,respectively,inInês'speech.

290

WMProduction %

S3 0/3 0

S4 2/26 7.69

S5 3/26 11.53

S6 6/87 6.89

S7 2/72 2.77

S8 1/34 2.9

S9 66/101 65.34

S10 151/187 80.74

Table75.Wordmarkerproduction(Inês)

Productionof/SW/ %

S3 0/4 0

S4 1/27 3.70

S5 1/27 3.70

S6 1/90 1.11

S7 3/73 4.10

S8 5/39 12.82

S9 82/140 58.57

S10 166/221 75.11

Table76.Productionof/SW/(Inês)

In Tables 75 and 76, we observe that the percentage rates in the word marker

productionandintheproductionof/SW/isverysimilaracrosssessions.Insession9,Inêsis

abletoproducethemajorityofwordmarkersandtrocheestarget‐like.

In (194)we present Inês' production of /SW/, before and after themastery of the

trochaicfoot.Theinstancesbelowshowthat, initially(untilsession8),Inêstruncates/SW/

into[S],shereduplicatesthestressedsyllableorsheaddsafillersoundbeforethestressed

syllable,thusgenerallycreating,not[SW],but[S]or[WS]words.

(194) Inês–productionof/SW/:

Orthogr. Gloss Target Output Agemanta 'blanket' /»må‚tå/ [»må‚] barco 'boat' /»baRku/ [»bQ:] 1;3.6(S4)

carro 'car' /»ka{u/ [»ka»ka] 1;4.9(S5)meia 'sock' /»måjå/ [»mE] banho 'bath' /»bå¯u/ [å»bå] 1;5.11(S6)

balde 'bucket' /»ba…dˆ/ [»pa] 1;6.11(S7)água 'water' /»agWå/ [»a:]banho 'bath' /»bå¯u/ [å»ba] 1;7.2(S8)

boa 'good' /»boå/ [»boå]tampa 'lid' /»tå‚på/ [»pata] 1;8.2(S9)

casa 'house' /»kazå/ [»katJå]banho 'bath' /»bå¯u/ [»baju]bolos 'cakes' /»boluS/ [»boluS]

1;9.19(S10)

Theinstancespresentedaboveshowthat,intheearlysessions(untilsession8),Inês

mainlytruncatestarget/SW/,deletingthefinalsyllable,reduplicatesthestressedsyllableor

291

usesafillersoundbeforethetruncatedstressedsyllable.Fromsession9onwards,thechild

produces/SW/target‐like.

Tables77 and78 show theproductionvalues for thewordmarkerproduction and

theproductionratesin/SW/byJoana.

WMProduction %

S7 0/16 0

S8 0/4 0

S9 1/12 8.33

S10 4/16 25

S11 15/43 34.88

S12 53/76 69.73

S13 37/53 69.81

S14 73/98 74.48

Table77.Wordmarkerproduction(Joana).

Productionof/SW/ %

S7 0/16 0

S8 0/4 0

S9 0/13 0

S10 2/16 12.5

S11 13/45 28.88

S12 57/91 62.63

S13 44/74 59.45

S14 108/163 66.25

Table78.Productionofdisyllabictrochees(Joana).

In Tables 77 and 78, we observe that there are simultaneous production rates for

both thewordmarkerand trochees. Joanaovercomes the50%of target‐likeproductionof

bothtrocheesandthewordmarkerapproximatelyinthesamesession,whichindicatesthat

shedoesnotacquirethewordmarkerbeforetrocheesorvice‐versa.Also,itisnoticeablethat

untilsession7,thereareno/SW/and,untilsession9,Joanaisnotabletocorrectlyproduce

any/SW/target‐like.

Theexamplesin(195)illustrateJoana'sproductionoftrochees,beforeandaftertheir

acquisition.

292

(195) Joana–productionof/SW/:

Orthogr. Gloss Target Output AgeCarla 'name' /»kaRlå/ [»ka]escola 'school' /S»kçlå/ [»kç] 1;8.4(S8)

pato 'duck' /»patu/ [pa»pa]escola 'school' /S»kçlå/ [kç»kç]sapato 'shoe' /så»patu/ [pa»pa]

1;9.25(S9)

luva 'glove' /»luvå/ [»bu:] 1;10.22(S10)Nando 'name' /»nå‚du/ [»¯å‚] 2;0.9(S11)seco 'dry' /»seku/ [»tjEku]outra 'other' /»otRå/ [»otJå] 2;2.9(S12)

mota 'motorbike' /»mçtå/ [»mç:tJå]fralda 'diaper' /»fRa…då/ [»fawdå]pedra 'stone' /»pedRå/ [»pEdå]

2;4.1(S13)

quadro 'picture' /»kWadRu/ [»kajdJu]pato 'duck' /»patu/ [»patJu] 2;6.24(S14)

In the instances above, we observe that truncation and reduplication are the

strategiesmostlyusedbyJoanatodealwithtargettrochees.Inanearlyperiod(untilsession

11), the final syllable in/SW/ isdeleted.Lateron (fromsession12onwards), trocheesare

producedtarget‐like.

Tables79and80showtheproductionvaluesof João for thewordmarker in/SW/

andthetarget‐likeratein/SW/.

WMProduction %

S7 2/15 13.33

S8 4/18 22.22

S9 6/13 46.15

S10 24/39 61.53

S11 3/17 17.64

S12 7/13 53.84

S13 18/23 78.26

S14 14/17 82.35

S15 4/7 57.14

Table79.WordMarkerproduction(João)

Productionof/SW/ %

S7 2/15 13.33

S8 1/18 5.55

S9 2/13 15.38

S10 5/39 12.82

S11 3/17 17.64

S12 4/12 33.33

S13 5/23 21.73

S14 8/17 47.05

S15 3/7 42.85

Table80.Productionof/SW/(João)

293

Inthesetables,wedidnotreachthesessioninwhichJoãoacquirestrochees(Session

16,at1;9.25), as Joãoreached thepointabove50% in thewordmarkerproductionbefore

that point (in session 12). In João, themastery of the wordmarker is attained before (in

session12)theacquisitionoftrochees(insession16).

Inthefollowingexamples,wewillshowJoão'sproductionsfor/SW/.

(196) João–productionof/SW/:

Orthogr. Gloss Target Output Ageágua 'water' /»agWå/ [a»ba˘] bola 'ball' /»bçlå/ [»bå˘w]panda 'pandabear' /»på‚då/ [»pa]

1;3.21(S7)

bolo 'cake' /»bolu/ [bo»u] 1;4.17(S8)patu 'duck' /»patu/ [»pa]uva 'grape' /»uvå/ [då»du] 1;5.12(S9)

meia 'sock' /»måjå/ [»må»må]bola 'ball' /»bçlå/ [bu»la˘] 1;5.26(S11)

bola 'ball' /»bçlå/ [»Båjå] 1;8.25(S15)Guida 'name' /»gidå/ [»dita]bolo 'cake' /»bolu/ [»bo¥u]panda 'pandabear' /»på‚då/ [»månå]pombo 'pigeon' /»po‚bu/ [»må˘mu]

1;9.25(S16)

chicha 'meatfam.' /»SiSå/ [»titå] zebra 'zebra' /»zebRå/ [»bibå] 1;10.11(S17)

Joãocorrectlyproducesthewordmarker,butincorrectlyproducestrocheesasiambs,

andtruncates/SW/tomonosyllables(untilsession15).Thesestrategies(stressshift,along

withtruncation)areabandonedinthecourseofdevelopmentandthechildwilllaterproduce

thewordmarkeranddisyllabictrocheestarget‐like(fromsession16onwards).However,it

isworthwhilementioning that thischild isable toproduce thewordmarkerwhenhedoes

notproducedisyllabictrocheestarget‐like,becauseheusedstressshiftinearlytrochees.The

data from this child suggest that the acquisition of trochees is not dependent on the

acquisitionofthewordmarker,ashecanproducethewordmarkerandstillusestressshift

ofthetype/SW/‐>[WS].

In (197) we provide some instances of stress shift in the early sessions of João's

speech.

294

(197) StressshiftinJoão

Orthogr. Gloss Target Output Ageágua 'water' /»agWå/ [å»Bå] 1;1.12(S3)água 'water' /»agWå/ [å»Bå] 1;2.13(S5)água 'water' /»agWå/ [a»ba˘] 1;3.21(S7)bolo 'cake' /»bolu/ [bo»u]água 'water' /»agWå/ [å»wa] 1;4.17(S8)

uva 'grape' /»uvå/ [du‚»a] 1;5.12(S10)bola 'ball' /»bçlå/ [bo»a]/[bu»la˘] 1;5.26(S12)

Tables81and82showLuma'sproductionvaluesforthewordmarkerandtrochees,

respectively.

WMProduction %

S25 1/30 3.22

S26 2/22 9.09

S27 9/17 52.94

S28 6/19 31.57

S29 18/36 50

S30 49/87 56.32

S31 58/103 56.31

S32 42/101 41.58

S33 88/161 54.65

S34 80/147 54.42

S35 75/115 65.21

Table81.Wordmarkerproduction(Luma)

Productionof/SW/ %

S25 1/30 3.33

S26 2/24 8.33

S27 0/18 0

S28 8/30 26.66

S29 9/38 23.68

S30 30/122 24.59

S31 42/118 35.59

S32 97/208 46.63

S33 130/215 60.46

S34 138/217 63.59

S35 101/146 69.17

Table82.Productionof/SW/(Luma)

Luma takes a pattern similar to João's, as she is able to produce thewordmarker

beforetheacquisitionofdisyllabictrochees,sincethischildalsousessomeinstancesofstress

shiftofthetype/SW/‐>[WS].However,thepercentagesofstressshiftofthetype/SW/‐>

[WS] inLumaaremorereduced than theonesobserved instressshiftof the type/WS/ ‐>

[SW](cf.Table60222).

Lumahasaratehigherthan50%forthewordmarkerproductioninSession30and

inSession33fortheproductionoftrochees,whichconstitutesanapproximateintervalofone

month. As João, Luma is able to produce the word marker before trochees are acquired,

222InChapter5,section5.1.3.4..

295

furthersuggestingthattheacquisitionoftrocheesmightnotbedependentoftheacquisition

ofthewordmarker.

The examples in (198) show the production of /SW/ where truncation to

monosyllablesandstressshiftareobservable,untilsession30.Fromsession31onwards,we

observethat/SW/areproducedtarget‐like.

(198) Luma–productionof/SW/:

Orthogr. Gloss Target Output Agegato 'cat' /»gatu/ [»ta˘]meias 'socks' /»måjåS/ [»nåj]polvo 'octopus' /»po…vu/ [bo»bo]

2;0.13(S25)

Tito 'name' /»titu/ [ti»to] 2;0.27(S26)carro 'car' /»ka{u/ [kA»{o] 2;1.10(S27)casa 'house' /»kazå/ [»ka] 2;3.26(S30)tinta 'ink' /»ti‚tå/ [»ti‚tå]linda 'beautiful' /»li‚då/ [»ni‚då]alto 'tall' /»a…tu/ [»atu] sapo 'frog' /»sapu/ [»fapu]

2;4.11(S31)

outro 'other' /»otRu/ [»otu] 2;4.25(S32)pedra 'stone' /»pEdRå/ [»pEdå] 2;5.15(S33)

Like João, Luma presents a period of word production where stress shift is found

(until session 30). In an early period (until session 30, at 2;3.26), Luma does not have a

consistent production of /SW/223, she has mostly target trochees truncated to

monosyllables224, reduplicated [WS]words or stress shift in non‐reduplicatedwords (until

session30).Afterthisperiod(fromsession31onwards),Lumacorrectlyproducestheword

markerandtargettrochees.

The observation of the developmental path pursued by the five children analyzed

withrespecttotheacquisitionof/SW/andthewordmarkerindicatesthatbothappeartobe

related.However,twoofthechildren(JoãoandLuma)couldproductivelyproducetheword

marker before the acquisition of trochees, through the recursion to stress shift of the type

/SW/‐>[WS].Despitethefactthattwoofthechildrenobserveddisplaystressshifttowards

iambs(see,forinstance,João'srenditionsshownin(174)),itisworthsayingthatstressshift

(/SW/‐>[WS]) in non‐reduplicated forms was not common in the speech productions of

Portuguesechildreningeneral225.

223Cf.Chapter5,Table32,forthepercentageoffaithful/SW/inLuma.224Cf.Chapter5,Table49forthepercentageoftruncationof/WS/and/SW/tomonosyllables(Luma).225Cf.Chapter5,section5.1.3.4..

296

Inordertofurtherinvestigatetheexistenceofarelationshipbetweentheacquisition

oftrocheesandtheacquisitionofgendercontrastsinanearlystage,weadditionallyprovide

information on the strategies used by the observed children towards target trochees not

bearing the word marker. Notice that the frequency of target trochaic words without the

wordmarker forms in child speech (as in the adult language) is scarce, as attested by the

reducednumberofwordsshowedinthetablesbelow.Recall,also,thatintheassumptionof

anearlyiambicfootandintheassumptionofalatemasteryoftrocheesduetomorphological

interaction (namely, to the late mastery of gender contrasts), we would expect that

/SW/wordswithoutgendermarkerwereproducedasiambs.

Inthefollowingtables,wewillshowalltheearlyproductionsoftargettrochaicwords

withoutwordmarkerinthefivechildrenobserved.In(199),weshowClara'srenditionsfor

theonlytargettrocheewithoutwordmarkerselectedbythechild,Noddy.

(199) Clara–productionof/SW/withoutwordmarker:

Orthogr. Gloss Target Output Age[»nçtHi][mˆ»nçti] 1;8.20(S10)

[nå»nç][nå»nçti][»nçti][»nçki][nç»nç]

1;9.23(S11)Noddy 'name' /»nçdi/

[no»now][nç»nç][»çti]

1;10.15(S12)

InClara's speech,weobserve that, until the endof the observationperiod (session

12),thereisanunstableproductionofthewordNoddy/»nçdi/'name'.Itisproducedtarget‐

like([»nçtHi],insession10),asareduplicatediamb([nç»nç],insession11and12)orevenasa

trisyllable ([mˆ»nçti], in session 10). Clara's productions for /SW/non‐verbswithoutword

markerisveryvariableacrosstheobservationperiod.

Theexamplesin(200)showtheproductionofBambi/»bå‚bi/'name'andlápis/»lapiS/

'pencil',theonlytwotrochaicwordswithoutwordmarkerselectedbyInês.

297

(200) Inês–productionof/SW/withoutwordmarker:

Orthogr. Gloss Target Output AgeBambi 'name' /»bå‚bi/ [»baba»ba]

[Bå»bå][»bab][E»be][»bi][pe»be][be»bE]

1;3.6(S4)

Bambi 'name' /»bå‚bi/ [bå][å»bå][ˆ»bå][»ba»a»ba]

1;4.9(S5)

Bambi 'name'' /»bå‚bi/ [»Bå] 1;6.11(S7)lápis 'pencil' /»lapiS/ [»patH]

[Qba»tu][»patu][»bato]

1;8.2(S9)

Bambi 'name' /»bå‚bi/ [»babi] 1;10.29(S11)Bambi 'name' /»bå‚bi/ [»babi] 2;0.11(S12)lápis 'pencil' /»lapiS/ [»api] 2;0.11(S12)

InInês'speech,weobservethat,untilsession7,trochaicwordswithoutwordmarker

areeithertruncatedtomonosyllables(e.g.,Bambiproducedas[»bi], insession4,or[bå], in

session 5), reduplicated (e.g., Bambi produced as [»baba»ba] or [Bå»bå], in session 4), or

producedwithcircumscriptionofthestressedsyllableprecededbyafillersound(e.g.,[å»bå]

and [ˆ»bå], in session 5). From session 9 onwards, /‐SW/wordswithoutwordmarker are

producedas trocheesandthestrategiesusedbeforeareno longerobserved.As inClara, in

Inêswe observe that /SW/non‐verbswithoutwordmarker go through an initialmoment

where no fixed shape is produced (they can take the form of a monosyllable, a [WS]

reduplication, be produced with a filler sound at the left‐edge or, yet, as a multiple

reduplication).Later(aftersession9),theyareproducedtarget‐like.

In(201),weshowJoana'searlyrenditionsfor/SW/wordswithoutwordmarker.The

words attempted by this childwereDinky/»di‚ki/ 'name',César/»sEzaR/ 'name',Heidi/»åjdi/

'name'andBarbie/»baRbi/'name'.

298

(201) Joana–productionof/SW/withoutwordmarker:

Orthogr. Gloss Target Output AgeDinky 'name' /»di‚ki/ [»ki‚˘]

[»di‚] 1;9.25(S9)

César 'name' /»sEzaR/ [»tSE] 2;0.9(S11)Heidi 'name' /»åjdi/ [»atS]

[»adH´] 2;4.1(S13)

César 'name' /»sEzaR/ [»SE] 2;4.1(S13)César 'name' /»sEzaR/ [»CeΩå] 2;6.24(S14)Barbie 'name' /»baRbi/ [»pabi] 2;6.24(S14)

The first/SW/wordwithoutwordmarker is selected in session9,by Joana (Dinky

/»di‚ki/ 'name'). In session 9 and 11, thesewords aremainly truncated to amonosyllable.

From session 13 onwards, we observe that Joana is able to produce /SW/wordswithout

word marker maintaining the target stress pattern (e.g.,Heidi /Ȍjdi/ 'name' produced as

[»adH´], in session13,César /»sEzaR/ 'name'producedas [»CeΩå]andBarbie /»baRbi/ 'name'

producedas[»pabi],insession14).InJoana,/SW/non‐verbswithoutwordmarkerarenever

producedasiambs.

In(202),wepresentJoão'srenditionsforthe/SW/wordswithoutwordmarker.

(202) João–productionof/SW/withoutwordmarker:

Orthogr. Gloss Target Output AgeNoddy 'name' /»nçdi/ [»didi]

[di»di] 1;9.25(S16)

Noddy 'name' /»nçdi/ [a»didi][»di»di][di»di]

1;10.11(S17)

táxi 'taxi' /»taksi/ [»tati] 1;10.26(S18)Noddy 'name' /»nçdi/ [å»dçdi] 1;10.26(S18)Noddy 'name' /»nçdi/ [å»dadi]

[»dadi][»dçdi]

1;11.10(S19)

táxi 'taxi' /»taksi/ [»tati] 1;11.10(S19)ténis 'snicker' /»tEniS/ [»tEni] 1;11.10(S19)

LikeJoana,Joãoselects/SW/wordswithoutwordmarkerlaterindevelopment.The

firstwordformofthistypeoccursinsession17(Noddy/»nçdi/'name').Insession16and17,

Joãoshowsanunstableproductionoftrocheeswithoutwordmarker,producingitastrochaic

299

oraniambicreduplications(e.g.,[»didi]or[di»di],insession16),asadisyllableprecededbya

filler sound ([a»didi], in session 17) or with level stress (e.g., [»di»di] in session 17). From

session 18 onwards, the three /SW/ types without wordmarker selected by João (Noddy

/»nçdi/ 'name',táxi/»taksi/ 'taxi'andténis/»tEniS/ 'snicker')keepthetargetstresspattern,

thoughtheycanbeprecededbyafillersound(e.g.,[å»dçdi]and[å»dadi], insessions18and

19,respectively).Asinthechildrenpreviouslyobserved,Joãodoesnotdisplayaninitialclear

preferenceforaniambicshapein/SW/non‐verbswithoutwordmarker.

The examples in (203) show Luma's production of target trochaic words without

wordmarker.

(203) Luma–productionof/SW/withoutwordmarker:

Orthogr. Gloss Target Output AgeNoddy 'name' /»nçdi/ [»ˆ»¯å»¥i]

[tç»tçR¥i][»nç¥i][»nån][»n廯i][»nåne][»¯å¯e˘][»nåni]

1;11.1(S22)

Noddy 'name' /»nçdi/ [ttAR»¥i][tåR»¥ç¥i] 1;11.15(S23)

Noddy 'name' /»nçdi/ [tç»tçR¥i˘][doR»¥oRi] 1;11.29(S24)

Noddy 'name' /»nçdi/ [toR»¥oRi] 2;0.13(S25)Heidi 'name' /»åjdi/ [»a»di]

[»adi] 2;2.4(S28)

Noddy 'name' /»nçdi/ [tçR»¥çRi˘][tçR»dçRi][nçR»¥çRi]

2;2.4(S28)

Noddy 'name' /»nçdi/ [toR»¥oRi] 2;2.22(S29)Goldy 'name' /»go…di/ [»dodi]

[»dodi˘][»gi˘]

2;3.26(S30)

Noddy 'name' /»nçdi/ [tçR¥ç»Ri][tçR¥çR»¥i][doR¥o»Ri˘][»¯ç»di˘][»dç»di˘]

2;3.26(S30)

Noddy 'name' /»nçdi/ [»nçdi˘] 2;4.25(S32)

300

[»nçdi]Noddy 'name' /»nçdi/ [»nçdi] 2;5.15(S33)Noddy 'name' /»nçdi/ [»nçdi] 2;6.6(S35)

InLuma'sspeech,/SW/wordswithoutwordmarkerareselectedlateindevelopment

(like in Joana and João). At the beginning, when the child attempts those words, variable

productions are observed (e.g.,Noddy 'name' produced as [»n廯i], [»nåne] or [tç»tçR¥i], in

session 22). The wordHeidi /Ȍjdi/ 'name is produced accordingly or with level stress in

session 28 (e.g., [»adi] and [»a»di]). The word Noddy /»nçdi/ 'name' is subject to a great

variationuntil session32.Until thissession,Lumaproduces itasa trisyllable,with finalor

penultimate stress (e.g., [tçR¥çR»¥i] or [toR»¥oRi]). In session 32, the child consistently

produced/SW/wordswithoutwordmarkertarget‐like.

Insum,theobservationoftargettrocheeswiththewordmarkersuggeststhat,infact,

theacquisitionoftrocheesmightberelatedtotheacquisitionofthewordmarker.However,

theresultsarenotclear‐cut:

(i) Clara has an initial accurate production of both /SW/ non‐verbs and theword

marker;

(ii) InêsandJoanahaveasimultaneousacquisitionpathforbothstructures;

(iii) JoãoandLumaacquirethewordmarkerbefore/SW/non‐verbs.

Thesefindingssuggestthattheobservationoftheacquisitionofthegendercontrast

and the acquisition of trochees is related. From the tables and instances presented, we

observedthatthewordmarkercouldbeinitiallyproduced,though,insomechildren(mainly

Inês, Joana, João and Luma), their production was not necessarily within a trochaic word

([SW]).

Despite the relationship that canbeestablishedbetween theacquisitionof trochaic

targetwords and thewordmarker, however, any causality between the acquisition of one

and the other should not be presumed, as /SW/ non‐verbs without the wordmarker (i.e.

wordsthatarethemselvesstems)displayedthesameproductionstrategiesandacquisition

paththan/SW/non‐verbswithwordmarker(i.e.,wordsthatarecomposedbyastem+word

marker).The results from the analysis on/SW/non‐verbswithoutwordmarker indicated

thattherearenodifferencesinthewordshapesandstresspatternsproducedbythechildren

forthetwostructures.Inboth,truncationtomonosyllables,reduplication(as[WS]),target‐

likeproductions([SW])andstressshiftarepossible.Additionally,/SW/non‐verbswithout

301

word marker were not systematically produced as iambs in the beginning, especially as

[CV1'CV2](e.g.,Noddy'name'/»nçdi/‐>[nç»di]).

Insum, theseresultsareconsistentwithan initialstagewherenoclear tendency is

observed and a later stage where a trochaic foot is under children's attention. Despite an

apparentacquisitionoftrocheesinteractingofmorphology(i.e.,trocheesareacquiredwhen

thewordmarker is acquired), the earlierdeviantproductionof trocheeswith andwithout

wordmarkersuggests that the lateacquisitionof trochees is irrespectiveof themasteryof

morphology.

6.1.1.3.Lookingcloserattheverbparadigm

Inthissection,wewillmainlyfocusonthedevelopmentalpathundertakenbythefive

childrenobservedtowardsstresspatternsindifferentverbtenses.

Ananalysisontheverbtensesisnecessary,sincetheverb'ssysteminPortuguesehas

beenanalyzeddifferentlyfromthenon‐verb'ssystem.Duetothereducednumberoftokens

ofverb formsand,especially,due to thereducednumberofverb formsperverb tense,we

will present the emergence of verbs tenses only. We will consider that a verb tense has

emergedwhenitisproducedtarget‐likeatleast2timesinonesession,withcontinuityonthe

followingsessions.Sincethenumberoftokensperchildissometimestooscarce,theresults

must,therefore,beinterpretedcautiously.

According to the properties of the target language (Lee, 1995; Bisol, 1992, 1993;

Wetzels, 2006), we expect that verbs forms that conform to a trochaic foot and are

constituted by a stem + a theme vowel are earlier acquired. First, because they have the

unmarked stresspattern forverb forms in the target languageand, secondly,because they

arelesscomplexfromamorphologicalconstituencypointofview(especiallywhencompared

to verb forms that are composed by the stem+theme vowel+tense/mood suffix

person/number suffix ‐ e.g., fal]á]va]mos 'we spoke, past imp.' /få»lavåmuS/). The earlier

verbformshould,therefore,be3rdp.sg.ofthePresent,whichcoincideswiththeverbformof

theImperative226.

Despitenon‐finite forms(Infinitive)wereearlierobserved inpreviousworksonBP

(Kato,1998;Santos,2007)wedonotexpecttofindthesestructures,astheyconformtothe

markedstresspatternassumedforverbs,i.e.,theyconformtoaniambicfoot.

In the following tables (83‐87), wewill show the emergence of verb tenses in the

observed children.Wewill take into account the verb form and its stress pattern.Wewill

226Cf.AppendixA.

302

consideras'correctproduction',aproductionwherethetensemarkerisrecognizableandthe

stresspatterniskept.Reduplicationswillbeconsideredasincorrectproductions.

First,wewillshowtheratesofproductionaccordingtothetarget,asfarasthetarget

stresspatternsare concernedand, later in this section,wewill further show the strategies

usedbythechildrenwhentheydonotproducetheverbformstarget‐like.

InTable83,wepresentClara'sresultsfortheemergenceofverbstenses:

Session PresentIndicative(3rdp.sg.)/Imperative­/SW/

Types SimplePast(3rdp.sg)­/WS/ #Types

S1 ‐ ‐ ‐ ‐S2 ‐ ‐ ‐ ‐S3 0%(0/1) 1 ‐ ‐S4 ‐ ‐ ‐ ‐S5 ‐ ‐ ‐ ‐S6 ‐ ‐ ‐ ‐S7 ‐ ‐ ‐ ‐S8 ‐ ‐ ‐ ‐S9 ‐ ‐ ‐ ‐S10 ‐ ‐ ‐ ‐S11 90%(9/10) 3 100%(2/2) 2S12 100%(21/21) 2 80%(4/5) 1

Table83.Emergenceofverbtenses(Clara)

Clara's speech was only analyzed until 1;10 (session 12). For that reason, the

description of results for Clara's speech regarding the emergence of verb tenses is

inconclusive.However,itispossibletoobservethatPresent(/‐SW/),Imperative(/‐SW/)and

SimplePastforms(/‐WS/)seemtoemergesimultaneously(insession11),assummarizedin

theschemebelow.

(204) OrderofemergenceofverbtensesinClara'sspeech:

Pres.Ind.(3rdp.sg.)/Imp.~Simp.Past(3rdp.sg)

The instances in (205) illustrate thedevelopmental path for verb tenses and stress

patterns undertaken by Clara (3rdp.sg. of the Present/Imperative and the 3rdp.sg. of the

SimplePast).

SW WS

303

(205) Clara–productionofverbstenses(perorderofemergence):

Orthogr. Gloss Target Output Ageabre 'open' /»abRˆ/ [»a»bˆ˘] 1;1.3(S3)olha 'look!' /»ç¥å/ [»ç˘la] gosta 's/helikes' /»gçStå/ [»gç˘tå]abre 's/heopens' /»abRˆ/ [»aBi]/[a˘»bi]

1;9.23(S11)

a. Present(3rdp.sg.)/Imp. ‐/SW/

abre 's/heopens' /»abRˆ/ [»a˘bi] 1;10.15(S12)acabou 'itfinished' /åkå»bo/ [kå»bo˘]mordeu 's/hebit' /muR»dew/ [mu»¯ew] 1;9.23(S11)

b. Simple Past(3rdp.sg.)‐/WS/

acabou 'itfinished' /åkå»bo/ [kå»bo] 1;10.15(S12)

InClara'sverbsproduction,weobservethat,initially(cf.session3and11),thechild

mainlyselectsthe3rdp.sg.ofthePresent(/‐SW/).Regardingthestresspattern,thechildhas

anunstableproductionofthetargetwordandalternatesbetweenatarget‐likeproduction(in

session11and12), levelstress(insession3)andiambicproductions(insession11).From

session11onwards,the3rdp.sg.oftheSimplePast(/‐WS/)emergesandsheproduces[WS]

verb forms.Though trisyllabic targetsmaynot beproduced as such, thedisyllabic outputs

carry the tense marker (e.g., /åkå»bo/ produced as [kå»bo]), providing evidence for the

acquisitionoftenseinflection.

InTable84,wepresenttheorderofemergenceofverbtensesinInês'speech.

304

Session

PresentIndicative(3rdp.sg.)/Imperative­

/SW/

TypesInfinitive­/WS/

Types

SimplePast

(3rdp.sg)­/WS/

TypesPastImperfect(3rdp.sg.)­/SW/

Types

S1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐S2 0%(0/2) 1 ‐ ‐ ‐ ‐ ‐ ‐S3 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐S4 0%(0/1) 1 ‐ ‐ ‐ ‐ ‐ ‐S5 0%(0/1) 1 ‐ ‐ ‐ ‐ ‐ ‐S6 0%(0/3) 1 0%(0/1) 1 ‐ ‐ ‐ ‐S7 ‐ ‐ 0%(0/1) 1 ‐ ‐ ‐ ‐S8 0%(0/2) 2 0%(0/1) 1 ‐ ‐ ‐ ‐S9 71.43%(5/7) 4 100%

(2/2)2 66.67%

(4/6)4 ‐ ‐

S10 66.67%(18/27)

6 60%(3/5) 5 ‐ ‐ ‐ ‐

S11 87.7%(30/35)

13 41.67%(5/12)

6 100%(3/3)

2 ‐ ‐

S12 79.2%(19/24)

9 36.36%(8/22)

12 90%(9/10)

5 100%(2/2) 2

S13 82.45%(47/57)

21 59.25%(16/27)

15 92%(23/25)

10 100%(6/6) 3

S14 63.63%(28/44)

12 50%(9/18)

10 93.75%(15/16)

7 100%(4/4) 2

S15 78.13%(25/32)

11 57.14%(16/28)

17 100%(8/8)

6 100%(5/5) 2

S16 69.04%(29/42)

12 67.64%(23/34)

11 83.33%(5/6)

5 88.89%(8/9) 3

S17 65.38%(17/26)

10 27.58%(8/29)

8 100%(8/8)

4 100%(1/1) 1

S18 78.26%(18/23)

15 45.83%(11/24)

17 50%(1/2)

2 86.67%(13/15)

5

Table84.Emergenceofverbtenses(Inês)

In Inês's speech,weobserve that, though the formof the3rdp.sg. of thePresent (/‐

SW/)andtheformoftheImperative(/‐SW/)areselectedbeforetheformoftheInfinitive(/‐

WS/).Fromsession2until session8, Inêsattempts the formsof thePresent tense,butshe

doesnotproducethemadult‐like227.Thesameoccursfromsession6tosession8,inInfinitive

forms. The emergence of target‐like productions for Present and Infinitive forms is

simultaneousinInês'speech,andoccursinsession9.The3rdp.sg.oftheSimplePast(/‐WS/)

emergesinsession11,thoughitisselectedbefore,insession9.Theformsofthe3rdp.sg.of

the Past Imperfect (/‐SW/) are the latest to emerge (in session 12), until the end of the

observationalperiodofInês'speech.

Theschemein(206)summarizesInês'emergencepathforverbstenses.

227Furtherinthissectionwewillanalyzethechildren'sstrategiesinverbswiththedifferentstresspatterns.

305

(206) OrderofemergenceofverbtensesinInês'speech:

Pres.Ind.(3rdp.sg.)/Imp.~Infinitive>>Simp.Past(3rdp.sg.)>>Past.Imperf.(3rdp.sg.)

Theinstancesin(207)showInês'renditions,mirroringtheorderofemergenceofthe

different verb tenses and stress patterns (3rdp.sg. of thePresent and Imperative, Infinitive,

3rdp.sg.oftheSimplePastand3rdp.sg.ofthePastImperfect).

(207) Inês–productionofverbstenses(perorderofemergence):

Orthogr. Gloss Target Output Agetoma 'takeimp.' /»tçmå/ [»tç]/[»a»dJç] 1;0.25(S2)tapa 'coverimp.' /»tapå/ [»pa»pa] 1;3.6(S4)mostra 'showimp.' /»mçStRå/ [»mç] 1;4.9(S5)ajuda 'helpimp.' /å»Zudå/ [»dJudJå]anda 'comeon' /»å‚då/ [»åtå]olha 'lookimp.' /»ç¥å/ [»çjå]

1;8.2(S9)

tira 'take(it)off' /»tiRå/ [´»tiå] 1;9.19(S10)

a. Present(3rdp.sg.)/Imp.‐/SW/

tira 'take(it)off' /»tiRå/ [»tiRå] 1;10.29(S11)cortar 'tocut' /kuR»taR/ [kå»ka] 1;5.11(S6)limpar 'toclean' /li‚»paR/ [pa»pa] 1;6.11(S7)tirar 'totakeoff' /ti»RaR/ [»tJQ] 1;7.2(S8)cantar 'tosing' /kå‚»taR/ [kå»Ba]

b. Infinitive ‐/WS/

tomar(banho)

'to have (ashower)'

/tu»maR/ [p´»ta] 1;9.19(S10)

caiu 's/hefelloff' /kå»iw/ [»kiw]/[tå»kJiw]vestiu 's/hedressed' /vˆS»tiw/ [å˘bˆ»tiw]fugiu 's/he ran

away'/fu»Ziw/ [i»di]

1;8.2(S9)

c. Simple Past(3rdp.sg.) ‐/WS/

partiu 's/hebroke' /påR»tiw/ [t´»iw] 1;10.29(S11)estava 's/hewas' /S»tavå/ [»tabå] 2;4.18(S16)d. Past

Imperfect(3rdp.sg.) ‐/SW/

tinha 's/hehad' /»ti¯å/ [»tiå] 2;7.16(S18)

The instancespresentedabove illustrate theearlierelectionandemergenceofverb

formsofthe3rdp.sg.ofthePresentandImperative(/‐SW/).Inthebeginning(session2,4and

5), however, these forms are not produced target‐like (they are mostly truncated to a

monosyllable or they are reduplicated and produced with level stress). From session 9

onwards,theytendtobeproducedtarget‐like.

SW WS WS SW

306

The second verb form to be selected and to emerge is the Infinitive (/‐WS/), in

session 6. At the beginning, these verb forms are producedwith the target stress pattern

([WS]), but they mostly consist in reduplicated productions (cf. sessions 6 and 7). After

session 10, they tend to be produced with the target stress pattern and non‐reduplicated

syllables.

Theother/‐WS/verb form(the3rdp.sg.of theSimplePast) isselected insession9.

When they are attempted, at the beginning, these verb forms might be truncated to

monosyllables(cf.session9),butthetargetstresspatternismostlyproducedtarget‐like.

Inês is the only child productively selecting and producing the 3rdp.sg. of the Past

Imperfect (/‐SW/). When the child selects these forms, she mostly produces them

accordingly.

Inthefollowingtable(Table85),wewillshowJoana'semergenceofverbstenses.

Session

PresentIndicative(3rdp.sg.)/Imperative.­

/SW/

TypesSimplePast(3rdp.sg)­/WS/

Types Infinitive­/WS/

Types

S1 ‐ ‐ ‐ ‐ ‐ ‐S2 ‐ ‐ ‐ ‐ ‐ ‐S3 ‐ ‐ ‐ ‐ ‐ ‐S4 ‐ ‐ ‐ ‐ ‐ ‐S5 ‐ ‐ ‐ ‐ ‐ ‐S6 ‐ ‐ ‐ ‐ ‐ ‐S7 ‐ ‐ ‐ ‐ ‐ ‐S8 ‐ ‐ ‐ ‐ ‐ ‐S9 ‐ ‐ ‐ ‐ ‐ ‐S10 ‐ ‐ ‐ ‐ ‐ ‐S11 ‐ ‐ 100%(1/1) 1 ‐ ‐S12 100%(4/4) 1 0%(0/1) 1 100%(1/1) 1S13 100%(4/4) 3 50%(2/4) 2 0%(0/3) 2S14 67.64%(23/34) 10 75%(3/4) 2 57.14%(4/7) 7

Table85.Emergenceofverbtenses(Joana)

InJoana'sspeech,weobservethatthe3rdp.sg.ofthePresenttense(/‐SW/)isthefirst

verbformtoemerge(insession12)andthefirstproductionsarefaithfullyproduced,though

inthelastsessionadecreasinginthetarget‐likeproductionrateisobserved(67.64%).The

forms of the 3rdp.sg. of the Simple Past (/‐WS/) emerge in session 13, but the rate of

production and the absolute values are reduced, until the end of the observational period

(there isamaximumof4 tokensand2 typespersession).The lastverb formtoemerge in

Joana's speech is the Infinitive (/‐WS/). The Infinitive is selected in session 12, but in the

307

following session no Infinitive targets are attempted. Infinitive forms are not produced

target‐likeuntilsession14.

Theschemein(208)summarizestheemergenceofverbtensesinJoana'sspeech.

(208) OrderofemergenceofverbtensesinJoana'sspeech:

Pres.Ind.(3rdp.sg.)/Imp.>>Simp.Past(3rdp.sg)>>Infinitive

In (209), we present Joana's renditions for verb tenses with reference to stress

patterns, in the order of emergence of each paradigm (3rdp.sg. of the Present/Imperative,

3rdp.sg.oftheSimplePastandInfinitive).

(209) Joana–productionofverbstenses(perorderofemergence):

Orthogr. Gloss Target Output Ageolha 'look!' /»ç¥å/ [»çjå] 2;2.19(S12)pode 's/hecan' /»pçdˆ/ [»pçdˆ]

a. Present(3rdp.sg.)/Imp. ‐/SW/ anda 'comeon!' /»å‚då/ [»å‚då] 2;4.1(S13)

b. Simple Past(3rdp.sg.)‐/WS/

fugiu 's/he ranaway'

/fu»Ziw/ [fu»Ziw] 2;4.1(S13)

limpar 'toclean' /li‚»paR/ [e‚»paj] filmar 'tofilm' /fi…»maR/ [fu‚»aj]

c.Infinitive‐/WS/

brincar 'toplay' /bRi‚»kaR/ [pi‚»kaj]2;6.24(S14)

In Joana's speech, despite the earlier selection of /‐WS/ verb forms (3rdp.sg. of the

SimplePast),the/‐SW/verbsformsofthe3rdp.sg.PresentandImperativehaveaconsistent

productionaccordingtothetargetthantheformer.WhenthechildselectsthePresentforms

(/‐SW/),sheproducesthemwiththecorrectstresspattern(insession12).TheSimplePast

andtheInfinitiveforms(/‐WS/)mightbeselectedfirst(insession11and12,respectively),

but their emergence and target‐like production only occurs later (in session 13 and 14,

respectively).ItisworthmentioningthatJoanareplacestheInfinitivemarker(/‐R/)withthe

glide[j]228.

Table86showstheresultsfortheproductionandemergenceofverbstensesinJoão's

speech.

228Thisstrategy isacommonstrategyusedbythechildren intheproductionof targetword‐final liquids.ForadescriptiononthesegmentalprocessesimpliedintheacquisitionofEP,cf.Freitas(1997)andCosta(inprep.).

