person hierarchies in algonkian and inuktitut

16
Person hierarchies in Algonkian and Inuktitut 1 JOHN HEWSON Abstract Much has been written in recent years on grammatical persons and the hierarchical roles they play in a variety of linguistic structures, verbal and nominal. This paper makes explicit the complex nature of third-person roles, which accounts for unusual agreement patterning in certain Inuktitut forms, which is in turn paralleled by the proximate vs. obviative distinction in Algonkian languages. The question of a common morphology for POS- SESSION and VERBAL AGREEMENT, common to a great variety of languages, is also addressed in terms of the universal iconic hierarchies of person and a justification proposed for this motivated syncretism. 1. Introduction 11 is vrctt known (AAtetv \%4; Seilet I9&3b; Ulvvtkg 1987* 40) that in a great variety of languages there is a common morphology for the pro- nominal elements that mark (a) nominal POSSESSION and (b) verbal AGREEMENT (that is, the verbal actors). There are also language families, such as Algonkian, where there is a strict hierarchy of pronomi- nal persons, leading to a contrast between DIRECT forms (action down- ward through the hierarchy) and INVERSE forms (action upward through the hierarchy) in the transitive verb paradigms. These hierarchies have been shown to be largely iconic, being based not only on the natural agentivity of the first and second persons (Silverstein 1976; Comrie 1978; Dixon 1979) but on the central role or VIEWPOINT established by the speech-act participants (DeLancey 1981; Ebert 1987) which may play a substantial role in any discourse. The purpose of this paper is to show (a) that there is an iconic justification for linking the grammar of POSSES- SION with that of TRANSITIVITY; (b) that these hierarchies are also reflected in the data of languages, such as Inuktitut, that have no ostensi- Linguistics 29 (1991), 861-875 0024-3949/91/0029-0861 $2.00 © Walter de Gruyter Brought to you by | University of Sussex (University of Authenticated | 172.16.1.226 Download Date | 7/18/12 10:44 PM

Upload: john

Post on 09-Aug-2016

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Person hierarchies in Algonkian and Inuktitut

Person hierarchies in Algonkian and Inuktitut1

JOHN HEWSON

Abstract

Much has been written in recent years on grammatical persons and thehierarchical roles they play in a variety of linguistic structures, verbal andnominal. This paper makes explicit the complex nature of third-personroles, which accounts for unusual agreement patterning in certain Inuktitutforms, which is in turn paralleled by the proximate vs. obviative distinctionin Algonkian languages. The question of a common morphology for POS-SESSION and VERBAL AGREEMENT, common to a great variety oflanguages, is also addressed in terms of the universal iconic hierarchies ofperson and a justification proposed for this motivated syncretism.

1. Introduction

11 is vrctt known (AAtetv \%4; Seilet I9&3b; Ulvvtkg 1987* 40) that in agreat variety of languages there is a common morphology for the pro-nominal elements that mark (a) nominal POSSESSION and (b) verbalAGREEMENT (that is, the verbal actors). There are also languagefamilies, such as Algonkian, where there is a strict hierarchy of pronomi-nal persons, leading to a contrast between DIRECT forms (action down-ward through the hierarchy) and INVERSE forms (action upward throughthe hierarchy) in the transitive verb paradigms. These hierarchies havebeen shown to be largely iconic, being based not only on the naturalagentivity of the first and second persons (Silverstein 1976; Comrie 1978;Dixon 1979) but on the central role or VIEWPOINT established by thespeech-act participants (DeLancey 1981; Ebert 1987) which may play asubstantial role in any discourse. The purpose of this paper is to show(a) that there is an iconic justification for linking the grammar of POSSES-SION with that of TRANSITIVITY; (b) that these hierarchies are alsoreflected in the data of languages, such as Inuktitut, that have no ostensi-

Linguistics 29 (1991), 861-875 0024-3949/91/0029-0861 $2.00© Walter de Gruyter

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 2: Person hierarchies in Algonkian and Inuktitut

862 J. Hewson

ble direct and inverse forms and, although ergative, show no evidence ofsplit ergativity; and (c) that some of the mysteries of Inuktitut morphology(see for example the title of Ulving 1987) begin to make sense once therole of these iconic hierarchies is perceived. Vaxtin has commented (1979:282), apropos of Asiatic Eskimo, for example, that 'agreement patterningin the 3S-3O configuration is rather difficult to analyse, mainly becauseit is impossible to identify separate subject and object markers'. In thispaper we shall show exactly what is marked, and why.

2. Transitivity and possession

Alongside the Eskimo evidence, which had already been mentionedamong the languages cited by Allen, Ulving cites Hungarian haz-om, haz-od, haz-ja 'my, thy, his house' versus ir-om, ir-od, ir-ja Ί, thou, he writes(something definite)'. To this we may add examples from Algonkianlanguages, such as the following from Ojibway, where the pronominalelements are prefixed: ni-wa:panta:n, ki-wa:panta:n, o-wa:panta:n Ί, thou,he sees (something inanimate)' versus ni-nikk, ki-nikk, o-nikk 'my, thy,his arm'.

