variation in child language by j l roberts

7
Jabeur M (1987). A sociolinguistic study in Tunisia: Rades. Ph.D. diss. University of Reading, UK. Jassem Z (1993). Impact of Arab-Israeli wars on language and social change in the Arab World: the case of Syrian Arabic. Kuala Lampur: Pustaka Antara. Lahlou M (1992). A morpho-syntactic study of code- switching between Moroccan Arabic and French. Ph.D. diss., Austin, University of Texas. Rouchdy A (ed.) (2002) Language contact and language conflict in Arabic. USA and Canada: Routledge Curzon. Sadiqi F (2002). ‘The language of introduction in the city of Fes: the gender-identity interaction.’ In Rouchdy A (ed.). 116–132. Versteegh K (1997). The Arabic language. Edinburgh: Edinburgh University Press. Variation in Child Language J L Roberts, University of Vermont, Burlington, VT, USA ß 2006 Elsevier Ltd. All rights reserved. The area of child language acquisition, for the most part, began in the very late 20th century, 40 years after Labov’s groundbreaking work investigating the social implications of language variation and change. However, this statement may imply that sociolinguistics is alone in confining language study to adults, when this is not at all the case. Linguistics as a whole, although containing many diverging, and sometimes conflicting, theoretical perspectives, has been virtually united as a field with adults-only con- tent. The purpose of this article is to situate the child in linguistic history, summarize current findings in child language variation, and argue that despite the difficulties and shortcomings of this work, there is interesting and important information to be learned about language from its not-fully-formed speakers, in addition to the more obvious and empirically demon- strated premise that there is interesting and important information to be learned about children from their language. Historical Overview Linguistics in the early 20th century was dominated by structuralism. The ways of looking at language structure, which for the most part meant phonologi- cal structure, varied, but the emphasis on structure as opposed to history differentiates the linguistics of this period from that of the previous century (Hymes and Fought, 1975: 11). Although language was seen as a ‘human function’ and a ‘cultural function,’ the focus was not so much on the speakers as on what was in essence the ‘‘product of long-continued social usage’’ (Sapir, 1921: 4). As such, children were seen as the acquirers of the language of their culture, not as contributors; their acquisition of language was the study of another field – psychology. It is not the case that children were never men- tioned, however. Bloomfield (1933: 29) speculated on language learning, which he called ‘‘the greatest intellectual feat any one of us is ever required to perform.’’ Following the highly popular learning the- ory, he likened the process of language acquisition to forming a ‘habit’ based on imitation of caretakers’ speech. This habit was later shaped and refined by corrections from competent speakers when the child made an error. Bloomfield (1933: 32), in fact, was quite clear on the interests of linguists in this process and noted the following: In the division of scientific labor, the linguist deals only with the speech signal (r ... . s); he is not competent to deal with problems of physiology or psychology. The findings of the linguist, who studies the speech- signal, will be all the more valuable for the psychologist if they are not distorted by any pre-possessions about psychology. Dialect geographers were similarly occupied with language structure – in this case, its variation across geographic area, whether lexical or phonological or, more rarely, grammatical. Kurath’s (1939a, 1939b) large project in the United States was predicated on the assumptions that geographic variation was best reflected in the adult speakers of the community, particularly those who were nonmobile, older, rural, and male (later termed NORMs by Chambers and Trudgill (1980)). Although the origin of these features was of critical importance at the time, it was most likely to be found in the historical settlement patterns and geographical barriers and boundaries of the areas studied (Kurath, 1939a) and not in those young speakers acquiring the dialects. Chomsky (1957, 1959, 1965) brought children to prominence in the context of learnability. That is, in early versions of generative grammar, the paucity of input to young children was taken as support for the heritability of language. However, these children, like the adult speakers discussed by generative linguists, were more theoretical than real, and it was not until 344 Variation in Arabic Languages

Upload: nandini1008

Post on 11-Dec-2015

4 views

Category:

Documents


1 download

DESCRIPTION

The area of child language acquisition, for the mostpart, began in the very late 20th century, 40 yearsafter Labov’s groundbreaking work investigatingthe social implications of language variation andchange. However, this statement may imply thatsociolinguistics is alone in confining language studyto adults, when this is not at all the case. Linguisticsas a whole, although containing many diverging, andsometimes conflicting, theoretical perspectives, hasbeen virtually united as a field with adults-only content.The purpose of this article is to situate the childin linguistic history, summarize current findings inchild language variation, and argue that despite thedifficulties and shortcomings of this work, there isinteresting and important information to be learnedabout language from its not-fully-formed speakers, inaddition to the more obvious and empirically demonstratedpremise that there is interesting and importantinformation to be learned about children from theirlanguage.