SW WS WS

308

Session

PresentIndicative(3p.sg.)/

Imperative­/SW/

TypesSimplePast(3p.sg.)­/WS/

Types Infinitive­/WS/

Types

S1 ‐ ‐ ‐ ‐ ‐ ‐S2 ‐ ‐ ‐ ‐ ‐ ‐S3 ‐ ‐ ‐ ‐ ‐ ‐S4 ‐ ‐ ‐ ‐ ‐ ‐S5 ‐ ‐ ‐ ‐ ‐ ‐S6 ‐ ‐ ‐ ‐ ‐ ‐S7 ‐ ‐ ‐ ‐ ‐ ‐S8 ‐ ‐ ‐ ‐ ‐ ‐S9 ‐ ‐ 0%(0/2) 1 ‐ ‐S10 ‐ ‐ 0%(0/1) 1 ‐ ‐S11 ‐ ‐ ‐ ‐ ‐ ‐S12 ‐ ‐ 100%(1/1) 1 ‐ ‐S13 ‐ ‐ ‐ ‐ ‐ ‐S14 ‐ ‐ 100%(1/1) 1 ‐ ‐S15 0%(0/1) 1 ‐ ‐ ‐ ‐S16 44.44%(4/9) 3 0%(0/6) 1 ‐ ‐S17 100%(1/1) 1 100%(1/1) 1 100%(2/2) 2S18 95.45%(21/22) 5 ‐ ‐ 100%(3/3) 1S19 77.78%(7/9) 2 83.33%(5/6) 5 ‐ ‐S20 87.5%(7/8) 3 ‐ ‐ 100%(1/1) 1S21 100%(3/3) 3 ‐ ‐ ‐ ‐S22 87.5%(7/8) 2 0%(0/1) 1 0%(0/1) 1

Table86.Target­likeproductionofverbtenses(João)

The emergence of verb tenses in João's speech is slightly different from the one

observedintheotherchildren,asthechildselectsandproducesthePastform(/‐WS/)rather

earlier thanPresentandImperative forms(/‐SW/). Joãoattempts the3rdp.sg.of theSimple

Past(/‐WS/) insession9, thoughhedoesnotproducethemtarget‐like. Insession12he is

able to produce Past verb forms target‐like, though inconsistently across the following

sessions.Untiltheendoftheobservationperiodthechildhasveryfewtokens(andtypes)in

thePasttense,thusbeinginconclusivetodrawarobustconclusionontheemergenceofthese

structures.ThesameappliestoInfinitiveforms(/‐WS/).ThechildattemptsInfinitiveforms

insession17butthereisaninconsistentproductionofthesestructuresacrosssessions,until

theendoftheobservationperiod.InJoão,onlythe3rdp.sg.ofthePresenttenseemergesand

is acquired. João selects Present and Imperative forms (/‐SW/) in session 15 but only in

session 16 he is able to produce them target‐like.When he starts producing forms of the

PresentandImperativetenseandmood,hedoesitwithhighpercentages(cf.ratevaluesin

sessions17‐22).

309

In(210)wepresenttheschemethatsummarizestheemergenceoftheverbtensesin

João'sspeech.

(210) OrderofemergenceofverbtensesinJoão'sspeech:

Pres.Ind.(3rdp.sg.)/Imp.>>Infinitive~Simp.Past(3rdp.sg.)

Therenditionsin(211)illustratetheemergenceofverbstensesinJoão(3rdp.sg.ofthe

Present, Imperative, Infinitive and3rdp.sg. of the Simple Past),with reference to the target

stresspatterns.

(211) João–productionofverbstenses(perorderofemergence):

Orthogr. Gloss Target Output Ageatende 'pickup' /å»te‚dˆ/ [»tå‚dˆ] 1;9.25(S16)dança 'dance' /»då‚så/ [»katå] olha 'look' /»ç¥å/ [»a¥å]ajuda 'help' /å»Zudå/ [»dudå]

1;10.11(S17)

a. Present (3rdp.sg.)/Imp.‐/SW/

toma 'takeit' /»tçmå/ [»pçmå] 1.10.26(S18)trabalhar 'towork' /tRåb廥aR/ [bibi»ja] 1;10.11(S17)limpar 'toclean' /li‚»paR/ [tˆ»ta˘]papar 'to eat

(fam.)'/på»paR/ [på»pa]

b.Infinitive‐/WS/

trabalhar 'towork' /tRåb廥aR/ [båbå»ja]

1;11.19(S19)

saltou 's/hejumped'

/sa…»to/ [tå»to]

pulou 's/hejumped'

/pu»lo/ [pu»jo]

c.SimplePast(3rdp.sg.)‐/WS/

parou 's/hestopped'

/på»Ro/ [på»jo]

1;11.10(S19)

The instances presented in (211) show that, when the forms of the Present and

Imperative (/‐SW/) are selected (mostly, from session 16 onwards), they are mostly

produced with the correct stress pattern ([SW], though they might be truncated if they

correspond to a /WSW/ trisyllable (e.g., atende 'pick up' /å»te‚dˆ/ produced as [»tå‚dˆ], and

ajuda'help'/å»Zudå/producedas[»dudå]).

Insession17, the Infinitive(/‐WS/)verb formsareselectedandproducedwith the

target stresspattern, thoughnomoodmarker is attested.The formsof theSimplePast (/‐

SW WS WS

310

WS/)arealsoproducedwith the targetstresspatternbut, in thiscase, the tensemarker is

noticeable(e.g.,pulou/pu»lo/producedas[pu»jo],insession19).

InTable87,weshowthepercentageoftarget‐likeproductions inthedifferentverb

tensesinthespeechofLuma.

Session

PresentIndicative(3p.sg.)/

Imperative­/SW/

Types SimplePast(3p.sg.)­/WS/

Types Infinitive­/WS/

Types

S1‐13 ‐ ‐ ‐ ‐ ‐ ‐S14 ‐ ‐ ‐ ‐ 0%(0/6) 1S15 ‐ ‐ ‐ ‐ 0%(0/3) 1S16 ‐ ‐ ‐ ‐ 0%(0/36) 1

S17‐S19 ‐ ‐ ‐ ‐ ‐ ‐S20 0%(0/12) 1 ‐ ‐ ‐ ‐S21 0%(0/7) 1 ‐ ‐ ‐ ‐S22 0%(0/14) 1 ‐ ‐ ‐ ‐S23 0%(0/13) 1 ‐ ‐ ‐ ‐S24 0%(0/3) 1 ‐ ‐ ‐ ‐S25 ‐ ‐ ‐ ‐ ‐ ‐S26 0%(0/2) 1 ‐ ‐ ‐ ‐S27 0%(0/1) 1 ‐ ‐ ‐ ‐S28 ‐ ‐ ‐ ‐ ‐ ‐S29 ‐ ‐ 100%(1/1) 1 ‐ ‐S30 37.5%(3/8) 3 ‐ ‐ ‐ ‐S31 45.45%(5/11) 7 33.3%(1/3) 2 ‐ ‐S32 76.04%(73/96) 29 ‐ ‐ ‐ ‐S33 87.80%(36/41) 20 100%(4/4) 3 ‐ ‐S34 98%(49/50) 17 95.83%(23/24) 9 100%(3/3) 3S35 55.76%(29/52) 19 84.62%(11/13) 5 100%(4/4) 3S36 83.33%(45/54) 17 75%(6/8) 6 30%(6/20) 17S37 69.23%(18/26) 17 66.67%(4/6) 3 55.56%(10/18) 12

Table87.Target­likeproductionofverbtenses(Luma)

InthetableaboveweobservethatLumaselects Infinitive forms(/‐WS/)earlier(in

sessions14,15and16)thanPresent(/‐SW/)orPastforms(/‐WS/).However,inthatperiod,

they are not with the target stress pattern. Verb forms in the Imperative are selected in

session 20 but they are not producedwith the target stress pattern until session 30. Both

InfinitiveandImperativeearlyverbsformsaretheproductoftheproductionofonesimple

tokenacross thesession:marchar/»maRSaR/ 'tomarch'andolha/»ç¥å/ 'look', respectively.

Theformsofthe3rdp.sg.oftheSimplePastareselectedinsession29.Whenverbformsare

produced consistently we observe that the forms of the Present (/‐SW/) are produced

faithfullyearlier(insession30)thantheremainderverbforms.VerbformsoftheSimplePast

311

(/‐WS/)areproducedconsistentlyandfaithfully insession33andInfinitiveforms(/‐WS/)

areproducedcorrectlyinsession34.

Theschemein(212)accountsfortheemergenceofverbformsinLuma'sspeech.

(212) OrderofemergenceofverbtensesinLuma'sspeech:

Pres.Ind(3rdp.sg.)>>Simp.Past(3rdp.sg.)>>Infinitive

In (213)wepresent Luma's renditions for thedifferent verb forms, in the order of

emergenceaccountedforinTable87.

(213) Luma–productionofverbstenses(perorderofemergence):

Orthogr. Gloss Target Output Ageanda 'comeon' /»å‚då/ [»å‚då]toma 'take(it)' /»tçmå/ [»pçmå˘]toma 'take(it)' /»tçmå/ [»pçmå]

2;4.11(S30)

passa 's/hepasses' /»paså/ [»paCå]

a.Present(3rdp.sg.)/Imp.‐/SW/

toma 'take(it)' /»tçmå/ [»pçmå]2;4.25(S31)

saiu 's/heleft' /så»iw/ [Så»iw] 2;4.11(S30)caiu 's/hefelloff' /kå»iw/ [kå»iw]chocou 's/hehit' /Su»ko/ [Cu»ko˘]

c. SimplePast (3rdp.sg.)‐/WS/

fugiu 's/he ranaway'

/fu»Ziw/ [fufu»Ziw]

2;5.15(S33)

morder 'tobite' /muR»deR/ [mu»de] 2;5.20(S34)chover 'torain' /Su»veR/ [fu»ve] andar 'towalk' /å‚»daR/ [a‚»da]

2;6.6(S35)

brincar 'toplay' /bRi‚»kaR/ [bi‚»kåj] 2;6.20(S36)

b.Infinitive‐/WS/

chover 'torain' /Su»veR/ [Cu»ve] 2;6.27(S37)

InLuma,weobservethat,formsession30onwards,thePresentandImperativeverb

forms(/‐SW/)areselectedandthestresspatternisproducedtarget‐like.

Insession33,theSimplePastverbforms(/‐WS/)areproducedwiththetargetstress

pattern and the inflection marker is noticeable. Likewise, from session 34 onwards, the

Infinitiveformsareproducedwiththetargetstresspatternconsistently,althoughthemood

marker /‐R/ is still not produced (it can be replaced by the glide /j/ or not have any

realizationatall).

SW WS WS

312

Insum,theemergenceofinflecteddi‐orpolisyllabicverbformsoccurslateinthefive

observed children. Verb forms of the 3rdp.sg. of the Present tense and the Imperative verb

forms(/‐SW/)arethefirsttoemergeinallchildren.However,somevariationisattestedin

theemergenceoftheSimplePast(/‐WS/)andInfinitive(/‐WS/)verbforms.Inês,Joanaand

Lumahaveanearlieremergenceofthe3rdp.sg.oftheSimplePast,whereasJoãohasanearlier

emergenceoftheInfinitive.Note,however,thatthenumberofdi‐andpolisyllabicverbforms

isreduced(muchmorereducedthannon‐verbs)inallchildren.

Recall that the production of the 2ndp.sg. or the 2ndp.pl. was very infrequent in all

childrenanddoesnotallowfortheestablishmentofanacquisitionpath229.The2ndp.sg.has

mainly a /‐SW/ stress pattern (e.g.,amas 'you love' /»åmåS/, comeste 'you ate' /ku»meStˆ/,

dormias 'you slept, Past Imperf.' /duR»miåS/), but are have amore complexmorphological

structureasafinalpersonandnumbersuffixisaddedtotheverbtheme(‐/s/).The2ndp.pl.

mighthavea/‐SW/ora/‐SWW/stresspattern,dependingontheverbtense(e.g.,fazemos

'we do' /få»zemuS/ but fizéssemos 'we did, Subj.' /fi»zEsˆmuS/230). As described by the

analysesonstressintheverbparadigmandsupportedbythedatanowfound,theformsof

the2ndp.pl.witha/‐SWW/stresspatternareconsideredasmarked(Andrade,1988/1992;

Bisol,1992;1993).

Thus far we have shown the emergence of di‐ and polisyllabic verb forms in the

speechofthefivechildrenobservedandrelateditwiththeemergenceofstresspatterns.In

the following paragraphs, we will demonstrate the strategies that children use in the

productionofverbformswith/‐SW/and/‐WS/stresspatterns.

DatafromClarawillnotbepresented,astheyarenotconclusivewithrespecttoverb

inflection,whichshefirstattemptsinSession3(cf.Table83).ThoughClaraisabletoselect

verb forms in session 3, she only produces verbs consistently in the two last observation

sessions.SheonlyhaspolisyllabicverbformsfromSession11onwards.However,whenshe

startsproducingdi‐andtrisyllabicverbs,sheproducesthemaccordingly.

Thetablesbelow(88and89)arerelativetoInês'productionsoftrochaicandiambic

targetverbs,respectively.Giventhereducednumberoftokensatthebeginning,wedecided

topresenttheabsolutenumberoftokensperwordshape(targettrocheesandtargetiambs)

andhighlighttheimportantresultswithagreyverticalcolumn231.

229Cf.AppendixD.230Cf.AppendixA.231Thismethodwillbeusedinthetablesconcerningtheotherchildrenaswell.

313

[SW] [WS] [S] [SS] [WSW] OtherTOTAL/­SW/

S2 1(filler) 1 2S3 ‐S4 1 1S5 1 1S6 1(filler) 1 2S7 ‐ ‐ ‐ ‐ ‐ ‐ ‐S8 3 1 4S9 7 1 8S10 22 1 2 1 3 29S11 33 2 2 1 2 3 43S12 22 2 2 3 29S13 70 1 2 2 75S14 60 60S15 52 1 53S16 77 3 80S17 48 3 51S18 76 6 82

Table88./­SW/verbsproduction(Inês)

[SW] [WS] [S] [SS] [WSW] OtherTOTAL/­WS/

S6 1(redupl.) 1S7 1(redupl.) 1S8 1 1S9 4 2 2 8S10 4(ep.)232 3 7S11 5(ep.) 7 2 1 15S12 8(ep.) 18 4 2 32S13 13(ep.) 37 1 2 53S14 11(ep.) 22 1 34S15 11(ep.) 28 2 41S16 4(ep.) 33 6 5 49S17 11(ep.) 26 5 42S18 10(ep.) 17 4 5 36

Table89./­WS/verbsproduction(Inês)

InTables88and89, it isnoticeable that/‐SW/appearsearlier than/‐WS/ in Inês'

speech. Both seem to be earlier interpreted as /S/ (until session 8) and later they are

produced target‐like (from session 9 onwards). In /‐WS/ verbs, epenthesis can occur,

creatinganapparenttrochaic([‐SW])pattern,aftersession10.NoticethattheInfinitiveverb

forms(whichare/‐WS/)cruciallyhaveaword‐final/‐R/ inPortuguese (e.g.,cortar 'tocut'

[kuR»taR],comer 'toeat' [ku»meR],dormir 'to sleep' [duR»miR]).Aswewill see further in this

chapter, Inês is the only child systematically using an epenthetic vowel in word‐final CLiq.

232 Epenthesis in Inês is prosodically conditioned (always after a word‐final liquid consonant, at the end ofPhonologicalPhraseorUtterance),anditisnotexactlystressshift,asstressismaintainedinthesamesyllablebutanadditionalsyllable,attherightedgeoftheword,isproduced.

314

These results are consistentwithFreitas (1997), according towhichword‐final epenthesis

was possible in the acquisition path of word‐final liquids in EP. Word‐final sonorant

consonantsinPortuguesemaysurfaceastheonsetofanempty‐headedsyllable(/.‐VCSon#/‐>

[V.CSonV]). It is worth notice, however, that the output of the application of this strategy

creates a different stress pattern. Since word‐final liquids tend to be present in stressed

syllables,iambictargetssuchas/CV.'CVCLiq/arerealizedastrochaicwords(/CV.'CV.CLiqV/).

In (214) and (215), we will present Inês' productions of /‐SW/ and /‐WS/ verbs,

respectively.

(214) Inês–productionof/‐SW/verbs:

Orthogr. Gloss Target Output Agetoma 'take(it)' /»tçmå/ [å»dJç] 1;0.25(S2)tapa 'cover(it)' /»tapå/ [»pa»pa] 1;3.6(S4)mostra 'show(me)' /»mçStRå/ [»mç] 1;4.9(S5)corta 'cut(it)' /»kçRtå/ [´»ko] 1;5.11(S6)acho 'Ithink' /»aSu/ [»atJi] 1;8.2(S9)ouviste 'youheard' /o»viStˆ/ [»titiS]queres 'youwant' /»kERˆS/ [»EdisJ] 1;9.19(S10)

(215) Inês–productionsof/‐WS/verbs:

Orthogr. Gloss Target Output Agecortar 'tocut' /kuR»taR/ [ka»ka] 1;5.11(S6)limpar 'toclean' /li‚»paR/ [pa»pa] 1;6.11(S7)vestiu 's/hedressed' /vˆS»tiw/ [å˘Bˆ»tiw]caiu 's/hefelloff' /kå»iw [tå»kiw]fugiu 's/heranaway' /fu»Ziw/ [i»di]limpar 'toclean' /li‚»paR/ [ge»pa]

1;8.2(S9)

cair 'tofalloff' /kå»iR/ [»kiRˆ] cantar 'tosing' /kå‚»taR/ [kå»Ba]guardar 'tokeep' /gWaR»daR/ [ƒå»daRi]tomar(banho) 'tohave(ashower)' /tu»maR/ [p´»ta]beber 'todrink' /bˆ»beR/ [»bedi˘]

1;9.19(S10)

EarlyverbsinInêsarereduplicatedortruncatedandprecededbyfillersounds(until

session 9), either they have a /‐SW/ or a /‐WS/ stress pattern. Later (from session 9

onwards), the child produces target‐like verbs with occasional recursion to epenthesis in

target/‐WS/(e.g.,guardar'tokeep'/gWaR»daR/producedas[ƒå»daRi]).

315

InTables90and91,wewillshowJoana'sacquisitionstrategiesin/‐SW/and/‐WS/

verbs.

[SW] [WS] [S] [SS] [WSW] OtherTOTAL/­SW/

S9 1 1S10 ‐ ‐ ‐ ‐ ‐ ‐ ‐S11 1 1S12 4 4S13 6 1 7S14 34 7 2 43

Table90./­SW/verbsproduction(Joana)

[SW] [WS] [S] [SS] [WSW] OtherTOTAL/­WS/

S10 1 1S11 1 1S12 1 1 2S13 2 5 1 8S14 3 7 10

Table91./­WS/verbsproduction(Joana)

AsshowninTables90and91,verbsareveryscarce in Joana'searlyspeech./‐SW/

and /‐WS/ verbs only appear in session 9 and 10, respectively. Like Inês, Joana truncates

early /‐SW/ to monosyllables (until session 11) and realizes them accordingly, in a later

period(fromsession12onwards).Insession11Joanaproducesone/‐WS/verbformtarget‐

likeand,inthefollowingsession,thechildisabletocorrectlyproducetheseverbformsmore

productively.

The instances in (216) and (217) illustrate Joana's productionof /‐SW/and /‐WS/

verbsforms,respectively.

(216) Joana–productionof/‐SW/verbs:

Orthogr. Gloss Target Output Agegosto 'Ilike' /»gçStu/ [»gç] 1;9.25(S9)gosto 'Ilike' /»gçStu/ [»gç] 2;0.9(S11)olha 'look' /»ç¥å/ [»çjå] 2;2.19(S12)pode 's/hecan' /»pçdˆ/ [»pçdˆ]deixas 'youleave' /»dåjSåS/ [»dåjsås˘]anda 'comeon' /»å‚då/ [»å‚då]

2;4.1(S13)

316

(217) Joana–productionof/‐WS/verbs:

Orthogr. Gloss Target Output Ageouvi 'Iheard' /o»vi/ [»Bi˘] 1;10.22(S10)acabou 'itisover' /åka»bo/ [kˆ»bo] 2;0.9(S11)caiu 's/hefelloff' /kå»iw/ [»kiw]lavar 'towash' /lå»vaR/ [gå»va] 2;2.19(S12)

caiu 's/hefelloff' /kå»iw/ [»kiw]fugiu 's/heranaway' /fu»Ziw/ [fu»Ziw]chover 'torain' /Su»veR/ [»vej]

2;4.1(S13)

In these exampleswe see that, at thebeginning (until session11), the childmostly

truncatestheverbformstomonosyllables.Earlyverbsformsconsistin3p.sg.ofthePresent

Tenseor Imperative inflectedwords, though few forms from the SimplePast areobserved

(untilsession12).Aftersession12,JoanaproducesSimplesPast(3p.sg.)andInfinitiveforms

more productively. Contrary to /‐SW/, /‐WS/ verb forms are truncated to [S] until late in

development.

Tables 92 and 93 summarize João's productions for /‐SW/ and /‐WS/ verbs,

respectively.

[SW] [WS] [S] [SS] [WSW] OtherTOTAL/­SW/

S15 1 1S16 4 5 9S17 1 1S18 22 22S19 8 1 9S20 9 9S21 3 3S22 10 10

Table92./­SW/verbsproduction(João)

317

[SW] [WS] [S] [SS] [WSW] OtherTOTAL/­WS/

S9 2 2S10 1 1S11 ‐ ‐ ‐ ‐ ‐ ‐ ‐S12 1 1S13 ‐ ‐ ‐ ‐ ‐ ‐ ‐S14 1 1S15 ‐ ‐ ‐ ‐ ‐ ‐ ‐S16 2(hiatus) 4 6S17 3 3S18 3 3S19 3 1 1 5S20 1 1S21 ‐ ‐ ‐ ‐ ‐ ‐ ‐S22 1(hiatus) 1 2

Table93./­WS/verbsproduction(João)

Target /‐WS/ verbs are earlier selected than /‐SW/ in João's speech, though in a

reducednumber.InJoão'sspeech,trochaicverbswerealmostabsentinearlysessions(until

session 15). When he starts producing /‐SW/ verb forms (from session 15 onwards), he

producesthemaccordingly./‐WS/verbsaretruncatedto[S]orproducedtarget‐likeinearly

production(untilsession16).Aftersession16,theyareconsistentlyproducedtarget‐like.

In (218) and (219), we show João's rendition of /‐SW/ and /‐WS/ verb forms,

respectively.

(218) João–productionof/‐SW/verbs:

Orthogr. Gloss Target Output Ageajuda 'help(me)' /å»Zudå/ [»du»a] 1;8.25(S15)atende 'pickup(thephone)' /å»te‚dˆ/ [»tå‚dˆ]dança 'dance' /»då‚så/ [»katå]olha 'look' /»ç¥å/ [ç»a¥å]

1;9.25(S16)

ajuda 'help' /å»Zudå/ [»dudå] 1;10.11(S17)senta 'sitdown' /»se‚tå/ [»te‚tå]ajuda 'help' /å»Zudå/ [å»dudå]toma 'take(it)' /»tçmå/ [»pçmå]adora 's/headores' /å»dçRå/ [»dçjå]

1;10.26(S18)

318

(219) João–productionof/‐WS/verbs:

Orthogr. Gloss Target Output Ageacabou 'itisover' /åkå»bo/ [»bo˘w] 1;5.12(S9)acabou 'itisover' /åkå»bo/ [»bo] 1;5.26(S10)acabou 'itisover' /åkå»bo/ [bå»wo˘] 1;7.0(S12)acabou 'itisover' /åkå»bo/ [å‚bu»bo] 1;8:24(S14)acabou 'itisover' /åkå»bo/ [»bo]/[»boo] 1;9.25(S16)acabou 'itisover' /åkå»bo/ [å»bo] trabalhar 'towork' /tRåb廥aR/ [bibi»ja]limpar 'toclean' /li‚»paR/ [tˆ»ta˘]papar 'toeat(fam.)' /på»paR/ [på»pa]

1;10.11(S17)

Tables94and95showtheproductionvalues for/‐SW/and/‐WS/,respectively, in

Luma.Thenumbersmarkedwithastar(*)correspondtoproductionsofonesingletype,the

wordolha'look'/»ç¥å/,intarget/SW/,andthewordmarchar'tomarch'/måR»SaR/,intarget

/WS/.

[SW] [WS] [S] [SS] [WSW] OtherTOTAL/­SW/

S20 12 12*S21 7 7*S22 14 14*S23 13 13*S24 3 3*S25 ‐ ‐ ‐ ‐ ‐ ‐ ‐S26 2 2*S27 1 1*S28 ‐ ‐ ‐ ‐ ‐ ‐ ‐S29 ‐ ‐ ‐ ‐ ‐ ‐ ‐S30 5 2 1 8S31 5 6 11S32 87 4 2 3 96S33 37 4 41S34 46 2 1 1 50S35 46 2 1 3 52S36 53 1 54S37 24 1 1 26

Table94./­SW/verbsproduction(Luma)

319

[SW] [WS] [S] [SS] [WSW] OtherTOTAL/­WS/

S14 5 1 6*S15 3 3*S16 35 1 36*

S17‐S28 ‐ ‐ ‐ ‐ ‐ ‐ ‐S29 1 1S30 ‐S31 2 1 3S32 ‐ ‐ ‐ ‐ ‐ ‐ ‐S33 4 4S34 2 25 27S35 15 1 1 17S36 10 17 1 1 29S37 10 13 23

Table95./­WS/verbproduction(Luma)

Luma appears to select both /‐WS/ and /‐SW/ verbs forms at the onset of word

production./‐SW/and/‐WS/aretruncatedto[S]untillateindevelopment(aroundsession

30).Fromsession30onwards,both[SW]and[WS]areproducedfor/‐SW/and/‐WS/.

Instances in (220) and (221) refer to the production of /‐SW and /‐WS/ verbs,

respectively,inLuma'sspeech.

(220) Luma–productionof/‐SW/verbs:

Orthogr. Gloss Target Output Ageolha 'look' /»ç¥å/ [»a] 1;9.29‐2;1.10(S20‐S27)abre 'open(it)' /»abRˆ/ [a»bi]anda 'come' /»å‚då/ [»å‚då]toma 'take(it)' /»tçmå/ [»pçmå˘]

2;3.26(S30)

ajuda 'help(me)' /å»Zudå/ [zJu»da]salta 'jump' /»sa…tå/ [Så»ta]toma 'take(it)' /»tçmå/ [»pçmå]magoa 'ithurts' /må»goå/ [»goå] tira 'takeit' /»tiRå/ [»tiå]passa 's/hepasses' /»paså/ [»paCå]

2;4.11(S31)

toma 'take(it)' /»tçmå/ [»pçmå]manda 'throwit' /»må‚då/ [»må‚då] 2;4.25(S32)

320

(221) Luma–productionof/‐WS/verbs:

Orthogr. Gloss Target Output Agemarchar 'tomarch' /måR»SaR/ [»Sa] 1;6.20‐1;7.19(S14‐S16)sujou 's/hemadedirty' /su»Zo/ [Cu»Zo] 2;2.22(S29)saiu 's/heleft' /så»iw/ [Så»iw] empurrou 's/hepushed' /e‚pu»{o/ [å»p{oå]/[»p{oå] 2;4.11(S31)

caiu 's/hefelloff' /kå»iw/ [kå»iw] 2;5.15(S33)

Both /‐SW/ and /‐WS/ verbs aremostly truncated to [S] at the beginning (before

approximatelysession30).Fromsession20to27,/‐SW/verbsconcernonesingletype,olha

'look',whichisproducedinadoubtfulmanner:[»a]233.Thehighfrequencyof/‐WS/verbsis

onlyapparentaswell,asitisrelativetoonesingletypetoo(marchar'tomarch').Insession

30‐32,/‐SW/and/‐WS/verbscanbeproducedas[SW]or[WS].Fromsession33onwards,

theyareconsistentlyproducedtarget‐like.

Table96summarizes theobservedchildren'sbehavior towards stresspatternsand

verb inflection in the observed children's speech. In this table, '>>' marks a subsequent

emergencepathand'~'meansthesimultaneousemergenceofagivenstructure.

233InTable94,theseinstancesaremarkedwithastar(*).

321

Emergenceofverbtenses/stresspatterns Strategies

ClaraPres.Ind.(3rdp.sg.)/Imp.(/SW/)~PastPerf.

(3rdp.sg)(/WS/)

/‐SW/:notanalyzed

(insufficientdata)

/‐WS/:notanalyzed

(insufficientdata)

Inês

Pres.Ind.(3rdp.sg.)/Imp.(/SW/)~Infinitive

(/WS/)>>PastPerf.(3rdp.sg.)(/WS/)>>

Past.Imper.(3rdp.sg.)(/WS/)

/­SW/:[S]>>[WS]

(redupl.)

/­WS/:[WS](redupl.)>>

[WS]>>[SW]

JoanaPres.Ind.(3rdp.sg.)/Imp.(/SW/)>>Past

Perf.(3rdp.sg)(/WS/)>>Infinitive(/WS/)

/­SW/:[S]>>[SW]

/­WS/:[S]~[WS]

João

Pres.Ind.(3rdp.sg.)/Imp.(/SW/)>>

Infinitive(/WS/)~Simp.Past(3rdp.sg.)

(/WS/)

/­SW/:[SW]

/­WS/:[S]>>[WS]

LumaPres.Ind.(3rdp.sg.)(/SW/)>>PastPerf.

(3rdp.sg.)(/WS/)>>Infinitive(/WS/)

/­SW/:[S]>>[SW]~[WS]

>>[SW]

/­WS/:[S]>>[WS]>>[SW]

Table96.Summaryfortheemergenceofverbinflectionandstresspatternsinverbs

The analysis conducted on the speech productions of the children in /‐SW/ and /‐

WS/verbsshowedthat:

(i) Inthespeechofthechildrenobserved,theImperativeandPresentforms,namely

3rdp.sg.ofthePres.Ind.,whichdisplaya/‐SW/patternandwhereastem+theme

vowel is observable, emerged earlier than the Infinitive or Simple Past(Inês is

exceptional,asSimplePresentandInfinitiveemergedsimultaneously);

(ii) TheformsfromtheSimplePast(3rdp.sg.)andfromtheInfinitive,whichdisplaya

/‐WS/patternandaperson/numberandtense/moodsuffix,respectively,maybe

selected earlier than the Present verb forms (as in João and Luma), but their

emergenceonlyoccurredaftertheacquisitionofthePresentforms;

(iii) Initially,childrengenerally truncated/‐SW/to [S]andproduce/‐WS/as [S]or

[WS],thoughthereducednumberoftokensatthebeginningdoesnotallowusto

drawacleartendency;

322

(iv) Later,childrentendtoproduce/‐SW/and/‐WS/verbstarget‐like.

Also, the results found in this section indicated that the verb formswith amarked

stresspattern(namely, the3rdp.sg.of theSimplePastor the Infinitive,whichbearstress in

the last syllable ‐ in the person/number and the tense/mood suffix, respectively) are

acquiredlater.

From the results presented in this section, we conclude that, though verbs are

selected and produced later than [+nouns], until the end of the observation period, stress

acquisitioninverbsseemstobegovernedbythesamemetricprinciplesobservedinChapter

5:S>>SW,WS>>SW.

6.1.1.4.Summaryformorphologyinteraction

In this section we analyzed both noun and verb inflection. First, we showed the

distributionofwordclasses(non‐verbsandverbs)perstresspattern(monosyllables/‐SW/

and /‐WS/).We observed a heterogeneous distribution ofword classes per stress pattern

across development in the children's intake. At the beginning, all children mainly select

monosyllabic words (either non‐verbs and verbs), and di‐ and polisyllabic non‐verbs.

Polisyllabicverbs,onthecontrary,emergedlaterinthechildren'sintake,asillustratedinthe

schemepresentedbelow.

(217) Developmentofwordclassesperwordshapeandstresspattern:

These findings suggest that prosodic constraints, namely, the unmastery of the

algorithm forwordstress, rather thanmorphosyntacticones,mightbepreventingchildren

from producing polisyllabic verb forms. The fact that children are able to produce

monosyllabic and polisyllabic non‐verbs andmonosyllabic, but not polisyllabic verb forms

(either/SW/and/WS/),mightindicatethat,infact,wordstressissensitivetowordclasses.

After that,we accounted for the acquisition of thewordmarker in target trochees,

aiming at investigating whether morphological information was playing a role in the

[+N][+V]Monosyl.

[+N]/‐SW/>/‐WS/

[+V]/‐SW/>/‐WS/

323

acquisitionoftrochees.Weanalyzedtheacquisitionoftargettrocheeswithandwithoutword

marker and found that at the beginning, trochees with word marker were not produced

accordingly(confirmingthegeneralresultsfoundinChapter5,for/SW/words).Truncation,

reduplications (both to [WS] andmultiple reduplications) and epenthesis on the leftwere

strategies observed in the deviant productions of /SW/ non‐verbs with word marker.

However,whenanalyzingtheproductionof/SW/withoutwordmarker,whereweexpected

tofindatendencyfor[WS],nosuchtendencywasobserved.Attheearlystages,theobserved

children used the same strategies used with /SW/with wordmarker and also target‐like

productionswerenoticed.Intheearlystages,theresultspointedtoagreatvariability.These

findingssuggestthatmorphologicalcontrastssuchastheacquisitionofthegendercontrast

might not be preventing trochees frombeing acquired, but instead, childrenmight not yet

havelearnedthewordstressalgorithm.

In the last part of this section,we investigated the acquisition of stress patterns in

verb forms. We analyzed the verb forms uttered productively by the observed children

(3rdp.sg. of thePresent, Imperative, 3rdp.sg. of the SimplePast, Infinitive and3rdp.sg. of the

PastImperfect234).Theresultsindicatedthattheearlyverbswereunmarkedforms,i.e.,they

mainlyconsistedinverbformswhichconformtoatrochaicfootandwhichonlybearastem,

plusa themevowel.These formsare the3rdp.sg.of thePresentand the Imperative,which

have a /‐SW/ stress pattern. Later, the verbs formswith a /‐WS/ stress pattern emerged,

namelythe3rdp.sg.oftheSimplePastandtheInfinitive.

At thebeginning,whenverb formswenotproduced target‐like, the strategiesused

were mainly truncation to [S] and reduplication. Later, all verb forms were produced

accordingly.

6.1.2.Ontheroleofweightintheacquisitionofwordstress

In the present section we will present the results for the role of weight in the

acquisitionofwordstressinEP.Twoaspectsareworthanalyzing,asfarasweight‐sensitivity

is concerned in Portuguese acquisition. The first is to investigate whether children show

weight‐sensitivity in their productions. The second (depending on the former) is to know

whendotheylearnweight.

Branching Nuclei and Branching Rhymes (/VN/, /VC/, /VNC/, /VG/ and /VGN/235)

arenormallyconsideredheavycross‐linguistically.In(52)236,below,werecallthefrequency

234The3rdp.sg.ofthePastImperfectwasonlyfoundproductivelyinInês'speech.Cf.AppendixD.235Herewedisregardsyllabletypeswherethe/‐s/pluralmarkerispresent.236Chapter1,section1.2.4..

324

values for stressed and unstressed types of Rhymes for the target language, presented in

Chapter1,whereweobservethatthesyllabletypes.

(52) Distribution of Rhymes (%) per stress position (adapted from Vigário, Martins &

Frota,2006):

Rhymestructure Stressed Unstressed TOTAL

(C)V 19.1 45.26 64.36

(C)VC 4.87 9.16 14.03

(C)VN 5.31 2.71 8.02

(C)VGN 4.36 1.26 5.62

(C)VG 3.43 0.74 4.17

(C)VGC 1.20 0.01 1.21

Other 2.59

The data showed above suggest that syllables with nasal vowels (VN), Branching

Nuclei(VG,VGN)andsimultaneousBranchingRhymeandBranchingNucleus(VGC)tendto

be stressed. Despite the higher frequency values for VC Rhymes in unstressed than in

stressed position, it isworth noticing that no distinction between themorphological Coda

and the lexical Coda /s/ was considered in this account237. However, the frequency study

conducted in Bisol (1992) indicated that 78% of the words ending in a consonant are

oxytonicand22%areparoxytonic.

Specifically,wewilltestwhethersyllablesthattendtobeconsideredasheavycross‐

linguistically and in Portuguese (/VN/, /VNC/, /VG/, /VGC/, /VGN/ and /VGNC/ ‐ (Bisol,

1992,1993andWetzels,2006)behaveasheavyandarestressedinthePortuguesechildren's

productions. We will evaluate the potential role of weight in stress assignment during

acquisition,whichcontrastsheavysyllabletypeswithlightsyllabletypes.

Since syllable weight might only be relevant for non‐verbs (Wetzels, 2006238) and

verb form in the early speech of Portuguese children were scarce, we did not take into

account any verb forms in this section. Any conclusions as to weight‐sensitivity in the

Portuguesechildrenobservedshouldapplytonon‐verbsonly.

Inthetargetwordsunderanalysisinthissectionnowordclassandnumbermarkers

arepresent,andtheyallconsist instems(e.g.,café] 'coffee',canal] 'channel', lápis] 'pencil').

237Cf.footnote63.238Cf.Chapter1,section1.2.5.2.

325

Therefore,wordssuchascasas'houses'[»kazåS]willnotbeconsideredwithinthe/'CV.CVC/

paradigm, since they are composed by a stem (cas‐), a word marker (‐a‐), and a number

marker(‐s).

We will consider as 'correct' a target‐like production in which the number of

syllables, the stresspatternandsyllableheavinessaremaintained. In the/CV.'CV/pattern,

the word olá [ç»la] 'hello', reduplications and the children's nicknames which are

reduplicatedformsweredisregarded.Oláisawordthatisfrequentlyimitatedfromtheadult

speechandchild‐directedspeechandittendstodisappear,bothfromthechildandthechild‐

directedspeechaftertheinitialsessions239.Sincewearetestingweightsensitivityinchildren

acquiring Portuguese as the target language, reduplications and children's reduplicated

nicknamesweredisregardedforthreemainreasons:

(i) the unstressed syllable of a reduplication is not mapable onto the unstressed

syllableofthetargetword;

(ii) reduplicatedwordsinthelexiconofthetargetsystemareveryinfrequent;

(iii) reduplicatedwordsmightbeinterpretedasarepetitionofasyllable(cf.Chapter

5),andnotaswords.

In the following tables, and since both closed syllables and syllables ending in a

diphthongtenddoattractstressinthetargetlanguage,wewillrepresentfinalheavysyllables

as 'CVC', irrespective of the syllable structure therein. Therefore, /CV.'CVC/ words will

account for /CV.'CVN/ (e.g., patim 'skate' /på»ti‚/), /CV.'CVC/ (e.g. rapaz 'boy' /{å»paS/),

/CV.CVG/(e.g.,chapéu'hat'/Så»pEw/)and/CV.CVGN/(e.g.,balão'balloon'/bå»lå‚w‚/)words.

Inaddition,sinceinformationintheonsetisirrelevantforsyllableweightpurposes,

/'CV.CVC/wordsmight correspond to /'V.CVC/ (e.g., hábil 'skillful' /»abi…/) and /CV.'CVC/

mightcorrespondto/V.'CVC/(e.g.,anis'anise'/å»niS/).

Inordertoprovideevidenceforweight‐sensitivity,wewillinvestigatewhether:

(i) Unstressed light syllables are more prone to deletion than unstressed heavy

syllables (assuming that stressed syllables are rarely prone to deletion in the

acquisition data) (section 6.2.2.1.). If syllable preservation in unstressed heavy

syllables is higher than in unstressed light syllables, then we will provide

evidenceforsyllableweightduringEPacquisition;

239Cf.,forinstance,João'slexiconinAppendixD.

326

(ii) /CV.'CVC/ and /CV.'CV/ words are produced correctly earlier than /'CV.CVC/

(section 6.2.2.2.). If heavy unstressed syllables are acquired later than heavy

stressedsyllables,thenevidenceforitsmarkedcharacterwillbeprovided;

(iii) Stress shift occurs in the direction of heavy syllables (/'CV.CVC/ ‐> [CV.'CVC])

(section 6.2.2.3.). If this pattern of stress shift is found, evidence for stress

attractiontoheavysyllablesisprovided.