The word for 'arm' in Ojibway is inanimate, however, and a furtherelement is called into play when the possessed noun is animate: ni-naw,ki-naw, o-naw-an 'my, thy, his cheek', again with an exact parallel in theparadigms of the transitive verb: ni-wa:pama:, ki-wa:pama.\ o-wa:pama:-/?, Ί thou, he sees (something animate)', where the suffixed -(a)n in bothcases marks the obviative singular. We may analyze the person relation-ships in these paradigms as, follows (atv=proximate, obv = obviative):

ninikkkinikkonikk

ninawkinaw kiwa:pama: 2 >3.an.pronawan owa :pama:n 3.an.pr>3.an.obv

3. Person hierarchies

In these paradigms elements of three Algonkian hierarchies are veryclearly demonstrated. The possessor is seen as the determining elementin the relationship of possession and is therefore required to be of ahigher grammatical rank than the possessee. In Algonkian person hierar-

niwa:panta:nkiwa:panta:nowa :panta:nniwa:pama:

123. an1

>3.inan>3.inan>3.inan>3.an.pr

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 3: Person hierarchies in Algonkian and Inuktitut

Person hierarchies in Algonkian 863

chies, first and second person rank higher than third; it can also be seenthat animate ranks higher than inanimate, and that proximate rankshigher than obviative. The proximate/obviative distinction only comesinto play when both actors are otherwise of an equal status (namelythird-person animate): its principal purpose appears to be the 'back-grounding' (Seiler 1983a: 24) of the possessee, thereby foregrounding thedetermining role of the possessor.

The notion of obviative, by itself, is secondary or derived: there canbe no obviative unless there is first a proximate, just as in baseballfreedom to advance to second base is contingent on touching first base,which thereby becomes a determining or controlling factor of the second-ary element. Proximate, in short, determines obviative: proximate estab-lishes a new (third-person) viewpoint or focus, to which all other thirdpersons are secondary or notionally subordinate.

The animate/inanimate hierarchy has similar values. Universally inlanguages animates are appropriate for agentive active roles, such asagent of a transitive verb, which controls or determines the action,whereas inanimates are appropriate for inactive roles such as patient.These are really two macroroles which cover a range of functions, aspointed out by Fillmore, who thinks it unlikely that 'the notions agentand patient are in any sense satisfactory semantic primitives' (1968: 11).The issue is further discussed in Foley and Van Valin, who present themas two macroroles to be called 'actor' and 'undergoer' (1984: 28ff.). Thediscussion is taken further by Klaiman (1988), who also uses the terms'controller' and 'affected entity' to indicate the scope of these roles andto suggest that they cover a continuum in which the controller may beaffected and the affected entity have some COTYITO\.

If A hits B, for example, A is the controlling and determining factor,who often has free choice, whereas B has little control or choice — he ishit whether he likes it or not. It is noteworthy that animates in languagessuch as Spanish (phenomenon of leismo), Gascon, Rumanian, and else-where in the world require special prepositional or other marking to playthe role of patient, mark of their iconic unsuitability to play such a role.

The person hierarchy, wherein first and second outrank third, is basedon the following factors: (a) first and second persons are always animate,and almost exclusively human, since only humans are normally capableof speaking, listening, and understanding (roles that fall to first andsecond person by definition), whereas third persons can be either animateor inanimate; (b) the SAP (speech-act participants) establish a VIEW-POINT (DeLancey 1981) for the discourse which determines directional-ity (such as come/go) and focus. There is, in fact, a natural hierarchy, asGustave Guillaume has observed (1987: 100), in the very act of language:

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 4: Person hierarchies in Algonkian and Inuktitut

864 /. Hewson

the speaker is the fully active person, creating discourse and causing itto vibrate on the airwaves, the hearer is in a medio-passive position,passive in regard to receiving the message, but active in interpreting andunderstanding it. The third person, by contrast, is inactive, not beingrequired to speak or listen. In this regard the hierarchy of person maybe seen to be just as iconic as those of animacy and obviation.

These same factors are also at play in the paradigm of possession: firstand second persons, for example, may possess third-person elements, butthird-person elements cannot be possessors of elements that are first andsecond person: we can have my dog, your dog, but we cannot have *dog'sme or *dog 's you. In the paradigms of possession in Algonkian languages,therefore, we can see the person hierarchies operating in every form:SAP> 3 is an irreversible relation of determination, and 3 > 3 must alwaysbe interpreted as either animate > inanimate (dominance of agentivity) orproximate > obviative (dominance of viewpoint).