TRANSCRIPT

Page 1: Variation in Child Language by J L Roberts

Jabeur M (1987). A sociolinguistic study in Tunisia: Rades.Ph.D. diss. University of Reading, UK.

Jassem Z (1993). Impact of Arab-Israeli wars on languageand social change in the Arab World: the case of SyrianArabic. Kuala Lampur: Pustaka Antara.

Lahlou M (1992). A morpho-syntactic study of code-switching between Moroccan Arabic and French. Ph.D.diss., Austin, University of Texas.

Rouchdy A (ed.) (2002) Language contact and languageconflict in Arabic. USA and Canada: Routledge Curzon.

Sadiqi F (2002). ‘The language of introduction in the city ofFes: the gender-identity interaction.’ In Rouchdy A (ed.).116–132.

Versteegh K (1997). The Arabic language. Edinburgh:Edinburgh University Press.

344 Variation in Arabic Languages

Variation in Child Language

J L Roberts, University of Vermont, Burlington,

VT, USA

� 2006 Elsevier Ltd. All rights reserved.

The area of child language acquisition, for the mostpart, began in the very late 20th century, 40 yearsafter Labov’s groundbreaking work investigatingthe social implications of language variation andchange. However, this statement may imply thatsociolinguistics is alone in confining language studyto adults, when this is not at all the case. Linguisticsas a whole, although containing many diverging, andsometimes conflicting, theoretical perspectives, hasbeen virtually united as a field with adults-only con-tent. The purpose of this article is to situate the childin linguistic history, summarize current findings inchild language variation, and argue that despite thedifficulties and shortcomings of this work, there isinteresting and important information to be learnedabout language from its not-fully-formed speakers, inaddition to the more obvious and empirically demon-strated premise that there is interesting and importantinformation to be learned about children from theirlanguage.

Historical Overview

Linguistics in the early 20th century was dominatedby structuralism. The ways of looking at languagestructure, which for the most part meant phonologi-cal structure, varied, but the emphasis on structureas opposed to history differentiates the linguistics ofthis period from that of the previous century (Hymesand Fought, 1975: 11). Although language was seenas a ‘human function’ and a ‘cultural function,’ thefocus was not so much on the speakers as on whatwas in essence the ‘‘product of long-continued socialusage’’ (Sapir, 1921: 4). As such, children were seen asthe acquirers of the language of their culture, not ascontributors; their acquisition of language was thestudy of another field – psychology.

It is not the case that children were never men-tioned, however. Bloomfield (1933: 29) speculatedon language learning, which he called ‘‘the greatestintellectual feat any one of us is ever required toperform.’’ Following the highly popular learning the-ory, he likened the process of language acquisition toforming a ‘habit’ based on imitation of caretakers’speech. This habit was later shaped and refined bycorrections from competent speakers when the childmade an error. Bloomfield (1933: 32), in fact, wasquite clear on the interests of linguists in this processand noted the following:

In the division of scientific labor, the linguist deals onlywith the speech signal (r . . .. s); he is not competentto deal with problems of physiology or psychology.The findings of the linguist, who studies the speech-signal, will be all the more valuable for the psychologistif they are not distorted by any pre-possessions aboutpsychology.

Dialect geographers were similarly occupied withlanguage structure – in this case, its variation acrossgeographic area, whether lexical or phonological or,more rarely, grammatical. Kurath’s (1939a, 1939b)large project in the United States was predicated onthe assumptions that geographic variation was bestreflected in the adult speakers of the community,particularly those who were nonmobile, older, rural,and male (later termed NORMs by Chambers andTrudgill (1980)). Although the origin of these featureswas of critical importance at the time, it was mostlikely to be found in the historical settlement patternsand geographical barriers and boundaries of the areasstudied (Kurath, 1939a) and not in those youngspeakers acquiring the dialects.

Chomsky (1957, 1959, 1965) brought children toprominence in the context of learnability. That is, inearly versions of generative grammar, the paucity ofinput to young children was taken as support for theheritability of language. However, these children, likethe adult speakers discussed by generative linguists,were more theoretical than real, and it was not until

Page 2: Variation in Child Language by J L Roberts

Variation in Child Language 345

considerably later that data from real children weresought (Marcus et al., 1990).