6.1.2.1.Deletionofunstressedsyllables

Given the infrequent and potentiallymarked status of unstressed complex Rhymes

andComplexNuclei inthetarget language, theexpectancyof findingthesesyllabletypes in

unstressed position, in the early speech of Portuguese children's speech, will be low.

Additionally, if light unstressed light syllables are more prone to deletion than heavy

unstressedsyllables,then,evidenceforweightsensitivitywillbeprovided.

Inthefollowingtable(Table97),wewillshowthegeneralresultsforthepercentage

ofdeletionofheavyandlightsyllablesinunstressedposition.Wedonotpresenttheresults

forClara,sincetheonlyrelevanttokenproducedofanunstressedheavysyllablewasinthe

wordpinguim'penguin'/pi‚»gWi‚/.Withonesingletokenwithunstressedheavysyllables,any

comparisonastoweightsensitivitythroughtheobservationofdeletioninunstressedheavy

andlightsyllables(in/SW/and/WS/)wouldnotbeclarifying.

Light/SW/e.g.,[»kazå]

Heavy/SW/e.g.,[»lapiS]

Light/WS/e.g.,[bå»lå‚w‚]

Heavy/WS/e.g.,[pi‚»gWi‚]

Inês 43.9%(101/230)

11.5%(3/26)

39.6%(19/48)

3.85%(1/26)

Joana 43.9%(54/123)

50%(2/4)

43.8%(28/64)

33.3%(2/6)

João 41.8%(69/165)

0%(0/0)

43.3%(26/60)

14.3%(1/7)

Luma 19%(58/313)

100%(1/1)

43%(19/44)

0%(0/6)

Table97.Percentageofdeletionofheavyandlightsyllablesinunstressedposition(/SW/and/WS/)

327

FromthedatashowninTable97,weobservethat:

(i) InInêsandJoão'sspeech,lightsyllablesaremorepronetodeletioninunstressed

position;

(ii) Joana and Luma have higher percentage rates of deletion in unstressed heavy

syllables than inunstressed lightsyllables (in/SW/only),but thatseemstobe

mostly due to the reduced number of tokens with heavy unstressed syllables

selectedin/SW/;

(iii) The number of unstressed heavy syllables selected by the four children under

analysisisreduced.

The higher tendency to preserve unstressed heavy syllables than unstressed light

syllablesandthereducednumberoftokenswithunstressedheavysyllablesinthechildren's

intake suggest, one the one hand, that Portuguese children might be sensitive to syllable

weight and, on the other hand, it might indicate that unstressed heavy syllables have a

markedcharacterduringprosodicacquisitioninEP.

6.1.2.2.Wordswithheavystressedandheavyunstressedsyllables:emergencepath

Inthissectionwewillinvestigatethedevelopmentalpathforwordswithfinalheavy

stressed and unstressed syllables (/CV.'CVC/ and /'CV.CVC/, respectively). We will also

present the results for final stressed light syllables (/CV.'CV/). An earlier production and

acquisitionof/CV'CV/(e.g.,café[kå»fE]'coffee')and/CV.'CVC/(e.g.,canal[kå»na…]'channel',

than /'CV.CVC/ (e.g., lápis [»lapiS] 'pencil'), will provide evidence for weight‐sensitivity.

Conversely,if/'CV.CVC/(e.g.,lápis)wordsareproducedandacquiredearlierthan/CV.'CVC/

(e.g.,canal),evidenceagainstweight‐sensitivityisgiven.

InTables98and99,weaccountforthedevelopmentalpathundertakenbyClara in

the production of for /CV.'CVC/ and /CV.'CV/, respectively. /'CV.CVC/ words were never

selectedbythischild,untiltheendoftheobservationperiod.

328

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 ‐ ‐Session6 ‐ ‐Session7 77.8%(7/9) 1Session8 ‐ ‐Session9 25%(1/4) 1Session10 40%(4/10) 2Session11 71.11%(32/45) 3Session12 75.6%(65/86) 3

Table98.Percentageof/CV.'CV/wordsproducedtarget­like(Clara)

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 0%(0/1) 1Session6 ‐ ‐Session7 ‐ ‐Session8 0%(0/1) 1Session9 50%(5/10) 1Session10 82.35%(14/17) 1Session11 85.71%(18/21) 1Session12 75%(12/16) 2

Table99.Percentageof/CV.'CVC/wordsproducedtarget­like(Clara)

InTable98,weobservethat/CV.'CV/wordsareselectedareproducedtarget‐likein

session7, thoughinsession8notargetsareselected,andinsession9and10lowervalues

(25% and 40%, respectively) are observed. From session 11 onwards, higher target‐like

(>75%)ratesaredetected./CV.'CVC/words(Table99)areearlierselected(insession5)but

theyarenotproducedtarget‐likeuntilsession9.Insession9,/CV.'CVC/wordsareproduced

with a 50% rate of target‐like production and in the following sessions higher values

(between75%and85%)arefound.ThenumberoftypesoftheCV.'CVandCV.'CVCformsis

veryreducedinClara'sspeech.

In (218), we present Clara's renditions of /CV.'CV/ and /CV.'CVC/ words, across

sessions.

329

(218) Clara–productionof/CV.'CVC/and/CV.'CV/words:

Orthogr. Gloss Target Output AgeJoão 'name' /Zu»å‚w‚/ [zu»å‚w‚]/[ju»å‚w‚] 1;7.11(S9)João 'name' /Zu»å‚w‚/ [du»å‚w‚]/[do»å‚w‚] 1;8.20(S10)João 'name' /Zu»å‚w‚/ [bu»å‚w‚] 1;9.23(S11)

/CV.'CVC/

João 'name' /Zu»å‚w‚/ [zu»å‚w‚] 1;10.15(S12)aqui 'here' /å»ki/ [å»ki]/[a»ki˘] 1;5.16(S7)aqui 'here' /å»ki/ [»ki]/[å»ki]avô 'grandfather' /å»vo/ [ˆ»vo]/[»u]/[»v˘u] 1;8.20(S10)

aqui 'here' /å»ki/ [å»ki]avô 'grandfather' /å»vo/ [å»vo]/[»vo]aqui 'here' /å»ki/ [å»ki]

1;9.23(S11)

/CV.'CV/

avô 'grandfather' /å»vo/ [a»bo] 1;10.15(S12)

Intheseproductionsweobservethat,in/CV.'CVC/words,theheavycharacterofthe

finalsyllablesispreservedfromthebeginning.

InTables100,101and102,weshowInês'percentageratesfor/CV.'CV/,/CV.'CVC/

and/'CV.CVC/words,respectively.

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 0%(0/4) 2Session4 ‐ ‐Session5 0%(0/3) 2Session6 9.09%(1/11) 2Session7 9.09%(1/11) 1Session8 38.46%(10/26) 4Session9 55.55%(10/18) 2Session10 36.67%(11/30) 1Session11 61.22%(30/49) 3Session12 44.03%(48/109) 4Session13 64.29%(27/42) 7Session14 68.29%(56/82) 5Session15 56.74%%(21/37) 4Session16 56.6%(30/53) 5Session17 46.15%(12/26) 5Session18 57.44%(27/47) 6

Table100.Percentageof/CV.'CV/wordsproducedtarget­like(Inês)

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 0%(0/19) 2Session4 0%(0/1) 1Session5 0%(0/1) 1Session6 0%(0/3) 2Session7 0%(0/11) 2Session8 0%(0/12) 4Session9 40%(8/20) 5Session10 72.7%(8/11) 5Session11 46.2%(12/26) 9Session12 35%(7/20) 8Session13 48.3%(14/29) 17Session14 33.3%(7/21) 9Session15 65%(13/20) 11Session16 37.5%(12/32) 12Session17 47.4%(9/19) 12Session18 68.4%(26/38) 17

Table101.Percentageof/CV.'CVC/wordsproducedtarget­like(Inês)

330

Tables100and101showthat/CV.'CV/and/CV.'CVC/wordsareacquiredbyInêsin

session 9 and 10, respectively, though they are selected from session 3 onwards.We see,

however, that the percentages of both structures do not go beyond 70%. The late

stabilizationof/CV.'CV/and/CV.'CVC/,however,seemstobeonlyapparent.

Ontheonehand,thepercentagesin/CV.'CV/arehighlyinfluencedbythefrequency

ofthewordaqui[å»ki]'here',whichisoftenprecededbyolha[»ç¥å]'lookimp.'orestá[S»ta]'it

is'.Thechild'soutputofthewordaquiinthiscontextisrepeatedlya'pseudo‐truncated'form,

resulting from theapocope:olhaaqui [»çja»ki] 'lookathere' orestáaqui [»ta»ki] 'it ishere'.

Though it was coded as truncation, this output form is consistentwith spontaneous adult

speech.

Ontheotherhand,/CV.'CVC/wordsfrequentlyhaveaword‐final‐Ror‐l (e.g.,doutor

[do»toR] 'doctor', hospital [çSpi»ta…] 'hospital'). Inêsmainly produces these formswith final

epenthesis(e.g.,[»toli],[pi»talˆ]).Asseenintheprevioussection,thestrategyofepenthesisin

Inêswaspreviouslyobservedinthe/‐WS/formsoftheInfinitive,whichcruciallyendin[‐R]

inPortuguese.Inês'datasuggestthatfinalepenthesisisindependentfromstressassignment

andmorphosyntactic aspects,but rather it seems related to the syllableand the segmental

structure,namelywiththeprocessofword‐finalepenthesisafteraword‐finalsonorant(cf.,

forinstance,theexamplesin(220),below,wheresimultaneousproductionswithword‐final

epenthesisafterasonorant,butnotaftertheobstruent,areobserved).

In(219)wepresentInês'renditionfor/CV.'CV/and/CV.'CVC/words.

(219) Inês–productionof/CV.'CV/and/CV.'CVC/words:

Orthogr. Gloss Target Output Ageaqui 'here' /å»ki/ [å»ki]ali 'there' /å»li/ [a»li]/[a»di] 2;0.1(S9)

café 'coffee' /kå»fE/ [kå»pE]aqui 'here' /å»ki/ [å»ki˘]sofá 'couch' /su»fa/ [tu»ta]

2;1.10(S10)

daqui 'fromhere' /då»ki/ [då»ki]café 'coffee' /kå»fE/ [gå»pE] 2;2.1(S11)

/CV.'CV/

boné 'hat' /bç»nE/ [bç»nE] 2;5.24(S17)chapéu 'hat' /Så»pew/ [pe»pEw]leão 'lion' /li»å‚w‚/ [i»å‚w‚] 1;9.19(S10)

João 'name' /Zu»å‚w‚/ [du»å‚w‚]jardim 'garden' /ZåR»di‚/ [dˆ»di‚] 1;10.29(S11)

balão 'balloon' /bå»lå‚w‚/ [bå»lå‚w‚]chapéu 'hat' /Så»pEw/ [tå»pEw] 2;0.11(S12)

/CV.'CVC/

também 'also' /tå‚»bå‚j‚/ [tå»må‚j‚] 2;1.10(S13)

331

In (219), we observe that /CV.'CV/ words are produced target‐like in session 9,

though theyaremostly restricted to two types:aqui 'here' andali 'there'. Fromsession10

onwards,both/CV'CV/and/CV.'CVC/wordsareproducedaccordingly,includingtheirfinal

heavy syllable,which is produced as heavy. It isworth noticing that the childwas able to

keep, both the stress pattern, the number of syllables of the target form and syllable

structure.

InTable102,wepresentthepercentageoftarget‐likeproductionof/'CV.CVC/words

inInês'speech.

Session %(absolutevalues) TypesSession1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 ‐ ‐Session6 ‐ ‐Session7 ‐ ‐Session8 ‐ ‐Session9 0%(0/7) 1Session10 ‐ ‐Session11 ‐ ‐Session12 0%(0/2) 1Session13 18.2%(2/11)* 2Session14 0%(0/1) 1Session15 100%(2/2) 2Session16 0%(0/1)* 1Session17 ‐ ‐Session18 50%(1/2) 1

Table102.Percentageof/'CV.CVC/wordsproducedtarget­like(Inês)

In Inês' speech, thenumberof /'CV.CVC/word is significantly reduced, both taking

intoaccountthenumberoftypesandthenumberoftokenspersession.Inallsessions,three

types with the structure /'CV.CVC/ were selected: lápis [»lapiS] 'pencil', açúcar [å»sukaR]

'sugar' andMickey [»mikåj] 'name'.When these formswere attempted, theywereprone to

greatinstability,untiltheendoftheobservationalperiod.

In(220),wepresentInês'renditionsfor/'CV.CVC/words.

332

(220) Inês–/'CV.CVC/words:

Orthogr. Gloss Target Output Agelápis 'pencil' /»lapiS/ [»patH]/[Q»batu]lápis 'pencil' /»lapiS/ [»patu] 1;8.2(S9)

lápis 'pencil' /»lapiS/ [»api] 2;0.11(S12)açúcar 'sugar' /å»sukaR/ [å»tukåli]Mickey 'name' /»mikåj/ [»mikåj] 2;1.10(S13)

lápis 'pencil' /»lapiS/ [»lapi] 2;2.1(S14)lápis 'pencil' /»lapiS/ [»lapiS]Mickey 'name' /»mikåj/ [»mikåj] 2;3.8(S15)

açúcar 'sugar' /å»sukaR/ [å»tukåli] 2;4.18(S16)lápis 'pencil' /»lapiS/ [»lapiS] 2;7.16(S18)

Intheinstancesaboveweshowthat,when/'CV.CVC/wordsareselected(insession

9),theyareinitiallyproducedas[SW]words,withafinallightsyllable.Untilsession12,the

child does not produce the target syllable‐final consonant, but she correctly produced the

target stresspattern.Fromsession13onwards,açúcar 'sugar' isproducedwithword‐final

epenthesis,creatingadifferentstresspattern ‐/'CV.CVC/‐> ['CV.CV.CV].Thoughtheuseof

thisstrategy(finalepenthesis)mightindicatethatthechildwaschangingthestresspattern,

thecorrectproductionofother/'CV.CVC/words, like lápis 'pencil' andMickey 'name',with

the target stress pattern indicates that that is not the case. In session 15, Inês is able to

correctlyproducea/SW/patternwithafinalheavysyllable.However,thereducednumber

of selected targets and thevariability found in theproductionof these structures seems to

reflect:(i)theirinfrequentstatusinthetargetsystemand(ii)themarkedcharacterofthese

structuresinthechildsystem.

In Tables 103 and 104, we show Joana's developmental path for /CV.'CV/ and

/CV.'CVC/words,respectively.

333

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 ‐ ‐Session6 ‐ ‐Session7 100%(1/1) 1Session8 0%(0/1) 1Session9 33.3%(1/3) 2Session10 57.1%(4/8) 1Session11 33.3%(2/6) 3Session12 10%(1/10) 3Session13 76.92%(10/13) 3Session14 37.5%(6/16) 4

Table103.Percentageof/CV.'CV/words(Joana)

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 ‐ ‐Session6 ‐ ‐Session7 ‐ ‐Session8 ‐ ‐Session9 10%(1/10) 2Session10 11.1%(1/9) 6Session11 13%(3/23) 10Session12 21.4%(3/14) 13Session13 10%(1/10) 5Session14 54.2%(13/24) 15

Table104.Percentageof/CV.'CVC/words(Joana)

Table 103 indicates that /CV.'CV/ words are prone to some instability across all

sessions, alternatingbetweenhighand lowpercentagesof target‐likeproductionalong the

observationperiod./CV.'CVC/words(Table104)wereselectedlaterthan/CV.'CV/words(in

session9),andtheygothroughagradualincreasingtarget‐likeproductionpathuntiltheend

of the observation period. As in Inês, the unstable behavior of /CV.'CV/ words is only

apparent,sinceoneofthetypesofthesestructuresisthewordaqui[å»ki]'here',whichtends

tobetruncatedinspontaneousspeech(bothinchildandinadultone).

In(221),weshowinstancesof/CV.'CV/and/CV.'CVC/words,fromJoana.

334

(221) Joana–productionof/CV.'CV/and/CV.'CVC/words:

Orthogr. Gloss Target Output Ageavó 'grandmother' /å»vç/ [a»Bo] 1;6.24(S7)café 'coffee' /kå»fE/ [ki»kE] 1;9.25(S9)avó 'grandmother' /å»vç/ [å»bç˘]/[å»vç˘]/[»fç] 1;10.22(S10)avô 'grandfather' /å»vo/ [»vo]avó 'grandmother' /å»vç/ [»vç]/[å»vç] 2;2.19(S12)

avô 'grandfather' /å»vo/ [å»vo]

/CV.'CV/

aqui 'here' /å»ki/ [å»ki] 2;4.1(S13)

avião 'plane' /åvi»å‚w‚/ [å»wå‚w‚] 1;10.22(S10)balão 'balloon' /bå»lå‚w‚/ [m´»¯å‚w‚]caracóis 'snails' /kåRå»kçjS/ [kç»kç˘S] 2;0.9(S11)

nariz 'nose' /nå»RiS/ [å»RiS]papéis 'papers' /på»pEjS/ [bE»bEjS] 2;2.19(S12)

Jesus 'name' /Zˆ»zuS/ [Zu»ZuS] 2;4.1(S13)prisão 'prison' /pRi»zå‚w‚/ [pi»Zåw]

/CV.'CVC/

chapéu 'hat' /Så»pEw/ [Z´»pEw] 2;6.24(S14)

In(221),weobserveinstancesoftarget‐likeandtruncatedformsfor/CV.'CV/words,

producedbyJoana.TruncatedformsarepossibleuntillateinJoana'sspeech.When/CV.'CVC/

wordsareselected,thefinalheavysyllableisproducedasheavy(i.e.,eitherwithBranching

Rhymeand/orBranchingNucleus)andthestresspatternmightbepreserved.

Table 105 shows the percentage of target‐like production of /'CV.CVC/ words in

Joana'sspeech.

Session %(absolutevalues) TypesSession1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 ‐ ‐Session6 ‐ ‐Session7 ‐ ‐Session8 ‐ ‐Session9 ‐ ‐Session10 ‐ ‐Session11 0%(0/1) 1Session12 ‐ ‐Session13 0%(0/1) 1Session14 0%(0/2) 2

Table105.Percentageof/'CV.CVC/words(Joana)

335

Thevalues inTable105 indicate that/'CV.CVC/wordsarerarelyselectedby Joana,

until the end of the observation period. The number of both types and tokens with the

/'CV.CVC/ structure is somewhat reduced aswell. Until the end of the observation period,

Joanadoesnotcorrectlyproducethe/'CV.CVC/words,asshownin(222).

(222) Joana–productionof/'CV.CVC/words:

Orthogr. Gloss Target Output AgeCésar 'name' /»sEzaR/ [»tSE] 2;0.9(S11)César 'name' /»sEzaR/ [»SE] 2;4.1(S13)pónei 'poney' /»pçnåj/ [»pç¯i]César 'name' /»sEzaR/ [»CeΩå] 2;6.24(S14)

In(222),weseethatJoanainitiallytruncates/'CV.CVC/wordsto[S](session11and

13)andlaterproducesthemas[SW]words(session14),withafinallightsyllable.AsinInês'

speech,theheavycharacterofthefinalsyllablesisnotpreserved.

Tables106and107showJoão'sacquisitionpathfor/CV.'CV/and/CV.'CVC/words.

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 ‐ ‐Session6 ‐ ‐Session7 ‐ ‐Session8 25%(1/4) 2Session9 0%(0/2) 2Session10 ‐ ‐Session11 ‐ ‐Session12 100%(1/1) 1Session13 0%(0/2) 2Session14 ‐ ‐Session15 0%(0/1) 1Session16 50%(3/6) 4Session17 62.5%(5/8) 2Session18 83.3%(5/6) 3Session19 75%(3/4) 2Session20 80%(8/10) 5Session21 100%(4/4) 2Session22 70%(7/10) 5

Table106.Percentageof/CV.'CV/words(João)

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 ‐ ‐Session6 ‐ ‐Session7 ‐ ‐Session8 42.9%(3/7) 3Session9 0%(0/10) 2Session10 0%(0/3) 2Session11 0%(0/4) 2Session12 0%(0/2) 2Session13 0%(0/8) 2Session14 0%(0/1) 1Session15 25%(1/4) 2Session16 38.5%(5/13) 10Session17 33.3%(1/3) 2Session18 63.6%(14/22) 5Session19 30%(3/10) 5Session20 23.1%(3/13) 8Session21 63.6%(7/11) 7Session22 33.3%(3/9) 8

Table107.Percentageof/CV.'CVC/words(João)

336

In Tables 106 and 107, we observe that both /CV.'CV/ and /CV.'CVC/ words are

acquired in session 16, though they are earlier selected. /CV.'CV/ words are selected in

session 8, though they are intermittently attempted until session 16, with low target‐like

productionrates./CV.'CVC/wordsareselectedinsession8aswell,andareattemptedinall

sessionswith low target‐like production rates until session 16. Both structures reach high

target‐likeproductionratesaftersession16,and/CV'CVC/wordsbecomemorefrequent,in

termsoftypesandtokens.

In(223),wepresentJoão'srenditionsof/CV.'CV/and/CV.'CVC/words.

(223) João–/CV.'CV/and/CV.'CVC/words:

Orthogr. Gloss Target Output Ageavó 'grandmother' /å»vç/ [å»bA] 1;4.17(S8)avô 'grandfather' /å»vo/ [pu»bu] 1;7.0(S12)café 'coffee' /kå»fE/ [tå»tE] 1;9.25(S16)avó 'grandmother' /å»vç/ [å»bç]café 'coffee' /kå»fE/ [pe»tHE] 1;10.11(S17)

kiwi 'kiwi' /ki»vi/ [bi»bi]kiwi 'kiwi' /ki»vi/ [di»di] 1;10.26(S18)

ali 'there /å»li/ [å»di] 1;11.19(S19)

/CV.'CV/

kiwi 'kiwi' /ki»vi/ [ti»ni] 2;0.20(S22)Jesus 'name' /Zˆ»zuS/ [ju»ju]avestruz 'ostrich' /åvˆS»tRuS/ [tu»tu]pinguim 'penguin' /pi‚»gWi‚/ [i»di‚]arroz 'rice' /å»{oS/ [a»jo˘]balão 'balloon' /bå»lå‚w‚/ [å»law]

1;9.25(S16)

avião 'plane' /åvi»å‚w‚/ [ti»jå‚˘w‚] 1;10.11(S17)maçã 'apple' /må»så‚/ [å»tå‚]limão 'lemon' /li»må‚w‚/ [mi»må‚w‚] 1;10.26(S18)

biberon 'milkbottle' /bibˆ»Ro‚/ [bi»bo‚] 1;11.20(S19)pinhais 'pinetreefield' /pi»¯ajS/ [pi»ajS]

/CV.'CVC/

jardim 'garden' /ZåR»di‚/ [då»di‚] 1;11.29(S20)

TheinstancesaboveshowthatJoãoisabletosimultaneouslyproduceboth/CV.'CV/

and/CV.'CVC/.Mostofthe/CV.'CVC/wordsproducedareheavyattheNucleuslevel.

Joãodoesnotselectany/'CV.CVC/wordsuntilthelastobservedsession.

InTables108and109,wepresentLuma's target‐likeproductionrates for/CV.'CV/

and/CV.'CVC/words.

337

Session %(absolutevalues)

Types

Session1‐19 ‐ ‐Session20 100%(17/17) 2Session21 100%(7/7) 1Session22 ‐ ‐Session23 0%(0/11) 2Session24 ‐ ‐Session25 83.3%(5/6) 2Session26 0%(0/2) 2Session27 0%(0/2) 1Session28 75%(9/12) 3Session29 0%(0/2) 2Session30 43.8%(7/16) 4Session31 76.9%(10/13) 2Session32 60.4%(29/48) 2Session33 18.2%(6/33) 4Session34 33.3%(2/6) 2Session35 30.8%(4/13) 2Session36 56.1%(23/41) 2Session37 30.8%(4/10) 3

Table108.Percentageof/CV.'CV/words(Luma)

Session %(absolutevalues)

Types

Session1‐10 ‐ ‐Session11 0%(0/1) 1

Session12‐22 ‐ ‐Session23 50%(2/4) 1Session24 0%(0/1) 1Session25 100%(2/2) 1Session26 100%(1/1) 1Session27 100%(1/1) 1Session28 ‐ ‐Session29 11.1%(1/9) 2Session30 50%(1/2) 2Session31 25%(3/12) 8Session32 66.7%(2/3) 1Session33 30.8%(4/13) 7Session34 45.5%(10/22) 7Session35 14.3%(1/7) 3Session36 66.7%(18/27) 10Session37 76.9%(10/13) 7

Table109.Percentageof/CV.'CVC/words(Luma)

InTables108and109,weobservethat/CV.'CV/and/CV.'CVC/wordsareacquired

approximatelyinthesamesession:/CV.'CV/wordsgobeyond50%oftarget‐likeproduction

rateinsession31,whereasin/CV.'CVC/,thatvalueisreachedinsession30.Bothstructures

wereattemptedbefore./CV.'CV/wordswerefirstselectedandproducedinsession20with

hightarget‐likepercentages.However,greatpercentagevariabilityintheproductionofthese

forms is observed up until session 30. /CV.'CVC/ words are first produced target‐like in

session11, but they are not selecteduntil session23. From session23 to session29, high

percentagesareobserved in theproductionof /CV.'CVC/words, though there is a reduced

numberoftokens(andtypes)persession.Untilthelastobservedsession,both/CV.'CV/and

/CV.'CVC/wordsareproducedwithvariablehighandlowpercentages.

In(224),weshowLuma'srenditionsof/CV.'CV/and/CV.'CVC/words.

338

(224) Luma–/CV.'CV/and/CV.'CVC/words:

Orthogr. Gloss Target Output AgePati 'name' /pa»ti/ [te»ti] 1;9.29(S20)Pati 'name' /pa»ti/ [ta»ti] 1;10.18(S21)avô 'grandfather' /å»vo/ [to»to] 1;11.15(S23)Pati 'name' /pa»ti/ [pa»ti˘] 2;0.13(S25)Soni 'name' /sç»ni/ [fç»ni] 2;3.26(S30)

/CV.'CV/

sofá 'couch' /su»fa/ [fu»fa] 2;5.15(S33)Miguel 'name' /mi»ge…/ [ni»å‚] 1;11.15(S23)Miguel 'name' /mi»ge…/ [ni»E˘] 2;0.13(S25)avião 'plane' /åvi»å‚w‚/ [ˆvi»å‚w‚] 2;3.26(S30)andor 'walk' /å‚»doR/ [å‚»doj]balão 'balloon' /bå»lå‚w‚/ [bu‚»å‚w‚] 2;5.15(S33)

balão 'balloon' /bå»lå‚w‚/ [b廥å‚w‚] 2;5.20(S34)

/CV.'CVC/

chapéu 'hat' /Så»pEw/ [tå»pEw] 2;6.20(S36)

As for /'CV.CVC/words, Luma only selects onewordwith this structure, íris [»iRiS]

'iris',insession33.Shetruncatedthewordto[»li].

Table 110 summarizes the emergence path for /CV.'CV/, /CV.'CVC/ and /'CV.CVC/

wordsinallchildren:

StageI StageII StageIII

Clara /CV.'CVC/ /CV.'CV/

Inês /CV.'CV/ /CV.'CVC/ /'CV.CVC/

Joana /CV.'CV/ /CV.'CVC/ /'CV.CVC/

João /CV.'CV/ /CV.'CVC/

Luma /CV.'CVC/~/CV.'CV/ /'CV.CVC/

Table110.Summaryoftheemergenceof/CV.'CV/,/CV.'CVC/and/'CV.CVC/words

Thecomparativeresultsof/CV'CV/,/CV.'CVC/and/'CV.CVC/wordsindicatedthat:

(i) ClaraandLumaacquired/CV.'CVC/before/CV.'CV/;

(ii) Inês,JoanaandJoãoacquire/CV.'CV/wordsbefore/CV.'CVC/;

(iii) Thefinalheavysyllableof/CV.'CVC/wordsisnormallyproducedasheavybyall

children;

(iv) Theemergenceof/'CV.CVC/wordsisonlyobservedinInês,JoanaandLuma;

(v) Theacquisitionof/'CV.CVC/wordsisnotattestedinanyofthechildren'sspeech;

339

(vi) Whenchildrenselect/'CV.CVC/words,whichoccurrs late indevelopment,they

mighttruncatethemtoamonosyllableorproducethemwithafinallightsyllable;

ThealternativepathundertakenbyClaraandLuma,ontheonehand(theyacquired

/CV.'CVC/before/CV.'CV/),andInês,JoanaandJoão,ontheotherhand(theyfirstacquired

/CV.'CV/ before /CV.'CVC/), as well as the data presented in this section showing that

/CV.'CV/and/CV.'CVC/wordsareacquiredapproximatelyinthesamesessionbyallchildren

indicatethatthesestructuresmightbeinterpretedbychildrenassimilarprosodicstructures.

Our analysis on /'CV.CVC/ words also showed that these structures were rarely

selected and produced in the speech of the five children observed. Clara and João did not

selectany/'CV.CVC/wordsandonlyinInêsandJoana'sspeecheswasitpossibletoobserve

thebehaviorofthesestructuresbutnocorrectproductionwasattested.InInês,weobserved

averyunstablebehaviorintheproductionof/'CV.CVC/forms,whichwereeitherproduced

withfinalepenthesisorwithafinallightsyllable.InJoana'sspeech,theselectionof/'CV.CVC/

wordswas very rare and no correct productionwas attested,mostly because thesewords

were truncated or produced with a final light syllable. Contrary to what occurred in

/CV.'CVC/words,whosefinalheavysyllablewasproducedasheavyfromthebeginning,Inês

and Joana produced /'CV.CVC/ words with a final light syllable. Luma produced one

/'CV.CVC/wordinthelatersessions(insession33),butshetruncatedittoa['CV]word.

Inallchildren/'CV.CVC/wordsshowedacleardistinctbehaviorfromboth/CV.'CV/

and /CV.'CVC/ words, which might indicate that the former have a marked status, when

comparedtothelatter.

6.1.2.3.Wordswithheavystressedandheavyunstressedsyllables:stressshift

Inthepresentsectionwewillshowtheresultsforstressshiftin/CV.'CV/,/CV.'CVC/

and/'CV.CVC/wordsinthespeechofthefiveobservedchildren.

Recall that, based on previous reports in the literature on Portuguese acquisition

(Bonilha,2005;Santos,2001,2007)wepredictedthat:

(i) Stressshiftofthetype/'CV.CVC/‐>[CV.'CVC]wouldprovideevidenceforweigh‐

sensitivity;

(ii) Stress shiftof the type/CV.'CVC/ ‐> ['CV.CVC]will indicate thatheavysyllables

areirrelevantforstresspurposes,thusprovidingevidenceforsyllableweightin

wordstress;

340

Stress shiftof the type/CV.'CV/ ‐> ['CV.CV]alonedoesnotmakeanyprediction for

weight‐sensitivitybutitshould,nevertheless,becomparedwiththetwopreviousstructures,

in order o verify whether the patterns observed are similar to the ones of /CV.'CVC/ or

/'CV.CVC/words. If /CV.'CV/wordsbehave like/CV.'CVC/,evidence fora similar structure

mightbeprovided(namely, fora finalheavysyllableand, thus,weight‐sensitivity),and the

sameisvalidif/CV.'CV/displaysabehaviorsimilarto/'CV.CVC/instressshift(inthiscase,

noweight‐sensitivitymightbeassumed).

In the following tables we will present the percentage rates for stress shift in

/CV.'CV/, /CV.'CVC/ and /'CV.CVC/ words. We will show the tables for all children all

together,sincesimilarresultswerefoundforallthechildren.It isworthnoticingthatClara

andJoãodidnotselectany/'CV.CVC/wordsandLumahadonlyonewordwiththisstructure.

Tables111and112presentClara'sresultsforstressshiftin/CV.'CV/and/CV.'CVC/

words.

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 ‐ ‐Session6 ‐ ‐Session7 0%(0/9) 1Session8 ‐ ‐Session9 0%(0/4) 1Session10 10%(1/10) 2Session11 2.22%(1/45) 3Session12 0%(0/85) 3

Table111.Stressshiftin/CV.'CV/words(Clara)

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 0%(0/1) 1Session6 ‐ ‐Session7 ‐ ‐Session8 0%(0/1) 1Session9 10%(1/10) 1Session10 5.88%(1/17) 1Session11 9.52%(2/21) 1Session12 0%(0/16) 2

Table112.Stressshiftin/CV.'CVC/words(Clara)

In Clara, we observe a reduced percentage of stress shift in both /CV.'CV/ and

/CV.'CVC/wordsuntil the lastobservedsession. It isalsonoticeable that thechildselected

fewtypesofthesestructures.

Tables113,114and115depict Inês's results forstressshift in/CV.'CV/,/'CV.CVC/

and/CV.'CVC/words,respectively. Inthesetables, '*'standsforepenthesisafterword‐final

sonorant.

341

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 25%(1/4) 2Session4 ‐ ‐Session5 33.33%(1/3) 2Session6 0%(0/11) 2Session7 0%(0/11) 1Session8 7.69%(2/26) 4Session9 0%(0/18) 2Session10 0%(0/30) 1Session11 6.12%(3/49) 3Session12 0%(0/109) 4Session13 0%(0/42) 7Session14 0%(0/82) 5Session15 0%(0/37) 4Session16 0%(0/53) 5Session17 0%(0/26) 5Session18 0%(0/47) 6

Table113.Stressshiftin/CV.'CV/words(Inês)

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 0%(0/19) 2Session4 0%(0/1) 1Session5 0%(0/1) 1Session6 0%(0/3) 2Session7 0%(0/11) 2Session8 8.33%(1/12) 4Session9 0%(0/20) 5Session10 16.67%(1/6)* 5Session11 15.38%(4/26)* 9Session12 10%(2/20) 8Session13 34.48%(10/29)* 17Session14 38.1%(8/21)* 9Session15 0%(0/20) 11Session16 15.53%(5/32)* 12Session17 31.58%(6/19)* 12Session18 21.05%(8/38)* 17

Table114.Stressshiftin/CV.'CVC/words(Inês)

In Inês' data, we observe that stress shift is rare in /CV.'CV/ words. In /CV.'CVC/,

stressshiftisscarceaswell,untilsession12.Whenadifferenceinthestresspatternoccursin

/CV.'CVC/(mainlyaftersession13),itistheproductofvowelepenthesisattheright‐edgeof

wordsendinginasonorant.

342

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 ‐ ‐Session6 ‐ ‐Session7 ‐ ‐Session8 ‐ ‐Session9 14.29%(1/7) 1Session10 ‐ ‐Session11 0%(0/1) 1Session12 0%(0/2) 1Session13 63.64%(7/11)* 2Session14 0%(0/1) 1Session15 0%(0/2) 2Session16 100%(1/1)* 1Session17 ‐ ‐Session18 0%(0/3) 2

Table115.Stressshiftin/'CV.CVC/words(Inês)

Table115indicatesthat/'CV.CVC/wordsarerarelyselectedand,whenadifference

in the target stress pattern occurs, it is the also result of vowel epenthesis. In fact, these

productionsarenot instancesof stress shift,but rather the resultof a change in the target

stresspattern.

InTables116,117and118,wepresent the results for Joanawith respect to stress

shiftin/CV.'CV/,/CV.'CVC/and/'CV.CVC/words,respectively.

343

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 ‐ ‐Session6 ‐ ‐Session7 0%(0/1) 1Session8 0%(0/1) 1Session9 0%(0/3) 2Session10 0%(0/8) 1Session11 0%(0/6) 3Session12 0%(0/10) 3Session13 0%(0/13) 3Session14 0%(0/16) 4

Table116.Stressshiftin/CV.'CV/words(Joana)

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 ‐ ‐Session6 ‐ ‐Session7 ‐ ‐Session8 ‐ ‐Session9 0%(0/10) 2Session10 22.2%(2/9) 6Session11 8.7%(2/23) 10Session12 0%(0/14) 13Session13 10%(1/10) 5Session14 12.5%(3/24) 15

Table117.Stressshiftin/CV.'CVC/words(Joana)

In Joana'sspeech,stressshiftwas inexistent in/CV.'CV/words. In/CV.'CVC/,stress

shiftwasmorefrequentthanin/CV.'CV/words,butitwasnotverycommon,untiltheendof

theobservationperiod.

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 ‐ ‐Session6 ‐ ‐Session7 ‐ ‐Session8 ‐ ‐Session9 ‐ ‐Session10 ‐ ‐Session11 0%(0/1) 1Session12 ‐ ‐Session13 0%(0/1) 1Session14 0%(0/2) 2

Table118.Stressshiftin/'CV.CVC/words(Joana)

/'CV.CVC/wordswereselectedlateandinproducedinveryreducedamounts(they

were selected in session 11). No stress shift was observed in /'CV.CVC/ words in Joana's

speech.

Tables 119 and 120 show the percentage of stress shift in /CV.'CV/ and /CV.'CVC/

words,inJoão'sspeech,respectively.Joãodidnotselectany/'CV.CVC/words.

344

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 ‐ ‐Session6 ‐ ‐Session7 ‐ ‐Session8 0%(0/4) 2Session9 0%(0/2) 2Session10 ‐ ‐Session11 ‐ ‐Session12 0%(0/1) 1Session13 0%(0/2) 2Session14 ‐ ‐Session15 0%(0/1) 1Session16 16.7%(1/6) 4Session17 0%(0/8) 2Session18 0%(0/6) 3Session19 0%(0/4) 2Session20 0%(0/10) 5Session21 0%(0/4) 2Session22 30%(3/10) 5

Table119.Stressshiftin/CV.'CV/words(João)

Session %(absolutevalues)

Types

Session1 ‐ ‐Session2 ‐ ‐Session3 ‐ ‐Session4 ‐ ‐Session5 ‐ ‐Session6 ‐ ‐Session7 ‐ ‐Session8 0%(0/7) 3Session9 0%(0/10) 2Session10 0%(0/3) 2Session11 25%(1/4) 2Session12 0%(0/2) 2Session13 12.5%(1/8) 2Session14 0%(0/1) 1Session15 0%(0/4) 2Session16 7.7%(1/13) 10Session17 33.3%(1/3) 2Session18 18.2%(4/22) 5Session19 0%(0/10) 5Session20 61.5%(8/13) 8Session21 27.3%(3/11) 7Session22 22.2%(2/9) 8

Table120.Stressshiftin/CV.'CVC/words(João)

InJoão'sspeech,thepercentageofstressshiftin/CV.'CV/wordsisreduced(16.7%in

session16and30%insession22)./CV.'CVC/wordsare, ingeneral,moresubject tostress

shift than/CV.'CV/.However, in theearly sessions (until session16), stress shift is rare (3

productions in 53 tokens). From session 17 onwards, the percentages are higher than the

ones observed until then, but can bemostly biased by the reduced number of targets (for

instance, in session17, the33%correspond to1productionof stress shift in3 tokens). In

session20, thechildhasahighrateof stressshift (61.5%),but thatvaluesdecrease in the

followingsessions(session21and22)to27.3%and22.2%.

In Tables 121 and 122, we show Luma's results for stress shift in /CV.'CV/ and

/CV.'CVC/words.

345

Session %(absolutevalues)

Types

Session1‐19 ‐ ‐Session20 0%(0/17) 2Session21 0%(0/7) 1Session22 ‐ ‐Session23 9.1%(1/11) 2Session24 ‐ ‐Session25 0%(0/6) 2Session26 0%(0/2) 2Session27 0%(0/2) 1Session28 16.7%(2/12) 3Session29 0%(0/2) 2Session30 6.3%(1/16) 4Session31 0%(0/13) 2Session32 0%(0/48) 2Session33 3%(1/33) 4Session34 0%(0/6) 2Session35 0%(0/13) 2Session36 0%(0/41) 2Session37 0%(0/10) 3

Table121.Stressshiftin/CV.'CV/words(Luma)

Session %(absolutevalues)

Types

Session1‐10 ‐ ‐Session11 0%(0/1) 1

Session12‐22 ‐ ‐Session23 0%(0/4) 1Session24 0%(0/1) 1Session25 0%(0/2) 1Session26 0%(0/1) 1Session27 0%(0/1) 1Session28 0%(0/1) 1Session29 11.1%(1/9) 2Session30 0%(0/2) 2Session31 8.3%(1/12) 8Session32 0%(0/3) 1Session33 42.9%(3/7) 7Session34 4.5%(1/22) 7Session35 14.3%(1/7) 3Session36 0%(0/27) 10Session37 23.1%(3/13) 7

Table122.Stressshiftin/CV.'CVC/words(Luma)

InLuma'sspeech, thepercentageof stressshift in/CV.'CV/and/CV.'CVC/words is

reduced.Insession33,thechildhas42.9%ofstressshiftin/CV.'CVC/wordsthatcorrespond

to 3 productions in 7 tokens. In the following sessions, however, the percentages have a

decreasingagain(valuesvaryfrom0%to23.1%).