4. Direct and inverse forms

It would appear that the majority of languages that have person hierar-chies or grammatical directionality (that is, direct and inverse forms)follow the natural hierarchy seen in the quote above from Guillaume andmade explicit by Seiler (1983a: 46):

lst>2nd>3rd human > 3rd animate > 3rd inanimate,or else treat the two SAPs as equal elements, as ranking equal in animacy^Comnc \*$Wr. 61). The Algonkian family, in fact, almost without excep-tion, presents the following hierarchy:2nd> lst>3rd an proximate > 3rd an obviative > 3rd inanimatewhere there is prominence given to second person over first. The possiblecultural overtones of this difference may be of interest to anthropologists,and indeed Speck (1935) has discussed at length the fact that among theNaskapi it is felt that one's mista:pe:w (literally 'great man' or 'spirit')may not be as powerful as that of one's interlocutor, to whom one must,as a consequence, always give deference. There is also the fact that thespeaker, as an SAP, is also a listener, and that there are two listeners andonly one speaker in any discourse. There is also the fact that the patientiverole is the unmarked role in the transitive relationship of an ergativelanguage.

Whatever the case, the Algonkian TA (transitive with animate goal)verb paradigms show two sets of direct and inverse forms, as in the

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 5: Person hierarchies in Algonkian and Inuktitut

Person hierarchies in Algonkian 865

following paradigms from Ojibway, where the second set represents theinterplay between first and second persons. In these paradigms 3 = 3rdproximate and 3' = 3rd obviate, lp= 1st plural exclusive s/he and I, 12 =1st plural inclusive you and I. The direct forms are on the left, the inverseforms on the right, so that niwa:pama: means Ί see him' whereasniwa:pamik means 'he sees me'.

DIRECT INVERSE1 > 3 ni-wa:pam-a: 1 < 3 ni-wa:pam-ik2>3 ki-wa:pam-a: 2<3 ki-wa:pam-ik3 > 3' o-wa:pam- a:-n 3 < 3' o- wa:pam-iko:-nlp>3 ni-wa:pam-a:-na:n lp<3 ni-wa:pam-iko-na:n12 > 3 ki-wa:pam-a:-na:n 12 < 3 ki-wa:pam-iko-na:n2p > 3 ki-wa:pam-a:-wa: 2p < 3 ki-wa:pam-iko-wa:3p > 3' o-wa:pam- a:-wa-n 3p < 3' o- wa:pam-iko-wa:-n2 > 1 ki-wa:pam-i 2 < 1 ki-wa:pam-in2p > 1 ki-wa:pam-i-m 2p < 1 ki-wa:pam-in-im2(p) > l p ki-wa:pam-i-min 2(p) < l p ki-wa:pam-in-imin

The last line is to be interpreted as follows: 2(p)> lp = 2nd person (sg.or pi.) acting on 1st plural: kiwa:pamimin 'you (sg or pi) see us'; 2(p)<Ip = 1st plural acting on 2nd (sg or pi): kiwa:paminimin 'we see you (sgor pi)'. All the inverse forms are to be interpreted according to thispattern. Note that in the inclusive form the ki- prefix of second personis used, indicating the hierarchical ranking of second person over first,and the ki- prefix is also found throughout the 2 > 1 and 2 < 1 forms,where second person takes hierarchical precedence over first,

In an extensive articie on person hierarchies (1977) Zwicty makes theclaim that a 'hierarchy of reference' 1 > 2 > 3 is a universal, and that theordering in Algonkian languages is morphosyntactic, not referential, andcan consequently be dealt with at the morphosyntactic level withoutaffecting the universal. There are several problems with this approach.First of all it is misguided to claim that the 'meaning of a particularmorpheme can be expressed as a list of all the reference sets covered bythat morpheme' (Zwicky 1977: 717). In terms of linguistic method, thisis putting the cart before the horse: a particular morpheme has to meansomething BEFORE it can be used for reference: the meaning, therefore,cannot be in the reference. Second, the only role of the morphosyntax isto mark meanings; the morphosyntax is the perceivable vehicle thatconveys the meaning, without which the meaning cannot be conveyed.The separation of syntax and semantics is a remnant of an outmodedbehaviorist doctrine which claims that in order to be 'scientific' one hasto separate the 'signaling system' (which is directly observable, that is,

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 6: Person hierarchies in Algonkian and Inuktitut

866 /. Hewson

perceivable) from the 'semantics' (which is only conceivable and from abehaviorist point of view 'not scientific'). In Saussure's game of chess thesystem does not lie in the chess pieces, for all that they are the perceivablereality; likewise in languages the system is not in the morphosyntax, butin what is marked by the morphosyntax. This is why the morphosyntaxcan be irregular, when the cognitive system that it marks is quite regular:mice and men are just as regularly plural, in cognitive terms, as cats, dogs,and horses. Hjelmslev noted long ago (1971 [1959]: 28) that it is possibleto replace a lost chess piece by 'any conceivable object of a convenientsize', without changing the fundamental system of the game in any way.To attempt to find the system in the morphosyntax, therefore, which isnothing more than a generally imperfect reflection of the cognitive system,is a methodological confusion that is a direct result of a positivist bias.