Given this historical context, it is not surprisingthat when the study of language variation and changeas a social, as well as regional, phenomenon began inthe mid-20th century, it was with adult or adolescentspeakers (Labov, 1966). Despite the very early workof Fischer (1958), whose sample comprised 3-, 5-,and 10-year-old speakers, it was not until more than20 years later that children younger than the age of 5years appeared in variation research again. Even then,this work can more accurately be described as a trick-le than a flood. In 1981, Kovac and Adamson ex-plored the deletion of finite be in African-AmericanVernacular English (AAVE). This feature of AAVE,along with others, was (and is) a popular topic ofvariation study (Baugh, 1986; Labov, 1969; Rickfordet al., 1991; Wolfram, 1969). Kovac and Adamson,however, were the first to examine the question ofdevelopmental versus dialectal variation, a particular-ly difficult issue to differentiate because the feature inquestion involved a deletion process that is difficultto distinguish from nonacquisition but important inthe study of early dialect learning. The researchersfound that although all of their speakers (aged 3, 5,and 7 years) demonstrated deletion, the constraintson deletion were more difficult to acquire. In fact, thechildren mastered the constraints on copula contrac-tion before those of deletion, which were not fullyacquired at age 7 years (Kovac and Adamson, 1981).

Wolfram (1989) studied even younger speakers ofAAVE, aged 18–54 months. He too examined a dele-tion process – in this case, deletion of final nasals withnasalization of the preceding vowel. He found thatbefore age 3 years, the children produced virtually nofinal nasal segments; after that, however, their acqui-sitional pattern bore strong relation to that of AAVE,as spoken by adults. That is, /n/ was preferred as atarget of deletion over /m/ or /N/, and utterance final/n/ was more frequently deleted than that followed byanother word. Finally, the children’s patterns of dele-tion tended to plateau at age 3 years and hold steady,suggesting an acquired and stabilized deletion pat-tern. Both of these early studies found that the fea-tures in question were not clearly apparent in theyoungest children. That is, it was not possible todetermine if the sparseness of the feature wasdue to an emerging deletion pattern or an emerginggrammatical (e.g., finite be) or phonological (e.g., thephoneme /n/) form.

Although these early studies cannot be said to rep-resent a trend toward the study of child languagevariation, they did set the stage for examining twoimportant questions: What can the study of childlanguage variation tell us about children and the

language acquisition process? and What can thestudy of child language acquisition tell us about lan-guage? The first question is far more easily dealt withthan the second. The evidence for acquisition can, attimes, be directly observed, and although languageacquisition is not instantaneous, its progress in indivi-duals is more rapid that the course of language change.Consequently, there are intriguing results that tell usat least the basics of dialect acquisition, whereas thediscussion of any implications for language as a wholemust be made up primarily of questions.

The Study of Variation and the LanguageAcquisition Process

Acquisition of language (particularly phonology) isassumed to be a process of movement from babblingto segment and/or rule acquisition. Locke (1983),however, noted that although this process appears tobe inevitable in normal acquisition, it is far less easyto document the point of passage from one stage toanother with certainty. There is clear agreement thatbabbling is a natural, physiological process. Deafchildren, who have no access to auditory input fromcaretakers or auditory feedback from their own pro-ductions, nevertheless produce babbled sounds in thefirst year of life. In fact, Locke noted that the move-ment from babbling to speech is generally assumedto be heavily dependent on ‘naturalness’ in that thosesegments that are easier to produce from a physiolog-ical standpoint and more frequent in the target lan-guage (not necessarily a coincidental pairing) areamong the first to be acquired by the child. However,therein lies the complication as well; as Locke (1983:158) noted,

If phonological segments and rules can be so ‘natural,’it must be difficult indeed to know when a child has‘acquired’ a rule, ‘learned’ an articulation. . . . Such pho-nological naturalness may blur the separate contribu-tions of biologically innate mechanisms, environmentalstimulation, and their modes of interaction.

There is a parallel dilemma in the study of childlanguage variation: at what point does a process suchas deletion stop being a stage on the way to mastery ofa phoneme or morpheme and become an acquired,but variable, dialectal process? As noted previously,Kovac and Adamson (1981) found that for deletionof finite be, the process was still not complete by age7 years. Their study, however, as well as later ones,demonstrated a possible venue for exploring this ‘nat-uralness’ problem. A segment or its deletion may benatural and/or developmental, but the constraintson that process are in some cases language or dialectspecific. One definition of acquisition could then be

Page 3: Variation in Child Language by J L Roberts

346 Variation in Child Language

the presence, absence, or modification of that seg-ment as the speech community requires, not theadult-like articulation of a given segment.