Inthetablespresentedabove,weobservethatstressshiftin/CV.'CV/,/CV.'CVC/and

/'CV.CVC/wordsisnotfrequent.SomecasesexistinJoãofor/CV.'CVC/words,asattestedin

Table120,butintheotherchildren,achangeinstressplacementin/CV.'CVC/wordsisnota

usedstrategy.

Asshownintheprevioussection,/'CV.CVC/wordsarerarelyselectedandstressshift

wasrarelyattested.ClaraandJoãodidnotselectany/'CV.CVC/wordsandLumaselectedone

singletoken.

AsshowninChapter5,stressshiftisnotastrategyusedbythePortuguesechildren

indealingwithword stress.When the childrendonot produce a targetword and a target

stress pattern accordingly, the most common strategies are reduplications, filler insertion

andtruncation240.

240Cf.Chapter5,section5.1.3..

346

6.1.2.4.Summaryforweight

In section 6.1.2., we testedweight‐sensitivity in the speech productions of the five

childreninourcorpus.

Weanalyzedthedeletionrateinheavyandlightunstressedsyllables,theacquisition

pathfor/CV.'CVC/,/'CV.CVC/and/CV.'CV/words,andfinally,weaccountedfortherateof

stressshiftin/CV'CVC/and/'CV.CVC/words.

Ourresultsonweightsensitivityshowthat:

(i) deletion inunstressed light syllables tends tobemoredeleted thanunstressed

heavy syllables (cf. Table 97, for a summary), though the reduced number of

unstressedheavysyllablesinthechildren'sintakeisabsent(inthecaseofClara)

orreduced(inthecaseofLumaandJoão);

(ii) inClara, /CV.'CVC/words emerge earlier than /CV.'CV/words; Inês, Joana and

João have the opposite path; Luma has a simultaneous emergence of the two

structures(cf.Table110,forasummary);

(iii) theemergencepathfor/CV.'CVC/and/CV.'CV/wordsopposestotheacquisition

path for /'CV.CVC/ words: the former is earlier acquired than the latter (cf.

Tables98‐109);

(iv) /'CV.CVC/words are not selected by Clara and João, and Luma has one single

tokenofthisstructure;InêsandJoanahaveaveryreducednumberof/'CV.CVC/

wordsintheirintake(cf.Tables102and105);

(v) /'CV.CVC/wordsmightbetruncateduntillateand,contraryto/CV.'CVC/words,

theycanbeproducedwithafinallightsyllable(cf.examplesin(219)and(220),

forInês,andin(221)and(222),forJoana);

(vi) stressshift in/CV.'CV/,/CV.'CVC/and/'CV.CVC/wordsisa infrequentstrategy

(cf.Tables111‐122).However, Inêstendstouseepenthesisat therightedgeof

words ending in a sonorant (/CV.'CVC/ and /'CV.CVC/), not exactlyperforming

stress shift but changing the target stress pattern of the words. João had one

session(session20)inwhichhighratesofstressshiftin/CV.'CVC/isobserved.

347

In general, three main facts suggest that children process differently words with

heavy syllables in stressed and in unstressed position, providing evidence for weight‐

sensitivityintheearlyspeechofPortuguesechildren:

(i) The asymmetry found in the deletion of unstressed heavy and light syllables

(withhigherdeletionratesinthelatter);

(ii) Theasymmetryfoundintheemergencepathfor/CV.'CVC/,and/'CV.CVC/words

(withanearlieremergenceoftheformer);

(iii) The fact that the two children who selected and produced /'CV.CVC/ words,

truncatedthesestructuresuntillateandproducedthemwithafinallightsyllable,

when/CV.'CVC/wordsweremainlyproducedwithafinalheavysyllable;

(iv) Thepaucityof/'CV.CVC/wordsinthechildren'sintake.

Despite some differences in the percentages of stress shift in /CV.'CV/ and in

/CV.'CVC/words,/CV.'CVC/and/CV.'CV/wordsseemtobehavemoresimilarlywithrespect

tostressassignmentamongthem,than/'CV.CVC/words.

6.1.3.Summaryofthemainfindings

In section 6.1., we have presented the results of our analysis on morphology‐

dependenceandweight‐sensitivityinthepathofwordstressacquisitioninEP.

As far as the interaction between morphology and word stress is concerned, we

observed that, at the early stages, children's productions mostly consist in monosyllabic

words (either [+nouns]or [+verbs]) and indi‐ andpolisyllabic [+nouns].Polisyllabicverbs

emergedlaterinthespeechofallobservedchildren(cf.Tables67‐71).

With respect to non‐verbs, we observed that the acquisition of trochees was, in

general,parallel totheacquisitionof thewordmarker, though, in JoãoandLuma,theword

markerwasnoticeablebeforetrocheesareacquired,mainly,throughtheproductionofstress

shift241.However,wealsoshowedthattrocheeswithoutwordmarkercouldbeproducedas

trochees or be produced with the same strategies as target non‐reduplicated iambs and

trochees with word marker (cf. examples in (199)‐(203)), i.e., truncation (/SW/‐>[S]),

241‐cf.instances(197)and(198).

348

reduplication (/'CV1.CV2/‐>[CV1.'CV1]) and epenthesis at the left‐edge of the circumscribed

syllable (/SW/‐>[fσ]). Our findings for non‐verbs inflection (namely, gender inflection)

suggest that the presence of the word marker might not explain the apparent iambic

tendencyfoundintheearlyspeechofPortuguesechildren,astheymainlyusetruncationor

other above‐mentioned strategies at an early stage, irrespectively of the morphological

constituencyofthetargetwordandthetargetstresspattern.

In [+verbs], data showed that the early multisyllabic verb forms emerging were

consistent with the unmarked stress pattern for verbs (/SW/). Our results indicated an

earlieremergenceofthe3rdp.sg.ofthePresentIndicativeandtheImperative,whichdisplaya

/SW/stresspattern(cf.Tables83‐87andTable96,forasummary)andmainlyconsistsina

verbtheme(stem+themevowel‐cf.examples(205)‐(213)).Person/numberortense/mood

suffixes,namelywitha /WS/ stresspattern,wereonlyproduced later. InClaraand Inês, a

simultaneous acquisitionof /SW/and/WS/verb forms (Present andSimplePast inClara;

PresentandInfinitiveinInês)occurred.InJoãoandJoana,adistinctacquisitionmomentwas

foundin/SW/and/WS/verbforms,withanearlieracquisitionofthePresentIndicativeand

Imperative with a /SW/ stress pattern and a later acquisition of the Simple Past and

Infinitive,which have a /WS/ stress pattern. In all children the acquisition of verb tenses,

either/SW/or/WS/,occurredafterthetrochaicfootwasthepredominantpattern.Untilthat

moment, childrenmostly truncatedor reduplicated /SW/and/WS/verb forms (cf. Tables

88‐95).

The asymmetry found in the selection and production of monosyllabic and

multisyllabicverbs,andthelateremergenceofverbformsingeneral,andinparticular,ofthe

verb forms which do not conform to a trochee and a verb theme (where, in general, the

markedpositionsofwordstressinverbsoccurs)suggestthatstressassignmentintheverb

paradigmmightbeadifficulttaskforchildren.

Asfarassyllableweightandwordstressareconcerned,ourdataofshowedthatlight

unstressedsyllablesweremorepronetodeletionthanheavyunstressedsyllables(cf.Table

97).However,theasymmetryfoundintheamountoftokensineachvariableanalyzed(heavy

syllable in unstressed position and light syllable in unstressed position, both in /SW/ and

/WS/)didnotallowforrobustconclusions.Itsuggests,nevertheless,thatheavyunstressed

syllablesareveryinfrequentinthechildren'sintake(asitisintheadultspeech242).

Thecomparisonbetweentheemergenceof/CV.'CVC/,/CV.'CV/and/'CV.CVC/words

indicatedthat/CV.'CVC/and/CV.'CV/emergedearlierthan/'CV.CVC/(cf.Tables98‐109).In

the speech of three children (Inês, Joana and João), /CV.'CV/ words emerged earlier than

242Cf.Chapter1,section1.2.4.,informationin(52).

349

/CV.'CVC/./CV.'CVC/wordsemergedearlierinClara'sspeechandLumahadasimultaneous

emergenceof/CV.'CV/and/CV.'CVC/words(cf.Table110,forasummay)./'CV.CVC/words

were absent in the intake of Clara and João, and they were rarely selected and produced

target‐likebyInês,JoanaandLuma(cf.Tables102and105,forInêsandJoana,respectively).

When /'CV.CVC/ words were selected (which occurred late in the speech of the three

children),theyweretruncatedorproducedwithafinallightsyllable.

Stressshiftdidnotoccurproductivelyinthespeechofthefiveobservedchildren(cf.

Tables111‐122).Joãowastheonlychildwithhighratesofstressshiftin/CV.'CVC/(cf.Table

120), though ithappenedmainly inonesession. Inêsmainlyusedastrategy that implieda

changeinthetargetstresspattern(vowelepenthesisattheright‐edgeofwordsendingina

sonorant consonant ‐ cf. examples in (220)), but this strategydoesnotnecessarily imply a

shift in the target stress position. The remainder children did not use stress shift

productively.

Inthefollowingsection,wewilldiscussthetargetanalysesonwordstress,basedon

theresultsbroughtupbyourresultsonmorphology‐dependenceandsyllableweightduring

theacquisitionofwordstressinEP.

6.2.Discussion

In the firstsectionof thischapter,wehavepresentedthedescriptionsandanalyses

on Portuguese word stress, and we further showed the proposals on BP word stress

acquisition from Santos (2001, 2007) ‐ supporting a morphology‐based word stress

algorithm in Portuguese ‐ and Bonilha (2005) ‐ supporting a weight‐based word stress

algorithminPortuguese.

In the following sections, we will discuss the target analyses purported for

Portuguese,onthelightofourresults.

6.2.1.Discussingtheinteractionofwordstresswithmorphology

As presented in Chapter 1, Lee (1995) defends that the domain for stress in

Portuguesenon‐verbsisthestemandthatthedefaultfootinPortugueseisaniamb.Indeed,

the morphology‐based approaches in Portuguese often have a contradictory factor: since

they consider the stem as the domain for word stress, causing the word marker to be

extrametrical, word stress is normally at the right‐most syllable of the stress domain,

indicatinganiambicfoot.

350

If one looks at the early metrical representations on word stress proposed by

Andrade (1992) and Andrade & Laks (1992), one observes that there are inconsistent

postulates: on the onehand, the domain for stress is the stemand, on the other hand, the

languagehasabinarytrochaic foot("Feetarebinaryandprominentonthe left" ‐Andrade,

1992:109). In themetrical representations presented in Andrade (1992), we observe that

morphological constituency is relevant, as it is necessary to distinguish between thematic

(e.g.,cabel]o 'hair')andathematic(jacaré] 'aligator')non‐verbs,forstresspurposes.Innon‐

verbswithwordmarker,stressfallsonthepenultimatesyllableofthelexicalword('feetare

binaryandprominentontheleft'),whereasinwordswithoutwordmarker,stressfallsonthe

lastsyllableofthe lexicalword.Thatsyllableconstitutes itselfa foot,sinceit 'standsalone',

though no motivation for why the 'element standing alone' is a foot, a unary one, is

postulated. Additional motivation for the unary character of the word‐final foot is, thus,

required.

Andrade&Laks(1992:17)statedthat"mainstressfallsonthelastvowelofthestem,

oron thepenultimate, if there is anextrametrical vowel".However, the rhythmicprinciple

postulated("Rhythmicwave,peak‐troughanchoredattheright,firstpeakstrong"‐Andrade

&Laks,1992:19)couldnotapplytothestem.Infact,therepresentationspresentinAndrade

&Laks(1992)clearlyshowatrochaicrhythm.However,ontheassumptionthatoneshould

stressthelastsyllableofthestem,thestemmustbeconsideredthedomainforstressandthe

word marker is explicitly considered as extrametrical. It is not clear, though, why the

rhythmic principle is applying to the lexical word and word prominence is always at the

right‐edgeofthestressdomain.Sincethewordmarkerisextrametrical,itshouldbeinvisible

to the rhythmic principle and for stress assignment. It remains unclear how the author(s)

reconcile(s)theapplicationofthemetricalandrhythmicprinciplesinsidethestressdomain

(thestem)withametricalizationofmorphologicalelementsthatareoutsidethestem(word

markers)innon‐verbs.

Pereira(1999)suggeststhatPortuguesewordstressismorphology‐dependentanda

trochaicsystem (Pereira,1999:135). Inorder todefenda trochaicrhythmin the language,

theauthorarguesthat:

(i) StressinPortugueseoverwhelminglyfallsonthepenultimatesyllableofprosodic

words;

(ii) Stress echoes affect the even syllables at the left of the stressed syllable in an

alternatingSWrhythm,fromrighttoleft;

351

(iii) Portugueseevolvedfromatrochaicsystem:Latin;

(iv) Iambicsystemsarealwaysweight‐sensitive.

Within Idsardi's (1992) metrical framework, Pereira (1999) presents the metrical

and rhythmic principles that allow for the construction of prominences grid in Portuguese

non‐verbs.Theprinciplesestablisheddefinethat,inPortuguese,mainstressisontheright‐

most edge of its domain, the stem. However, the nature of the constituent thatmarks the

domain boundary ‐whether it is a foot, a degenerate one, or a syllable ‐, remains unclear.

Furthermore, the word marker is not present in the representations showed in Pereira

(1999),theauthordoesnotmentionwhatisthemetricalstatusofthewordmarker.Oneof

thehypotheseswouldbethatthewordmarkerisextrametrical.Asfarasextrametricalityin

non‐verbs is concerned, the author states that the stress domain‐final syllable is lexically

marked with the information of 'do not stress' within a specific syllable (e.g., súplica

'supplication', âmbar 'amber'). The author seems to assume that only within the stress

domainitisnecessarytospecifyextrametricality.Onthisassumption,wordmarkersarenot

extrametrical.Theyaresimplyinexistentforstressassignmentpurposes.Ontheassumption

that Portugueseword stress, at least in non‐verbs, crucially falls on the last syllable of its

domain,wefurthermoreneedtostatethatwordprominenceinPortugueseisright‐headed(‐

WS),evenifwedonotassumethatthefootplaysanyroleinthelanguagephonology.Insum,

morphology‐based approaches fail in accounting for the trochaic rhythm of the target

language.Withthisrespect,Mateus&Andrade(2000)clarifythat:

"It [the derivational stem] is, we assume, the basic domain of stress. The apparent

complexity of the facts, as one can see, can be reduced to an essential principle: main

stress falls on the last vowel of the stem, or on the one before the last, if there is an

extrametricalvowel.

In this analysis, the right‐to‐left rhythmicwave cannot be considered a primitive of the

stress systembut only as a derived effect. Therefore, it is not possible to formalize the

regularityaccordingtowhichmainstressfallssystematicallyonthefirstvoweltotheleft

ofthesyllablethatcontainsanextrametricalvowel."(Mateus&Andrade,2000:121)

According toMateus & Andrade (2000), assuming that the stem is the domain for

stress, there isnowayofaccounting for the trochaicrhythmofPortuguese. Ifoneassumes

that "the right‐to‐left rhythmicwave cannotbe consideredaprimitiveof the stress system

butonlyaderivedeffect",thereforestressisnot"thelinguisticmanifestationoftherhythmic

structure",assuggestedbyHayes(1995:8),but, instead, itsmotivator.Putdifferently, ifwe

352

assume that word stress is generally domain‐final (stress the last syllable of the stem),

therefore,thetrochaicrhythmofPortugueseisaconsequenceofstress,andnotthecontrary.

Additionally, the authors clearly claim that the final syllable of non‐verbs bearing

wordmarkers isextrametrical,which, fromaphonologicalacquisitionperspective, raisesa

problem:beingconsideredextrametrical,wordswiththewordmarkershouldbeconsidered

markedstructuresand,thus,shouldbesubjectedtoalateacquisition.

Santos (2007) supported Lee's (1995) analysis for the adult language, according to

whomthedefaultfootinnon‐verbsisaniambandthedomainforstressisthestem.Santos

(2007)observedthatiambs,butnottrochees,wereearlierproducedandacquiredinBP.The

highfrequencyofreduplicatedwordswithan[WS]shape,wherenomorphologicalendings

were present, and the truncation of /SW/ words to [S] lead the author to defend that

Brazilian children have a default iambic foot, and that Brazilian children only acquire

trocheeswhenmorphological contrasts aremastered, namely gender contrasts, which are

extrametrical.

Inverbs, Santos (2007)didnot supportLee's analysis. Lee (1995)defends that the

stressdomainisthelexicalwordandthedefaultfootisatrochee,sinceinmostverbforms

stress falls on the penultimate syllable of the lexical word. Santos (2007) found a similar

tendency as the one found in nouns, i.e., Brazilian children tended to produce /WS/ forms

(mainlyInfinitives)earlierthan/SW/.

Based on Santos (2007), according to whom /SW/ would not be acquired until

extrametricality(associatedtothewordmarker)wasacquired,insection6.1.1.,weanalyzed:

(i) the distribution of word classes (non‐verbs and verbs) per word shape

(monosyllables,/‐SW/and/‐WS/);

(ii) theacquisitionofthewordmarkerandtrochees;

(iii) the acquisition of verb tenses and strategies used in the production of verb

forms.

Wehypothesizedthat:

A. Given the results found in previous analyses on the adult language (Andrade,

1988/1992; Andrade & Laks, 1992; Mateus, 1983; Mateus & Andrade, 2000;

Pereira, 1999) and on word stress acquisition in Portuguese (Santos, 2007),

morphologicalacquisitioninteractswithwordstressandwordshapeacquisition

inEP.

353

IfPortuguesewordstressismorphology‐dependent(Lee,1995;Pereira,1999,among

others),andassumingthattheword‐finalinformationin/SW/[+nouns],butnotin[+verbs],

is extrametrical (as proposed by Lee, 1995 and supported by Santos, 2007 with BP

acquisitiondata),thenwewouldexpectthatchildrenacquire/WS/in[+nouns]and/SW/in

[+verbs].

Theresultsofouranalysisontheinteractionofmorphologywithintheacquisitionof

trochaic[+nouns]–theonlywordsbearingmorphological informationintheearlystages–

showthat:

(i) multisyllabic verbs are selected later than multisyllabic non‐verbs and

moosyllabicverbs(cf.Tables67‐71);

(ii) theacquisitionofthetrochaicstresspatternandthewordmarkerin/SW/non‐

verbsisingeneralsimultaneous,thoughtwoofthechildrenobservedareableto

producethewordmarkerbeforetheacquisitionoftrochees(cf.Tables73‐82);

(iii) /SW/ non‐verbswithout thewordmarker are subject to the same production

strategies as /SW/ non‐verbs with the word marker (cf. examples in (199)‐

(203));

(iv) /SW/[+nouns]withwordmarkeraresubjecttothesamestrategiesobservedfor

non‐reduplicated /WS/words and targetmonosyllables observed in Chapter 5

(cf.examplesin(117)‐(121));

(v) in verbs, trochaic forms emerge earlier than iambic forms and these mainly

consisteinverbthemes(Tables83‐87andexamplesin(205)‐(213)).

The results found in non‐verbs seem to confirm Santos' (2007) findings: the

acquisition of trochees in non‐verbs appears to be dependent on the acquisition of gender

contrasts. Before the acquisition of the word marker, trochees are rarely observed in the

speechofthePortuguesechildren;whenmorphologicalcontrastsareacquired,trocheesare

producedaswell.In(225),werecalltheexamplesfromInêsandJoanafortheacquisitionof

trochees,wherethewordmarkerisnotpresent,inthebeginning,andlaterisproduced.

354

(225) Inês–initialproductionof/SW/:

Orthogr. Gloss Target Output Agemanta 'blanket' /»må‚tå/ [»må‚] 1;3.6(S4)banho 'bath' /»bå¯u/ [»på] 1;3.6(S4)barco 'boat' /»baRku/ [»bQ:] 1;3.6(S4)carro 'car' /»ka{u/ [»ka»ka] 1;4.9(S5)meia 'sock' /»måjå/ [»mE] 1;5.11(S6)banho 'bath' /»bå¯u/ [å»bå] 1;5.11(S6)balde 'bucket' /»ba…dˆ/ [»pa] 1;6.11(S7)cartas 'cards' /»kaRtåS/ [»ka] 1;6. 11(S7)

(186) Inês–laterproductionof/SW/:

Orthogr. Gloss Target Output Ageboa 'good' /»boå/ [»boå] 1;8.2(S9)tampa 'lid' /»tå‚på/ [»pata] 1;8.2(S9)casa 'house' /»kazå/ [»katJå] 1;9.19(S10)banho 'bath' /»bå¯u/ [»baju] 1;9.19(S10)bolos 'cakes' /»boluS/ [»boluS] 1;9.19(S10)

InInês'productions,weobservethat,beforesession9,shedoesnotproducetrochees

target‐like.Untilsession8,thechildcircumscribesthestressedsyllableandusestruncation,

reduplicationandfillerinsertion.Aftersession8,sheisabletocorrectlyproducetrochees.

WeobservetheexactsamepatterninJoana.

(226) Joana–initialproductionof/SW/:

Orthogr. Gloss Target Output AgeCarla 'name' /»kaRlå/ [»ka] 1;8.4(S8)escola 'school' /S»kçlå/ [»kç] 1;8.4(S8)pato 'duck' /»patu/ [pa»pa] 1;9.25(S9)escola 'school' /S»kçlå/ [kç»kç] 1;9.25(S9)sapato 'shoe' /så»patu/ [pa»pa] 1;9.25(S9)luva 'glove' /»luvå/ [»bu:] 1;10.22(S10)Nando 'name' /»nå‚du/ [»¯å‚] 2;0.9(S11)

355

(187) Joana–laterproductionof/SW/:

Orthogr. Gloss Target Output Agemota 'motorbike' /»mçtå/ [»mç:tJå] 2;4.1(S13)fralda 'diaper' /»fRa…då/ [»fawdå] 2;4.1(S13)pedra 'stone' /»pedRå/ [»pEdå] 2;4.1(S13)quadro 'picture' /»kWadRu/ [»kajdJu] 2;6.24(S14)pato 'duck' /»patu/ [»patJu] 2;6.24(S14)

In Joana's speech, truncation and reduplication are observed, in dealingwith /SW/

[+nouns]intheearlystages(untilsession12).Aftersession12,sheisabletoproduce/SW/

[+nouns]target‐like.

Apparently,theseresultsconfirmSantos(2007)as,indeed,atthebeginningchildren

mainlyproducewordswith a [WS] shapeandnomorphological contrasts (gender, person,

tense,number, etc.) arepresent in the children'sproductions.However, it ishard to argue

that morphological constraints, rather than prosodic constraints, are conditioning the

acquisitionofstresspatternsinnon‐verbs,whennoevidencefortheiambicfootisfound.In

thefollowingparagraphs,wewillshowwhytheresultsofthiscomparison,aimingattesting

previousanalyses(Santos,2007)mightbemisleading.

Thehypothesisaccordingtowhichthelateacquisitionoftrocheeswasdependenton,

andrelatedtothepresenceofword‐finalmorphological(extrametrical)informationishardly

falsifiable. Trocheeswithoutwordmarker (/SW/ ‐ e.g., lápis 'pencil' /»lapiS/) are inmuch

lesseramountthantrocheeswithwordmarker(/S+W/‐e.g.,gat]o'cat'/»gatu/,cas]a'house'

/»kazå/). Since the wordmarker is always in unstressed final position and since children

initially circumscribe the stressed syllable of the target word (irrespective of the stress

pattern), it is likely that, during the acquisition of trochees, the preservation of the last

syllable of the stem (which, in general, coincides with the stressed syllable) occurs, along

withthedeletionoftheunstressedsyllable,wherethewordmarkerislocated.Therefore,the

target‐likeproductionoftrocheesandthewordmarkerare,expectedly,highlycorrelated.On

theotherhand,non‐reduplicatediambsshouldbecomparedwithtrochees(withandwithout

wordmarker),inordertoverifywhethermorphologicalorprosodicconstraintswereacting

duringacquisitionofwordstress.Furthermore, it isobservedthat theearlydeletionof the

wordmarkeraloneisnotattested,aschildrendonotproducethewordmarkerbecause, in

fact,theydeletetheentireunstressedsyllableoftargettrochees.

Indeed,theresultsfoundinthischapterindicatethat:

356

(i) at an early stage, trochees in general, and trochees without word marker in

particular,arepronetothesamestrategiesasnon‐reduplicatedtargetiambs(cf.

therenditionsin(227)and(229),below);

(ii) trochees without word marker were prone to the same strategies and

acquisitionpath as trocheeswithwordmarker (cf. the renditions in (228) and

(229),below).

In(227)and(228),weprovidesomeexamplesfromInêsshowingthattrochees(with

and without word marker, respectively) and iambs (in (229)) are prone to the same

productionstrategies,namelyreduplication,epenthesisandtruncation.

(227) Inês–productionof/SW/withwordmarker:

Orthogr. Gloss Target Output AgeMário 'name' /»maRiw/ [´»må] Epenthesischupeta 'pacifier' /Su»petå/ [å»be]chupeta 'pacifier' /Su»petå/ [»pe]

1;1.30(S3)

babete 'bib' /bå»bEtˆ/ [ba»ba˘]/[»bE»bE»bE]Teresa 'name' /tˆ»Rezå/ [ti»tJi]bóia 'buoy' /»bçjå/ [Ba»B´]

Redupl.

banho 'bath' /»bå¯u/ [bå»båba]

1;3.6(S4)

manta 'blanket' /»må‚tå/ [»m˘å‚] 1;3.6(S4)pêlo 'hair' /»pelu/ [»pe] 1;4.9(S5)fralda 'diaper' /»fRa…då/ [»ka˘] 1;5.11(S6)

Truncation

cartas 'cards' /»kaRtåS/ [»ka] 1;6.11(S7)

In this table, we observe that Inês mostly uses reduplication, truncation and filler

insertionattheleft‐edgeofthestressedsyllable,whensheattemptstrochaicwords[+nouns]

withwordmarker.Thesameisobservableintrochaic[+nouns]withoutthewordmarker.

357

(228) Inês–productionof/SW/withoutwordmarker:

Orthogr. Gloss Target Output AgeBambi 'name' /»bå‚bi/ [E»be] 1;3.6(S4)Bambi 'name' /»bå‚bi/ [å»bå]

[ˆ»bå]1;4.9(S5)

Epenthesis

lápis 'pencil' /»lapiS/ [Qba»tu] 1;8.2(S9)Bambi 'name' /»bå‚bi/ [»baba»ba]

[Bå»bå][Bå»bå][pe»be][be»bE]

1;3.6(S4)

Redupl.

Bambi 'name' /»bå‚bi/ [»ba»a»ba] 1;4.9(S5)Bambi 'name' /»bå‚bi/ [»bi]

[»bab]1;3.6(S4)

Bambi 'name' /»bå‚bi/ [»bå] 1;4.9(S5)Bambi 'name'' /»bå‚bi/ [»Bå] 1;6.11(S7)

Truncation

lápis 'pencil' /»lapiS/ [»patH] 1;8.2(S9)

Thechildusesepenthesis, reduplicationandtruncationto[S], todealwithtrochees

without the word marker. Even though, these words are far less frequent in the child's

speech,sheusesthesamestrategies,irrespectiveofthetargetstresspatternattempted.

Iniambicwords,reduplication,truncationandfillerinsertionisalsoobserved.

358

(229) Inês–productionofnon‐reduplicated/WS/:

Orthogr. Gloss Target Output AgeIsabel 'name' /izå»bE…/ [E»bE]/[´»BE] 1;1.30(S3)

Epenthesisbalão 'balloon' /bå»lå‚w‚/ [å»Bå] 1;6.11(S7)chapéu 'hat' /Så»pEw/ [»pa»pa]

[»bE»bE»bE][bå»bQ]

1;1.30(S3)

biberon 'milkbottle' /bibˆ»Ro‚/ [»ƒ´»ƒA] 1;5.11(S6)balão 'balloon' /bå»lå‚w‚/ [bå»bå]/[Bo»Bo]

Redupl.

colher 'spoon' /ku»¥ER/ [kE»kE] 1;6.11(S7)

ali 'there' /å»li/ [»i˘] 1;1.30(S3)chapéu 'hat' /Så»pEw/ [»pHå] 1;1.30(S3)Isabel 'name' /izå»bE…/ [»bE] 1;1.30(S3)João 'name' /Zu»å‚w‚/ [»dJa˘w˘] 1;4.9(S5)champô 'shampoo' /Så‚»po/ [»po] 1;5.11(S6)balão 'balloon' /bå»lå‚w‚/ [»b´] 1;6.11(S7)colher 'spoon' /ku»¥ER/ [»kE] 1;6.11(S7)avô 'grandfather' /å»vo/ [»do] 1;7.2(S8)

Truncation

chapéu 'hat' /Så»pEw/ [»pHE] 1;7.2(S8)

It is worthwhile noticing, also, that the use of the same strategies (epenthesis,

reduplication and truncation) implies the same production patterns, i.e., mainly

monosyllablesand[WS]words.However,thetruncationofnon‐reduplicatediambsprovides

evidenceagainsttheearlyprocessingofaniambicfoot.

The examples in (227)‐(229), alongwith the data presented in Chapter 5, indicate

thatthereisnoclearpreferenceforiambicwords.

Recall that the authors claiming for an early iambic tendency in BP (Santos, 2007;

Stoel‐Gammon,1976)hadintoaccountthechildren'searlyreduplications,whicharemostly

[WS].InChapter5,wehavedemonstratedthatearly[WS]wordsinthespeechofPortuguese

childrendonotnecessarilyindicatetheprocessingandproductionofaniambicfoot.Infact,

wedefended that the Portuguese children's early productions are consistentwith an early

processingofthesyllable,andnotthefoot,since:

(i) both /SW/ and /WS/ words are truncated to [S] (the stressed syllable is

overwhelmingly circumscribed, either in /SW/ with word marker, in /SW/

withoutwordmarkerandin/WS/)‐cf.Chapter5,section5.1.3.2.;

(ii) /WSW/ can be truncated either to /SW/ or /WS/, with a slight tendency for

/SW/(cf.Chapter5,section5.1.3.3.);

359

(iii) when the observed children start producing trochees and iambs, they initially

truncate them, or use reduplication and epenthesis. That is irrespective of the

targetstresspattern(monosyllableswereactuallypronetothesamestrategies).

The emergence of stress patterns in verbs showed that the early verb forms to

consistently emergewere in general trochaic (3rdp.sg. of the Present Indicative), though a

simultaneous emergence of /WS/ verb forms, such as the Infinitive or the 3rdp.sg. of the

SimplePastispossible(inInêsandClara),asshowninTable96,whichwenowrecall:

Emergenceofverbtenses/stresspatterns Strategies

ClaraPres.Ind.(3rdp.sg.)/Imp.(/SW/)~PastPerf.

(3rdp.sg)(/WS/)

/‐SW/:notanalyzed

(insufficientdata)

/‐WS/:notanalyzed

(insufficientdata)

Inês

Pres.Ind.(3rdp.sg.)/Imp.(/SW/)~Infinitive

(/WS/)>>PastPerf.(3rdp.sg.)(/WS/)>>

Past.Imper.(3rdp.sg.)(/WS/)

/­SW/:[S]>>[WS]

(redupl.)

/­WS/:[WS](redupl.)>>

[WS]>>[SW]

JoanaPres.Ind.(3rdp.sg.)/Imp.(/SW/)>>Past

Perf.(3rdp.sg)(/WS/)>>Infinitive(/WS/)

/­SW/:[S]>>[SW]

/­WS/:[S]~[WS]

João

Pres.Ind.(3rdp.sg.)/Imp.(/SW/)>>

Infinitive(/WS/)~Simp.Past(3rdp.sg.)

(/WS/)

/­SW/:[SW]

/­WS/:[S]>>[WS]

LumaPres.Ind.(3rdp.sg.)(/SW/)>>PastPerf.

(3rdp.sg.)(/WS/)>>Infinitive(/WS/)

/­SW/:[S]>>[SW]~[WS]

>>[SW]

/­WS/:[S]>>[WS]>>[SW]

Table96.Summaryfortheemergenceofverbinflectionandstresspatternsinverbs.

Our results on the distribution of word classes per stress pattern (Tables 67‐71)

furthermoreindicatedthat:

360

(i) monosyllabic verbs (dá 'give' /»da/, é 'it is' /»E/, há 'there is' /»a/, quer 'want'

/»kER/)arefrequentinthespeechofPortuguesechildren;

(ii) mono‐ and multisyllabic non‐verbs are frequent in the speech of Portuguese

children;

(iii) di‐ and multisyllabic verbs are selected and emerge later than mono‐, di‐ and

multisyllabicnon‐verbs.

Theseresultssuggestthat:

(i) non‐verbsandverbsmay,indeed,besubjecttodifferentstressalgorithms;

(ii) prosodicandrhythmicconstraints,namely,constraintsontheacquisitionofthe

algorithmforwordstressinverbs,mightbepreventingchildrenfromproducing

verbformswithmorethantwosyllables.

Inourdata,wedidnotfindevidencetoconsiderthatthedefaultfootinEPisaniamb,

neitherthedatashowedthatthewordmarkerisanextrametricalelement.Instead,assoon

aswordprominence isbeingprocessed (at Stage III) andprosodicorganizationwithin the

worddomainemerged,Portuguesechildrenseemtoacknowledgethatatrochaicpatternisat

stake and produce the rightmost syllables of /SW/ non‐verbs, including thewordmarker.

Theexceptionalcharacterofextrametricalityisconfirmed,since/SWW/words,whosefinal

syllable is extrametrical ‐ butnot /SW/or /WSW/withwordmarker ‐ are rarely selected

and,therefore,notacquireduntiltheendoftheobservationperiod(2;6).

Overall results onmorphology suggest that, at the early stages, children select the

stressedsyllableofthetarget‐wordandapplythestrategiesoftruncationtomonosyllables,

reduplication and epenthesis, irrespective of the target stress pattern. In the early stages,

thereisnoevidencetoconsiderthatchildrenareprocessingmorphologicalinformationand

morphologicalcontrasts,sinceearlyproductionsmostlyconsistinthestressedsyllableofthe

targetword,optionallyprecededbysomethingelse(eitherarepetitionofthatsyllable,ora

fillersound).

In sum, our results do not infirm or confirm an initial interaction between

morphologyandwordstressinEP.Instead,thedatafromthischaptersuggestthattheearly

word's representation inEP isnot an iambic foot and that theearlyword's representation

doesnotseemtobedrivenbyanykindofmorphologicalconstraintbut,rather,byprosodic

ones. It is worth mentioning, however, that even though our results disconfirm the early

processing of an iambic foot, they do not totally disconfirm a morphology‐dependence of

wordstressinPortuguese.Inourdata,weobservedthat:

361

(i) Monosyllabicverbsareselectedandproducedfromtheonsetofwordproduction

(cf.Tables67‐71);

(ii) Multisyllabic verbs are not selected and produced until late in development cf.

Tables67‐71);

(iii) Verbs formswhereperson/numberor tense/moodsuffixesarepresent tend to

beacquiredlaterthanverbformsinwhichweobservetheverbthemeonly(cf.

Tables83‐87andexamplesin(205)‐(213)).

Thedifferencesfoundintheacquisitionofdisyllabicnon‐verbsandverbs,ontheone

hand,andthedifferencesfoundintheproductionofmonosyllabicverbsanddisyllabicverbs,

ontheotherhand,suggestthatPortuguesechildrenmightbesensitivetothefactthatnon‐

verbsandverbsaresubjecttoadifferentstressalgorithm.

6.2.2.Discussingweight

Inthissection,wewilldiscussthetargetanalysesarguingforaweight‐sensitiveword

stressalgorithm,basedonthedatafoundinthischapter.Wewillgivespecialattentiontothe

early stages,whenmostly non‐verbswere found in the speech of the Portuguese children

observed. That is due, on the one hand, to the greater frequency of non‐verbs in the early

speechofthePortuguesechildrenobservedand,ontheotherhand,tothecontroversialissue

onthenatureandshapeofthedefaultfootinPortuguesenon‐verbs.

InChapter5,weshowed that, afteran initialperiodwhenchildrenmostlyproduce

monosyllables and a few number of real disyllabic forms (Stage I and II, respectively),

Portuguesechildrenstartproducing[SW]words(while theystill truncate/WS/words).At

thisstage,/WSW/wordsweremainlytruncatedto[SW]aswell.

In thepresentchapterweanalyzed/SW/and/WS/words,having intoaccount the

syllable structure of the stressed and unstressed syllables, in the cases where weight‐

sensitivityshouldbenoticed.Inparticular,weanalyzed:

362

(i) Words with final stressed heavy syllables ‐ /CV.'CVC/ (e.g. amor [å»moR]

'channel');

(ii) Words with final unstressed heavy syllables ‐ /'CV.CVC/ (e.g. lápis [»lapiS]

'pencil');

(iii) Wordswithfinalstressedlightsyllable‐/CV.'CV/(e.g.café[kå»fE]'coffee').

According to Bisol (1992, 1993, 1999) andWetzels (2006), the general, unmarked

ruleforstressassignmentinPortugueseis:stressthepenultimateorthefinalsyllableifitis

heavy.Thisholdstruefornon‐verbs243.AnyBranchingRhyme(orNucleus)countsasheavy.

Themarked rule for stress predicts the presence of a final extrametrical syllable inwords

withantepenultimatestress(e.g.,pássaro'bird'[»pasåRu])orafinalextrametricalconsonant

(Bisol, 1993, 1999) in words with penultimate stress but a final heavy syllable (e.g., lápis

'pencil' [»lapiS]). The stress algorithm based on weight predicts that the domain for word

stressisthelexicalword.

According to Fikkert (1994) for Dutch acquisition, children learned that their

language was quantity‐sensitive when words like /'CV.(C)VC/, with a final extrametrical

syllable, were producedwith level stress (['CV.'(C)VC]). Since extrametricalitywas not yet

acquiredbuttheprosodicwordforthechildrenatthatstagewasadisyllable,theyproduced

the twosyllablesof theword,putting thesameamountofprominence inbothsyllables(in

the heavy and in the penultimate syllable). Furthermore, Dutch children never produced

stress shift in /CV.'(C)VC/ words, showing that they knew that a heavy syllable must be

stressedinthelanguage.

InBP,Bonilha(2005)foundevidenceforweightearlyacquisitionsincetheBrazilian

child observed (i) correctly produced /CV.'CVG/ words from the beginning and (ii) had a

differentacquisitionpathforwordswiththeunmarkedstresspattern(penultimatesyllable

or the finalsyllable if it isheavy)and themarkedstresspattern(penultimatesyllablewith

final heavy syllable or antepenultimate syllable). Bonilha (2005) supported and provided

empirical evidence for the algorithms proposed by Bisol (1992, 1993, 1999) andWetzels

(2006) forword stress acquisition in non‐verbs. According to the authors, the domain for

stress assignment in non‐verbs is the lexical word. Feet are binary and trochaic. The

unmarked stress rule states that one should stress the penultimate syllable of the lexical

243Foradetaileddescriptiononaweight‐basedruleforwordstresscf.Chapter1,section1.2.3.and,inparticular,Wetzels (2006).Verbs are subject to adifferent stress rule,whichwewill not refer to in this discussion, sinceearlywordsinPortuguesechildrenweremostly[+nouns],andtheverbtensesproducedbythechildrendidnotallowforanyfurtherconclusionsonanalgorithmforwordstress.