Finally, given the nature of human consciousness, it is fundamental tomake an experiential distinction between the me and the not-me, andconsequently quite natural for this distinction to form the cognitive baseof systems of person. But given the arbitrary nature of the linguistic sign,which affects all aspects of language, there is no requirement for this me-centered cognitive basis to be universal, to the exclusion of all others.There is ample evidence from Cree, Blackfoot, Micmac, Ojibway, Fox,Menominee, etc., that in Algonkian languages it is what we (from ourown ethnocentric viewpoint) would call the second person that is thecognitive basis of the system of person in these languages.

In the Ojibway TA paradigm above, the direct and inverse markersshown in Figure 1 may be observed. These markers represent thefollowing functions: (1) -a:- = direct action on subordinate 3rd person;(1) -ik{oj- = in\trst; action of subordinate 3rd person on others; (3) -i- =direct action of 2nd person on 1st; -in- = inverse action of 1st person on2nd. The figure below may profitably be compared to that in Hockett

in

ik(o) ik(o)

Figure 1. Direct and inverse markers in Ojibwa

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 7: Person hierarchies in Algonkian and Inuktitut

Person hierarchies in Algonkian 867

(1966), where the status of direct and inverse forms is discussed atlength.

The Algonkian verb, as I have argued elsewhere (Hewson 1987), isfundamentally ergative in function; the direct forms, for example, agreewith the patient in both gender (an/inan) and number (sg/pl), as thefollowing Ojibway forms show: niwa:pama: Ί see him/her'; niwa:panta:nΊ see it'; niwa:pama:k Ί see them (an)': in this latter form the -k marksthe animate plural. In the inverse forms, of course, because of the inver-sion, such a marker relates to the agent: niwa:pamik 's/he sees me';niwa:pamiko:k 'they see me' (where -(o:)k still marks animate plural). Itwould be quite wrong, however, to conclude that the inverse forms arenominative in function: to interpret them in this way one would have toignore the role of -ik(o), the inverse marker, which simply inverts theergative roles, already established. If the direct forms were not fundamen-tally ergative, the inverse marker would not produce the effect that itdoes.

This point has to be made because of the confusions that surround thequestion of split ergativity, and because there is also the possibility ofinterpreting the Inuktitut U-shifted and 'inverse' forms (see next section),where the morphology is much less explicit, as being 'nominative ratherthan ergative based' (Vaxtin 1979: 283). Such an interpretation may beseen to be misleading once the role of the pronominal elements in theInuktitut transitive verb is fully understood.

Here again, however, there is the necessity of a caveat. If one thinksof ergativity as being simply a morphosyntactic category, one may wellnot only have difficulty acepting that a direct/inverse system can behmdaiTOTrtatty etgatwe, but eveu that there can be 'ergativity' without acase system. The point of view adopted here is functional and cognitive:ergtivity is seen as the procedure whereby the transitive verb is predicateddirectly of the patient rather than of the agent, and the morphosyntacticagreements are simply markers of this predication (for explicit details,with discussion of nominative, passive, and antipassive, see Hewson1987). Case systems are only one way of marking this functional strategy.Likewise there is more than one way of marking inverse forms, as Comriehas shown (1980). In short, a given functional strategy may not onlyoperate in different ways, but also be marked in different ways.

5. Transitivity and possession in Inuktitut

Since Inuktitut has dual number as well as singular and plural, theparadigms of the transitive verb, when laid out to fill all possible permuta-

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 8: Person hierarchies in Algonkian and Inuktitut

868 J. Hewson

tions of actor, are very complex. Such a paradigm may be seen in thethree pages presenting the 63 possible forms of the indicative in Smith(1977: 61-63): syncretism is ignored in Smith's presentation, the appropri-ate form being given for every possible combination of actors. All 63forms are compressed onto one page in Lowe's grammars of the dialectsof the Beaufort Sea region (for example, 1985a: 117, 1985b: 121, 1985c:130).

An important and recurrent pattern emerges from these Inuktituttransitive paradigms: third-person proximate is regularly unmarked. Thismeans that in the following items it is the argument that is italicized thathas no marker in the Inuktitut forms:

taku-va-ga I see himtaku-va-t You see himtaku-va-a He sees him

In these forms -va marks the verb as transitive, and -ga and -/ first- andsecond-person transitive agents respectively: the third-person patient isnot marked. In the remaining form, however, the -a has to be interpreted,as we shall see, as the goal of the transitive verb; this time it is the third-person agent that is unmarked.