The process of (-t, d) deletion, a subset of finalconsonant cluster reduction, provides an example.This process, as students of variation well know,allows the deletion of final /t/ or /d/ in word-final con-sonant clusters. However, the process is constrainedby phonological and grammatical environmentalconditions, most crucially, in this case, following seg-ment and grammatical form. Although most of thefollowing segment constraints can be seen as a formof simplification, and (following Locke) a childmay not be accurately credited for the acquisition ofa rule, the constraints of following pause and gram-matical form are not examples of simplification. Guyand Boyd’s (1990) examination of the grammaticalconstraint and Roberts’s (1997a) study of both pho-nological and grammatical constraints explored thisrule in preschool children. Roberts found that thechildren age 3 years and older had acquired all thephonological constraints, including that of followingpause, and although the specific findings differedslightly, both studies found the children to be in themidst of learning the grammatical constraints. Eventhe 4-year-olds had not completely acquired the adultrule (Roberts, 1997a).

A study on vowel acquisition revealed similarresults. In this case, 3- and 4-year-old children inPhiladelphia were studied to determine their progressin learning the short a or (æh) vowel (Roberts andLabov, 1995). This feature is complex, with lexical,phonological, and grammatical factors conditioningthe raising of short a tokens. As was the case with(-t, d) deletion, the children were making excellentprogress and had acquired several constraints onshort a. However, some of the constraints continuedto be realized in an inconsistent manner, demon-strating that this phoneme was partially learned, al-though its articulation was completely mastered byall the children.

As the previous examples demonstrate, the conceptof acquisition goes beyond the production of thesounds of speech. This point is not unfamiliar topsycholinguists researching the acquisition of phonol-ogy. Locke (1983: 181), in fact, noted that ‘‘one mayacquire and use forms without necessarily learningtheir precise surface characteristics.’’ He proposed astudy in which adults would be compared on theirability to learn a phonetically based and a nonphone-tically based rule created by the researchers. How-ever, a rule that is unpredictable unless one knows thedialect of the speaker would seem to be a far bettercandidate for study than a hypothetical one, andchildren learning their first language would be more

natural subjects. Children who learn that in theirspeech community /t/ and /d/ are rarely deleted beforea pause and short a is raised before /S/ but not before/d/ except in the words ‘mad,’ ‘bad,’ and ‘glad’ arelearning a linguistic rule or pattern, not relying on aphysiological imperative. Systematic variation pro-vides naturally occurring learning and researchopportunities that can be of value to both fields ofstudy.

The Language Learning Environment

That children acquire phonemes, rules, constraints,and patterns is a critical point of departure for thestudy of child variation. However, the outcome of thelearning process – a rule acquired or a pattern mas-tered – is only one facet of this study. It is also impor-tant to explore the process by which this learningoccurs. What is the language environment fromwhich the child learns his or her community norms?How much of this environment does the child attendto and use?

There is a large body of work on the subject oflanguage learning environments in Western culture(Ferguson, 1977) and others (Ochs and Schieffelin,1984). In addition to the rich descriptive work thesestudies provide, they also help to answer the questionof the necessity or benefit of a modified languageenvironment for children. This continues to be asubject of debate, but the cross-cultural work citedpreviously does cast doubt on the necessity of a par-ticular (i.e., mainstream, Western) language learningenvironment. Whether particular environmentalattributes are beneficial to language acquisition is, atleast to some extent, dependent on the cultural valuesof the community. For example, there is evidence thatmodification of the child’s environment as with child-directed speech may lead to faster language learning(Clarke-Stewart, 1973), but whether faster is better isa question to which the answer can vary from onesociety to another. It is not the purpose of this articleto enter into this particular debate, but variationstudies can shed light on another aspect of the lan-guage learning environment: How does the linguis-tic environment that children encounter affect theirlanguage and its acquisition?

Again, the study of variation can augment whatis already known through the study of children’s lan-guage learning environment. Foulkes et al. (1999)studied glottalization in forty 2-, 3-, and 4-year-oldchildren from Newcastle Upon Tyne in England.Their findings on the acquisition of glottalizationwere similar to those on other variables discussedpreviously. The children had made progress but notcompleted their acquisition of the complicated glottal

Page 4: Variation in Child Language by J L Roberts

Variation in Child Language 347

stop replacement and reinforcement of /t/. In addi-tion, they found that the children demonstrated highdegrees of preaspiration when /t/ was in utterance-final position, a pattern that cannot be explainedby simplification. Rather, the authors’ interpretationwas socially based in that they noted that preaspira-tion was the dominant pattern of young womenspeakers in the area – those who were also primarilyresponsible for child rearing.