363

word(/‐'CV(C).CV/)orthepenultimatesyllableifitisheavy(/‐CV.'CVC/244).Inthemarked

cases ‐ antepenultimate stress or penultimate stress with final heavy syllable ‐,

extrametricality was assumed in the final syllable or in the final consonant, respectively.

Bonilha (2005) found that /‐'CV.CV/, /CV.'CVC/ and /CV.'CV/wordswere acquired earlier

than /'CV.CV.CV/ and /'CV.CVC/words, providing evidence for the unmarked character of

penultimatestressandfinalstresswhenthelastsyllableisheavy,andthemarkedcharacter

ofantepenultimatestressandpenultimatestresswhenthefinalsyllableisheavy.

Inthebeginningofthischapter,wehypothesizedthat:

B. Giventheresultsfoundinthepreviouschapterandonpreviousanalysesforthe

childdataandthetargetlanguage(e.g.,Bonilha,2005,Bisol,1993,1999;Wetzels,

2002,2006),wordstressinEPacquisitionreliesonaweight‐basedalgorithm.

Inparticularwepredicted that, ifPortuguese isweight sensitive, it isexpected that

children obey the general weight‐based rule for stress assignment (Bisol, 1992, 1994;

Wetzels,2006):stressthepenultimatesyllableofthelexicalword,orthefinalsyllableifitis

heavy). Evidence for stress‐attraction in heavy syllables would be found, through earlier

acquisition of the unmarked stress pattern (penultimate stress, or final stress if the last

syllable isheavy)anda lateracquisitionofthemarkedstresspattern(penultimatestress if

thefinalsyllableisheavyandantepenultimatestress).

Ouranalysisonweight‐sensitivityinthespeechoffivePortuguesechildrenindicated

that:

(i) Light unstressed syllables are slightly more prone to deletion than heavy

unstressedsyllablesin/‐WS/words(cf.Table97);

(ii) Words with final stress (/CV.'CV/ and /CV.'CVC/) are acquired earlier than

/'CV.CVC/, but they are subject to an unstable behavior until the end of the

observationperiod(cf.Tables98‐109);

(iii) Stressshiftin/CV.'CVC/israre(cf.Tables112,114,117,120and122);

(iv) /SW/wordswith a final heavy syllable (/'CV.CVC/) are rarely selected by the

children(Tables102and105);

244RecallthatBisol(1993)suggeststhat/CV.'CV/wordsbearafinalabstractconsonant,basedonderivedwords‐e.g.,café'coffee'[kå»fE],cafeteira'coffeekettle'[kåfˆ»tåjRå].

364

(v) when/'CV.CVC/wordsareselected,theyareinitiallytruncatedorproducedwith

a final light syllable, whereas the final heavy syllable of /CV.'CVC/ words is

producedasheavyfromthebeginning.

The higher frequency of deletion in unstressed light syllables than in heavy

unstressed syllables, the different acquisition paths for the unmarked stress position

(/CV.'CVC/)andthemarked/'CV.CVC/stresspattern,alongwiththepaucityofcasesofstress

shift in/CV.'CVC/suggestthat, indeed,childrenperceivewhatcountsasaheavysyllable in

their language (i.e., any Branching Rhyme or Nucleus ‐ /VN/, /VG/, /VGN/ and /VC/) and

cruciallystressthosesyllables.

The data from Chapter 5 indicated that, at an initial stage (Stage I), children's

productionsareconsistentwiththeprocessingofasyllable(monosyllabicwordsprevailed,

targetmonosyllablesaremainlyproducedtarget‐likeanddisyllablesaremainly[WS],though

thatistheresultofreduplicationorfillerinsertion‐cf.Figures29‐33).AtStageII,disyllabic

words emerge, though in very reduced amounts and they are still prone to truncation. At

Stage III,weargued for theprocessingofa trochee,since the target‐likeproductionrate in

/SW/ become higher than 50%, and it overcomes the target‐like production rate in /WS/.

Additionally,/WSW/wordsaremostlytruncatedto[SW].

These developmental stages indicate that, as soon as any prosodic and metrical

organizationemergeinthePortuguesechildren'sspeech,thetrochaicfootisbeingprocessed.

However, /SW/ and /WS/ emerge simultaneously (Tables 34‐38) and the acquisition of

/SW/and /WS/is very close in time.This ledus to argue for a neutral start.Weobserved

that, as soon as the trochaic foot (penultimate stress) is available, final stress emerges,

though/WS/wordskeptbeingmorepronetotruncationthan/SW/(cf.,forinstance,João's

truncation patterns in Table 48). This developmental path cannot be accounted for in a

morphology‐based algorithm,which predicts that /SW/words, but not /WS/, have a final

extrametricalwordmarkerwhichwouldbeacquiredlater.

These data suggest that the early prosodic and metrical organization in the

Portuguese children's speech is the product of the application of a general rhythmic and

weight‐basedalgorithmforwordstressinthelanguage:stressthepenultimatesyllableofthe

lexicalwordorthe finalsyllable, if it isheavy.Assoonastrochees(penultimatestress)are

acquired, Portuguese children learn and produce word stress in words with final heavy

syllables,sincethosewordsobeythedefaultstressalgorithm.

Thus,ourfindingsarepartlycomparabletotheonesfoundinBonilha(2005):thereis

adistinctacquisitionpathforunmarked/CV.'CVC/andmarked/'CV.CVC/words.However,

365

contrary to Bonilha (2005), we do not find this pattern since the early stages of word

production,butonlyafterthePortuguesechildrenobservedwereprocessingthetrochee.

On theassumption that childrenare showingweight‐sensitivity,one intriguing fact,

however,isthesimultaneousacquisitionpathfor/CV.'CV/and/CV.'CVC/words,whichcould

provideevidenceagainstweight. Indeed,assumingthatchildrenalreadyknowthataheavy

syllable must be /VN, /VG/ or /VC/, they should distinguish between the two structures

(/CV.'CV/ and /CV.'CVC/), as the former has a light stressed syllable and the latter has a

heavystressedsyllable.

Thealternativeaccounts (i) eitherpostulateaword‐final abstract consonant (Bisol,

1992,1993,1999)or(ii)simplystatethatthosewordsareofforeignorigin(Wetzels,2006).

As for (i), it is true that inmost non‐verbs ending in a stressed light syllable, a consonant

surfaces inderivedcontexts(e.g.café 'coffee'[kå»fE]~cafeteira 'coffeekettle' [kåfˆ»tåjRå]~

cafezal 'coffee field' [kåfE»za…]). It is also true that most of the /CV.'CV/ words have an

indigenous or French origin (as indicated in Wetzels, 2006). However, words like avô

'grandfather' [å»vo], avó 'grandmother' [å»vç] or aqui 'here' [å»ki] have an undeniable

Latin/Portugueseorigin, andwords like rubi 'ruby' [{u»bi] orperu 'turkey/Peru' [pˆ»Ru] do

notderivewordscontaininganyabstract consonant (rubiáceo 'ruby‐colored' [{ubi»asiu]or

peruano'Peruvian'[pˆRu»ånu]?

Acloseranalysisontheproductionofthethreestresspatterns(/CV.'CV/,/CV.'CVC/

and /'CV.CVC/) is required, in order to clarify the heavy character of the final stressed

syllablein/CV.'CV/words.Noticethatinallthereferredstresspatterns,thelastvowelisthe

lastofthestem.Recall,also,that,inourdata,/CV.'CV/and/CV.'CVC/hadasimilarbehavior

and are acquired earlier, whereas /'CV.CVC/ words have a distinct development and are

acquired later. The question one might ask at this point is whether the similarity in the

development found for /CV.'CVC/ and /CV.'CV/ words were due to the light or heavy

characterofthefinalsyllable.

Assumingthat/CV.'CVC/wordsareunmarkedwithrespecttostresspositioninthe

adultandinPortuguesechildlanguage,andthat/'CV.CVC/wordsaremarked(asourresults

further confirmed), the similaritiesbetween/CV.'CVC/and/CV.'CV/words shouldprovide

evidence for children being processing these two structures as similarmetrical structures,

that is, as both having a final heavy (stressed) syllable. Additionally, since children often

produced ['CV.CV] as an output for /'CV.CVC/ words, it was expected that stress shift in

/CV'CV/wordsoccurred.However,thatwasnotthecase.In(230)and(231),weshowsome

productions attested in our data for each stress pattern, in the only two children who

produced/'CV.CVC/words(inInêsandJoana,respectively).

366

(230) Inês–renditionsfor/CV.'CV/,/CV.'CVC/and/'CV.CVC/words:

Orthogr. Gloss Target Output Age/CV.'CV/ café 'coffee' /kå»fE/ [kå»pE] 2;1.10(S10)/CV.'CVC/ chapéu 'hat' /Så»pEw/ [tå»pEw] 2;0.11(S12)/'CV.CVC/ lápis 'pencil' /»lapiS/ [»api]

[»lapi]2;0.11(S12)2;2.1(S14)

(231) Joana–renditionsfor/'CV.CVC/,/CV.'CV/and/CV.'CVC/words:

Orthogr. Gloss Target Output Age/CV.'CVC/ nariz 'nose' /nå»RiS/ [å»RiS] 2;2.19(S12)/CV.'CV/ aqui 'here' /å»ki/ [å»ki] 2;4.1(S13)/'CV.CVC/ pónei 'pony' /»pçnåj/ [»pç¯i] 2;6.24(S14)

As illustrated in (230) and (231), and further shown in section 6.1.2.2., /'CV.CVC/

words are mostly produced as monosyllables, or with a final light syllable but with

penultimate stress (as ['CV.CV], without the final consonant or glide), even though final

stressedsyllablesarealreadyproducedin/CV.'CVC/words.

On the assumption that (i) by the time children select /'CV.CVC/ words, they are

processinga trochaic foot (/SW/wordshadhigher target‐likeproduction rates than/WS/

and /WSW/ are truncated to [SW]) and (ii) syllable‐weight is irrelevant to children, then

stressshift in/CV.'CV/wordswouldbehighlyexpected.Ourdata,however,donotconfirm

thisprediction(cf.Tables111,113,116,119and121).Stressshiftin/CV.'CV/wordsrarely

occurrs in the speech of the five observed children. Moreover, since children produce

/'CV.CVC/wordswithafinallightsyllable,itappearsthatchildrenaremisleadbythestress

placementandproducethefinalsyllableaslight(e.g.,pónei 'pony'/»pçnåj/‐>[»pç¯i]Joana,

2;6;lápispencil'/»lapiS/‐>[»patu],[»lapi],Inês,1;8and2;2,respectively).Ontheassumption

of a morphology‐based algorithm, as soon as children start producing word stress in the

targetpositionin/'CV.CVC/,theyshouldbeabletoproducewordslikepóneiorlápiswitha

finalheavysyllable,as theydo inchapéu 'hat'/Så»pEw/‐>[tå»pEw](Inês,2;0)and[Z´»pEw]

(Joana,2;6).However, in theearly stages, children'sproductionof /'CV.CVC/words, either

aretruncated,orproducedwithafinalunstressedlightsyllable(['CV.CV]).

Our data suggest that children perceive the heavy character of a final syllable in

/CV.'CVC/ words and, consequently, place stress on it. Our results support Bisol's (1993,

1999)andWetzels'(2006)proposals,accordingtowhichtheunmarkedpositionofstressin

367

Portugueseisthepenultimatesyllableofthelexicalword,orthefinalsyllableifitisheavy.On

thisassumption,thedomainforwordstressisthelexicalword.

Assoonasthe(trochaic)footisacquired,childrenareabletocorrectlyassignstress,

astheystressthepenultimatesyllableofthelexicalword([‐'CV.CV]).Atthisstage,binaryfeet

are built left‐headed. /‐WS/ words are possible after children realize that words can be

longerthana footandthatstress fallsonthe lastsyllableof theword, if it isheavy.At this

stage, also, children know that feet in their language are quantity‐sensitive (on the final

syllable)andthatanyBranchingRhymeorNucleus(oralornasal)andnasalNon‐Branching

Nucleuscountsasheavy.

IfPortuguesechildrenwereinsensitivetoweight,wordsofthetype/'CV.CVC/(lápis

'pencil',íris'iris',pónei'poney',móvel'chest',ténis'sneakers',etc.)shouldnotbeproblematic,

especiallywhen/‐SW/wordsarethemostfrequentwordsinthelanguageandtheevidence

for the trochaic foot is attested in the children's productions (namely, through the high

target‐likeproduction rate for/SW/wordsand the truncation for [SW] in/WSW/words).

Also,ifsyllableweightwas'invisible'tochildren,therewouldbenoreasonforthedisparate

acquisitionof/'CV.CVC/words,whencompared to theother (unmarked)structures(/SW/

wordsingeneral,and/WS/).Thoughtheproductionof/CV.'CVC/and/CV.'CV/wassubject

tosomeinstabilityandvariability,aparallelacquisitionpathandtheabsenceofstressshiftof

the type /CV'CV/ ‐> ['CV.CV] suggest that these are, in fact, two instances of the same

structure: /CV.'CVC/. Additionally, if weight was not relevant, /'CV.CVC/ words should be

produced correctly from the beginning and therewould be no reason for their infrequent

(marked)characterinthelanguage.

Our data confirmed themarkednature of /'CV.CVC/ and /'CV.CV.CV/words,which

bearextrametricalinformationand,therefore,areacquiredlaterbyEP‐speakingchildren.

When non‐reduplicated disyllables are selected, no preference for any foot type is

attested./SW/and/WS/wordsareequallyprone to truncationand/WSW/wordscanbe

truncatedto[SW]or[WS],withaslighttendencyfor[SW]).Whenchildrenstartproducing

the trochaic foot, they already know the general word stress algorithm: stress the

penultimatesyllableofthelexicalword(232a.),orthefinalsyllableifitisheavy(232b.),as

showninthefollowingmetricalrepresentationsforthe/SW/wordcasa 'house'[»kazå]and

the/WS/wordbalão'balloon'/bå»lå‚w‚/‐>[m´»¯å‚w‚].

368

(232) Metrical representationof /SW/and/WS/wordsproducedby the childrenwhen

thebasicstressalgorithmisacquired:

*

(* .)

a./»kazå/‐> [»ka tå]

*

. (*)

b./bå»lå‚w‚/‐> [m´ »¯å‚w‚]

The prosodic representation of word prominences for the non‐verbs casa 'house'

/»kazå/ and balão 'balloon' /bå»lå‚w‚/, produced by the children as [»katå] and [m´»¯å‚w‚],

respectively,wouldbetheonepresentedin(233a.)and(233b.).

(233) Prosodicrepresentationof/SW/and/WS/wordsproducedby thechildrenwhen

thebasicstressalgorithmisacquired:

Whenboth/SW/and/WS/wordsarecorrectlyproduced,childrenhavelearnedthe

basic algorithm forword stress in their language, even thoughextrametricality (present in

/'CV.CVC/andin/SWW/words)isnotyetacquired.

Theresultsfoundinthischapterindicatethat,indeed,childrenappeartobesensitive

to syllable weight during the acquisition of words stress and, in particular, to the heavy

characterofthefinalsyllableinlexicalwords.Theasymmetryfoundbetween/CV.'CVC/and

a.e.g.,casa'house'/»kazå/‐>[»katå]

/SW/ω

g

Σ

2

σsσw [ka tå]

b.e.g.,balão'baloon'/bå»lå‚w‚/‐>[m´»¯å‚w‚]

/WS/ω

3

Σ

2

σs(σw)

[m´ »¯å‚w‚]

369

/'CV.CVC/words, attested in the emergence path, in the children's productions and in the

amountoftokensselected,suggestedthat,indeed,finalstressedsyllablesinEPareheavyand

thatthewordstressalgorithmmightbesensitivetosyllableweight.

6.3.Summary

In this chapterwe analyzed twoputative stress‐related aspects in the spontaneous

speechoffivePortuguesechildren:theinteractionwithmorphologyandtheroleofsyllable

weightinwordstressacquisition.

Firstly, we analyzed the distribution of word classes per word shape and stress

pattern in the children's intake. We accounted for the amount of monosyllabic and

multisyllabic (/‐SW/ and /‐WS/) non‐verbs and verbs. The results showed that some

childrenearlierselected/‐WS/words(Clara, InêsandJoana)andothersearlierselected/‐

SW/words(JoãoandLuma).Allchildrenhadgreateramountsofmonosyllabictargets,inthe

early stages. The results further showed an asymmetry between (i) monosyllabic and

multisyllabic verbs and (ii) disyllabic non‐verbs and verbs. Initially, children selected both

verbsandnon‐verbs,butmostlymonosyllables.Multisyllabicnon‐verbswereselected later

and, finally,multisyllabic verbswere selected aswell. The presence ofmonosyllabic verbs

andbothmono‐andmultisyllabicnon‐verbsinthechildren'sintakesuggestedthatprosodic

or rhythmic constraints, rather than morphological ones, were preventing children from

producingmultisyllabic verbs. These data however, indicated that childrenmay indeed be

sensitivetodifferentstressalgorithmsinthenon‐verbandintheverbs'system.

In the following section, we investigated the acquisition of trochees and the word

marker, aiming to test whether morphological information was interacting with the

acquisitionoftrochaicnon‐verbs,andtotestwhethertherewasevidenceforadefaultiambic

foot.Weaccountedfortheacquisitionofthewordmarkerandtheacquisitionoftrocheesand

concludedthattherewasageneralsimultaneousacquisitiononboth,suggestingthat,infact,

the production of a /SW/ stress pattern might be motivated by the acquisition of gender

contrast.However,twochildrenwereabletoproducethewordmarkerbeforetheyacquired

thetrochaicpattern,byrecurringtostressshiftintargettrochees(e.g.,carro'car'/»ka{u/‐>

carró[kA»{o],Luma,2;1;bola'ball'/»bçlå/‐>[bu»la˘],João,1;5).Moreover,trocheeswithout

gender marker, which do not bear morphological information and would be produced as

iambs,wereproducedaccordinglyortheywerepronetothesamestrategiesasbothtrochees

withwordmarkerandiambs.Infact,asobservedinChapter5,muchvariabilityintheearly

stages was attested in the children's production patterns, at the beginning. Target‐like

370

productions, truncation, reduplication and epenthesis were strategies available for the

children,whendealingwithnon‐reduplicated iambs, target trochees,with andwithout the

wordmarker.Therefore,wefoundnoindisputableevidenceforadefaultiambicfoot,neither

atrochaicone,orforaninteractionofmorphologyinwordstressacquisitioninEP.

As for the acquisition of stress patterns in verbs, we found that, in general, when

multisyllabic verb forms were selected and produced – which occurred later than for

multisyllabic no‐verbs andmonosyllabic verbs – children preferred /‐SW/ verb forms and

verb forms that tend to conform to a theme (stem+theme vowel). Inflection markers

(person/number or tense/mood), especially with a /‐WS/ stress pattern, tend to emerge

later in the children's productions. At the early stages, children mainly use the same

strategies used for non‐verbs, irrespective of the target stress pattern (mostly truncations

andreduplications)and,later,theyproducetargetverbformsaccordingly.Theresultsfrom

the observation of the acquisition of verb forms indicated that the early stress patterns

producedinverbsaretrochaicandconformtoaverbtheme.

Inthesecondmainsectionofthischapter,weinvestigatedtheroleofsyllableweight

intheacquisitionofstresspatterninthespeechofthefivechildreninourcorpus.

Weanalyzedthedeletionrateinunstressedlightandheavysyllablesinthespeechof

4ofthechildrenandconcludedthat(i)unstressedlightsyllableswereslightlymoreproneto

deletionthanunstressedheavysyllables;(ii)unstressedheavysyllableswereinmuchlesser

amountsthanunstressedlightsyllables,inthechildren'sintake.

Afterwards, we compared the emergence path for words with stressed and

unstressedheavysyllable(/CV.'CVC/and/'CV.CVC/,respectively),andwordswithstressed

CVfinalsyllables(/CV.'CV/).Inallcases,nomorphologicalconstituentsapartfromthestem

were present (e.g., anel] 'ring' /å»nE…/, lápis] 'pencil' /»lapiS/ and café] 'coffee' /kå»fE/,

respectively).Theresultsfoundindicatedthattherewasadifferenceintheacquisitionpath

for /CV.'CVC/and/'CV.CVC/,noticeableeither in theamountof tokens selectedand in the

production patterns. /CV.'CVC/ words emerged and were acquired much earlier than

/'CV.CVC/.When/'CV.CVC/wordsareselected,theyareinitiallytruncatedorproducedwith

a final unstressed light syllable, contrary to /CV.'CVC/ words, which tend to maintain the

Rhymalstructureofthetarget,i.e.,theirheavycharacter.Thesedatasuggestedthatchildren

were sensitive to the different stress pattern and prosodic structure of /CV.'CVC/ and

/'CV.CVC/words,andtothefactthat,in/'CV.CVC/words,stresswasnotinthefinalsyllable,

which, in that case, should not be heavy. /CV.'CV/ words pursued an emergence and

acquisition path similar to the one observed in /CV.'CVC/words,which lead us to assume

thatthefinalsyllablesofboth/CV.'CVC/and/CV.'CV/wordswerebeingprocessedasheavy

andstressed.Infrequentstressshiftin/CV.'CV/words,inaperiodofchildren'sdevelopment

371

wheretrocheeswereacquired,supportedtheassumptionthatthe finalsyllable in/CV.'CV/

wordswas being processed as heavy. As referred to in section 6.1.2., since no verb forms

were analyzed with respect to syllable weight (mainly due to its paucity and later

productionsinthechildren'sspeech),anyconclusionsastotheroleofsyllableweightduring

theacquisitionofwordstressbornoutfromourdatashouldonlyapplytonon‐verbs.

Overall results in this chapter supported the claim that children are sensitive to

syllableweightduringwordstressacquisitionand that the interactionofmorphologywith

wordstressintheearlystagesisunclear.

372

373

7.Summaryandconclusions

InthisdissertationweinvestigatedtheacquisitionofprimarywordstressinEP.We

aimed to describe the acquisition path undertaken by Portuguese childrenwith respect to

primarywordstress.Specifically,inthisresearchproject,weintendedto:

(i) identify thedevelopmentalpath for theacquisitionof stresspatternsandword

shapeinEP;

(ii) provide empirical evidence for the ongoing discussion on the nature of word

stressinthetargetlanguage;

(iii) compare the acquisition of stress patterns in EPwith the acquisition of stress

patternsandwordshapeinother languages(namelyDutch,EnglishandBP), in

order to contribute for a discussion on the nature of this prosodic aspect in

humanlanguages.

Inthefirstchapter,wedescribedthegeneralpropertiesofwordstress,anddefinedit

asaprosodicpropertyofasubsetoflanguagesoftheworld.Despitebeingalanguage‐specific

aspect,word stress is governedby general principles that are common cross‐linguistically.

Stressisdefinedas"thelinguisticmanifestationoftherhythmicstructure"(Hayes,1995:8).

Instresslanguages,stressplacementispredictableandassignedonthebasisonanalgorithm

(fixed stress) or not (free stress). Stress is, also, a phonological aspect that is related to

rhythmic,prosodicandmetricalpropertiesofthelanguages.

InPortuguese,wordstresshasbeenunderdebateinthelast4decades.Mainly,two

differentanalyseshavebeenpurportedaimingatdescribingandexplaininghowstressworks

inthelanguage.

Ontheonehand,authorsdefendthatprimarywordstressisassignedonthebasisof

morphological information (Lee, 1995; Mateus, 1983; Mateus & Andrade, 2000; Pereira,

1999).Thedefaultrulefornon‐verbspredictsthatstressfallsonthelastsyllableofthestem

(e.g.,gat]o 'cat' [»gatu],calor] 'heat' [kå»loR]). Innon‐verbs the stem is, thus, thedomainof

word stress. In verbs, word stress is assigned on the basis of morphological constituency

alone, with different stress rules applying to different tenses (Present, Past and Future),

374

depending on themorphological constituency of the word. In verbs, the domain for word

stressisthelexicalword.

On the other hand, authors defend that stress is assigned on the basis of syllable

weight (Bisol, 1992,1993,1999;Wetzels, 2006). In this case, thedefault rulepredicts that

stress isonthepenultimatesyllableor inthe finalsyllable, if it isheavy(at the levelof the

Rhyme/VC/, /VNC/ ‐ or at the level of theNucleus ‐ /VN/, /VG/, /VGN/).Thedomain for

wordstressisthelexicalword,bothinnon‐verbsandinverbs.

Besides the conflicting descriptions above‐mentioned, the pertinacity of feet in the

languagephonology is not clear either (Mateuset al., 2003;Vigário, 2003).Apart from the

putativeroleoffeetinstressassignment,thefootisonlyreferredtointwootherprocesses:

the Spondaic and the Dactylic Lowering (Wetzels, 1992, 1995). Neither of these two

processesprovidesevidenceforthetrochaicfoot.

In the secondchapter,we reviewed the literatureon theacquisitionofword stress

and word shape. This literature on the topic focusses on three main aspects: the

representationofearlywords, thedevelopmentofstresspatternsacrossdevelopment,and

theroleoffillersoundsandreduplicationsintheprosodicdevelopment.

InmostGermanic languages (Dutch,EnglishandGerman), thedevelopmentalpaths

found are consistent with an early production of monosyllabic words (Stage 0) (Fikkert,

1994;Demuth,1995,1996;Lleó&Demuth,1999).Infact,Germaniclanguagesareknownto

have a longer stage in which mostly monosyllables are produced (Demuth & Lleó, 1999;

Roark&Demuth,2000).ParticularlyinDutch,alanguagethathasbeenexhaustivelystudied

from the stress acquisition perspective, it was observed that, during the early stages of

disyllabic productions (Stage I), correctly children produce target trochees and mainly

truncate target iambs to monosyllables. At Stage II, children no longer truncate iambs to

monosyllables, and tend to transform iambs into trochees. In later stages (Stage III), the

production of longer words (with more than one foot) is attained, though some cases of

deviant productions are yet attested for words longer than two syllables. Dutch children

acquire weight when they realize that heavy syllables attract stress and produce two

prominentsyllablesinwordswithafinalheavysyllable.Inthelaterstage(StageIV),stress

patterns are fully acquired and words are produced target‐like. The results for Germanic

languages indicate that children are sensitive to the properties of the language they are

acquiring.

DespitethegreatagreementfoundinGermaniclanguages, inthedatafromchildren

speaking other languages, different results are observed. Children acquiring trochaic

languagestendtomirrorthetrochaicrhythmofthetargetsystem(e.g.,Demuth,2001band

Gennari&Demuth,1997, forSpanish;Prieto,2006, forCatalan). InFrench,a languagewith

375

phrase final accent, children's early productions alternate betweenmono‐ and disyllables,

andthereisaheavytendencyforreduplicationswithafinalstressedsyllable(Braud,2003;

Demuth& Johnson, 2003; Rose, 2000;Wauquier‐Gravelines, 2003). InHebrew, a language

withamajorityofiambicfeet,atrochaictendencyisobserved(Adam&Bat‐El,2008,2009).

In Greek, also a trochaic,weight‐insensitive language, no general preference for an earlier

acquisitionofiambsortrocheesisobserved,andagreatvariabilitybetweenchildrenisfound

inthetruncationpatternsofbothtrochaicandiambicwords,suggestinganeutralstartinthe

acquisitionofwordstress(Tzakosta,2004).Fromthiscross‐linguisticanalysis,itispossible

toverifythat,despitesomevariationobserved, inmostcases,childrenmirrortherhythmic

tendency attested in the target language (Hebrew is, probably, the only exception and the

patternfavoredwastrochaic‐Adam&Bat‐El,2008,2009).

Earlyproductionswithfillersoundsattheleft‐edgeofwordsandreduplicationsare

reportedinsomelanguages(namely,English,French,German,Spanish,SwedishandBP).The

nature and role of filler sounds and reduplications, their lexical and prosodic status is,

however,farfrombeingclear.Theliteraturesuggeststhatchildrenmayusefillersoundsand

reduplications for a number of reasons, which are mainly related to the prosodic and

morphosyntactic requirements (e.g., Lleó, 1997, 2001a,b; Peters, 2001b; Santos & Scarpa,

2005;Scarpa,1998,forfillers;e.g.,Demuth&Johnson,2003;Fee&Ingram,1982;Ferguson,

1983;Moskowitz,1973;Wauquier‐Gravelines,2002,forreduplications).Theresultsfoundin

somestudiessuggest,though,thattheyarenormallyproducedinordertomatcheitherfeet,

wordorphrasaltemplates.

Cross‐linguistic investigation on word stress acquisition further suggests that

frequencyinformationandtheprosodicstructureofthelanguage(tendencyforproclitization

orencliticization,minimalityandmaximalityrequirementsandtherhythmictendency)may

provide children with evidence for specific word shapes as well. The frequency of word

shapes found in the language may provide children with evidence for larger or shorter

prosodic words, which children will mirror in the early productions. For instance, in

languageswhere a high number ofmonosyllabicwords are present, children tend to keep

monosyllabicword shapes for longer time (e.g., Catalan and English vs. Spanish ‐ Roark&

Demuth,2000;Prieto,2006).InSpanish,aprocliticlanguage,childrentendtoproducelonger

wordsearlier,duetothelanguage'sprosodicproperties,namelyduetothehighfrequencyof

multisyllabic phrases provided by sequences of determiner+nouns. In French, a language

withnominimalityrequirementsanda languagewithphrase‐finalaccent, inwhichfeetare

irrelevant for prominence, alternating productions between mono‐ and disyllables, a high

frequency of reduplications and a predominance of early iambic words are observed.

376

Therefore, the earlyword shape does not necessarilymatch a foot, and itmight verywell

fulfillawordorphrasetemplate.

TheworksonBPwordstress (e.g.,Bonilha,2005;Santos,2001,2007) suggest that

themajorityofearlywordsinBrazilianchildrenhavea[WS]shape.Santos(2001)describes

word stress acquisition from an intonational perspective and observed an inital [WS]

tendency,mainlyduetophrasalstress,whichisleft‐to‐rightinBP.Thewordstressalgorithm

is only mastered when extrametrical syllables (in /SWW/ words) are produced. Santos

(2007)defends thatBrazilian childrenhave adefault iambic foot and they cannotproduce

andacquiretrocheesduetotheinteractionwithmorphology,sincetrochaic–butnotiambic

– non‐verbs bear a final wordmarker. These results support Lee's (1995) description for

Portuguesewordstress,accordingtowhichwordmarkersareextrametricalandthedefault

foot innon‐verbs is an iamb (e.g., sapát]o 'shoe'; e.g.,colár] 'necklace'). Inverbs, an iambic

tendencyisalsoobservedinBP(Santos,2007),thoughintheadultverbsystem(Lee,1995),a

trochaicdefaultfootisassumed.

Based on the reports and findings in the literature, both on Portuguese and other

languages,weadressedthreemainresearchquestionsinthisdissertation:

(i) HowiswordstressacquiredinPortuguese?

(ii) What can thedevelopmentofword shapeand stresspatterns tell us about the

stressalgorithminPortuguese?

In particular, we aimed at discussing how and when do children acquire primary

word stress, whether children's early words match a default monosyllabic or a disyllabic

template,andwhetherchildren'searlydisyllabicproductionsmatchadefaulttrochaicfoot.

InChapter3,adescriptiononthemethodusedinthisdissertationwasprovided,by

introducingtheinformationaboutthechildren,datacollectionandphonetictranscriptions.

In Chapters 4, 5 and 6, we presented and discussed the results for word stress

acquisitioninEP.

Chapter4consistedinapreliminaryacousticstudybasedonaspeechsampleoftwo

of the children in our corpus (Inês and Joana). In this chapter,we carried out an analysis,

aimingatmeasuring theacousticparameters that are relevant forword stress,both cross‐

linguistically (fundamental frequency, intensity and duration) and in EP (intensity and

duration‐DelgadoMartins,1977,1986,1988).Frequentperceptionoflevelandfinalstress

leadus toanalyze fundamental frequency, intensityandduration in theearlydisyllablesof

twochildren,inordertoidentifywordprominence.Ourresultsconfirmedthat,attheonset

ofwordproduction,anduntilaround2;0,childrenmightnotmastertheacousticparameters

377

used to assign word prominence. Great variability is found in the early stages of word

production, in the use of fundamental frequency, intensity and duration. Variability is also

observedintheproductionoftheacousticparameters,bothbetweenchildrenandwithinthe

samechild.Acloseranalysisondurationand intensity in thechildren'sdisyllablesshowed

that, in fact, Joana uses frequent recursion to level stress and Inês prefers final stress,

althoughcasesof levelstressarealsoattested inbothchildren.Fromthethreeparameters

analyzed,onlydurationin/‐WS/wordsshowsomeconsistency,intheearlyspeechofoneof

the children observed. Inês normally lengthens the second vowel in /‐WS/. Joana shows a

greatearlyvariation inallacousticparameters. In the lastanalyzedsessions,bothchildren

areabletocorrectlyproducethetargetstresspattern.Overallresultsoftheacousticanalysis

suggest that word stress in the early stages of word production might not be mastered

(Pollock,Brammer&Hageman,1993;Kehoe,Stoel‐Gammon&Buder,1995).

In Chapter 5, we conducted an analysis on the word shapes and stress patterns

producedby thechildren,hypothesizingonanearly trochaic tendency inEP.We identified

thechildren'sproductionpatterns,andanalyzed thedata taking intoaccount thedegreeof

faithfulness, and the strategies used by the five children in our corpus. We investigated

monosyllables, disyllables and trisyllables, both in the children's production and in the

children's intake. The results for the production patterns show that, at the onset of word

production,thereisamajorityofmonosyllabicformsandmonosyllableshavehighertarget‐

like production rates. Early [WS] word shapes are frequent as well, and they are

overwhelminglyreduplicatedformsorproductionsresultingfromepenthesisattheleft‐edge

of words. Great variability is found in this early stage, both in the strategies used by the

children(someshowapreferenceformonosyllabicforms,othersusemorereduplicationsor

epentheses)and in the faithfulproductions. Ina secondstage,non‐reduplicateddisyllables

producedtarget‐likeemergeandnopreferencefortrochaicoriambicformsisobserved.Both

targetformscanbeoccasionallyproducedaccordinglyorbetruncated.Atthisstage,faithful

productionsof targettrisyllablesemerge insomechildren'sspeech,buttruncation isalsoa

frequentstrategy.Truncationof/WSW/wordsinstageIIdoesnotfavoratrochaicoriambic

foot.Somechildrenprefertotruncate/WSW/to[SW],othersto[WS]andothersstillprefer

[S].Atathirdstage,targettrocheesaregenerallyproducedtarget‐likeandwithlowerratesof

truncation than target iambs, and/WSW/ trisyllables arepreferably truncated to trochees.

Trisyllablesareonlyacquiredlaterindevelopment,atafourthstage,andonly/WSW/show

anacquisitionpath,since/WWS/and/SWW/havealateemergence.

The results showed that, at the early stages (stage I and II), no general fixedword

shape and stress pattern is observed in the children's speech, suggesting that (i) word

prominenceisnotacquiredandthestressalgorithmisnotmasteredand(ii)theremightbe

378

no prosodic organization within the prosodic word domain. Reduplications and filler

syllables, which are frequent in these stages, are optional elements within the word, and

might actually provide evidence for the stressed syllable being the target of children's

attention.Atthebeginningofspeechproductions,thePortuguesechildrenobservedproduce

onlythestressedsyllableand,optionally,arepetitionofthestressedsyllableorafillersound

at the left‐edge of words. Prosodic organization appears when a trochaic preference is

attested,bothby thehigher ratesof target‐likeproduction in trochees, ahigher truncation

rateintargetiambsandapreferencefortruncationto[SW]intrisyllablesofthetype/WSW/.

Both/WWS/and/SWW/arenotacquireduntil theendof theobservationperiod,/WWS/

having,ingeneral,highertarget‐likeratesthan/SWW/.

Ourhypothesiswaspartlyconfirmedsince,indeed,atrochaictendencywasattested

in the data, though, initially, no word shape was favored categorically, contrary to what

occurs, for instance, in Dutch. The developmental path accounted for in Chapter 5 suggest

that, despite the early differences found between EP and other Germanic languages, EP

behaveslikeothertrochaiclanguages,suchasGreekorSpanish.Indisputableevidenceforan

early iambic foot is not found sincemost non‐reduplicated target iambs are scarce at the

beginningandarepronetotruncation,justliketrochees.Theclaimforanearlyiambicfootin

BP(Santos,2007)mightbebiasedbythetendencyforreduplicationsandepenthesis,which

tend to conform to a [WS] pattern (though not only, asmultiple reduplicationswith level

stressarealsofound).Theseearlyproductionsmightnotbe iambic feetand, like inFrench

(Braud, 2003), they might be used to match prominences assigned from higher prosodic

levels,namelyphrasalstress(Santos,2001;Santos&Scarpa,2005;Vihman,DePaolis&Davis,

1998).Thedata found inChapter5 suggesta syllableprominenceaccount,withaputative

role of phrasal prominence in early word production, and children's sensitivity for the

rhythmicpropertiesofthelanguageduringtheacquisitionofwordstress.

InChapter6,wetestedtheacquisitionofmorphologicalaspectsandsyllableweight

inrelationshipwiththeacquisitionofstresspatterns.Weanalyzedthedistributionofword

classes (non‐verbs and verbs) per word shape (monosyllables, di‐ and trisyllabic words

endingin/‐SW/and/‐WS/).Inrespectwiththenon‐verbparadigm,ourmainpurposewas

totesttheinteractionofmorphologyandwordstress.Weanalyzedtheacquisitionofword

markers and trochees (e.g., casa 'house' /»kazå/, gato 'cat' /»gatu/) and the production

strategies introcheeswherethewordmarker isanemptycategory inthetarget form(e.g.,

lápis 'pencil' /»lapiS/). Moreover, we investigated the acquisition of stress patterns and

inflection in verbs, by accounting for the acquisition of verb tenses andword shapes. The

resultsfromtheinteractionbetweenmorphologyandtheacquisitionofwordstressinnon‐

verbssuggestedthataninteractionbetweenwordstressacquisitionandthemasteryofsome

379

morphologicalaspects(namely,nounandverbinflection)isarguableandhardytestable,as

thePortuguesechildrenobservedtendtodeleteunstressedsyllablesandnotspecificallythe

wordmarker.

Regarding theverbsystem,ourdata indicatedsomesensitivity to thewordclass in

theasymmetricaldistributionofwordclassesperwordshape.Mono‐andmultisyllabicnon‐

verbsarepresentinchildren'searlyintake,butonlymonosyllabicverbsareselectedatthat

moment. The early verb forms produced target‐like conform, in general, to a /SW/ stress

pattern and a verb theme (namely, through the production of the 3rd p.sg. of the

Present/Imperative).Inflectionsuffixesarenotfoundinthechildren'searlyproductionsand

iambic verb forms tend to emerge later in the observed children's speech. Though verbs

emerge and are acquired later than non‐verbs, the order of emergence and acquisition of

stress patterns in non‐verbs and verbs is similar ([S] > [SW],[WS] > [SW]). These results

indicatedthatlongerverbforms,whicharenormallymorecomplex,areacquiredlater.

Generalresultsoftheinteractionbetweenthemasteryofsomemorphologicalaspects

and word stress acquisition suggested that, despite some differences observed in the

emergenceandacquisitiontimings,thedevelopmentalpathforstresspatternsinnon‐verbs

andverbsproceedsverymuchalike:initially,nowordshapeispreferred,laterapreference

fortrochaicwordsismostlyobservedand,finally,iambicverbformsemerge.Also,evidence

formorphological contrasts (either thewordmarker or verb inflection) being produced is

foundatthelaterstagesofwordproduction.

With respect to the potential effect of syllable weight carried on in Chapter 6, we

investigated theemergenceand thedeletionrate inwordswith finalunstressedheavyand

light syllables (/CV.'CVC/ and /'CV.CVC/ words – e.g., amór 'love' and lápis 'pencil',

respectively),andinwordswithafinalstressedlightsyllable(/CV.'CV/words–café'coffee').