We have already seen that this is precisely the pattern followed in theAlgonkian transitive animate paradigms, where the third-person proxi-mate makes the same shift from the role of patient with SAP subjects toagent in 3>3 forms (pr = proximate; U = unmarked):

A Igonkian Inuktitut1 >3pr 1 >3prU1 >3pr 2 >3prU3pr > 3obv 3prU > 3obv

In the transitive paradigms of Inuktitut, as we have seen, the stem ismarked for transitivity, but only the agent is marked in inflections of the1>3 and 2>3 forms, whereas in the 3pr>3obv forms only the patientis marked, as the following forms from Lowe (1985b: 121) clearly show.

utaqqi-ga-a 'he waits for him'utaqqi-ga-i 'he waits for them'utaqqi-ga-a-t 'they wait for him'utaqqi-ga-i-t 'they wait for them'

In this dialect -ga is the transitive marker, -a marks obviative singular,-/ marks obviative plural, and -t marks proximate plural, proximatesingular being unmarked. I shall call this recurrent pattern the U-shift(U = unmarked), and it will be found to apply not only to transitives, but

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 9: Person hierarchies in Algonkian and Inuktitut

Person hierarchies in Algonkian 869

also to possessives and to the marking of singular and plural, since theproximate plural marker -t (as in illu-t 'house-s') is not used for theobviative.

The syncretism that we have already mentioned as being a feature ofthe transitive paradigms is revealing. For ldual>3U and 2dual>3Uforms there is leveling of number in the patient, whereas in 3U > 3 formsthere is leveling of number in the agent, as the following forms fromSmith's Labrador Inuttut data (1977: 62) clearly show (3 = prox; 3' =obv):

Id >3/3d/3p takuvavuk 'we two see him/them both/them all'2d > 3/3d/3p takuvatik 'you two see him/them both/them all'3/3d/3p > 3' takuvauk 'he/they both/they all see him'3/3d/3p > 3'd takuvaagik 'he/they both/they all see them both'3/3d/3p > 3'p takuvait 'he/they both/they all see them all'

These patterns of leveling (1/2>3U vs. 3U>3', where 3U is unmarkedfor number) persist throughout the paradigm, but with less consistencythan in the examples given, showing that third person consistently behavesdifferently from first and second, indicating the different status of thirdperson from first and second. The unmarked status of 3U, for example,shows very clearly in taku-va-tik, which only differs from faku-vu-tik 'youboth see (intransitive)' by the fact that -va- is the verbal marker oftransitivity, whereas -vu- is the marker of intransitivity. In both formsthere is only one person marker: the -tik- that marks 2nd person dual.

Ulving (1987: 40-41) also points out that the possessive markers forthird person differ significantly from those for first and second and reportsthe contusion that this difference has caused among analysts of thelanguage. His Greenlandic data, using the noun illu-(t) 'house-(s)', showsthat whereas the suffixes mark the possessor in first- and second-personforms, the suffixes -a and -/ mark the possessum, 'while the 3sg. possessoris marked by zero' (1987: 41):

1 > 3U illu-ga (< *ka) 1 > 3Up illu-kka (< *illu-t-ka)2>3U illu-t 2>3Up illu-t-it3U>3' illu-a 3U>3'p illu-i3Up>3' illu-a-t 3pU>3'p illu-i-(t)Here again we can observe the effect of the U-shift. In the form illua 'hishouse', a- marks the noun as obviative singular. In illui 'his houses', -/marks the noun as obviative plural. In both cases the proximate possessoris unmarked, and need not be marked, since one can only have anobviative where there is already a proximate. In the forms illuat and illuitthe final -/ marks the proximate possessor as plural, whereas in illukka

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 10: Person hierarchies in Algonkian and Inuktitut

870 J. Hewson

(<*-tkd) and illutit it marks the proximate possessum as plural. This -tis in fact the regular proximate plural marker that is found on illut'houses' when the noun itself is proximate, not obviative. The shiftingrole of-/ in this paradigm is entirely dependent on the U-shift; it regularlypluralizes the unmarked proximate, whether this latter be possessor orpossessum.

6. Direct and inverse forms in Inuktitut

The Inuktitut transitive morphology of SAP > 3, when compared to 3>SAP, also shows that third person patterns differently from first andsecond, the speech-act participants. From an Algonkianist perspective, itis quite obvious that Inuktitut, in a rather subtle way, shows forms thatare similar to direct and inverse forms. Ulving gives us the followinginflections from Greenlandic:

1 > 3U taku-va-ga Ί see him' 3 > 1 taku-va-a-nga 'he sees me'2> 3U taku-va-t 'thou seest him' 3 > 2 taku-va-a-tit 'he sees thee'3U>3' taku-va-a 'he sees him'3U>3'pl taku-va-i 'he sees them'3U.pl >3obv taku-va-a-t 'they see him'3U.pl > 3obv.pl taku-va-i-t 'they see them'In these forms we find root taku- 'see', -va- transitive-verb marker, andin the 1 >3U and 2>3U forms agent markers that are already familiarfrom the possessive paradigms. We note, as well, that the 3prox patientin these forms is unmarked, as is, also the third ptoxvmate agenl vn the3 > 3' forms, a pattern that can now be seen to be a commonplace, andthat now distinguishes, as we shall see, the 'direct' forms of transitivityin Inuktitut from those we might call inverse forms (in the 'inverse' forms3prox surfaces and is marked). We also note the -a and -/, obviativesingular and plural markers, and -/, proximate plural marker, alreadyfamiliar from the possessive paradigms.