Roberts (2002) also found evidence of variablepatterns in the language environments of young chil-dren. This study examined the speech of four whitemiddle-class mothers from Tennessee talking withtheir toddlers, aged 18 or 19 months, and with aninterviewer. These mothers were all speakers ofSouthern American English, which has as one of itsfeatures the pronunciation of long /aI/ as in ‘time’ as afronted monophthong, or closer to [tæm]. All fourmothers used both types of /aI/ but produced more ofthe monophthong /aI/ in their speech to the adultinterviewer than with their children. The children,therefore, heard child-directed speech that compriseda mix of productions, even though their mothers mayhave been presenting them with more examples of thenonlocal variant than they typically used with adults.They heard examples of variation: what soundswithin their dialect were malleable and in whatways. The previous results suggest that child-directedspeech contains the elements necessary for supportingproducers of variation at home in their own speechcommunities.

Although caretaker speech has captured the inter-est of researchers for decades, one concern about thisarea of study that has recently become more relevantis the effect of peers, particularly day care peers, onchildren’s speech. This is a particularly importantquestion for variationists because peer effects arebasic to variation theory and have always been con-sidered and applied to adolescents and adults. It isthis point that has provided the underpinnings for thenetwork analysis by Milroy and others, who haveused increasingly specific methods to determine ex-actly who is the speaker’s network. It seems question-able, at best, to work from the assumption that achild’s, even a very young child’s, network consistssolely, or even primarily, of his or her mother. Back-ground information to this effect should be gatheredand included in articles written about acquisitionof variation. Also advisable are studies that focus onthe relative effects of day care and caretaker influ-ence, difficult as these may be to tease apart in theday-to-day existence of children and families.

Early evidence on this point is mixed but generallyleaning toward less, rather than more, parental influ-ence. For example, Kerswill (1996) found variable

influences in 4-year-olds he studied in Milton Keynes.Some of the children did demonstrate parental effectson dialect learning; others were more influenced byolder peers. In any event, the parental influence ap-peared to disappear by age 8 years. Roberts (1997b)found that parental effects were seen most easily incomplicated vowel patterns because children who didnot have two native Philadelphia parents did notacquire the complex short a pattern. Even on simplerpatterns, however, parental effects can be inferredfrom the finding that children learned most effectivelythe Philadelphia vowel changes most prominentin their mothers, as opposed to those changes thathave been found to be led by males. It is also relevantthat the only child to make very limited progress inthe acquisition of any of the Philadelphia vowels stud-ied was one with Italian-speaking parents. However,the fact that this child did sound like a speaker ofAmerican English, not Italian-influenced English,suggests that peer and other community influenceswere already at work; his progress on the Philadelphiapatterns was simply not as rapid as that seen in theother 3- and 4-year-old children.

Starks and Bayard (2002) also explored this issuewith four young speakers of New Zealand English.Their findings echoed many of the previously dis-cussed findings in that the children showed markedlydifferent patterns of acquisition of postvocalic /r/ de-letion. Two of the children were virtually nonrhoticfrom a very early age and two showed considerablymore rhoticity, even though all four had rhotic par-ents and nonrhotic peers. Similar results were foundwith regard to their acquisition of several NewZealand vowels. The authors suggested that age ofentry into day care (and exposure to peers) as well assibling influence may have been factors in the acqui-sition differences. Although the study was limited bya small amount of data, it is notable that Starks andBayard found it essential to perform word-by-wordanalyses to make sense of the highly variable data.

Some Additional Data

A study of linguistic variation in Vermont (Roberts,2005) provided an opportunity to explore the daycare vs. home linguistic influence on children’s re-placement of /t/ with glottal stop [ ]. This apparentlylocal variant of /t/ is both stigmatized and, unlikeother localized productions, enduring in the face ofdialect leveling. The data discussed here are a subsetof a larger set representing the speech of children andadults. The speech of seven 3- to 5-year-olds wascoded as to the use of glottal stop to replace medialand final /t/ in words. All the children were residentsof Vermont. Three of them had parents who were not

Page 5: Variation in Child Language by J L Roberts

Table 1 Glottal replacement of /t/ in children with parents who

use / / and parents who do not use / /

Child Total N / /N % / /

/ / parents

Alice 229 48 20

Jessie 114 15 13

David 55 13 23

Emmy 174 26 14

Non / / parents

Jean 121 6 4

Marie 214 55 25

Cal 174 17 9

348 Variation in Child Language

glottal stop users. One child’s parents were fromFrance and were learning English; one child’s singleparent was from the mid-Atlantic region of the UnitedStates; and one child’s parents were originally fromVermont but had extensive travel experience, ad-vanced educational degrees, and a clear dispreferencefor Vermont speech patterns. The parents of the otherfour children were Vermonters who, to varyingdegrees, demonstrated glottal stop replacement of /t/.