TheclassificationinEPontheheavycharacterofthesefinalsyllablesisnotconsensual.Our

analysis in this dissertation mainly aimed at analyzing them under the perspective that

childrenmightprocessthemasheavy,eventhoughadultsmightnot.Inaddition,weanalyzed

the frequencyof stress shift in theseword shapes. Suchwords consisted innon‐verbs and

neither of them contained any word marker or plural suffix. The results showed that

/'CV.CVC/ words are rarely selected by the children and that their developmental path is

different from the one observed in /CV.'CVC/ and /CV.'CV/ words, as the former emerge

much later thanthe latter.Thisasymmetryappears toconfirmtheexceptionalandmarked

characteroftheformer,whencomparedtothelatterand,moreover,itindicatesthat,infact,

Portuguese childrenmight be sensitive to syllableweight in the last syllable, duringword

stressacquisition. Indeed, even though they correctlyproduce the final stressed syllable in

wordslikebalão 'balloon'ouamór 'love',wordslikelápis 'pencil'ornúvem 'cloud'areoften

380

truncated or produced with a final light syllable. Stress shift, however, does not reveal

sensitivity for syllable weight, as this strategy is not frequent in none of the word forms

tested.

Inthisdissertation,wedefendthatPortuguesechildrenseemtoprocesswordstress

on the basis of the rhythmic properties of the target language. The data indicated that

children mirror the trochaic rhythm of the target language, furthermore supporting a

rhythmic word stress algorithm. Despite the variable and neutral productions at the

beginning,trochaicwordstendtobeacquiredearlierthannon‐reduplicatediambs,whichare

pronetotruncationuntillater.

The data did not show, however, robust conclusions as to the interaction of

morphologyduringwordstressacquisition.Childreninitiallydeleteunstressedsyllablesand

notnecessarilywordmarkersinnon‐verbs,andearlyverbsacquiredtendtobetrochaicand

conformtoaverbtheme.Earlyinflectionisnotnoticeable,probablyduetoitsunmasteryat

theonsetofwordproductionbut,morelikely,duetoprosodicconstraintsshapingchildren's

earlywords.However,multisyllabic–butnotmonosyllabic–verbsarenotearlyselectedby

children, even thoughmono‐ andmultisyllabic non‐verbs are produced, at the same stage.

This asymmetry suggests that, indeed, children might be sensitive to a different stress

algorithminverbsandnon‐verbs.

As the resultsof the interactionofmorphology inword stressacquisitionwerenot

conclusive,manyotheraspectswereleftaside,inthisdissertation.Wenowwishtolistthem,

inordertoprovidenewinsightsforfutureresearch.

Onefirstaspectisrelatedtoouroptiontoconsiderspeechproductionsuntil2;6only.

Thisoptionwas formerlybasedonthe importanceofdiscussingthepresumedproblematic

stagesofwordstressacquisitioninEP(theearlystages),andadditionally,onthefactthat3

of the five children in the database were only recorded and transcribed until that age.

However,thisdecisionclearlylimitedouranalysis,namelywithrespecttotheacquisitionof

trisyllables (especially /WWS/ and /SWW/), extrametricality and the full mastery of verb

inflection.Therefore,furtheranalysisonthelaterspeechofPortuguesechildrenisrequired,

inordertofullyaccountforwordstressacquisitioninEP.

Also, frequency data on heavy syllables attracting stress in the target system is

necessary, inorder to empirically support ordefinitely reject the roleof syllableweight in

Portuguese word stress. This information should consider spontaneous adult speech, take

into account both thenumber of types and tokens in the language andput apartwords in

which /‐s/ that does not correspond to the plural marker, as in words like rapaz 'boy'

/{å»paS/ or viés 'bias' /vi»ES/. Afterwards, and given the reduced amount of words with

unstressedheavysyllablesintheobservedchildren'sintake,cross‐sectionalandlongitudinal

381

experimentalstudiesonsyllableweightduringacquisitionarenecessary,inordertoprovide

morerobustevidencefororagainsttheroleofsyllableweight,bothinnon‐verbsandverbs,

duringtheacquisitionofEP.

Finally,one issue that is stronglyrelated towordstressandwhichrequires further

investigation is intonation. It is yet to be deeply studied the role and the nature of the

relationshipbetweenwordstressandintonation,inEPwordstressacquisition(inthelineof

the research conducted in Frota & Vigário, 2008 and Frota &Matos, 2009). This research

would allow, not only for a better understanding on the nature of the apparent iambic

tendency, early reduplications and epentheses, but also it would enable us to more

empirically ground the putative interaction between lower and the higher levels of the

prosodichierarchy. Itwouldfurthermoreshedlightonapossible interactivetop‐downand

bottom‐upprocessingofprosodicacquisition.

Insum,withthisresearchwork,wehopetohaveprovidedanextensivedescription

ontheacquisitionofwordstressinEP,bymeansofdevelopmentalstages.Ontheassumption

that the investigation on language acquisition contributes to build and validate theoretical

models on the representation of language, with this dissertation we hopefully have

contributedwithfurtherempiricalevidencetothediscussiononthelinguisticrepresentation

ofwordstress,inPortugueseandincross‐linguisticacquisition.

383

References

Abdi, H. (2007). Z‐scores. In N. J. Salkind, Encyclopedia and Measurments and Statistics.

ThousandOaks,CA:Sage.

Adam,G.,&Bat‐El,O. (2008).TheTrochaicBias isUniversal: Evidence fromHebrew. InA.

Gavarró,&M. J.Freitas (Ed.),LanguageAcquisitionandDevelopment:Proceedingsof

GALA2007(pp.12‐24).CambridgeScholarsPublishing.

Adam, G., & Bat‐El, O. (2009). When Do Universal Preferences Emerge in Language

Development? The Acquisition of Hebrew Stress. Brill's Annual of Afroasiatic

LanguagesandLinguistics,1,1‐28.

Allen,G.,&Hawkins, S. (1978).Thedevelopmentofphonological rhythm. InA.Bell,& J.B.

Hooper(Eds.),Syllablesandsegments(pp.173‐185).Amsterdam,NorthHolland.

Allen, G., & Hawkins, S. (1979). Trochaic rhythm in children's speech. In H. Hollien, & P.

Hollien, Current issues in the phonetic sciences (pp. 927–933). Amsterdam: John

BenjaminsB.V.

Allen,G.,&Hawkins,S.(1980).Phonologicalrhythm:Definitionanddevelopment.InG.Yeni‐

Komshian,J.Kavanagh,&C.Ferguson(Eds.),ChildPhonology(Vol.1,pp.227‐256).

Andrade,E.(1977).Aspectsdelaphonologie(générative)duportugais.Lisbon:INIC‐CLUL.

Andrade, E. (1988). O acento de palavra em Português. In J. Staczek (Ed.), Colloquium on

Spanish, Portuguese and Catalan Linguistics (pp. 17‐38). Washington: Georgetown

UniversityPress.

Andrade,E.(1992).TemasdeFonologia.Lisboa:Colibri.

Andrade, E. (1997). Some remarks about stress in Portuguese. In F. Martínez‐Gil, & A.

Morales‐Front,IssuesinthePhonologyandMorphologyoftheMajorIberianLanguages

(pp.343‐358).Washingtonn,D.C.:GeorgetownUniversityPress.

Andrade,E.,&Laks,B.(1988).Fonologiamétricaeanálisearitméticadaquantidade.Actasdo

3ºEncontrodaAssociaçãoPortuguesadeLinguística(pp.39‐50).Lisboa:APL.

Andrade,E.,&Laks,B.(1992).Nacristadaonda:oacentodepalavraemportuguês.Actasdo

VII Encontro Nacional da Associação Portuguesa de Linguística (pp. 15‐27). Lisbon:

APL.

Andrade, E., &Rodrigues, C. (1999). Das escolas e das culturas: história de uma sequência

consonântica.ActasdoXIVEncontroNacionaldaAssociaçãoPortuguesadeLinguística.

1,pp.117‐133.APL/Colibri.

384

Andrade,E.,&Viana,M.C.(1989).Aindasobreoacentoeoritmoemportuguês.Actasdo4º

EncontrodaAssociaçãoPortuguesadeLinguística(pp.3‐16).Lisboa:APL.

Araújo, I. (2004).A Percepção do Acento em Português. Descrição, Implicações e Aplicações

para o Ensino do Português como Língua Materna. Lisboa: Master's Dissertation:

UniversidadedeLisboa.

Archibald, J. (1995).TheAcquisitionofStress. In J.Archibald(Ed.),PhonologicalAcquisition

andPhonologicalTheory(pp.81‐109).Hillsdale,N.J.:LawrenceErlbaumAssociates.

Baia,M.F.(2006).Opétrocaicofrenteadadosinfantisdoportuguêsbrasileirocomolíngua

materna.LetrasdeHoje:AnaisdoVIIENAL.PUC‐RS.

Baia, M. F. (2008a). O Modelo Prosódico Inicial do Português Brasileiro: Uma Questão

Metodológica?Master'sDissertation:UniversidadedeSãoPaulo.

Baia,M.F.(2008b).EstudoExperimentalSobreoFormatoProsódicoInicialnaAquisiçãodo

PortuguêsBrasileiro.EstudosLinguísticos,37(2),27‐36.

Baia, M. F. (2008c). O formato prosódico na fala infantil: a produção de criações lexicais,

produções familiares e erros de acento em dados experimenatis e naturalísticos.

AnaisdoXENAPOL,(pp.1‐13).

Baia, M. F. (2008d). O formato prosódico inicial do Português Brasileiro: uma questão

metodológica?ReVEL,6(11).

Barbosa, J.M. (1965/1983).ÉtudesdePhonologiePortugaise (2nded.).Évora:Universidade

deÉvora.

Beckman,M.(1986).StressandNon­StressAccent(Vol.VII).Dordrecht:Foris.

Bernhardt, B., & Stemberger, J. (1998). Handbook of Phonological Development. From the

PerspectiveofConstrained­basedNonlinearPhonology.SanDiego:AcademicPress.

Bisol,L.(1992).Oacentoeopémétricobinário.CadernosdeEstudosLinguísticos,23,83‐101.

Bisol, L. (1993). The stress in Portuguese.Actas doWorkshop sobre Fonologia (pp. 19‐32).

Coimbra:AssociaçãoPortuguesadeFonologia.

Bisol, L. (1999). IntroduçãoaEstudos de Fonologia doPortuguêsBrasileiro (2nd ed.). Porto

Alegre:PUCRS.

Bonilha,G.(2005).AquisiçãoFonológicadoportuguêsbrasileiro:umaabordagemconexionista

daTeoriadaOtimalidade.PhDDissertation:PUC,RioGrandedoSul.

Braud,V.(2003).AcquisitiondelaProsodiechezlesEnfantsFrancophones.LesPhenomènesde

Troncations.PhDDissertation:UniversitédeNantes.

Broselow,E.(1982).Onpredictingtheinteractionofstressandepenthesis.Glossa ,16,115‐

132.

Broselow, E. (1995). Skeletal Positions and Moras. In J. Goldsmith, The Handbook of

PhonologicalTheory(pp.175‐205).Cambridge,Oxford:Blackwell.

385

Carvalho,J.B.(1987).Leportugaisdansladynamiqueibero­romane:troisexemplesdurôlede

la dérive syllabique dans les phénomènes de divergence dialectale. Paris: PhD

Dissertation:UniversitéParisIII.

Carvalho, J. B. (1988). Réduction vocalique, quantité et accentuation: pour une explication

structurale de la divergence entre portugais lusitanien et portugais brésilien. In

BoletimdeFilologia(Vol.XXXII,pp.5‐26).Lisbon.

Chomsky,N.(1986).OConhecimentodaLíngua.SuaNatureza,OrigemeUso.Lisboa:Caminho.

Chomsky,N.,&Halle,M.(1968).TheSoundPatternofEnglish.NewYork:Harper&Row.

Cintra,G.(1997).Distribuiçãodepadrõesacentuaisnovocábuloemportuguês.Confluência,5

(3),83‐92.

Clark,E.(2003).FirstLanguageAcquisition.Cambridge:CambridgeUniversityPress.

Comrie,B.(1981).Languageuniversalsandlinguistictypology.Chicago:UniversityofChicago

Press.

Connelly, M. (1984). Basotho children's acquisition of noun morphology. PhD dissertation:

UniversityofEssex,Cochester.

Costa, J.,&Freitas,M. J. (2001).Morphologicaland/orprosodicplaceholders.Talkgivenat

theMeetingStatisticalPhysics,PatternIdentificationandLanguageChange.Bielefeld.

Costa,T.(inprep.).TheAcquisitionoftheConsonantalSysteminEuropeanPortuguese:Focus

onthePlaceandMannerFeatureson.PhDDissertation:UniversidadedeLisboa.

Cuesta,P.V.,&Luz,M.A.(1971/1988).GramáticadaLínguaPortuguesa.Lisboa:Edições70.

Cunha, C., & Cintra, L. F. (1984).Nova Gramática do Português Contemporâneo (17th ed.).

Lisboa:EdiçõesSádaCosta.

Delattre,P. (1965).Comparingthephonetic featuresofEnglish,French,GermanandSpanish.

Heidelberg:JuliusGroosVerlag.

Delgado‐Martins,M.R.(1977).Aspectsdel'AccentenPortugais.PhDDissertation:Université

deStrasbourg,Strasbourg.

Delgado‐Martins,M. R. (1986). Sept Études sur la Perception. Lisbon: Instituto Nacional de

InvestigaçãoCientífica.

Delgado‐Martins, M. R. (1988). Ouvir Falar. Introdução à Fonética do Português. Lisbon:

Caminho.

Dell, F. (1984). L'accentuation des phrases en Français. In F. Dell, D. Hurst, & J. Vergnaud,

Formesonoredulangage.(pp.65‐122).Paris:Hermann.

Demuth,K. (1995).Markednessand theDevelopmentofProsodic Structure.Proceedingsof

theNorthEasternLinguisticSociety.25,pp.13‐25.Amherst,MA:GLSA,Universityof

Massachusetts.

386

Demuth,K.(1996a).Theprosodicstructureofearlywords.InJ.Morgan,&K.Demuth(Eds.),

SignaltoSyntax:BootstrappingfromSpeechtoGrammarinEarlyAcquisition(pp.171‐

184).Mahwah,N.J.:LawrenceErlbaumAssociates.

Demuth,K. (1996b).Stages in theacquisitionofprosodicstructure.Proceedingsof the27th

ChildLanguageResearchForum,(pp.39‐48).StanfordUniversity:CSLI.

Demuth,K.(1996c).Alignment,StressandParsinginEarlyPhonologicalWords.Proceedings

of the UBC International Conference on Phonological Acquisition (pp. 113‐125).

Somerville,MA:CascadillaPress.

Demuth,K.(2001).Prosodicconstraintsonmorphologicaldevelopment.InJ.Weissenborn,&

J.Höhle,ApproachestoBootstrapping:Phonological,SyntacticandNeurophysiological

AspectsofEarlyLanguageAcquisition.LanguageAcquisitionandLanguageDisorders

Series(Vol.24,pp.3‐21).Amsterdam:JohnBenjamins.

Demuth, K. (2007). Acquisition at the Prosody‐Morphology Interface. In A. Belikova, L.

Meroni, & M. Umeda (Ed.), Proceedings of the 2nd Conference on Generative

Approaches to Language Acquisition in North America (GALANA) (pp. 84‐91).

Somerville,MA:CascadillaPress.

Demuth, K., & Fee, J. (1995).MinimalWords in early phonological development.Ms, Brown

University&DalhousieUniversity.

Demuth,K.,&Johnson,M.(2003).TruncationtosubminimalwordsinearlyFrench.Canadian

JournalofLinguistics,48(3/4),211‐241.

Demuth, K., & Tremblay, A. (2008). Prosodically‐conditioned variability in children's

productionofFrenchdeterminers.JournalofChildLnaguage,35,99‐127.

Demuth,K.,Culbertson,J.,&Alter,J.(2006).Word‐minimality,EpenthesisandCodaLicensing

intheEarlyAcquisitionofEnglish.Language&Speech,49(2),137‐174.

Drachman,G.,&Malikouti‐Drachman,A.(1999).Greekwordaccent.InH.vanderHulst,Word

ProsodicSystemsinthelanguagesofEurope(pp.897‐945).Berlin:MoutondeGruyter.

Fee,J.(1996).SyllableStructureandMinimalWords.InB.Bernhardt,J.Gilbert,&D.Ingram

(Ed.), Proceedings of the UBC International Conference on Phonological Acquisition.

SommervilleMassachusetts:CascadillaPress.

Fee,J.,&Ingram,D.(1982).Reduplicationasastrategyofphonologicaldevelopment.Journal

ofChildLanguage,41‐54.

Ferguson, C. (1978). Historical background on universals research. In J. Greenberg, C.

Ferguson,&E.Moravcsik (Eds.),Universalsofhuman language (pp.7‐31). Stanford:

StanfordUniversity.

Ferguson,C.(1983).Reduplicationinchildphonology.JournalofChildLanguage,10,239‐243.

387

Ferguson,C.,&Macken,M.(1983).TheRoleofPlayinPhonologicalDevelopment.InN.Keith

(Ed.), Children's Language (Vol. 4, pp. 231‐245). New Jersey, London: Lawrence

ErlbaumAssociates.

Fikkert,P.(1994).OnTheAcquisitionofTheProsodicStructure.Dordrecht:HollandInstitute

ofGenerativeLinguistics.

Fikkert, P., & Levelt, C. (2008). How does Place fall into Place? The lexicon and emergent

constraintsinchildren’sdevelopingphonologicalgrammar.InP.Avery,E.Dresher,&

K.Rice,Contrast inPhonology.Theory,Perception,Acquisition. (pp.231‐270).Berlin:

MoutondeGruyter.

Freitas,M.J.(1996).OnsetsinEarlyProductions.InB.Bernhardt,J.Gilbert,&D.Ingram(Ed.),

Proceedingsof theUBC InternationalConferenceonPhonologicalAcquisition (pp.76‐

84).Sommerville:CascadillaPress.

Freitas, M. J. (1997). Aquisição da estrutura silábica do Português Europeu. PhD Thesis:

UniversidadedeLisboa,Lisboa.

Freitas,M. J.,&Miguel,M. (1998).Prosodicand syntactic interaction: theacquisitionofNP

functionalprojectionsinEuropeanPortuguese.ProceedingsoftheConSOLEVI(pp.27‐

44).Leiden:SOLE.

Freitas,M.J.,Miguel,M.,&Faria,I.(2001).InteractionbetweenProsodyandMorphosyntax:

PluralswithincodasintheacquisitionofEuropeanPortuguese.InJ.Weissenborn,&

B. Höhle, Approaches to Bootstrapping. Phonological, lexical, syntactic and

neurophysiological aspects of early language acquisition. (Vol. 2, pp. 45‐57).

Philadelphia,Amsterdam:JohnBenjamin's.

Frota, S. (2000). Prosody and focus in European Portuguese. Phonological phrasing and

intonation.NewYork:GarlandPublishing.

Frota,S.,&Matos,N.(2009).Otemponotempo:umestudododesenvolvimentodasdurações

a partir das primeiras palavras. Actas do XXIV Encontro Nacional da Associação

PortuguesadeLinguística­TextosSeleccionados.Lisbon:Colibri.

Frota, S. & Vigário, M. (2008). Early intonation in European Portuguese. Talk given at the

ThirdConferenceonToneandIntonation(TIE3).LisbonUniversity.

Frota,S.,Vigário,M.,&Martins,F.(2006).FreP:AnElectronicToolforExtractingFrequency

InformationofPhonologicalUnits fromPortugueseWrittenText.Proceedingsof the

5th InternationalConferenceonLanguageResourcesandEvaluation.Genoa.CD‐Rom

Edition.

Gennari,S.,&Demuth,K.(1997).SyllableomissionintheacquisitionofSpanish.Proceedings

ofthe21stAnnualBostonUniversityConferenceonPhonologicalAcquisition(pp.182‐

193).Somerville,MA:CascadillaPress.

388

Gerken,L.‐A.(1990).DoAdultsandChildrenHaveDifferentFeet?InM.Ziolkowski,M.Noske,

& K. Deaton (Ed.), Papers from the 26th Regional Meeting of the Chicago Linguistic

Society. 2. The parasession on the syllable in Phonetics and Phonology, pp. 113‐127.

Chicago:CLS.

Gerken, L. (1994). Ametrical template account of children’sweak syllable omissions from

multisyllabicwords.JournalofChildLanguage,21,565‐584.

Gerken,L.‐A.(1996).ProsodicStructureinYoungChildren'sLanguageProduction.Language,

72(4),683‐712.

Gerken,L.‐A.,&McIntosh,B.(1993).InterplayofFunctionMorphemesandProsodyinEarly

Language.DevelopmentalPsychology,29(3),448‐457.

Gerken, L.‐A., Landau, B., & Remez, R. (1990). Function Morphemes in Young Children's

SpeechPerceptionandProduction.DevelopmentalPsychology,26(2),204‐216.

Goedemans, R., & van derHulst, H. (2008).TheWorld Atlas of Language Structures Online.

Munich: Max Planck Digital Library. (M. Haspelmath, D. Matthew, G. David, & C.

Bernard,Editors)Retrievedjan4,2009,fromhttp://wals.info/feature/17

Goedmans, R., van der Hulst, H., & Visch, E. (1996). Stress Patterns of the World. Part I:

Background.TheHague:HollandAcademicGraphics.

Goffman, L. (1985). Some pragmatic consideration in phonetic and syllabic productions:

Influences,informativenessandstress.WestLafayette,IN:PurdueUniversity.

Goldsmith, J. (1989).Autosegmental and Metrical Phonology. Oxford, Cambridge, MA: Basil

Blackwell.

Gordon,M. (2002).Weight‐by‐positionadjunctionandsyllablestructure.Lingua ,112,901‐

931.

Greenberg,J.(1963).UniversalsofLanguages.Cambridge,Massachussets:MITPress.

Gussenhoven, C. (2004). The Phonology of Tone and Intonation. Cambridge: Cambridge

UniversityPress.

Guttman,L.(1980).Thebasisforscalogramanalysis.InS.Stoufer,L.Guttman,E.Suchman,P.

Lazarsfeld, S. Star, & J. Clausen (Eds.), Studies in Social Psychology inWorldWar II:

Measurement&Prediction(Vol.IV).Princeton,NJ:PrincetonUniversityPress.

Halle, M., & Idsardi, W. (1995). General Properties of Stress and Metrical Structure. In J.

Goldsmith, The Handbook of Phonological Theory (pp. 403‐443). Cambridge,

Massachusetts,Oxford:Blackwell.

Halle,M.,&Vergnaud,J.‐R.(1987).Anessayonstress.Cambridge,MA:MITPress.

Hammond,M.(1991).Parametersofmetricaltheoryandlearnability.InI.Roca,LogicalIssues

inLanguageAcquisititon(pp.47‐62).Dordrecht:Foris.

389

Harris,J.(1991).LaestructurasilábicayelacentoenEspanol.Análisisnolineal.Madrid:Visor

Distribuciones.

Hayes, B. (1995).Metrical Stress Theory. Principles and Case Studies. Chicago/London: The

UniversityofChicagoPress.

Hochberg, J. (1988a). First steps in the acquisition of Spanish stress. Journal of Child

Language,15,273‐292.

Hochberg,J.(1988b).LearningSpanishStress:DevelopmentalandTheoreticalPerspectives.

Language,64(4),683‐706.

Idsardi,W.(1992).TheComputationofProsody.Cambridge,Massachusetts:MIT.

Ingram,D.(1979).PhonologicalPatternsintheSpeechofYoungChildren.InP.Fletcher,&G.

Michael,LanguageAcquisition(pp.133‐148).Cambridge:CambridgeUniversityPress.

Ingram, D. (1992). Early Phonological Acquisition. A Cross‐Linguistic Perspective. In C.

Ferguson,L.Menn,&C.Stoel‐Gammon,PhonologicalDevelopment.Models,Research,

Implications(pp.423‐435).Timonium,Maryland:YorkPress.

Jakobson,R.(1941/1968).Langageenfantinetaphasie.Paris:LesÉditionsdeMinuit.

Jessen,M. (1999). German. InH. vanderHulst,WordProsodic Systems in the Languages of

Europe.(pp.515‐544).Berlin:MoutondeGruyter.

Johnson,J.,&Salidis,J.(1996).Theshapeofearlywords:Aprosodicdevelopmentalanalysis.

Proceedingsofthe20thannualBostonUniversityConferenceonLanguageDevelopment

(pp.386‐396).Sommerville,MA:CascadillaPress.

Jusczyk, P., Cutler, A., & Redanz, N. (1993). Infant's Preference for the Predominant Stress

PatternofEnglishWords.ChildDevelopment,64(3),675‐687.

Kager, R. (1989). A metrical theory of stressing and destressing in English and Dutch.

Dordrecht:Foris.

Kager, R. (1995). The Metrical Theory of Word Stress. In J. Goldsmith, The Handbook of

PhonologicalTheory(pp.367‐402).Cambridge,Massachussets,Oxford:Blackwell.

Kager, R. (2007). Feet andmetrical stress. In P. d. Lacy (Ed.),The Cambridge Handbook of

Phonology(pp.195‐228).Cambridge:CambridgeUniversityPress.

Kato,M.(1996).Raízesnão‐finitasnacriançaeaconstruçãodosujeito.CadernosdeEstudos

Linguísticos,29,119‐136.

Kehoe,M.(1998).Supportformetricalstresstheoryinstressacquisition.ClinicalLinguistics

&Phonetics,12(1),1‐23.

Kehoe,M.(2001).ProsodicPatternsinChildren’sMultisyllabicWordProductions.Language,

SpeechandHearingServicesinSchools,32,284‐194.

Kehoe, M. (2000). Truncation Without Shape Constraints: The Latter Stages of Prosodic

Acquisition.LanguageAcquisition,8(1),23‐67.

390

Kehoe,M., & Stoel‐Gammon, C. (1997). Truncation Patterns in English‐Speaking Children's

WordProduction.JournalofSpeech,LanguageandHearingResearch,40,526‐541.

Kehoe, M., Stoel‐Gammon, C., & Buder, E. (1995). Acoustic correlates of stress in young

children'sspeech.JournalofSpeechandHearingResearch,38,338‐350.

Kenstowicz,M.(1996).Quality‐sensitivestress.RevistadiLinguistica,9(1),157‐187.

Kenstowicz,M.(1994).Sonority­drivenstress.RutgersOptimalityArchive.

Lee, S.‐H. (1995). Morfologia e fonologia lexical do português do Brasil. Dissertatção de

Doutorado:UNICAMP,Campinas.

Lee, S.‐H. (2001). Acento do verbo do Português: uma análise à luz da OT. Actas do XVI

Encontro Nacional da Associação Portuguesa de Linguística. 1, pp. 277‐289. Lisbon:

APL/Colibri.

Lee, S.‐H. (2006). Teoria da Otimalidade e mudança linguística ‐ evolução do acento do

português.SCRIPTA,9(18),45‐61.

Lee, S.‐H. (2007). O acento primário no português: uma análise unificada na teoria da

otimalidade. In G. A. Araújo, O Acento em Português (pp. 121‐143). São Paulo:

Parábola.

Leroy,M.,&Morgenstern,A.(2005).Reduplicationbeforetwoyearsold.InH.Bernhard(Ed.),

StudiesonReduplication(pp.475‐492).Berlin:MoutondeGruyter.

Liberman,M.(1975).TheintonationalsysteminEnglish.MIT:DoctoralDissertation.

Liberman,M.,&Prince,A. (1977).Onstressand linguistic rhythm.Linguistic Inquiry,8 (1),

249‐336.

Lleó,C.(1997).FillerSyllables,Proto‐ArticlesandEarlyProsodicConstraintsinSpanishand

German.LanguageAcquisition:Knowledge,RepresentationandProcessing.Proceedings

ofGala'97,(pp.251‐256).

Lleó,C.(2001a).TheTransitionFromPrenominalFillerstoArticlesinSpanishandGerman.

InM.Almgren,A.Barrena,M.‐J. Ezeizabarrena, I. Idiazabal,&B.MacWhinney (Ed.),

Research on Child Language Acquisition. Preceedings of the 8th Conference of the

InternationalAssociationfortheStudyofChildLanguage.2,pp.713‐737.Sommerville:

CascadillaPress.

Lleó,C.(2001b).TheInterfaceofPhonologyandSyntax:Theemergenceofthearticleinthe

earlyacquisitionofSpanishandGerman.InJ.Weissenborn,&B.Höhle,Approachesto

Bootstrapping. Phonological, lexical, syntactic andneurophysiological aspects of early

languageacquisition(pp.23‐44).Amsterdam,Philadelphia:JohnBenjamin's.

Lleó,C.,&Arias,J.(2007).Foot,wordandphraseconstraintsinfirstlanguageacquisitionof

SpanishStress.InF.Martínez‐Gil,&S.Colina,Optimality­TheoreticStudiesinSpanish

391

Phonology (pp. 470‐496). Philadelphia/Amsterdam: John Benjamins Publishing

Company.

Lleó,C.,&Demuth,K.(1999).ProsodicConstraintsontheEmergenceoftheofGrammatical

Morphemes:CrosslinguisticEvidence fromGermanicandRomanceLanguages. InA.

Greenhill, H. Littlefield, & C. Tano (Ed.), Proceedings of the 23rd Annual Boston

UniversityConferenceonLanguageDevelopment(pp.407‐418).CascadillaPress.

Locke,J.L.(1983).PhonologicalAcquisitionandChange.NewYork:AcademicPress.

MacNeilage, P., &Davis, B. (1993).Motor Explanations of Babbling andEarly Speech. InB.

Boysson‐Bardies, S. Schonen, P. Jusczyk, P. MacNeilage, & J. Morton (Eds.),

Developmental Neurocognition: Speech and face processing in the first year of life.

Dordrecht:KluwerAcademic.

MacNeilage,P.,&Davis,B.(2000).OntheOriginofInternalStructureofWordForms.Science,

288,327‐330.

MacNeilage, P., & Davis, B. (2001). Motor mechanisms in speech ontogeny: phylogenetic,

neurobiological and linguistic implications.Current Opinion inNeurobiology, 11 (6),

696‐700.

Martins,F.,Vigário,M.,&Frota,S.(2004‐2007).FreP‐FrequencyinformationinPortuguese,

v.1.

Mateus,M.H.(1975/1982).AspectosdaFonologiadoPortuguês(2nded.).Lisboa:INIC.

Mateus, M. H. (1983). O acento de palavra em português: uma nova proposta. Boletim de

Filologia,TomoXXVIII,fasc.1­4,1,211‐230.

Mateus,M.H.,&d'Andrade,E.(2000).ThePhonologyofPortuguese.Oxford,NewYork:Oxford

UniversityPress.

Mateus,M.H.,Brito,A.M.,Duarte,I.,Faria,I.,Frota,S.,Matos,G.,etal.(2003).Gramáticada

LínguaPortuguesa.Lisbon:Caminho.

Matzenauer,C.L.(1990)245.Aaquisiçãodafonologiadoportuguês:estabelecimentodepadrões

combaseemtraçosdistintivos.PhDDissertation.PUCRS:PortoAlegre.

McCarthy, J., & Prince, A. (1986). Prosodic Morphology 1986. Manuscript reviewed and

publishedin1996.RutgersOptimalityArchive.

McCarthy,J.,&Prince,A.(1993).ProsodicMorphology.ConstraintInteractionandSatisfaction.

Manuscriptreviwedandpublishedin2001.RutgersOptimalityArchive.

McCarthy, J., & Prince, A. (1994). The emergence of the unmarked: Optimality in Prosodic

Morphology. Proceedings of the North­East Linguistic Society. 24, pp. 333‐379.

Amherst:GraduateLinguisticStudentAssociation.

245Untiltheyearof2000,thisauthorpublishedunderthenameHernandorenaorMatzenauer‐Hernandorena.

392

McCarthy, J., & Prince, A. (1995). Prosodic Morphology. In J. Goldsmith, The Handbook of

PhonologicalTheory(pp.318‐366).Cambridge,Massachusetts,Oxford:Blackwell.

Menn, L., & Stoel‐Gammon, C. (1996). Phonological Development. In P. Fletcher, & B.

MacWhinney, The Handbook of Child Language (pp. 335‐359). Cambridge,

Massachusetts,Oxford.

Moskowitz, A. (1973). Acquisition of phonology and syntax: a preliminary study. In I.

Hintikka, J. Moravcsik, & P. Suppes (Eds.), Approaches to natural language:

proceedings of the 1970 Stanford workshop on grammar and semantics. Dodrecht:

Reidel.

Nespor,M.,&Vogel,I.(1986).ProsodicPhonology.Dordrecht:Foris.

Pater, J. (1997).MinimalViolationandPhonologicalDevelopment.LanguageAcquisition , 6

(3),201‐253.

Pereira, I. (1990).DaProsódia:AnálisedaEvoluçãodoConceitodeProsódiaedasDiferentes

Abordagens Linguísticas das Questões Prosódicas.Master Dissertation: Universidade

deLisboa,Lisbon.

Pereira, I. (1997).Oacento latinoeoacentoemportuguês:dotroqueumoraicoaotroqueu

silábico.ActasdoXIIEncontroNacionaldaAssociaçãoPortuguesadeLinguística.1,pp.

269‐275.Lisboa:APL/Colibri.

Pereira,I.(1999).OAcentodePalavraemPortuguês.UmaAnáliseMétrica.PhDDissertation,

UniversidadedeCoimbra,Coimbra.

Pereira,I.(2007).Acentolatinoeacentoportuguês:queparentesco?InG.A.Araújo,OAcento

emPortuguês.AbordagensFonológicas(pp.61‐83).SãoPaulo:ParábolaEditorial.

Peters, A. (1993). The interdependence of social, cognitive and linguistic development:

Evidence fromavisually impaired child. InTager‐Flusberg,Constraints on language

acquisition:Studiesofatypicalchildren(pp.195‐219).Hillsdale,NJ:LawrenceErlbaum

Associates.

Peters,A.(1995).Strategies intheAcquisitionofSyntax. InP.Fletcher,&MacWhinney,The

HandbookofChildLanguage(pp.462‐482).Oxford:Blackwell.

Peters, A. (1997). Language typology, prosody and the acquisition of grammatical

morphemes.InD.Slobin,TheCrosslinguisticStudyofLanguageAcquisition.Hillsdale,

NJ:LawrenceErlbaumAssociates.

Peters,A.(2001a).FillersyllablesinEarlyLanguageDevelopment.InM.Almgren,A.Barrena,

M.‐J.Ezeizabarrena,I. Idiazabal,&B.MacWhinney(Ed.),ResearchonChildLanguage

Acquisition.Proceedingsof the8thConferenceof the InternationalAssociation for the

StudyofChildLanguage.2,pp.697‐700.Sommerville:CascadillaPress.

393

Peters,A.(2001b).Fillersyllables:Whatistheirstatusinemerginggrammar?JournalofChild

Language,28(1),229‐242.

Peters,A. (2001c).FromProsodytoGrammar inEnglish:Thedifferentiationofcatenatives,

modals,andauxiliariesfromasingleprotomorpheme.InJ.Weissenborn,&B.Höhle,

Approaches to Bootstrapping. Phonological, lexical, syntactic and neurophysiological

aspectsofearly languageacquisition (Vol.2,pp.121‐156).Amsterdam,Philadelphia:

JohnBenjamin's.

Peters, A., & Menn, L. (1993). False starts and filler syllables: ways to learn grammatical

morphemes.Language,69,742‐777.

Plénat, M. (1993). Observations sur le motminimal français. L’oralisation des sigles. In B.

Laks, & M. Plénat, De natura sonorum: Essais de phonologie (pp. 143–172). Saint‐

Denis:PressesUniversitairesdeVincennes.

Pollock,K.E.,Brammer,D.M.,&Hageman,C.(1993).Anacousticanalysisofyoungchildren's

productionofwordstress.JournalofPhonetics,21,183‐203.

Prieto,P. (2006).EarlyProsodicWordAcquisition inCatalan.LanguageandSpeech,49 (2),

231‐259.

Rapp,C. (1994).AElisãodeSílabasFracasnosEstágios IniciaisdaAquisiçãodaFonologiado

Português.DissertaçãodeMestrado:UFB.

Rice, K. (2007). Markedness in phonology. In P. De Lacy, The Cambridge Handbook of

Phonology.(pp.79‐98).Cambridge:CambridgeUniversityPress.

Roark,B.,&Demuth,K.(2000).Prosodicconstraintsandthelearner’senvironment:Acorpus

study. In H. S. Catherine, S. Fish, & T. Keith (Ed.), Proceedings of the 24th Annual

Boston University Conference on Language Development (pp. 597‐608). Somerville,

MA:CascadillaPress.

Roca,I.(1999).StressintheRomanceLanguages.InH.vanderHulst,WordProsodicSystems

in the Languages of Europe (Vol. 4, pp. 659‐812). Berlin/New York: Mouton de

Gruyter.

Rose, Y. (2000).Headedness and Prosodic Licensing in the L1 Acquisition of Phonology. PhD

Dissertation:McGillUniversity,Montreal.

Rose,Y.,&Champdoizeau,C.(2008).ThereisnoInnateTrochaicBias:AcousticEvidencefor

theNeutral StartHypothesis. InA. Gavarró,&M. J. Freitas (Ed.),Proceedings of the

Generative Approaches to Language Acquisition Conference 2007. Cambridge:

CambridgeScholarsPublishing.

Rose, Y., MacWhinney, B., Byrne, R., Hedlund, G., Maddocks, K., O'Brien, P., et al. (2006).

IntroducingPhon:ASoftwareSolutionfortheStudyofPhonologiclAcquisition.InD.

Bamman, T. Magnitskaia, & C. Zaller (Ed.), Proceedings of the 30th Annual Boston

394

University Conference on Language Development (pp. 489‐500). Sommerville, MA:

CascadillaPress.

Santos,R.(1995).UmainterfaceFonologia­Sintaxe:ousode"sonspreenchedores"dacategoria

funcional dos determinantes no processo de aquisição da linguagem. Master's

Dissertation:UNICAMP,Campinas.

Santos,R.(2001).AaquisiçãodoacentodepalavranoPortuguêsBrasileiro.PhDDissertation:

UNICAMP,Campinas.

Santos, R. (2003). Bootstrapping in the acquisition ofword stress in Brazilian Portuguese.

JournalofPortugueseLinguistics,2(1),93‐114.

Santos, R. (2005). Strategies for Word Stress acquisition in Brazilian Portuguese. (M.

Tzakosta, C. Levelt, & J. van de Weijer, Eds.) Developmental Paths in Phonological

Acquisition,SpecialIssueofLeidenPapersinLinguistics,2(1),71‐91.

Santos,R. (2007a).AAquisiçãoProsódicadoPortuguêsBrasileiro de 1 a 3 anos: Padrões de

Palavra e Processos de Sândi Externo. Tese de Livre Docência apresentada junto ao

Dep. de Linguística da Faculdade de Filosofia, Letras e Ciências Humanas da

UniversidadedeSãoPaulo.

Santos, R. (2007b). O acento e a aquisição da linguagem em português brasileiro. In G. A.

Araújo, O Acento em Português: abordagens fonológicas (pp. 225‐258). São Paulo:

Parábolaeditorial.

Santos,R.,&Scarpa,E.(2005).Thephonologicalbootstrappingofdeterminers.Linguisticsin

the Netherlands 2004 (pp. 165‐178). Philadelphia: John Benjamins Publishing

Company.

Scarpa,E.(1998).LearningExternalSandhi:EvidenceforaTop‐downHypothesisofProsodic

Acquisition.InA.Sorace,C.Heycock,&R.Shellcock(Ed.),ProceedingsoftheGALA'97

ConferenceonLanguageAcquisition(pp.272‐277).Edimburgh:EdinburghUniversity

Press.

Scarpa,E.(1999).Sonspreenchedoreseguardadoresdelugar:relaçõesentrefatossintáticos

efatosprosódicos.(E.Scarpa,Ed.)EstudosdeProsódia,253‐284.

Schwartz, R., & Leonard, L. (1983). Some Further Comments on Reduplication in Child

Phonology.JournalofChildLanguage,10,441‐448.

Schwartz,R.,Leonard,L.,Wilcox,M.,&Folger,M.(1980).AgainandAgain:Reduplicationin

ChildPhonology.JournalofChildLanguage,7,75‐87.