In the 3 > 1 (takuvaangd) and 3 > 2 (takuvaatit) forms the suffixes -ngafirst person and -tit second person are identical to those found in theintransitive paradigms, and the -a which marks the third-person agent isidentical in form to that which is found elsewhere as the third-personobviative marker. But the third-person element in these two verb formscannot be other than proximate (there is only one third person); this -ahere cannot be the obviative marker. The proof that third person here isproximate may be seen in the fact that takuvaanga in the Labrador dialectmeans 'he/they both/they all see me' and that takuvaatit means 'he/they

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 11: Person hierarchies in Algonkian and Inuktitut

Person hierarchies in Algonkian 871

both/they all see you' — the 3pr, as usual, is unmarked for number inthese forms, as it is elsewhere, and the obviative plural marker -/ is notto be found in this usage. Furthermore, in the West Greenlandic formscited by Ulving (1987: 46), where plural of the agent is marked, the pluralmarker is the -/ of the proximate plural: aki-va-a-tit 'he answered you';aki-va-a-t-tit 'they answered you'.

The syncretism that we see here would seem to stem from the fact thatthe third proximate element in these verbs is playing a marked role, asopposed to its normal unmarked role (the obviative being the markedform of the third person). The obviative is the third person that is differentfrom the unmarked proximate, and in this case the third person is playinga very different role, a role that is impossible in the paradigms of posses-sion in both Inuktitut and Algonkian, a role that leads to the creationof the Algonkian inverse paradigms as a result — the topsy-turvy formsof transitivity. The use of this marked morphology in Inuktitut is there-fore, as I have indicated above, a somewhat subtle parallel to the Algon-kian inverse forms.

In a morphology that is as convoluted as the Eskimo transitive para-digm, where 63 possible combinations of agent and patient have to berepresented, it is not surprising that there is syncretism where distinctionsof number are left unmarked. Equally it is not surprising that in twoinstances where third person plays a marked functional role, a commonmarked morphology may be shared in both cases, especially when thethird-person agent in the 'inverse' forms remains unmarked for number,true to its proximate status. This sharing of a common morphology isalso quite normal elsewhere: in the -k and -m forms of Yavapai (Kendall\915), for exampte, οτ \V\e t\o\m p\/possess\ve/vttb Sid sg of English(Hewson 1975, 1989), where the -s spelling marks a functional transcen-dence in all three categories (noun plural is any transcendence of thenotion of unit; possessive 's marks, as one grammarian put it 'the sphereof the possessor' [Curme 1931: 110]; and 3rd sg, as we have seen, tran-scends the phatic relationship of 1st and 2nd). We should not forget thatwhereas -s may be etymologically justified in the English genitive, it isless so in the noun plural and totally replaces etymological -th in theverbal paradigm: a common morphology has been found to be use-ful, and consequently its usage has been expanded over the course ofcenturies.

As Haiman comments (1980: 517), 'In much recent work, it has beenan article of faith that systematic syntactic homonymy is semanticallymotivated: similar morphological shape or syntactic behavior of (appa-rently disparate) categories may be an icon of their underlying semantichomogeneity.'

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 12: Person hierarchies in Algonkian and Inuktitut

872 /. Hewson

7. Conclusion

The problem of person is a true linguistic universal since, as Guillaumecomments, 'It is impossible to conceive of language without a certainsolution to the problem of person, seeing that language requires theconfrontation of an active, speaking person and a listener, active also bydint of his listening' (1987: 187; translation mine). Solutions to the prob-lem vary, but it is not surprising to find that third person, an elementtranscendent or exterior to the phatic relationship of first and second, isfrequently treated quite differently from these latter persons. It is alsonot surprising, given the necessarily always active role of first and second,to find them ranked in an iconically based hierarchy above third person.And the use of inverse forms, which require the representation to runcounter to these established hierarchies, can likewise be seen as a develop-ment to be expected, a necessary result once such highly iconic systemsof grammatical person have been established. We also find in languagesof the Tibeto-Burman family (one of the rare other families that hasdirect and inverse forms) an example of the 'split ergative type withergative syntax in transitive sentences with a third person subject andaccusative syntax with first or second person subjects' (Ebert 1987: 475),a situation that emphasizes the distinctive role of speech-act participants.