It will surprise no one who has read either theprevious discussion or the extensive literature on in-dividual differences in child language acquisition (seeBates et al., 1995) that although there were indica-tions of trends of influence present in the results, therewas nothing that approached significance in the vari-able rule analysis of the data. Table 1 shows theresults of the comparison between children with par-ents who demonstrated / / replacement of /t/ andthose with parents who did not. In general, the chil-dren who had glottalization input both at home andat school appeared to do more glottal replacementthan those who had this input only at school (wherethe input was mixed because the teacher was a Ver-monter but not all the children came from homes withthe same linguistic environment). Strikingly, however,the highest replacement percentage came from Marie,whose input for Vermont speech came from schoolonly. Marie’s linguistic background was also the mostdifferent from the others because she was the childwith French-speaking parents. Although Marie’sEnglish was quite good, her parents were insuffi-ciently comfortable in English to consent to be inter-viewed. This background suggested at least twopossibilities for interpreting her glottal stop data.The first has to do with possible overlap betweenthe languages. One of the more popular environmentsfor glottal replacement is word-final /t/ following /n/,as in the word ‘Vermont,’ which is derived fromFrench. French is well-known to include the deletionof final /n/ and final /n/ plus stop with the nasalizationof the previous vowel (Ruhlen, 1978; Tranel, 1968).

In Vermont speech, this word and others like it areproduced with a nasalized vowel and deleted /n/ fol-lowed by / / (Roberts, 2005). Although it is not clearfrom the previous research that / / follows the nasa-lized vowels in French, Locke (1983: 230) noted thatword-final nasalized vowels may be perceived, in anycase, as preceding a glottal stop. The second possibil-ity is that as Marie was learning the English languageas well as the Vermont dialect at school, she wasacquiring the Vermont patterns from her teacher (anative Vermonter) and her Vermont peers.

Conclusions are difficult to draw from such smallsamples. As psycholinguists have found, the processof language acquisition, like that of language change,is a complicated one. Slices of cross-sectional dataand group trends are sometimes sufficient only tohint at possible explanations. Continued examina-tions of even small amounts of data, such as thoseof Starks and Bayard (2002), are necessary to gaininsight into the complexity of child variation.

Conclusion

The study of child variation necessarily combines thedisciplines of both psycholinguistics and sociolinguis-tics. Both fields emphasize process as well as out-come, but the emphasis may take on a different tonein each. In sociolinguistics, the speech of fully formedindividuals is used to explore the process of languagechange. Speakers are assumed to be competent speak-ers of their particular varieties, and the emphasis ison the change in the variety. In psycholinguistics, fullyformed language is used to explore its developmentin speakers. Speakers (children) are assumed to bemoving toward competence, and the emphasis is onthe change in the speaker. The challenge for childvariationists is to bring together two entities, bothof which are in a state of change. It is a difficultchallenge because the sociolinguistic methodology,exemplified most often by the interview, necessarilycaptures a moment in time, both for the language andfor the speaker. On the other hand, the collaborationaffords the researcher an opportunity to make a con-tribution to both fields since the assumption thateither the language or the speaker is static is an un-proven one, at best. The psycholinguist is able tomake use of the stable language assumption by focus-ing on language forms that change so slowly as to bepragmatically immobile: tense and case markings andbasic word order. When the focus is on forms thatsociolinguists have found to vary (e.g., phoneme pro-duction and lexical forms), the more stable main-stream or standard forms are taken as the goal of thelanguage learner. The variationist is able to make useof the stable speaker assumption by focusing on

Page 6: Variation in Child Language by J L Roberts

Variation in Child Language 349

mature speakers in established speech communities,exemplified by the initial dialectology ideal, thegeographically isolated NORMs. The increasingmobility of the population at large and the focus onurban settings and increasingly less isolated ruralsettings have created a challenge to the stable speakerassumption. It is no longer clear that the speaker, eventhe mature speaker, resists the linguistic changesbrought about by increased mobility.