Selkirk,E.(1980).Prosodicdomains inphonology:Sanskritrevisited. InM.Aronoff,&M.‐L.

Kean(Eds.),Juncture(pp.107‐129).Saratoga,CA:AnmaLibri.

Simonsen,H. (2001).Phonological fillers:Data fromNorwegian. InM.Almgren,A.Barrena,

M.‐J.Ezeizabarrena,I. Idiazabal,&B.MacWhinney(Ed.),ResearchonChildLanguage

395

Acquisition. Proceeding of the 8th Conference of the International Association for the

StudyofChildLanguage.2,pp.701‐712.Sommerville:CascadillaPress.

Smith, N. (1973). The Acquisition of Phonology: A Case Study. Cambridge: Cambridge

UniversityPress.

Snow,D. (1998).Aprominence accountof syllable reduction in early speechdevelopment:

Thechild'sprosodicphonologyof"tiger"and"giraffe".JournalofSpeechandHearing

Research,41,1171‐1184.

Stampe,D.(1979).ADissertationonNaturalPhonology.NewYork:Garland.

Stoel‐Gammon,C.(1976).BabyTalkinBrazilianPortuguese.RevistaBrasileiradeLinguística,

3(1),22‐26.

Trubetzkoy,N.(1939).GrundzügederPhonologie.Prague:VandenhoeckandRuprecht.

Tzakosta, M. (2004). Multiple Parallel Grammars in the Acquisition of Stress in Greek L1.

Utrecht:LOT.

vandeVijver,R.(1998).Theiambicissue(Vol.37).Amsterdam:HILDissertation.

van der Hulst, H. (1999).Word Prosodic Systems in the Languages of Europe (Empirical

ApproachestoLanguageTypology/Eurotyp).Berlin/NewYork:MoutondeGruyter.

Veneziano, E. (2001). A System‐Approach to the Analysis of Fillers at the Transition to

Grammar. In M. Almgren, A. Barrena, M.‐J. Ezeizabarrena, I. Idiazabal, & B.

MacWhinney (Ed.), Research on Child Language Acquisition. Preceedings of the 8th

ConferenceoftheInternationalAssociationfortheStudyofChildLanguage.2,pp.738‐

759.Sommerville:CascadillaPress.

Veneziano,E.,&Sinclair,H. (1997).Fromthesurface inward:adiscontinuouscontinuity in

theemergenceofgrammaticalmorphemes.ArchivesdePsychologie,65,107‐116.

Veneziano, E., & Sinclair, H. (2000). The changing status of 'filler syllables' on the way to

grammaticalmorphemes.JournalofChildLanguage,27,1‐40.

Veneziano,E.,Sinclair,H.,&Berthoud,I.(1990).Fromonetotwowords:Repetitionpatterns

onthewaytostructuredspeech.JournalofChildLanguage,17,633‐650.

Vigário,M.(2003).TheProsodicWordinEuropeanPortuguese.Berlin/NewYork:Moutonde

Gruyter.

Vigário,M.,Freitas,M.J.,&Frota,S.(2006).Grammarandfrequencyeffectsintheacquisition

ofprosodicwordsinEuropeanPortuguese.LanguageandSpeech,49(2),175‐203.

Vigário,M.,Martins,F.,&Frota,S.(2006).AferramentaFrePeafrequênciadetipossilábicos

e classesde segmentosnoPortuguês.ActasdoXXIEncontroNacionaldaAssociação

PortuguesadeLinguística.Lisbon:APL/Colibri.

396

Vihman,M.M. (1992).EarlySyllablesandtheConstructionofPhonology. InC.Ferguson,L.

Menn,&C. Stoel‐Gammon,PhonologicalDevelopment.Models,Research, Implications

(pp.393‐422).Timonium,Maryland:YorkPress.

Vihman, M. (1996). Phonological Development. The Origins of Language in the Child.

Cambridge,Massachussetts:Blackwell.

Vihman,M.,&McCune,L.(1994).Whenisawordaword?JournalofChildLanguage,21,517‐

542.

Vihman,M.,DePaolis,R.,&Davis,B.(1998).Istherea"trochaicbias"inearlywordlearning?

Evidence from infant production in English and French. Child Development, 69 (4),

935‐949.

Villalva,A.(2008).MorfologiadoPortuguês.Lisboa:UniversidadeAberta.

Waterson,N.(1971).Childphonology:Aprosodicview.JournalofLinguistics,7,179‐211.

Wauquier‐Gravelines, S. (2003). Troncation et reduplication. Peut‐on parler de gabarits

morphologiques dans le lexique précoce? Les unités morphologiques. Silexicales 3.

UniversitédeLilleIII.

Weber,C.,Hahne,A.,Friedrich,M.,&Friederici,A. (2004).Discriminationofwordstress in

early infant perception: electrophysiological evidence.Cognitive Brain Research, 18,

pp.149‐161.

Weissenborn, J., & Höhle, B. (2001). Approaches to bootstrapping. Phonological, lexical,

syntactic and neurophysiological aspects of early language acquisition. (Vol. 1).

Philadelphia,Amsterdam:JohnBenjamin's.

Wetzels, L. (1992). Mid Vowel Neutralization in Brazilian Portuguese. Caderno de Estudos

Linguísticos,23,19‐55.

Wetzels,L.(1995).Mid‐vowelAlternations intheBrazilianPortugueseVerb.Phonology ,12

(2),281‐304.

Wetzels,L.(2003).OntheWeightIssueinPortuguese,aTypologicalInvestigation.Letrasde

Hoje,38(4),107‐133.

Wetzels,L.(2006).PrimaryWordStressinBrazilianPortugueseandtheWeightParameter.

(G.Elordieta,&M.Vigário,Eds.)JournalofPortugueseLinguistics,5(2),9‐58.

Wetzels,L.(1997).TheLexicalRepresentationofNasalityinBrazilianPortuguese.Probus,9,

203‐232.

Wiese,R.(1996).ThePhonologyofGerman.Oxford:ClarendonPress.

Zec,D. (2007).Thesyllable. InP.deLacy,TheCambridgeHandbookofPhonology (pp.161‐

194).Cambridge:CambridgeUniversityPress.

397

I

Appendices

II

III

AppendixA

RegularverbinflectioninPortugueseandstresspatterns

Cantar'tosing'=Cant]Root+a]TV+r]TM(TV/a/‐1stconjugation)

Indicative

PresentTenses PastTenses FutureTenses

Present PastPerfect Imperfect Pluperfect Future

CANto(SW)

CANtas(SW)

CANta(SW)

canTAmos(SW)

canTAIS(WS)

CANtam(SW)

canTEI(WS)

canTASte(SW)

canTOU(WS)

canTÁmos(SW)

canTAStes(SW)

canTAram(SW)

canTAva(SW)

canTAvas(SW)

canTAva(SW)

canTÁvamos(SWW)

canTÁveis(SW)

canTAvam(SW)

canTAra(SW)

canTAras(SW)

canTAra(SW)

canTÁramos(SWW)

canTAreis(SW)

canTAram(SW)

cantaREI(WS)

cantaRÁS(WS)

cantaRÁ(WS)

cantaREmos(SW)

cantaREIS(WS)

cantaRÃO(WS)

Infinitive PastParticiple Conditional

canTAR(WS) canTAdo(SW)

Imperative

CANta(SW)

canTAmos(SW)

Gerund

canTANdo(SW)

cantaRIa(SW)

cantaRIas(SW)

cantaRIa(SW)

cantaRIamos(SWW)

cantaRIeis(SW)

cantaRIam(SW)

Subjunctive

PresentTenses PastTenses

Present Imperfect

FAle(SW)

FAles(SW)

FAle(SW)

faLEmos(SW)

faLEIS(WS)

FAlem(SW)

faLAsse(SW)

faLAsses(SW)

faLAsse(SW)

faLÁssemos(SWW)

faLÁsseis(SW)

faLAssem(SW)

IV

Comer'toeat'=com]Root+e]TV+r]TM(TV/e/‐2ndconjugation)

Indicative

PresentTenses PastTenses FutureTenses

Present PastPerfect Imperfect Pluperfect Future

COmo(SW)

COmes(SW)

COme(SW)

coMEmos(SW)

coMEIS(WS)

COmem(SW)

coMI(WS)

coMESte(SW)

coMEU(WS)

coMEmos

(SW)

coMEStes

(SW)

coMEram

(SW)

coMIa(SW)

coMIas(SW)

coMIa(SW)

coMÍamos

(SWW)

coMÍeis(SW)

coMIam(SW)

coMEra(SW)

coMEras(SW)

coMEra(SW)

coMEramos(SWW)

coMEreis(SW)

coMEram(SW)

comeREI(WS)

comeRÁS(WS)

comeRÁ(WS)

comeREmos

(SW)

comeREIS(WS)

comeRÃO(WS)

Infinitive PastParticiple Conditional

coMER(WS) coMIdo(SW)

Imperative

COme(SW)

coMAmos(SW)

Gerund

coMENdo(SW)

comeRIa(SW)

comeRIas(SW)

cantaRIa(SW)

comeRIamos(SWW)

comeRIeis(SW)

comeRIam(SW)

Subjunctive

PresentTenses PastTenses

Present Imperfect

COma(SW)

COmas(SW)

COma(SW)

coMAmos(SW)

coMEIS(WS)

COmam(SW)

coMEsse(SW)

coMEsses(SW)

coMEsse(SW)

coMÊssemos(SWW)

coMÊsseis(SW)

coMEssem(SW)

V

Dormir'tosleep'=dorm]Root+i]TV+r]TM(TV/1/‐3rdconjugation)

Indicative

PresentTenses PastTenses FutureTenses

Present PastPerfect Imperfect Pluperfect Future

DURmo(SW)

DORmes(SW)

DORme(SW)

dorMImos(SW)

dorMIS(WS)

DORmem(SW)

dorMI(WS)

dorMISte(SW)

dorMIU(WS)

dorMImos(SW)

dorMIStes(SW)

dorMIram(SW)

dorMIa(SW)

dorMIas(SW)

dorMIa(SW)

dorMÍamos(SWW)

dorMÍeis(SW)

dorMIam(SW)

dorMIra(SW)

dorMIras(SW)

dorMIra(SW)

dorMÍramos(SWW)

dorMIreis(SW)

dorMIram(SW)

dormiREI(WS)

dormiRÁS(WS)

dormiRÁ(WS)

dormiREmos(SW)

dormiREIS(WS)

dormiRÃO(WS)

Infinitive PastParticiple Conditional

dorMIR(WS) dorMIdo(SW)

Imperative

DORme(SW)

dorMImos(SW)

Gerund

dorMINdo(SW)

dormiRIa(SW)

dormiRIas(SW)

dormiRIa(SW)

dormiRIamos(SWW)

dormiRIeis(SW)

dormiRIam(SW)

Subjunctive

PresentTenses PastTenses

Present Imperfect

DURma(SW)

DURmas(SW)

DURma(SW)

durMAmos(SW)

durMEIS(WS)

DURmam(SW)

durMIsse(SW)

durMIsses(SW)

durMIsse(SW)

durMÍssemos(SWW)

durMIsseis(SW)

durMIssem(SW)

VI

VII

AppendixB

Monosyllables

(faithfullproductionandstrategies)

Clara

/S/ [S] % [SW] %[SW] [WS] %[CV1CV1] [FS] %[fS] [CVCV...] %[CVCV...] %TOTALSession1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session2 1 0 0 0 0 0 0 1 100 0 0 100Session3 4 4 100.0 0 0 0 0 0 0 0 0 100Session4 7 6 85.7 0 0 1 14.3 0 0 0 0 100Session5 7 5 71.4 0 0 1 14.3 1 14.3 0 0 100Session6 13 6 46.2 1 7.7 0 0 5 38.5 1 7.7 100Session7 27 20 74.1 3 11.1 0 0 2 7.4 2 7.4 100Session8 17 11 64.7 5 29.4 0 0 0 0 1 5.9 100Session9 22 15 68.2 2 9.1 2 9.1 1 4.5 2 9.1 100Session10 34 32 94.1 1 2.9 1 2.9 0 0 0 0 100Session11 100 86 86.0 4 4.0 0 0 6 6.0 4 4.0 100Session12 87 82 94.3 2 2.3 3 3.4 0 0 0 0 100

VIII

Inês

/S/ [S] % [SW] %[SW] [WS] %[CV1CV1] [FS] %[fS] [CVCV...] %[CVCV...] %TOTALSession1 1 0 0 0 0 0 0 0 0 1 100 100Session2 22 13 59.1 0 0 0 0 6 27.3 3 13.6 100Session3 68 25 36.8 2 2.9 2 2.9 5 7.4 34 50.0 100Session4 112 45 40.2 1 0.9 15 13.4 3 2.7 48 42.9 100Session5 39 27 69.2 0 0 1 2.6 8 20.5 3 7.7 100Session6 129 81 62.8 1 0.8 4 3.1 25 19.4 18 14.0 100Session7 80 54 67.5 0 0 3 3.8 8 10.0 15 18.8 100Session8 89 64 71.9 13 14.6 1 1.1 3 3.4 8 9.0 100Session9 97 86 88.7 4 4.1 3 3.1 1 1.0 3 3.1 100Session10 136 119 87.5 8 5.9 1 0.7 8 5.9 0 0 100Session11 236 200 84.7 22 9.3 0 0 5 2.1 9 3.8 100Session12 410 374 91.2 18 4.4 1 0.2 0 0 17 4.1 100Session13 259 239 92.3 8 3.1 1 0.4 1 0.4 10 3.9 100Session14 316 288 91.1 11 3.5 2 0.6 1 0.3 14 4.4 100Session15 226 211 93.4 4 1.8 0 0 0 0 11 4.9 100Session16 454 416 91.6 13 2.9 0 0 2 0.4 23 5.1 100Session17 236 218 92.4 0 0.0 0 0 0 0 18 7.6 100Session18 428 358 83.6 21 4.9 1 0.2 0 0 48 11.2 100

IX

Joana

/S/ [S] % [SW] %[SW] [WS] %[CV1CV1] [FS] %[fS] [CVCV...] %[CVCV...] %TOTALSession1 2 2 100 0 0 0 0 0 0 0 0 100Session2 1 1 100 0 0 0 0 0 0 0 0 100Session3 1 1 100 0 0 0 0 0 0 0 0 100Session4 8 8 100 0 0 0 0 0 0 0 0 100Session5 3 3 100 0 0 0 0 0 0 0 0 100Session6 3 2 66.7 0 0 0 0 0 0 1 33.3 100Session7 7 7 100 0 0 0 0 0 0 0 0.0 100Session8 8 7 87.5 0 0 1 12.5 0 0 0 0 100Session9 21 18 85.7 2 9.5 1 4.8 0 0 0 0 100Session10 62 52 83.9 6 9.7 0 0 1 1.6 3 5.0 100Session11 57 48 84.2 2 3.5 1 1.8 4 7.0 2 4.0 100Session12 59 51 86.4 4 6.8 1 1.7 1 1.7 2 3.0 100Session13 55 49 89.1 0 0 1 1.8 1 1.8 4 7.0 100Session14 229 171 74.7 0 0 0 0 5 2.2 53 23.0 100

X

João

/S/ [S] % [SW] %[SW] [WS] %[CV1CV1] [FS] %[fS] [CVCV...] %[CVCV...] %TOTALSession1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session2 4 1 25.0 0 0 2 50.0 1 25.0 0 0 100Session3 1 0 0 0 0 0 0 1 100 0 0 100Session4 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session5 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session6 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session7 3 2 66.7 1 33.3 0 0 0 0 0 0 100Session8 1 1 100 0 0 0 0 0 0 0 0 100Session9 26 16 61.5 5 19.2 1 3.8 0 0 4 15.4 100Session10 13 10 76.9 1 7.7 2 15.4 0 0 0 0 100Session11 26 12 46.2 0 0 12 46.2 1 3.8 1 3.8 100Session12 39 13 33.3 0 0 21 53.8 1 2.6 4 10.3 100Session13 18 15 83.3 0 0 2 11.1 1 5.6 0 0 100Session14 12 10 83.3 0 0 1 8.3 1 8.3 0 0 100Session15 29 27 93.1 0 0 1 3.4 1 3.4 0 0 100Session16 41 35 85.4 2 4.9 0 0 2 4.9 2 4.9 100Session17 72 55 76.4 7 9.7 2 2.8 1 1.4 7 9.7 100Session18 31 25 80.6 3 9.7 3 9.7 0 0 0 0 100Session19 68 62 91.2 0 0 3 4.4 3 4.4 0 0.0 100Session20 52 48 92.3 3 5.8 0 0 0 0 1 1.9 100Session21 33 26 78.8 5 15.2 0 0 1 3.0 1 3.0 100Session22 71 63 88.7 4 5.6 1 1.4 2 2.8 1 1.4 100

XI

Luma

/S/ [S] % [SW] %[SW] [WS] %[CV1CV1] [FS] %[fS] [CVCV...] %[CVCV...] TOTALSession1 1 0 0 0 0 0 0 0 0 1 100 100Session2 2 1 50.0 0 0 0 0 1 50.0 0 0 100Session4 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session6 13 12 92.3 0 0 0 0 0 0 1 8 100Session8 36 12 33.3 0 0 2 5.6 11 30.6 11 31 100Session10 1 1 100 0 0 0 0 0 0 0 0 100Session12 5 3 60.0 0 0 2 40 0 0 0 0 100Session14 2 2 100 0 0 0 0 0 0 0 0 100Session16 7 2 28.6 0 0 4 57.1 0 0 1 14 100Session18 7 5 71.4 0 0 2 28.6 0 0 0 0 100Session19 22 11 50.0 0 0 7 31.8 2 9.1 2 9 100Session20 14 8 57.1 0 0 2 14.3 1 7.1 3 21 100Session21 7 2 28.6 0 0 3 42.9 2 28.6 0 0 100Session23 3 2 66.7 0 0 1 33.3 0 0 0 0 100Session25 14 10 71.4 0 0 3 21.4 0 0 1 7 100Session27 53 51 96.2 0 0 1 1.9 0 0 1 2 100Session29 89 85 95.5 0 0 3 3.4 0 0 1 1 100Session30 149 136 91.3 0 0 1 0.7 5 3.4 7 5 100Session32 260 227 87.3 11 4.2 4 1.5 5 1.9 13 5 100Session34 168 144 85.7 20 11.9 0 0 2 1.2 2 1 100Session36 183 162 88.5 13 7.1 4 2.2 0 0 4 2 100

XII

XIII

AppendixC

Truncationintrisyllables

Clara

/WSW/ [WSW] %[WSW] [SW] %[SW] [WS] %[WS] Other

Session1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session2 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session3 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session4 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session5 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session6 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session7 3 0 0 1 33.3 0 0 2Session8 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session9 4 2 50.0 2 50.0 0 0 0Session10 11 4 36.4 1 9.1 0 0 6Session11 21 12 57.1 4 19.0 0 0 5Session12 22 8 36.4 8 36.4 1 4.5 5

/WWS/ [WWS] %[WWS] [WS] %[WS] Other

Session1 ‐ ‐ ‐ ‐ ‐ ‐Session2 ‐ ‐ ‐ ‐ ‐ ‐Session3 ‐ ‐ ‐ ‐ ‐ ‐Session4 ‐ ‐ ‐ ‐ ‐ ‐Session5 ‐ ‐ ‐ ‐ ‐ ‐Session6 ‐ ‐ ‐ ‐ ‐ ‐Session7 ‐ ‐ ‐ ‐ ‐ ‐Session8 ‐ ‐ ‐ ‐ ‐ ‐Session9 ‐ ‐ ‐ ‐ ‐ ‐Session10 ‐ ‐ ‐ ‐ ‐ ‐Session11 1 0 0 1 100 0Session12 5 0 0 4 80 1

XIV

Inês

/WSW/ [WSW] %[WSW] [SW] %[SW] [WS] %[WS] [S] %[S] Other

Session1 1 0 0 0 0 1 100 0 0 0Session2 1 0 0 0 0 0 0 0 0 1Session3 12 0 0 1 8.3 0 0 6 50.0 5Session4 11 0 0 0 0 3 27.3 3 27.3 5Session5 13 0 0 3 23.1 1 7.7 7 53.8 3Session6 33 1 3.0 0 0 16 48.5 8 24.2 8Session7 35 1 2.9 0 0 16 45.7 4 11.4 14Session8 33 7 21.2 4 12.1 5 15.2 6 18.2 11Session9 28 7 25.0 8 28.6 4 14.3 1 3.6 8Session10 77 33 42.9 9 11.7 3 3.9 23 29.9 9Session11 85 49 57.6 12 14.1 3 3.5 7 8.2 14Session12 112 74 66.1 19 17.0 9 8.0 7 6.3 3Session13 125 78 62.4 12 9.6 9 7.2 1 0.8 25Session14 75 53 70.7 11 14.7 2 2.7 3 4.0 6Session15 112 77 68.8 5 4.5 4 3.6 1 0.9 25Session16 106 54 50.9 27 25.5 21 19.8 0 0.0 4Session17 69 37 53.6 12 17.4 11 15.9 4 5.8 5Session18 79 45 57.0 11 13.9 12 15.2 4 5.1 7

/WWS/ [WWS] %[WWS] [S] %[S] [WS] %[WS] Other

Session1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session2 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session3 16 0 0 16 100 0 0 0Session4 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session5 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session6 1 0 0 0 0 0 1Session7 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session8 2 0 0 0 0 1 50.0 1Session9 4 1 25.0 0 0 3 75.0 0Session10 2 0 0 0 0 2 100 0Session11 11 2 18.2 2 18.2 6 54.5 1Session12 6 1 16.7 0 0 5 83.3 0Session13 22 9 40.9 0 0 6 27.3 7Session14 14 7 50.0 0 0 0 0 7Session15 17 8 47.1 1 5.9 3 17.6 5Session16 12 6 50.0 0 0 2 16.7 4Session17 12 1 8.3 0 0 6 50.0 5Session18 14 5 35.7 0 0 3 21.4 6

XV

/SWW/ [SWW] %[SWW] [S] %[S] [SW] %[SW] Other

Session1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session2 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session3 7 0 0 4 57.1 0 0 3Session4 1 0 0 0 0 0 0 1Session5 2 0 0 1 50.0 0 0 1Session6 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session7 21 0 0 15 71.4 0 0 6Session8 3 0 0 1 33.3 0 0 2Session9 1 0 0 0 0 0 0 1Session10 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session11 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session12 7 2 28.5 1 14.3 3 42.9 1Session13 5 1 20.0 0 0 2 40.0 2Session14 3 0 0 0 0 2 66.7 1Session15 9 6 66.7 0 0 2 22.2 1Session16 5 0 0 0 0 0 0.0 5Session17 4 0 0 0 0 3 75.0 1Session18 3 1 33.3 0 0 2 66.7 0

XVI

Joana

/WSW/ [WSW] %[WSW] [SW] %[SW] [WS] %[WS] [S] %[S] Other

Session1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session2 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session3 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session4 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session5 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session6 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session7 2 0 0 0 0.00 0 0.00 1 50.0 1Session8 1 0 0 0 0.00 0 0.00 1 100 0Session9 11 0 0 0 0.00 5 45.45 4 36.4 2Session10 12 1 8.3 0 0.00 4 33.33 6 50.0 1Session11 31 5 16.1 2 6.45 4 12.90 13 41.9 7Session12 42 2 4.8 28 66.67 0 0.00 7 16.7 5Session13 44 5 11.4 15 34.09 5 11.36 5 11.4 14Session14 84 39 46.4 20 23.81 10 11.90 4 4.8 11

/WWS/ [WWS] %[WWS] [S] %[S] [WS] %[WS] Other

Session1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session2 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session3 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session4 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session5 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session6 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session7 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session8 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session9 1 0 0 0 0 0 0 1Session10 3 0 0 1 33.3 0 0 2Session11 2 0 0 0 0 2 100 0Session12 1 0 0 1 100 0 0 0Session13 1 0 0 1 100 0 0 0Session14 3 0 0 0 0 1 33.3 2

XVII

/SWW/ [SWW] %[SWW] [S] %[S] [SW] %[SW] Other

Session1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session2 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session3 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session4 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session5 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session6 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session7 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session8 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session9 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session10 3 0 0 3 100 0 0 0Session11 5 0 0 2 40 0 0 3Session12 2 0 0 1 50 1 50 0Session13 1 0 0 0 0 0 0 0Session14 12 1 8.3 3 25 8 66.7 0

João

/WSW/ [WSW] %[WSW] [SW] %[SW] [WS] %[WS] [S] %[S] Other

Session1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session2 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session3 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session4 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session5 7 1 14.3 2 28.6 3 42.9 1 14.3 0Session6 10 0 0.0 0 0 0 0 4 40 6Session7 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session8 5 0 0.0 0 0 0 0 0 0 5Session9 1 0 0.0 0 0 0 0 0 0 1Session10 1 0 0.0 0 0 0 0 1 100 0Session11 8 1 12.5 4 50 2 25.0 0 0 1Session12 7 1 14.3 2 28.6 3 42.9 1 14.3 0Session13 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session14 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session15 2 0 0.0 1 50.0 0 0 0 0 1Session16 48 27 56.3 3 6.3 3 6.3 1 2.1 14Session17 38 26 68.4 5 13.2 3 7.9 0 0 4Session18 62 39 62.9 17 27.4 0 0 0 0 6Session19 34 23 67.6 7 20.6 0 0 0 0 4Session20 49 28 57.1 15 30.6 1 2.0 0 0 5Session21 42 15 35.7 20 47.6 1 2.4 0 0 6Session22 62 48 77.4 9 14.5 1 1.6 0 0 4

XVIII

/WWS/ [WWS] %[WWS] [S] %[S] [WS] [WS] Other

Session1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session2 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session3 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session4 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session5 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session6 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session7 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session8 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session9 2 0 0 2 100 0 0 0Session10 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session11 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session12 1 0 0 0 0 1 100 0Session13 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session14 1 1 100 0 0 0 0 0Session15 3 0 0 3 100 0 0 0Session16 8 1 12.5 4 50.0 2 25.0 1Session17 4 1 25.0 0 0 1 25.0 2Session18 1 0 0 0 0 1 100 0Session19 9 3 33.3 1 11.1 3 33.3 2Session20 2 1 50.0 0 0 0 0 1Session21 4 2 50.0 0 0 2 50.0 0Session22 12 5 41.7 0 0 1 8.3 6

/SWW/ [SWW] %[SWW] [S] [S] [SW] %[SW] Other

Session1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session2 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session3 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session4 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session5 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session6 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session7 5 0 0 4 80 1 20 0Session8 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session9 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session10 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session11 3 0 0 3 100 0 0 0Session12 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session13 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session14 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session15 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session16 6 0 0 0 4 66.7 2Session17 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session18 2 0 0 0 0 0 0 2Session19 15 3 20.0 0 0 5 33.3 7Session20 8 1 12.5 0 0 2 25.0 5Session21 3 1 33.3 0 0 1 33.3 1Session22 3 0 0 0 0 0 0 3

XIX

Luma

/WSW/ [WSW] %[WSW] [SW] %[SW] [WS] %[WS] [S] %[S] Other

Session1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session2 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session3 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session4 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session5 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session6 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session7 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session8 3 0 0 3 100 0 0 0 0 0Session9 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session10 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session11 3 0 0 1 33.3 0 0 0 0 2Session12 7 0 0 1 14.3 2 28.6 0 0 4Session13 1 0 0 1 100 0 0 0 0 0Session14 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session15 10 0 0 7 70.0 0 0 1 10.0 3Session16 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session17 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session18 1 0 0 0 0 1 100 0 0 0Session19 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session20 2 0 0 0 0 1 50.0 0 0 1Session21 5 0 0 2 40.0 1 20.0 0 0 2Session22 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session23 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session24 1 0 0 0 0 0 0 1 100 0Session25 1 0 0 0 0 1 100 0 0 0Session26 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session27 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session28 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session29 4 0 0 3 75.0 1 25.0 0 0 0Session30 7 1 14.3 4 57.1 0 0 0 0 2Session31 22 6 27.3 7 31.8 4 18.2 0 0 5Session32 63 6 9.5 47 74.6 1 1.6 1 1.6 8Session33 33 5 15.2 26 78.8 0 0.0 2 6.1 0Session34 55 9 16.4 41 74.5 1 1.8 0 0.0 4Session35 84 43 51.2 38 45.2 2 2.4 1 1.2 0Session36 55 28 50.9 23 41.8 0 0 0 0 4Session37 64 49 76.6 13 20.3 2 3.1 0 0 0

XX

/WWS/ [WWS] %[WWS] [S] %[S] [WS] %[WS] Other

Session1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session2 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session3 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session4 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session5 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session6 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session7 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session8 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session9 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session10 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session11 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session12 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session13 3 0 0 0 0 1 33.3 2Session14 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session15 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session16 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session17 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session18 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session19 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session20 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session21 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session22 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session23 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session24 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session25 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session26 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session27 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session28 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session29 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session30 1 1 100 0 0 0 0 0Session31 2 0 0 0 0 0 0 2Session32 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session33 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session34 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session35 3 0 0 1 33.3 0 0 2Session36 11 2 18.2 0 0 7 63.6 2Session37 11 2 18.2 0 0 4 36.4 5

XXI

/SWW/ [SWW] %[SWW] [S] %[S] [SW] %[SW] Other

Session1 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session2 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session3 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session4 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session5 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session6 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session7 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session8 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session9 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session10 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session11 1 0 0 0 0 0 0 1Session12 1 0 0 1 100 0 0 0Session13 7 0 0 6 85.7 0 0 Session14 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session15 1 0 0 0 0 0 0 1Session16 2 0 0 1 50.0 0 0 1Session17 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session18 4 0 0 4 100 0 0 0Session19 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session20 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session21 2 0 0 1 50.0 0 0 1Session22 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session23 2 0 0 1 50.0 0 0 1Session24 1 0 0 0 0 1 100 0Session25 2 0 0 2 100 0 0 0Session26 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session27 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session28 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session29 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session30 1 0 0 0 0 0 0 1Session31 2 0 0 0 0 0 0 2Session32 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session33 2 2 100 0 0 0 0 0Session34 3 0 0 0 0 3 100 0Session35 3 0 0 0 0 3 100 0Session36 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐Session37 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐

XXII

XXIII

AppendixD

Children'slexiconperwordclassandwordshape

(Withnumberoftokens)

Clara

Monosyllables Trochees Iambs

+Verbs +Nouns +Verbs +Nouns +Verbs +NounsOther

S1 1papa 3

olá

S22

dá,está 1olá

S31dá 1

abre3

papa9

bebé3não

S43

dá,está 10bebé,papá

6não

S5 4água,carro

6bebé,

pinguim,cocó

9não,olá

S6 9cão,mãe 4

água

4bebé

6não,olá

S74

é,dá

18mãe,cão,pé,pai

8

Clara,mano,sapato,Aurora

6chichi,bebé

28não,olá,aqui

S86

é,está13

cão,mãe 6mano

10chapéu,bebé,papá

3ó,não

S92está

18pés,dois,pai

13Clara,Aurora,mana,papa

17bebé,João

10não,aqui,olá

S106

é,está

18mãe,pai,cão,Né

90menina,mana,Noddy,Uva,Clara,gato,pato,peixe,Aurora,Tita,outro,sapato,queijo,papa

23João,

cocó,avô

26não,ai,onde,aqui,cucu,olá

XXIV

S11

31é,tem,ver,põe,cai,sai,está

50mãe,cão,pé,chão,pai,mau,um

10olha,gosta,

abre

173Clara,chucha,Noddy,queijo,menina,papa,este,esta,Tita,Uva,isto,carro,casa,borboleta,tampa,pato,menino,bola,pata,outra,gato,uma,mano,lobo,

peixe,lua,ano,quatro,isso

2acabou,mordeu

35João,memé,mamã,avô,bebé,nariz,chichi

69não,já,aqui,aí

S1234

é,tem,dá,está

43mãe,cão,pai,sol,chão

21toma,abre,

143Clara,esta,

mano,macaco,peixe,menina,Noddy,sapato,uma,lua,carro,mana,gato,

tampa,Aurora,estrela,bola,Uva,outra,papa,batata,

Ruca

5acabou

50João,papá,nariz,avó,avô,piu‐piu,popó,chichi,cocó,bebé,mamã

118não,pum!já,

onde,aqui,olá

XXV

Inês

Monosyllables Trochees Iambs

+Verbs +Nouns +Verbs +Nouns +Verbs +NounsOthers

S11

dá

1

Fernanda

19

Inês,bebé,mamã,papá

S222

dá,está

2

toma

1

Fernanda

41

Inês,mamã,chichi

5

não,já

S3

31

dá,há,está

16

cão,má,mau,meu,mãe,pé

16

água,chupeta,queijinho

60

bebé,Inês,mamã,papá,Suzy,Isabel,chichi,chapéu

26

mais,não,já,aqui,ali

S4

31

há,dá,põe,tem,é,está

17

cão,mão,pé,céu

1

tapa

38

barco,cabelo,banho,

Bambi,bóia,bola,babete,fralda,carro,queijo,manta,

boneca,papa,sapato,Teresa,

cadeirinha

67

bebé,mamã,Inês,

bombom,chapéu,popó,quá‐

quá

65

cá,mais,não

S523

dá,é

5

cão,meu,um,Zé

1

mostra

39

Bambi,bola,Bela,macaco,carro,livro,Fernanda,Vanessa,

papa,sapato,pêlo,chupeta,porta,Teresa

53

bebé,Inês,mamã,papá,popó,João

14

não,aqui,ali

S6

18

dá,quer,é,está

52

cu,cão,mão,pá,pé

2

corta

124

água,cabelo,banho,bola,dedo,dentes,faca,fralda,carro,queijo,copocorda,creme,leite,

1

cortar

99

bombom,bebé,biberõ,Inês,cucu,cocó,mamã,papá,popó,champô,João

72

mais,não,já,aqui

XXVI

vitaminas,minha,meia,mamocas,banana,

boneca,papa,pêlo(s),penas,

chupeta,pele,vestido,leitinho,cueca,

passarinho,laranja

S7

20

dá,foi,é,está

18

cu,cão,má,

meu,pé,um

108

balde,boca,Bambi,bola,Bela,dedo,dente,fralda,cartas,colo,meia,boneca,ovo,papa,pato,pêlo,chupeta,prato,tio,urso,cavalo,casaco,

cuecas,gato,garfo,golo,dinheiro,girafa

1

limpar

81

bombom,balão,bebé,Inês,colher,cocó,mamã,pipi,Suzy,ão‐ão

56

mais,não,já,aqui

S8

10

há,dá,põe,é,está

28

có,mãe,mão,pai,pó,pé,um

4

puxam,tira,olha,gosto

69

água,cabelo,banho,cadeira,elefante,macaco,

Carlos,carta,copo,creme,Guilhermina,

neste,banana,Fernanda,

sapato,porta,vestido,tia,uma,urso,garfo,iogurte,

girafa,roda

1

tirar

73

bebé,dói‐dói,Inês,colher,cocó,leão,mamã,papel,popó,Suzy,avô,avó,canguru,chapéu

61

mais,não,já,tudo,aqui

S9 45 36 8 159 8 87 49

XXVII

há,dá,põe,

tem,vai,é,está

bom,Bé,dois,eu,cu,cão,meu,mão,pão,pó,pé

acho,tira,anda,olha,ajuda

água,cabelo,umbigo,boa,bolas,deste,dente,disco,duas,cadeira,este,fita,

flores,folha,isto,carta,

carro,escova,queijo,copo,quarto,lápis,livro,linda,Guilhermina,maminha,meia,manta,menino,

outro,sapato,pente,perna,prato,preta,apertado,

talco,vestida,Cristina,

leitinho,tira,tampa,uma,unha,urso,ouvido,Ana,esta,barulho,

jogo

fugiu,encontrou,

caiu,limpar,lavar,vestiu

bacio,bebé,dói‐dói,Inês,cocó,mamã,papá,popó,Suzy,sabão,Totó,Leonor,ó‐ó,chichi,chapéu,João

não,já,onde,aqui,ali

S10

78

há,dá,dou,quer,pôr,põe,tem,vão,é,está,estás

19

dois,flor,cão,mar,

meu(s),mãe,pá,pó,pé(s)

29

bebe,deita,queres,

senta,tira,ouviste,anda,olha

274

água,balde,bolos,banho,bóia,brincos,duas,deita,este(s),faca,fato,fita,fotografia,festa,casa,cama,colo,cola,copo,creme,livro,baloiço,

loura,minha,muita,meia,manta,

pequenina,bonecos,boneca(s),janela,Vanessa,outro(a),

papa,sapato,parque,pêlo(s),

7

beber,cair,cantar,lavar,tomar,guardar

89

bebé,Inês,colher,leão,mamã,maçã,papá,popó,Suzy,ó‐ó,parabéns,chapéu

97

bem,mais,

não,sim,já,onde,obrigada,agora

XXVIII

chupeta,praia,

prato(s),preta,sinos,piscina,

pulseira,tia,tira,uma,Ana,esta,gaita,

papagaio,garfo,igreja,grande,banheira,laranja,morena,

amarelo(a),roupa,

peixinhos,palhaço,coelhinho

S11

140

há,dá,ir,cai,quer,

põe,são,vai,é,és,está

46

dois,eu,flor,cão,mau,mar,

meu(s),mim,mãe,

mão,pó,pé(s),céu,Gil

43

abre,bebe,fecha,cabe,calça,canto,quero,queres,senta,sobe,

partimos,tira(s),toma,verte,

esmagaste,segura,gosta,ajuda

296

Tiago,água,bolo,boa,Bambi,comboio,bolas,

felicidades,dele,

crocodilo(s),Aladino,

bocadinhos,duas,cadeira,dela,dessas,desta,este,flores,elefante,telefone,fralda,isto,macaco,casa,escuro,cama,copo,queque,

aquela,creme,lado,lenço,lua,Legos,massas,minha,manta,penico,

menino(s),pequenina,nuvem,nua,banana,

15

buscar,beber,caiu,limpar,partiu,papar,abrir,jogar

136

bombom(ns),bebé,doutor,dói‐dói,Inês,colher,cocó,mamã,

memé,papá,popó,tau‐tau,avião,avó,ó‐ó,parabéns,chichi,

chapéu,João,jardim

143

lá,não,mé,já,ainda,onde,cima,agora,aqui,ali,assim

XXIX

boneco,boneca(s),janelas,outro(a),

papa(s),pato,leitinho,patatinha,batatinha,toda,tampa,ténis,uma,vaca(s),vela,coelho,olhos,ela,esta(s),papagaio,

garfo,grande,querida,barulho,orelhas,

laranja,rua,palhaços,sapato(s),chupeta,piupius,

portas,praia,prato,preta,

sala,dinossauros,princesa,triciclo,

sumo,sujo

S12

244

há,dá,cai,

quer,lê,leu,pôr,põe,pões,são,

tem,vai,ver,é,está,estão

53

dois,eu,cão,má,mau,meu,mãe,mão,pão,pé,teu,ti,tu,três,Gil

29

abre,deu‐lhe,fica,cabe,

encontra,mostra,preciso,estava,

tinha,tira,toma

338

água,barba,cabelo,

cabeça,boca,bolso,Bambi,bola(s),babete,

branco(a),Aladino,duas,duro,cadeira,dela,

este,almofada,fotografias,flores,telefone,

França,isto,macaco,casa,cama,colo,aquela(s),quarto,lápis,borboleta,lenço,

32

bateu,bater,brincar,dormir,

fazer,caiu,limpar,ouvir,papar,pegar,tirar,tocar,

estragou,abrir,

guardou,chorar,chegou

128

balão,bebé,biberõ,Inês,cocó,lençol,mamã,nariz,papá,Paris,popó,Suzy,Portugal,

Isabel,anões,anão,avô,avó,ó‐ó,

chapéu,João

317

cá,lá,mais,

não,trás,já,

dentro,nada,onde,tudo,agora,aqui,ali

XXX

livro(s),galinha,

limpo,lobo,loura,lua,

minha,meias,vermelho,manta,

Bernardo,nele,menino,menina,bonito,

nomes,nua,banana,boneca,Vanessa,neve,

outro(a/s),papa(s),patos(s),chupeta,

tapete,porco,peça,praia,princesa,sinos,pulseira,

insectos,sete,batata,castelo,partido,

todos,toda,castanhos,castelo,

estrelas,uma,verde(s),

Joana,olhos,cuecas,esta,estragado,gato,grilo,grande(s),dinheiro,tartarugas,amarelo(s),amarela,roto,

rua,capuchinho,chuva,palhaço,coelhinho,pijama

S13

161

há,dá,dar,

dou,dei,

54

dois,flor,cu,cão,luz,

75

abre,acho,brinca,

deito,fala,

358

água,cabelo,bicho,boca,boa,banho,

53

bateu,bebeu,beber,

106

biberõ,dói‐dói,Inês,colher,café,

125

bem,lá,mais,

não,sim,

XXXI

dói,faz,foi,cais,quer,lê,pôr,pôs,põe,

sou,são,tem,vai,ver,vou,

é,está(s),estou,estão

má,mau,meu,mim,pau,pé,teu,ti,tu,u,

chão,Gil

cabem,procura,cola,quero,lembro,mete,mexe,

enganaste,apaga,pode,posso,sabe(s),acende,precisa,consegue,estava,

tinha,tira,tenho,

toca,toma,olha,era,gosta,chega,cheira

Becas,babete,

branco(a),Fabrice,

deles,destes,cadeira,dela,caderno,desta,esse,este,doente,almofada,faca,

fotografia,folha,fumo,elefante,telefone(s),febre,isto,capa,casa,calça,aquele,escuro,cama,colo,cola,copo,quero,

laba,chocolate,borboleta,letra(s),

Filipe,livro,lixo,lobo,louça,luvas,

mesa,Mickey(s),amigo,

minha,muito,muitas,vermelho,vermelha,manta,amoras,menino(s),menina,

bonito,nome,boneco,bonecos,Vanessa,

neve,outro,outra,papa,pato,sapatos,pena,pena,tapete,Poupas,

porta,prato,sapo,

princesa,triciclo,circo,sono,açúcar,sumo,sete,

brincar,dormir,deitar,entrar,fazer,fechou,

encontrou,comer,acabou,caiu,calou,

responder,partiu,papar,apagou,pentear,tocar,tomar,tapar,

arrumar,chover,chorou

calor,mamã,Natal,nariz,papá,papel,sofá,senhor,tambor,televisão,

Isabel,anões,avô,avó,Tiagão,ó‐ó,coração,chapéu,cachecol,João

já,embora,dentro,como,nada,onde,cima,ena,

obrigado,agora,também,adeus,aqui,ali,assim,olá

XXXII

leitinho,todos,

pantufa(s),montanha,tantas,uma,cavalo(s),verde,sozinha,piano,Ana,olhos,Egas,

esta,papagaio,galos,gorro,iogurte,grande,nenhuma,tartarugas,laranja,

amarelo(a),roupa,rosa,chuva,palhaço,coelhinho,palhinha,beijinhos,jogo

S14

233

há,dar,dá,dou,dói,foi,quer,pôr,

põe,são,tem,vai,vou,é,está,estou,estão

43

bom,dois,eu,flor,cão,má,meu,mim,mão,pai,pé(s),teu,ti,tu,três

60

bebeste,sabia,

bebe,falta,cabem,cabe,come,quero,lembro,pinta,sabe(s),preciso,consigo,assopra,estava,vesti‐la,tira,

sentou‐se,estamos,toma,olha,gosta,chego,chama

271

água,cabelo,cabeça,Bicas,boca,boa(s),comboio,

botas,Becas,babete,braços,branca,crocodilo,Aladino,

duas,destas,este,doente,faca,folha,elefante,

telefone,isto,macaco,macaca,quente,

porquinho(s),copo,lápis,borboleta,

Filipe,minha,monstro,muito,

vermelho(a),manta,

34

buscar,bateu,bebeu,dormir,fazer,acordar,comer,acabou,caiu,cantar,ouvir,sentou,soprou,tirar,tirou,tocar,guiar,arrumar