There is, of course, no one system of personal pronouns that is univer-sal. What is universal, however, as Guillaume has indicated (1987: 188),is the distinction between what is immanent to the linguistic system, andwhat transcends it: third person is always immanent to the system; firstand second, however, because they are the two poles between which theact of language takes p\ace, a*e estabfehed outside the system and changewith every change of speaker. They are necessarily deictic elements, in away that third person is not. It is not surprising that in child languagethe child initially uses its own name (that is, third person) for self-reference; use of first- and second-person pronouns is a later development(Jones 1970: 10).

In Inuktitut, the identity of morphological markers of possession inthe noun and of agents of the transitive verb has led linguists to (a) viewtransitivity as a subcategory of possession (Thalbitzer 1911; Hammerich1970; Schmitt 1976; Johns 1987) or (b) to view possession as a subcategoryof transitivity (Mey 1970; Rischel 1971; Kalmar 1979). Rather thanpropose that one function determines another, however, it is more appro-priate, in sound linguistic method, to derive them both from a commonunderlying source than to derive each from the other. We do not, inhistorical linguistics, derive daughter language A from daughter languageB; we derive both A and B from a prehistoric source that needs to be

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 13: Person hierarchies in Algonkian and Inuktitut

Person hierarchies in Algonkian 873

reconstructed. As Seiler has pointed out (1983a), what we call 'possession'is in fact a simple determining relationship that covers relations asdifferent as my book (Seller's 'establishing relationship'), my mother (Sell-er's 'inherent relationship' — every speaker has a mother), my song (thesong they always sing for me), and so on. Clearly the role of 'controller'of a transitive verb and of 'possessor' of a noun are similarly broaddetermining relationships, so that it is profitable for many languages toexploit a common morphology for both functions. Ulving (1987) in hisfinal paragraph expresses a similar point of view and points out that hehas argued 'that the fundamental function of this form [i.e. possessive]is to express a relationship of some kind'.

Finally we should note the importance of unmarked categories suchas third proximate in Inuktitut. The fact that this unmarked element isthe patient of the transitive verb in 1 > 3 and 2 > 3 forms, but naturallyand expectedly shifts to being the agent in 3 > 3' forms, just as it does inAlgonkian languages, lies behind many of the oddities in the transitiveparadigms of Inuktitut. Once one understands the different roles playedby this unmarked element, and the significance of the pattern of U-shift,much of the transitive morphology of Inuktitut becomes quitetransparent.

It is consequently not surprising that Comrie comments (1980: 66) thatthe inverse marker ne- in Chukchee (distantly related to Inuktitut), whichis used with (a) third-person singular agent, first- or second-personpatient, and (b) third-person plural agent, any patient, 'is often referredto as a third person prefix (although it does not occur with intransitiveverbs, nor with third person direct objects when the subject is third personsingular)'. It is obvious that the third person in these inverse forms isquite simply unmarked (another example of 3U), as it often is elsewhere:if there is an inverse marker for action on a first or second person andthe agent is unmarked, the agentive role will automatically be attributedto the third person.

Once one understands that all person hierarchies are based on the sameuniversal cognitive experience, the role of the inverse marker ne- inChukchee becomes a commonplace. The mistaken analysis of this elementas a 'third-person prefix' only occurs when one attempts a purely formalanalysis without any understanding of the cognitive function of personhierarchies.

Received 11 September 1990 Memorial University ofRevised version received Newfoundland8 March 1991

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 14: Person hierarchies in Algonkian and Inuktitut

874 /. Hewson

Note

1. I would like to thank an anonymous reviewer for Linguistics not only for helpfulsuggestions but also for challenging comments which led me to make a more explicitstatement of my own analytical methods. Correspondence address: Deparment of Lin-guistics, Memorial University of Newfoundland, St. John's, Newfoundland, CanadaA1B 3X9.

References

Allen, W. Sidney (1964). Transitivity and possession. Language 40, 337-343.Comrie, Bernard (1978). Ergativity. In Syntactic Typology, Winfred P. Lehmann (ed.),

329-394. Austin: University of Texas Press.—(1980). Inverse verb forms from Siberia: evidence from Chukchee, Koryak, and Kamcha-

dal. Folia Linguistica Historica 1, 61-74.Curme, George O. (1931). A Grammar of the English Language, vol. 3. Boston: Heath.DeLancey, Scott (1981). An interpretation of split ergativity and related patterns. Language

57, 626-657.Dixon, Robert M. W. (1979). Ergativity. Language 55, 59-138.Ebert, Karen H. (1987). Grammatical marking of speech act participants in Tibeto-Burman.