Child variationists can provide a venue for explor-ing this new situation: changing language in changingspeakers. The studies discussed previously demon-strate both the challenges and the benefits of thiswork. In summary, we may formulate the followingkey points that may be useful to variationists:

1. The necessity for close examination of the data ofindividual speakers, even when group trends areexamined. The purpose would be not only to as-certain that individual children are following thedocumented group trends, although this is criticalwhen working with speakers who exhibit develop-mental as well as dialect variation. It is also impor-tant to explore the very complex pattern ofinfluences on these young speakers and whichinfluences are largest as well as most enduringand, therefore, most likely to influence the processof language change.

2. The importance of exploring the child’s networkof relationships to document the influence patternand separate fleeting from long-term influencesthat may affect not only the speaker but also thelanguage variety.

Of course, neither of these suggestions represents anew methodology. Many researchers examine theirdata closely to provide interpretive insights thatwould be lost in group patterns. In addition, networkanalysis has been practiced in some form since theearliest days of dialect study (Labov, 1966) and con-tinues to be refined (Milroy, 1987). The difference isthat the rapidly changing world demographics todaycoupled with the rapidly changing young speakersmake these aspects of method more clearly critical.We are a long way from identifying the ways in whichchildren may influence language change in their ownacquisition of that language. However, it is hopedthat variationists may benefit from both the method-ological and the theoretical results of the exploration.

See also: Age: Apparent Time and Real Time; Dialect

Atlases; Gender; Infancy: Sensitivity to Linguistic Form;

Social-Cognitive Basis of Language Development;

Sociophonetics; Teenagers, Variation, and Young

People’s Culture.

Bibliography

Bates E, Dale P S & Thal D (1995). ‘Individual differencesand their implications for theories of language develop-ment.’ In Fletcher P & MacWhinney B (eds.) The hand-book of child language. Oxford: Blackwell. 96–151.

Baugh J (1986). ‘A re-examination of the Black Englishcopula.’ In Labov W (ed.) Locating language in timeand space. New York: Academic Press. 83–195.

Bloomfield L (1933). Language. New York: Holt, Rinehart& Winston.

Chambers J K & Trudgill P (1980). Dialectology.Cambridge, UK: Cambridge University Press.

Chomsky N (1957). Syntactic structures. The Hague, TheNetherlands: Mouton.

Chomsky N (1959). ‘A review of B. F. Skinner’s VerbalBehavior 1957.’ Language 35, 26–58.

Chomsky N (1965). Aspects of the theory of syntax.Cambridge: MIT Press.

Clarke-Stewart K A (1973). ‘Interactions between mothersand their young children: characteristics and conse-quences.’ Monographs for the Society for Research inChild Development 38(153).

Ferguson C (1977). ‘Babytalk as a simplified register.’ InSnow C & Ferguson C (eds.) Talking to children.Cambridge, UK: Cambridge University Press. 209–235.

Fischer J (1958). ‘Social influence of a linguistic variant.’Word 14, 47–56.

Foulkes P, Docherty G & Watts D (1999). ‘Tracking theemergence of structured variation.’ In Leeds workingpapers in linguistics and phonetics. Leeds, UK: Universityof Leeds. 1–25.

Guy G & Boyd S (1990). ‘The development of a mor-phological class.’ Language Variation and Change 2,1–18.

Hymes D & Fought J (1975). American structuralism. TheHague, The Netherlands: Mouton.

Kerswill P (1996). ‘Children, adolescents, and languagechange.’ Language Variation and Change 8, 177–202.

Kovac C & Adamson H (1981). ‘Variation theory andfirst language acquisition.’ In Sankoff D & Cedergren H(eds.) Variation omnibus. Edmonton, Alberta, Canada:Linguistic Research. 403–410.

Kurath H (1939a). Handbook of the linguistic atlas ofNew England. Providence, RI: Brown University Press.

Kurath H (1939b). Linguistic atlas of New England.Providence, RI: Brown University Press.

Labov W (1966). The social stratification of English inNew York City. Washington, D.C.: Center for AppliedLinguistics.

Labov W (1969). ‘Contraction, deletion and inherent varia-bility of the English copula.’ Language 45, 716–762.

Locke J (1983). Phonological acquisition and change.New York: Academic Press.

Marcus G, Ullman M, Pinker S, Hollander M, Rosen T J &Xu F (1990). Overregularization. Occasional paperno. 41. Cambridge: MIT, Center for Cognitive Science.

Milroy L (1987). Language and social networks. Oxford:Blackwell.