89

botão,bebé,dói‐dói,Inês,colher,café,mamã,maçã,

Natal,hospital,papá,popó,televisão,avô,avó,coração,chapéu

196

cá,quem,lá,mais,não,sim,tão,já,dentro,como,quando,nada,onde,daqui,melhor,tinoni,também,aqui,ali,assim

XXXIII

manga,neste,Nica,menino,pequenino,meninos,menina,bonito,bonito,bonitas,

nisto,noutro,banana,boneco(s),boneca,Vanessa,outro(a),papa,

chupeta,pino,pilha,Poupas,peixe,

perna(s),prato,preta,

preso,prendas,sala,sapo,sapos,triciclo,salsicha,sopa,sujo,suja,cassete,cantiga,toda,

tantas,estrela,uma,cavalo,verde,piano,Joana,Egas,esta(s),estragada,garfo,iogurte,parede,tangerina,passarinho,tartaruga,barulho,orelhas,amarelo,

barriga,roda,rosas,Jorge

S15

159

há,faz,quer,leu,pôr,põe(s),sei,são,tem,

46

bons,eu,cu,má,mau,meu,

mão,pó,pé,teu,

53

penteá‐la,mudá‐la,dá‐me,

acordaste,deixa,fica,limpa,

287

água,cabelo,cabeço(s),cabeça,

borbulhas,bola,babete,braço,data,

41

buscar,bebeu,brincar,dormir,deitar,afiar,

50

balão,favor,feliz,cocó,leão,mamã,

Nuni,Carnaval,nariz,

94

bem,quem,mais,

não,sim,já,

dentro,

XXXIV

vai(s),ver,vou,vens,vão,é,está,estou,estão

ti,tu,Tó aleijo,leva,muda,morde,mexe,puxa,sabes,esqueci‐me,

tinham,tiro,tira,toma,

fazia,olha,era,eram,gosto,sujaste

dedo,Aladino,

cadeira,este,fita,

fotografia(s),flores,fralda,frente,isto,cabo,

complicado,caixa,

macaco,casa,carro,aquele,pequeno,

copo,aquela,creme,lápis,chocolate,borboleta,lenço,livro,lobo,maluca,Mickey,

Minie,Mina,minha(s),muito,muita(s),vermelho,manta,penico,

menino(s),menina,

bonito,nua,boneca(s),janela,nesta,outro(a),

pato,sapatos,palmas,parque,

pente,porco,perna,sala,dinossauro,salva,

princesa,triciclo,ursinho,

cinto,vestido,todos,toda,tua,estrela,uma,vela,horas,Joana,coelho,ela,

esta,estragado,gordo,

grande(s),tartaruga,barulho,

fazer,contar,

encontrou,encontrei,acordou,comi,acabou,escrever,limpar,mordeu,morder,papar,sentar,tirar,tocar,pentear,vestiu,andar,arrumar

pinguim,totó,avião,você,

parabéns,chichi,chapéu,

nada,onde,porque,para,cima,

tudo,daí,depois,melhor,também,aqui,ali,

XXXV

orelha(s),rabo,roupas,capuchinho,

cheio,bochechas,palhaço

S16

235

há,dá,dar,faz,foi,ir,quer,ler,pôr,põe,

sou,sei,são,tem,vai(s),vê,ver,vês,vi,vou,vem,vão,é,está(s),estar,estou

99

bom,dois,eu,flor,cu,cão,luz,meu,mim,pai,

pé(s),tu,ti,chão

80

abro,abre,dá‐me,dar‐me,disse,

deixa,dói‐me,falta,ia,cabe,come,quero,lembro,comemos,pago,paga,pode,espera,consigo,estava,tiro,tira,toma,veres,vamos,olha,era,eram,chama,chamem

293

água,cabeça,boa,bola,bruxa,dele(s),

escondida,dia,

bocadinho,dela,

desta(s),este,fada,elefante,fria,fruta,isto,casa,descalça,

carro,casca,quente,

autoclismo,cama,colo,copo,queque,gelado(s),livro,lixo,baloiço(s),Mina,amigo,minha,meia,primeiro,vermelho,amoras,menino(s),bonito,noite,nomes,

boneco(s),boneca,

nesta,outra,papa,sapatos,porta,pele,preta,

prenda,saia,sapo,centro,ursinho,salsicha,

açúcar,sumo,pulseira,apertado,castelo,partido,todo(s),

49

buscar,bebeu,brincar,dormir,deitei,fazer,

comprar,comer,caiu,

molhou,puxar,papar,pagar,parou,

abri,abrir,andar,andou,gostei,jogar

60

bebés,doutor,

favor,Inês,café,cocó,mulher,

mamã,maçã,Natal,papel,

popó,comercial,senhor,anéis,avó,chichi,chapéu,jardim

246

bem,cá,quê,

quem,lá,mais,

não,não,sim,já,dentro,como,comigo,nada,hoje,onde,porque,para,pronto,tudo,agora,então,amanhã,ninguém,adeus,aqui,ali,assim,olá

XXXVI

toda(s),tua,tanto,tantos,retrete,uma,cavalo,verde,cozinha,

Joana,cuecas,ela,esta,obrigado,estragada,iogurte,grande,

dinheiro(s),tangerina,Maria,caroço,tartaruga,laranja,amarelo,

carrinho,rua,escorrega(s),molhada

S17

132

há,fiz,fui,ir,quer,ler,pôr,põe,

pus,sei,tem,vai,são,vou,é,está,estou

43

dois,eu,cu,cão,luz,mal,meu(s),mim,

mãe,pé,sol,teu,ti,tu,chão

51

acho,perdemos,perdeu‐se,cabe,coube,procura,come,quero,morde,espera,sabe,estava,temos,estamos,toma,vamos,

anda,olha,gosta(s)

196

avariado,quarto,

toalha,bolos,banco,bola,

dele,podemos,deste(s),

duas,cadeira,ele,Julieta,este,faca,flores,isto,casa,carro,aquele,quente,aquilo,cubo(s),

escola,copo,aquela,chocolate,Filipe,livros,maluca,

longe,leite,Lego,Rometa,

comida,Mina,minha,almoço,

muito,mana,vermelho,penico,

pequenina,

42

buscar,beber,

bebi,fazer,esconder,contar,comeu,comer,

comi,caiu,morder,mandei,subir,tirou,engoliu,arranjar

25

boné,doutor,colher,café,lugar,mamã,

nariz,hospital,

papá,senhor,azul,

pinguim,cereais,chichi,cachecol

124

bem,cá,quê,lá,não,sim,sem,só,já,

dentro,ainda,quando,porque,tudo,agora,daqui,também,aí,aqui,ali,assim

XXXVII

cenoura,neve,outro,Diogo,espera,prato(s),preto(s),prenda,ursinho,

lacinho,sopa,professora,sumo,leitinho,todo(s),

tampa,uma,verde,

cafezinho,cozinha,

Joana,moeda,ela,esta,

garfo,grande,girafa,orelhas,laranja,amarelo,

roupa,resto,dinheiro

S18

206

dá,dói,faz,fez,fiz,foi,fui,ir,quer,lê,ler,pôr,põe,

sou,sei,são,tem,vai(s),ver,vês,vou,é,és,rir,está,estou,estão

123

bom,dois,dó,dez,eu,flor,cão,luz,má,mau,mar,

meu(s),mim,mãe,

mão,pá,pai,pé,seis,sol,teu,ti,tio,tu,três,um,reis,chão

82

sabia,cabia,procura,come,quero,lava,leio,levo,mostro,podes,posso,espera,sabes,acende,

sentar‐me,tinha,tenho,toma,

podemos,mordeu‐lhe,digo,deixa,

fazes,fica,foste,fura,ia,caíram,indo,cabe,coube,

261

avariada,alfaiates,bata,balde,barco,bibe,bolos,banco,bola,branco,brancas,

dedo,dele(s),deste(s),dia,mindinho,duma,duas,dela,eles,esse,este(s),almofada,Serafina,folha,frios,isto,casa,carta,coisa,compras,

conto,cama,escola,aquela,quatro,quarto,

gelado,lápis,rolava,

36

buscar,beber,brincar,dormir,fazer,fechar,

pendurar,fechou,comprei,contar,

encontrei,cortei,caí,escrever,escrevi,apagar,perdi,tirar,tapar,

estraguei,vestir,andar,segurar,segurou,arrumar,chorar

56

boné,doutor,favor,Inês,cartões,Lili,mamã,Manel,nenhum,hospital,

pincéis,papá,papéis,papel,senhor,Viseu,você(s),

amor,anel,avô,azul,

chorão,Juju,jardim

227

bem,cá,quê,

quem,lá,mais,

não,sim,só,já,dentro,ainda,como,nada,onde,cima,contigo,tudo,tanto,agora,

ninguém,também,aí,aqui,ali,

assim,atrás,epá

XXXVIII

traz‐me,trouxe,havia,viram,veio,vamos,levantas,olha,era(s),estragou‐se,segura,gosto,

segurá‐lo,ofereço,deixaram,rirem,arruma,chego,chega,

aleijaram,ajuda

livro(s),lixo,mala,Mina,Gulhermina,amigos,caminho,

minha,muito,meia(s),vermelha,caneta(s),neste,

menino,Nita,Anita,nomes,boneco,janelas,Vanessa,

nesta,outros,piolhos,

tapado,papa,sapatos,

pasta,pêlo,perna,praia,

salva,lacinhos,cinco,sete,leitinho,

todos,toda,tua,tanto,ténis,uma,

verde,casaco,vizinho,sozinha,Joana,

persiana,ela,esta,

empregada,obrigado,obrigados,segura,nenhuma,dinheiro,pendurado,parede,

amarelo(s),barriga,rosa,fechada,palhinha,Jorge

XXXIX

Joana

Monosyllables Trochees Iambs

+Verbs +Nouns +Verbs +Nouns +Verbs +NounsOthers

S1

1

mãe

1

não

S2

4

mamã

1

não

S32

é,está

2

meia

3

mamã

S44

há

2

pé,pai

3

bebé,mamã

2

não

S52

há,é

1

não

S6 2

pai

1

não

S74

é,há

1

pai

18

Carla,boneca,papa,gato,bigode

2

bebé,avó

3

não,olá

S84

há

2

mãe,pai

6

Carla,escola,gato,

capuchinho

2

avó,popó

3

não,cocorocó

S92

há,é

17

cão,mãe,pai,Zé,chá

1

gosto

23

água,Dinky,padrinho,Ricardo,

escola,Mila,Nando,papa,sapato,Elvira,Ana,Joana

27

bombom,Bobby,café,cocó,quá‐quá,piu‐piu,popó,avó,ó‐ó,carapau,Raquel,

chichi,João

4

não,olá,cocorocó

S10 8 43 29 1 27 9

XL

há,quer,é

flor,cão,mau,mão,

mãe,pai,pão,pé,Zé,rã,

chá,chão

água,bolo,macacos,Carla,carro,colo,escola,lobo,luva,vermelho,Nando,papa,sapato,Paula,cabritinhos,Elvira,Ana,coelho,esta

ouvi Fundão,nariz,

bombom,bebé,mamã,papel,popó,

titi,televisão,avião,avó,carapau,

Raquel,João

não

S11

16

quer,é,está,estou

24

flor,frio,luz,mau,Mi,mãe,pá,pai,pás,pão,Zé,rei,rã,chão,giz

1

gosto

87

água,balde,barco,cabelo,cabeça,bola,bolso,flores,feio,fralda,macaco,

calças,Carla,escola,

gelado,Lina,palitos,lindo,linda,lua,leite,Mila,

camisa,meio,mana,menino,menina,Nenuco,Nando,bonecos,bonecas,

piscina,César,sapatinho(s),mosqueteiros,tecto,uma,cavalos,vela,Rosinha,candeeiro,Ana,Daniela,Algarve,

Guito,bigode,grande,

dinheirinho,coroa,

peixinho(s),chucha,palhaço,

Borralheira

1

acabou

50

balão,bebé,dim‐dão,

colar,mamã,maçã,

memé,piu‐piu,pastor,popós,sofá,vovó,avô,avó,azul,ó‐ó,caracóis,Raquel,chapéu,João,Jesus

23

não,sim,já,aqui

S1214

há,cai,

23

dois,má,4 145

água,balde,

2

caiu,

32

caracóis,21

XLI

quer,tem,é

mãe,pá,pai,pão,véu,rei

olha barco,cabelos,boca,bolo,Lisboa,aberta,

Piedade,Dida,este,fada,fita,fotografias,flores,feia,isso,Inha,

isto,escadas,Ricardo,

Carla,Quina,gelados,Lina,Celeste,Rometa,manas,

menino(s),menina(s),bonito,Nando,Manecas,janela,outra(s),papa,pato,sapatos,

Paula,Gepeto,penso,pente,porta,pernas,

praia,rebuçado,saco,seco,piscina,vassoura,

sapatinho(s),pastilha,azeitonas,castelo,cavalos,

Elvira,vela,Zinha,piano,Ana,olhos,Daniela,esta,galo,gata,Guito,

dinheiros,Margarida,coroa,areia,fechada,

chucha,Xana,Borralheira,beijinho

lavar balão,dói‐dói,Fundão,esfregão,

colher,cocó,cristal,Luís,Manel,

maçã,nariz,papéis,

popó,anão,avô,avó,

chapéu,João

não,sim

S1316

foi,tem,

24

cão,mau,

7

deixas,

112

Heidi,barba,

8

fugiu,

27

bacio,cocó,

24

cá,não,

XLII

é,está,estou

meu,mãe,

mão,pai,pés,rei,giz

pode,anda,olha,grite,chore

cabeça,boa,botas,aberto,cabrinhas,dentes,Dido,Dida,estes,flores,

elefantes,ferro,fralda,fralda,isto,Ricardo,

Carla,Clara,colo,escola,quarto,Lina,livro,leite,Celeste,

comprimidos,comida,

missa,amiga,almoço,menino,Anita,Nenuco,boneco(s),Manela,

outro,padre,Pedro,porta,pedra,praia,primos,

prima,seco,sopa,César,batatas,tia,azeitonas,

mosqueteiros,pestana,

castelo,uma,cavalo,veado,

casaco,Rosinha,Ana,homem,

Daniela,essa,zangado,

Guito,bigode,grande(s),Clarinha,

amarelo,rato,fechado,peixinho,cachimbo,

palhaço,jogo,Joana

fazer,comer,caiu,chover

Quequé,maçã,Pepê,piu‐pius,

popó(s),ó‐ó,Carabás,chapéu,

João,Jesus,avô,azulavó

sim,aqui

S14147

dá,deu,diz,faz,

31

bons,dois,cão,

43

filma,mandam,

242

carruagem,Barbie,barba,

10

brincar,filmar,

31

bombons,irmã,

80

quê,não,sim,

XLIII

foi,pôs,põe,sai,tem,é,está

má,mau,más,meu,mãe,mão,mãos,pai,pão,pães,rei,rã,chá,giz

mora,mexe,partem,espera,estava,toma,anda,olha,gosta,parece,chama

barco,cabeça,boca,bolo,boa,banco,bolas,bota,Becas,Bela,brincos,

branca,Dida,duas,cadeira,madrinha,este(s),fada(s),golfinho,flores,

elefante,isto,casa,Ricardo,cara,esquilo,porquinho(s),

quinta,bicicleta,coisa(s),

quadro,laço,violino,livro(s),bolinhas,galinha(s),bolinhos,língua,lobo,baloiço,luvas,Legos,medo,aquecimento,comprimido,comida,amiga,

palminhas,muitas,vermelho,caneta(s),menino(s),menina,Nici,

Anita,Nenuco,Manecas,

chinelo,neve,osso,outros,pato,Paula,parque,peras,porco,pónei,

posta,prenda(s),prima,saco,adormecida,triciclo,vassoura,César,tia,ratinhos,

fazer,limpar,partiu,apanhar,pescar,pegou,varrer

chaminé,maçãs,

Natal,piu‐piu,postal,patins,popó,prisão,

computador,tractor,avião,avô,avó,avós,

azul,Romeu,Raquel,chapéu,João,Jesus

nada,onde,tumba,aqui,assim,

XLIV

sapatinho,sestinha,castelo,estrela(s),uma,uvas,vaca,cavalo,Elvira,vela,Rosinha,coisinha,avozinha,casinhas,

Susana,piano,Ana,Joana,viola,Olga,olhos,

Daniela,esta,estragado,iogurte,bigode,

Tamagoshi,grandes,querida,

coroa,cerejas,Maria,

morango,amarela,borracha,escorrega,capuchinho,jipe,beijinhos

XLV

João

Monosyllables Trochees Iambs

+Verbs +Nouns +Verbs +Nouns +Verbs +NounsOthers

S1 6

papa

S24

dá

1

olá

S31

dá

8

água

4

olá

S46

papa,água

2

mamã

4

olá

S5

14

bola,água,bolacha

S6

10

bolacha,chupeta

5

mamã,papá

3

olá

S73

pau

15

papa,água,bola,panda

18

mamã,papá

4

olá

S8

1

pau

23

água,bolo,bola,

chupeta,bolacha,comboio

11

limão,maçã,

melão,avô,avó

1

olá

S95

dá,está

20

mãe,pá,pau

14

bola,mota,pato,panda,uva,chupeta

2

acabou

14

limão,maçã,papá,tau‐tau,avô,avó,

2

mais

S103

dá

10

pá,pau,pé

39

água,boa,bola,meia,mano,panda

1

acabou

23

bebé,dlim‐dlão,limão,mamã,

maçã,papá,

6

olá

XLVI

popó,João

S1115

dá

7

pau

35

bola,mano,mota,pato,panda,banana

17

limão,maçã,mamã,memé,

papá,popó

2

não

S1220

dá,está

18

mã,pau,pão,pé

20

bola,meia,mano,bolacha,banana,sapato,girafa

1

acabou

26

limão,leão,mamã,

papá,popó,vovô,avô

3

mais,não,babau

S138

dá,está

11

pau

23

bola,mano,pato

54

limão,mamã,memé,

Natal,papá,sofá,avô

6

não,já

S142

está

4

mãe,pau,pé

17

bolo,bola,mano,mota,pato,uva

1

acabou

31

mamã,memé,

Natal,papá,popó,João

9

mais,não,já,olá

S15

4

quer,está

12

mãe,pai,pau

1

ajuda

10

bola,papa,peixe,panda,

verde,banana,Adriana

30

bebé,mamã,papá,

avestruz,javali

17

não,já,olá

S16

9

dá,cai,cais,está

17

frio,luz,mãe,pai,pau,Zé,chão

9

dança,olha,atende

87

bolo,bola,fruta,

quente,colo,mota,

Noddy,papa,pombo,

peixe,panda,porta,tia,Guida,roda,gira,banana,iogurte,

quentinho,

6

acabou

34

balão,bebé,café,leão,pião,

pinguim,popó,polar,avô,arroz,ó‐ó,José,João,Jesus,avestruz,avião,

computador

21

mais,não,sim,já,aqui,

olá

XLVII

menina,Susana,ursinho,artista,chupeta,girafa,

autocarro,bicicleta,barrigudo,crocodilo,joaninha,rinoceronte

S173

é

56

cão,meu,mã,mãe,pai,pau,pão,pé,três,chão

1

ajuda

109

água,balde,bolo,cama,quatro,Lila,mano,mota,Noddy,pombo,Pumba,

panda,porta,cinco,sopa,suja,urso,vaca,zebra,galo,gato,rua,chicha,alface,bolinho,comprido,cuidado,carnicha,cabeça,limpinho,leitinho,laranja,macaco,manteiga,Susana,sapato,artista,gatinho,chouriço,autocarro,capacete,crocodilo,proibido,torradinha,esfregona

3

limpar,trabalhar,acabou

22

café,mamã,papá,popó,avó,arroz,João,avião

17

ai,não,upa

S185

dá,é

21

luz,meu,

22

prova,senta,

104

água,bola,fruta,colo,

3

papar

50

dindão,kiwi,cocó,

20

não,upa,embora,

XLVIII

pai toma,adora,ajuda

livro,meia,mota,

Noddy,papa,pêra,pombo,panda,porta,perna,sopa,soro,tia,urso,vida,zebra,galo,rua,roda,Benfica,banana,barulho,comboio,cuidado,cozinha,cadeira,cavalo,casaco,querido,laranja,morango,orelha,

pombinho,papaia,Susana,sapato,tomate,girafa,

crocodilo,autocarro,telefone,gelatina

limão,mamã,maçã,memé,

papá,popó,polar,arroz,

João,jardim,biberõ

obrigada,aqui,ali

S19

19

há,dá,deu,dói,é,está

22

Bi,dois,meu,pai,pó,pé,céu,tu,Zé

9

anda,ajuda

117

água,balde,banho,

braço,dente,fruta,Lila,livro,mota,Noddy,

outra,pêra,pombo,Pumba,

porta,perna,prima,cinco,suja,sete,Tina,tampa,téni,trinta,uma,verde,Zoca,galo,gato,grande,rua,chave,banana,

6

pulou,parou,saltou,estragou,caiu

28

bebé,dim‐dão,dói‐dói,mamã,popó,polar,arroz,ó‐ó,chichi,João,jardim,biberõ,

construtor,computador

33

cá,mais,não,ão,

catrapumba,aqui,ão‐ão

XLIX

barriga,Benfica,Bernardo,cuidado,cadeira,laranja,mosquito,morango,menino,pijama,Susana,triciclo,girafa,

autocarro,bicicleta,capacete,Catarina,Portalegre,amarela,estragado,fotografia

S206

é

28

dois,meu,

mãe,pai,pau,seis,sol,Zé

9

pára,posso,senta,ajuda

139

balde,Bento,bolo,doce,doze,fino,festa,cacto,carro,mano,mota,papa,pombo,

panda,porta,perna,sopa,sumo,téni,trinta,vinte,zebra,Zoca,Guida,rua,jipe,batata,barriga,

brinquedo,ervilha,fechada,cavalo,carrinha,Lourenço,menina,mexido,papaia,Susana,sapato,Tiago,tapete,triciclo,vizinha,vassoura,

1

trabalhar

35

Abi,baton,favor,Miguel,mamã,pinhais,

papel,popó,polar,

senhor,avô,ó‐ó,João,jardim,fungagá

31

ai,mais,não,já,

pouco,cedo,tudo,

embora,obrigada,aqui,ali

L

chichinha,escola,

autocarro,bicicleta,bicharada,capacete,crocodilo,Margarida

S211

é

16

dois,flor,meu,

mão,pai,sol,céu,Zé

3

toma,anda,gosta

135

água,Bento,bolo,bola,este,fofo,

quarto,livro,lua,mano,mota,Noddy,parque,

pêra,pombo,Pumba,porta,tia,tampa,téni,uva,vida,

zebra,Guida,grande,rua,chave,jipe,giro,Benfica,banana,Diogo,comboio,carrinha,laranja,migalha,mantinha,menina,Susana,tomate,Teresa,triciclo,chupeta,escuro,estrela,girafa,bicicleta,borboleta,elefante,capacete,tartaruga,telefone,

rechonchudo

35

balões,balão,dim‐dão,kiwi,miau,mamã,

papá,popó,polar,real,chichi,João,jardim,

construtor,avião

18

não,embora,aqui

S2220

dá,cais,

30

meu,

10

anda,

117

Bento,bolo,

2

sentar,

38

balão,dói‐

31

ali,embora,

LI

sai,é,está

mãe,pai,pé,Zé

viste,gosto,ajuda

burro,bola,fofo,fruta,casa,colo,creme,

mano,mota,poucos,pomba,

panda,porta,tia,tigre,tonto,vida,zebra,Zinho,esta,galo,Guida,

Roque,rosa,jipe,

bombeiro,cuidado,koala,cabeça,cadeira,cavalo,carrinha,menina,papaia,pantufas,Susana,estrada,girafa,

autocarro,elefante,capacete,crocodilo,palhacinho,fotografia

acabou dói,favor,kiwi,cocó,miau,mamã,

popó,polar,real,chichi,

João,jardim,Beatriz,Barnabé,construtor,pontapé,tinoni,avião

obrigado,ão‐ão

LII

Luma

Monosyllables ­SW ­WS +Verbs +Nouns +Verbs +Nouns +Verbs +Nouns

Other

S11dá

1Frota

S22dá

S31dá

9Frota

S431

Hopla,Mara7

bebé,mamã

S5 5

Mara,pato,Hopla

S613dá

7bola,Mara,

pato

2mamã

S730dá

1cão

22bola,Frota,lua,Mara,pato,gato

1não

S835dá

1pau

29bola,Mara,banana,pato,

gato

2mamã

S98

dá,está1Po

13bola,Frota,carro

S101dá

2bola

2mamã

S111dá

5bola,banana,sapato,estrela

2balão,cocó

S125dá

9gato,banana

6mamã

S1321dá

12bola,Mara,banana,pato,

gato

3mamã

S14 8

bola,Mara,pato,gato

6marchar

14mami,mamã

1não

S1541dá

28harpa,bola,bolacha,borboleta,banana,Susana,papagaio,gato

3marchar

29cocó,mami,mamã

S16 8

1 36 33

LIII

dá,está bola marchar mami,mamã

S178dá

11bola,Mara,

Tito

17mami,mamã

S187

dá,está

15bola,Mara,banana,chata

21mami,mamã,chuchu

1não

S1922dá

4bola,pato

15mami,

mamã,Totô

S2012

dá,está

12olha(?)

22bola,Lala,bolachas,Mara,

piscina,Ana,amarelo

49bebé,mami,mimi,mamã,

Pati

6ó,já,aqui,ão‐

ão

S213

dá,está

5Bi,Pi,chá

7olha(?)

19bola,Mara,pato,tia,

Susana,Ana

57Bibi,mami,mamã,Pati,

vovô

S222dá 14

olha(?)

26bola,Mara,Noddy,papa,Tito,gata,

rechonchudo

30mami,mamã

S232dá

1chá

13olha(?)

30Mara,Noddy,

tia

52Miguel,mamã,

bisavó,avô

S242é 3

olha(?)

29bola,Mara,milho,menina,Noddy,

sapato,gato

40mami,Miguel,mamã

5não,mais,au‐

au

S255dá

9bom,pé

31bola,carro,baloiço,lua,Mara,meias,Noddy,pato,polvo,gato

21bebé,mami,Miguel,

mamã,Pati,bisavó,Totô

2au‐au

S261dá

14Bé,mão,pé,tio,três,um

2olha(?)

23bola,Mara,pato,Tito,verde,amarelo

42bebé,mami,Miguel,

mamã,Titi,Totô,bisavó,avô,chichi

3au‐au

S272dá

10pão,pé,três,

um,chá

1olha(?)

19Fimbo,carro,Mara,Pome,Tito,verde,

viva,amarelo

65mami,Miguel,mamã,

bisavó,ó‐ó,chichi

41não,sim

LIV

S281sai

14cão,mão,

pé,chão

30Heidi,carro,lobo,Mara,Noddy,Tito

60bebé,Inês,cocó,mami,mamã,Pati,sofá,vovó,ó‐

ó

45não,sim,aqui

S2913sai,vem

17mar,mão,pum,pé,tu,três,chão

46bamba,bola,queijo,Mara,menina,

Noddy,Tito,verde,gato,amarela

1sujou

75dói‐dói,mami,Miguel,mamã,pimpim,

sofá,tau‐tau,avô,ó‐ó,chinês

59não,sim

S306

tem,é,está

18fio,cão,mau,mão,pai,pum,pé,tu,três,chão

8abre,toma,anda

119alto,água,bola,dele,filho,fumo,isto,casa,carro,linda,lobo,lua,Mara,menina,

Noddy,olho,parvo,pêlo,Francisco,cinco,cinto,sujo,suja,santo,Tito,tinta,tonto,tonta,tua,Susana,gato,guizo,Goldy

86bebé,dói‐dói,cucu,mami,

mamã,Pati,pipi,Soni,tau‐tau,trovão,

avião,vovô,vovó,ó‐ó

147não,sim,tum,já,upa,amen,dim‐dom,aqui,

ali

S31

10sai,tem,vai,vou,vem,é,está

22chão,dois,dez,frio,mau,mão,pé,seis,sol,

três,um

11corre,passa,

salta,tira,toma,ajuda,magoa

134água,alto,fimbinho,banco,bola,brinco,

bruxa,dia,fotografia,isto,colo,lago,

lindo(a),lobo,Mara,mundo,menino,

banana,oito,outro(a),

papa,espaço,parvo,

prenda,sapo,cisne,cinco,sete,tenda,

Tito,

3empurrou,

saiu

71bebé,dói‐dói,favor,cocó,mami,Miguel,

Nené,pião,papéis,

trovão,vovó,azul

165ai,au,não,sim,sem,ó,já,nada,hoje,tanto,

aqui,ali,assim,olá

LV

cabritinho,tinta,tonto,tua,uma,casaco,Susana,

piano,Ana,gato,

grandes,virados,amarela,onze

S32

32quer,põe,sai,tem,vai,vê,ver,

vem,é,está

36dez,fio,mão,nó,pé,seis,ti,três,u,um

96abre,bate,brilha,brinca,dança,

fica,fecha,foge,mata,manda,mete,passa,pára,puxa,pega,apanha,tira,pode,pega,aperta,espera,senta,

toma,vira,voa,ando,anda,olha,ajuda

166baixo,

bicho(zinho),bola,

bocadinho,duas,isto,carro,colo,Mara,medo,mocho,menina,

Noddy,nove,oito,outro,pata,pinha,ponta,buraco,licença,

cinco,seta,sete,Tito,tanto(a)(s),toca,uma,golfe,grande,choque

59bebé,mami,mamã,piu‐piu,tau‐tau

253au,lá,mais,não,sim,só,ó,dentro,frente,pumba,cima,tudo,aqui,ali,

assim

S33

48fez,cai,cais,quer,põe,sai,tem,vai,está

22ar,dois,dez,cão,mau,pé,sol,ti,tu,três,u,um

41abre,

acha,bate,fecha,canta,muda,mexe,pára,apanha,pega,senta,sopra,tira,vira,voa,

levanta,anda,olha,ajuda,joga

203água,arco,baixo,balde,bolo,bola,Dina,Dingo,

duas,fotografia(s),íris,isto,carro,bicicleta,coisa,colo,quatro,livro,lindo(a),lobo,lua,

Mara,Mário,medo,mocho,comando,mota,

4fugiu,caiu,chocou

94balão,dói‐dói,favor,mami,mamã,

malhão,sofá,bisavó,televisão,avião,vovô,vovó,andor,

algum

151bem,cá,mais,não,sim,só,já,dentro,ainda,cima,tudo,

então,aqui,ali,olá

LVI

cogumelo,menina,Anita,banana,

Noddy,nova,nove,

outro(a),pata,peixe,porta,pedra,pernas,preza,licença,cinco,

sombra,Tito,tinta,toda,tanto(a)(s),uma,urso,vela,Susana,

anos,alguma,grande,

Teresa,gelo,ajuda,jeito

S34

48faz,fez,quer,põe,sai,tem,vem,está

41ar,dois,dez,frio,cor,mão,pum,pão,pó,pé,seis,três,um,chão

50brinca,deita,entra,falta,

larga,liga,leva,mete,mexe,puxa,aperta,empresta,sopra,tira,voa,levanta,anda

220água,alto,baixo,cabeça,blusa,bola(s),cuidado,bocadinho,doze,duas,força,isto,quente,bicicleta,escuro,

queijo,cama,copo,queda,caracolinho,

linda,relógio,

Mara,mosca,muito,muita,

meia,comando,manga,negro,menina,Anita,bananas,

Noddy,nove,oito,outro,outra,olho,onze,pilha,peças,cinco,

27fugiu,caiu,

morder,mexeu,partiu,saiu,tocou,voou,abriu,chover,chorou

108balão,bebé,favor,mami,Miguel,mamã,televisão,vovó,azul,Jesus

111cá,lá,mais,

não,sim,só,já,dentro,pumba,pronto,aqui,ali,assim

LVII

sujo,sete,Tito,

mantinha,festinha,bonitinho,saltitona,

torto,tampa,tanto,tecto,terra,uma,verde,

Susana,gota,grande,buraco,barulho,

laranja,rosa,escorrega,puxinho,ajuda,fechada

S35

60há,faz,ir,cai,quer,tem,vê,é,está

29bom,dois,flor,cor,mão,pá,pé,sol,

três,um

52dada,

falta,caía,come,leva,mora,pisa,

empurra,apanha(s),aperta,salta,

senta,tira,anda,agarra,chega,chupa,chora

184alto(a),

barco,bicho,sombrinha,cuidado,

sentadinha,este,Frota,isso,isto,casa,carro,casca,quente,sequinho,casquinha,fresquinha,coisa,corda,quatro,bolinha,linda,

baloiço,lua,Mara,medo,comida,formiga,

muito(a/as),manta,

cogumelo,meninos(as),perninha,banana,Noddy,

nossa,outro,outra,papa,tapete,pele,

praia,laranja,crescidos,crescida,

17trabalhar,dançou,correr,mordeu,tirou,andar,andou,arranjou

68bebé,favor,mami,

mamã,vovô,vum‐vum,avô,azul

134mais,mal,não,sim,sem,já,

hoje,depressa,adeus,aqui,assim

LVIII

sombra,Susana,sujo,Tito,antigo,terra,testa,estrela,

tronco,uma,verde,vento,cãozinho,sozinha,besouro,

anos,coelho,gato,grande,barriga,escorrega,bichinho(s),

ajuda

S36

122dá,deu,faz,ir,quer,pôr,põe,

ser,sei,são,tem,está

23cão,lei,mau,mar,mão,pum,ti,três,um

54rebenta,falta,fica,conta,mostra,mete,

empurra,puxa,pode,pega,espera,empresta,tira,

entorna,levanta,anda,olha

224arco,baixo,barco,cabeça,

bicho,boca,burro,aberta,brincos,Dina,dia,coitadinho,sentadinha,rodinha,disco,ele,esta,este,Sofia,isto,casa,carro,aquele,aquilo,

buraquinho,porquinhos,coisa,escola,estrelinha,lindo,língua,

lobo,pirilampo,Mara,medo,amigo,monstro,muito,menina,

novo,numa,Noddy,boneca,outro(a),

olho,parque,porco,pano,panda,porta,Francisco,Sophie,sujo,

29buscar,brincar,dormir,deitar,entrar,fazer,

escondeu,comer,caiu,limpar,levar,mostrar,pintou,partiu,pegar,sentar,

será,tirar,entornou,tocar,andar,regar,chorou,chegar

116bebé,dragão,favor,cocó,mami,

mamã,maçã,nenhum,piu‐piu,pum‐pum,popó,prisão,senhor,Tatá,avião,vovô,chichi,

chapéu,João

158à, ai, cá, lá,mais, não, sim,tum, ó, hoje,cima, tanto,agora, balalão,aí, aqui, ali,assim,olá,epá

LIX

Tito,pontinha,pestana,tanto,toca,uma,urso,vaca,vivo,ervinhas,velho,

Susana,esta,papagaio,grande,girafa,

cadeirinha,orelhas,

Mafarricos,chave,

chuchinha,ajuda

S37

70ir,

quer,pôs,põe,ser,tem,vai,vê,viu,é,rir,está

12Blu,Fá,flor,

cor,cão,mão,Po,pé

26rebenta,disse,deixa,

entra,fica,fecha,leva,

empurra,queria,puxa,apanha,pode,senta,anda,olha,chega,chora

207água,baixo,cabeça,burro,banho,babete,

sentadinho,sentadinha,este,fita,Afonso,elefante,

fresco,Frota,isto,casa,aquilo,

cobra,cola,aquela,lado,colagens,borboleta,colado,medo,amigos,flamingo,muito,

menino(s),menina,Anita,caniche,

novo,Noddy,boneco,trotinete,

parque,poça,porta,sapo,ursinha,

circo,cinco,sono,sujo,Catita,tigre,tanta,triste,

23encher,fazer,cortar,correr,caiu,pintar,

empurrar,empurrou,puxar,passou,sair,aturar,andar,regar,chover

81balão,bebé,feijão,calor,cocó,mami,mamã,capitão,pinguins,Tatá,vovô,vovó,avô,avó,regador

75mais, não, sim,tum, ó, xau, já,quase, nunca,onde, daqui,balalão, adeus,aqui,ali,assim

LX

trela,uma,gaveta,verde,casaco,tesoura,

Susana,esta,girafa,Teresa,tartaruga,orelhas,laranja,amarelo,esta,roxo,rosa,chuva,

cheio