Journal of Pragmatics 11, 473-482.Fillmore, Charles J. (1968). The case for case. Universals in Linguistic Theory. Emmon Bach

and Robert Harms (eds.), 1-88: Holt, Rinehart and Winston.Foley, William A., and Van Valin, Robert D., Jr. (1984). Functional Syntax and Universal

Grammar. Cambridge: Cambridge University Press.Guillaume, Gustave (1987). Le ons de linguistique 8. Grammaire particuliere du francais et

grammaire generale. Quebec: Presses de FUniversite Laval; Lille: Presses Universitaires.Haiman, John (1980). The inconicity of grammar. Language 56, 515-540.Hammerich, L. L. (1970). The Eskimo Language. Fridtjof Nansen Minneforelesninger 6.

OAo.Hewson, John (1975). Derivation and inflection in English. In Studies in English Grammar,

A. Joly (ed.), 77-104. Lille: Presses de I'Universite de Lille.—(1987). Are Algonquian languages ergative? Proceedings of the 18th Algonquian Confer-

ence, William Cowan (ed.), 147-153. Ottawa: Carleton University.—(1989). Motivated syncretism. Journal of the Atlantic Provinces Linguistic Association 11,

39-56.Hjelmslev, Louis (1971 [1959]). Essais linguistiques, 2nd ed. Paris: Editions de Minuit.Hockett, Charles F. (1966). What Algonquian is really like. International Journal of Ameri-

can Linguistics 32, 59-73.Johns, Alana (1987). Inter-referential predication in Eskimo. In Proceedings of the 17th

Annual Meeting of N ELS, GLSA. Amherst, MA: University of Amherst.Jones, Robert M. (1970). System in Child Language. Cardiff: University of Wales Press.Kalmar, Ivan (1979). Case and Context in Inuktitut (Eskimo). Canadian Ethnology Service

Paper 49. Ottawa: National Museums of Canada.Kendall, Martha B. (1975). The /-K/,/-M/ problem in Yavapai syntax. Inernational Journal

of American Linguistics 41, 1-9.Klaiman, M. H. (1988). Affectedness and control: a typology of voice systems. In Passive

and Voice, Masayoshi Shibatani (ed.), 25-83. Amsterdam: Benjamins.

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 15: Person hierarchies in Algonkian and Inuktitut

Person hierarchies in Algonkian 875

Lowe, Ronald (1985a). Kangiryuarmiut Uqauhingita Ilihautdjutihangit. Basic KangiryuarmiutEskimo Grammar. Inuvik, North West Territories: Committee for Original Peoples En-titlement.

—(1985b). Uummarmiut Vqalungiha Ilihaurrutikrangit. Basic Uummarmiut Eskimo Gram-mar. Inuvuk, North West Territories: Committee for Original Peoples Entitlement.

—(1985c). Siglit Inuvialuit Uqausiita Ilsarviksait. Basic Siglic Inuvialuit Eskimo Grammar.Inuvik, North West Territories: Committee for Original Peoples Entitlement.

Mey, Jacob (1970). Possessive and transitive in Eskimo. Journal of Linguistics 6, 47-56.Plank, Frans (ed.) (1979). Ergativity. London, New York: Academic Press.—(1985). Die Ordnung der Personen. Folia Linguistica 19, 111-176.Rischel, J0rgen (1971). Some characteristics of noun phrases in West Greenlandic. Ada

Linguislica Hafnensia 13, 213-245.Schmitt, Alfred (1976). Der nominale Character des sogennanten Verbums der Eskimo-

Sprache. Zeitschrift für Vergleichende Sprachforschung 73, 27-45.Seiler, Hansjakob (1983a). Possession as an Operational Dimension of Language. Tübingen:

Narr.—(1983b). Possessivity, subject and object. Studies in Language 7, 89-117.Silverstein, Michael (1976). Hierarchy and features of ergativity. In Grammatical Categories

in Australian Languages, Robert M. W. Dixon (ed.), 112-171. Canberra: AustralianInstitute of Aboriginal Studies.

Smith, Lawrence R. (1977). Some Grammatical Aspects of Labrador Inuttut (Eskimo). Mer-cury Series No. 37. Ottawa: National Museums of Canada.

Speck, Frank G. (1935). Naskapi. Norman: University of Oklahoma Press.Thalbitzer, William (1911). Eskimo: an illustrative sketch. BAE Bulletin 40 (1), 965-1069.Ulving, Tor (1987). The Eskimo suffixes -a and -/: markers of what? International Journal

of American Linguistics 53, 39-60.Vaxtin, N. B. (1979). Nominal and verbal ergativity in Asiatic Eskimo: splits in the person

and mood paradigms. In Ergativity, Frans Plank (ed.), 279-305. London: AcademicPress.

Zwicky, Arnold M. (1977). Hierarchies of person. Chicago Linguistic Society 13, 714-733.

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM

Page 16: Person hierarchies in Algonkian and Inuktitut

Brought to you by | University of Sussex (University of Sussex)Authenticated | 172.16.1.226

Download Date | 7/18/12 10:44 PM