Page 7: Variation in Child Language by J L Roberts

350 Variation in Child Language

Ochs E & Schieffelin B B (1984). ‘Language acquisition andsocialization: three developmental stories and their impli-cations.’ In Shweder R & Levine R (eds.) Culture theory:essays in mind, self and emotion. New York: CambridgeUniversity Press. 276–320.

Rickford J, Ball A, Blake R, Jackson R & Martin N (1991).‘Rappin’ on the copula coffin: theoretical and meth-odological issues in the analysis of copula variation inAfrican-American Vernacular English.’ Language Varia-tion and Change 3, 103–132.

Roberts J (1997a). ‘Acquisition of variable rules: a study of(-t, d) deletion.’ Journal of Child Language 24, 351–372.

Roberts J (1997b). ‘Hitting a moving target: acquisitionof sound change in progress by Philadelphia children.’Language Variation and Change 11, 249–266.

Roberts J (2002). ‘Child language variation.’ In Chambers JK, Trudgill P & Schilling-Estes N (eds.) The handbookof language variation and change. Oxford: Blackwell.333–438.

Roberts J (2005). ‘Withstanding the flatlanders: the case ofglottal stop in Vermont.’ Manuscript in preparation.

Variation in First Language AcquisE Lieven, Max Planck Institute for Evolutionary

Anthropology, Leipzig, Germany

� 2006 Elsevier Ltd. All rights reserved.

Variation among Children

Children differ greatly from each other in their ratesof learning language, and there are also many reportsof differences in the process of acquisition. Thesedifferences are relatively well documented for theacquisition of English, but are poorly documentedfor children learning other languages. In this article,references to variation among children learninglanguages other than English are to the articles inSlobin (1985) unless otherwise indicated. Lieven(1997) and Peters (1997) contain detailed surveysof the cross-linguistic evidence. Bates et al. (1988)report the most detailed longitudinal study of indi-vidual differences in the early learning of English, andFenson et al. (1994) provide the most comprehensivecross-sectional overview based on the MacarthurCommunicative Development Inventory, a parentalreport measure.

Variation in Early Comprehension and Production

Children vary considerably in when they first startto show signs of comprehension and when they firststart producing words. Because much depends onhow comprehension and production are defined, it is

Roberts J & Labov W (1995). ‘Learning to talkPhiladelphian.’ Language Variation and Change 7,101–122.

Ruhlen M (1978). ‘Nasal vowels.’ In Greenberg J H (ed.)Universals of human language, vol. 2: phonology.Stanford, CA: Stanford University Press.

Sapir E (1921). Language: an introduction to the study ofspeech. New York: Harcourt Brace Jovanovich.

Starks D & Bayard D (2002). ‘Individual variation in theacquisition of postvocalic /r/: day care and sibling orderas potential variables.’ American Speech 77(2), 184–194.

Tranel B (1968). Concreteness in generative phonology:evidence from French. Berkeley: University of CaliforniaPress.

Wolfram W (1969). A sociolinguistic description of DetroitNegro speech. Washington, DC: Center for AppliedLinguistics.

Wolfram W (1989). ‘Structural variability in phonologicaldevelopment: final nasals in Vernacular Black English.’ InFasold R W & Schiffrin D (eds.) Language change andvariation. Amsterdam: Benjamins. 301–332.

itiondifficult to give an exact age range for these activities.Comprehension of individual words such as ‘No’starts very early, and of course, children can use con-textual cues to interpret the utterances of thosearound them without necessarily parsing much ofwhat they hear. Although most well-controlled stud-ies of early word learning have found that compre-hension is in advance of production, the age range ofproduction is, if anything, even more varied, withsome children producing their first words (as reportedby parents) at around 10 months whereas othersmight not produce more than a few recognizablewords before 16–18 months (Fenson et al., 1994).For most children, progress in comprehension andin production is highly correlated, but there arereports of children whose comprehension outstripstheir production by much more than the normalextent (Bates et al., 1988). Children who are morethan 1 standard deviation (SD) or sometimes 1.5 SDbelow the mean on standardized tests of language areusually considered to be late talkers. Among thisgroup are children who ‘catch up’ and join the normalrange, as well as a group of children who go on tobe defined as having specific language impairment(Rescorla et al., 2000). Many studies, though notall, have found that girls tend to start to producewords earlier than boys (Fenson et al., 1994). Thereare also some studies indicating that first-born chil-dren start to produce words earlier than subsequent-born children (Fenson et al., 1994; Pine, 1995). It is