Inalienable Possession and Referential Coding in Choctaw

Inalienable Possession and Choctaw Referential CodingAuthor(s): William D. DaviesSource: International Journal of American Linguistics, Vol. 50, No. 4 (Oct., 1984), pp. 384-402Published by: The University of Chicago PressStable URL: http://www.jstor.org/stable/1265399 .

Accessed: 11/11/2014 22:34

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access toInternational Journal of American Linguistics.

http://www.jstor.org

This content downloaded from 128.255.6.125 on Tue, 11 Nov 2014 22:34:42 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=ucpress

http://www.jstor.org/stable/1265399?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


INALIENABLE POSSESSION AND CHOCTAW REFERENTIAL CODING'

WILLIAM D. DAVIES

CORNELL UNIVERSITY

0. Introduction. Languages have a wide range of phenomena for grammatically signaling relationships among clausal and sentential elements. A subclass of these, such as agreement, pronominalization, reflexives, switch-reference, and others, can be termed NP referential coding phenomena since information referring to particular attributes (e.g., person, number) of NPs is morphologically coded on other elements in the clause or sentence.

As in other languages, in Choctaw, a Muskogean language, coding rules generally reference the head noun of an NP. However, Choctaw provides systematic counterevidence to any claim that referential coding rules refer only to the head noun of an NP. In this article I examine verb agreement, switch-reference, and reflexive data in which a possessor may optionally supersede an inalienably possessed body part as the NP referenced by the coding rules.

In arguing for the account proposed here, I reject a Possessor Ascen- sion analysis of the form proposed for various languages (Allen et al., in preparation; Bell 1983; Crain 1979; and others). Rejection of an ascension analysis for the data under consideration is based on the independence of referencing possessors for purposes of verb agreement and switch-reference in the same sentence. This independence runs counter to predictions of a Possessor Ascension analysis.

1. Choctaw possessive morphology. Possession in Choctaw is most commonly expressed on the possessed noun by a prefix that marks the person and number of the possessor. The possessor, if it occurs overtly

I The Choctaw data presented in this article were elicited during fieldwork conducted in and around Broken Bow, Oklahoma, during 1980 and 1981-82 with a speaker from that area living in San Diego. This work was supported in part by the National Science Foundation through grant BNS 78-17498 to the University of California, San Diego, and by the American Philosophical Society through a Phillips Fund grant. I would like to thank Margaret Langdon, David Perlmutter, Leslie Saxon, and Pamela Munro for helpful comments and discussion of some of the material presented here. I alone am responsible for errors, omissions, and the conclusions drawn herein.

[IJAL, vol. 50, no. 4, October 1984, pp. 384-402] O 1984 by The University of Chicago. All rights reserved. 0020-7071/ 84/5004-0003$01.00

384



CHOCTAW REFERENTIAL CODING

as an independent noun or pronoun, precedes the possessed noun. Pos- session is marked by two sets of prefixes. Inalienable possession, asso- ciated with body parts and kin terms, is generally marked by a prefix from the set of accusative agreement markers (cf. 2).2

(la) sa-yyi 1 Poss-leg

'my leg' (Ib) chi-shki

2Poss-mother

'your mother'

(lc) hattak-ma ippashi man-DDt hair

'the man's hair'

Not all body parts and kin terms take possessive prefixes from the set used to denote inalienable possession, however. Certain words which have other uses in the language are marked by affixes from the set of dative verb agreement markers (cf. 2), which generally mark alienable possession. For instance, when such a prefix is affixed to either the "relational noun"3 ashaka 'behind' or the verb anokfila 'think', the words can function as body parts.

(2a) am-ashaka 1 Poss-behind

'my behind'

(2b) chim-anokfila 2Poss-think

'your mind'

2 Despite the fact that prefixes marking possession are isomorphic with accusative and dative verb agreement markers (and perhaps are in fact the same markers), I gloss these

prefixes as possessive to make clear their function in these contexts. No point of theory or

analysis relevant to the present discussion rests on this decision. In the presentation of Choctaw data, I use the traditional Choctaw orthography for

consonants only. For vowels I adopt the convention followed by many Muskogean scholars

(cf. Booker 1980, Munro and Gordon 1982, and Nicklas 1975) representing the phonemic vowels as i (front unrounded), o (back rounded), and a (back unrounded). In addition, I

mark length by the diacritic: and nasalization by - for ease of exposition. The abbrevia- tions used in the morphemic glosses are: 1 = first person, 2 = second person, 3 = third

person, Acc = accusative, Ben = benefactive, Comp = complementizer, Dat = dative, DDt = distal determiner, DS = different-subject marker, Dt = determiner, Nom = nomi-

native, PI = plural, Poss = possessive, Pred = predicative, Pst = past tense, Q = question particle, Refl = reflexive, Sg = singular, and SS = same-subject marker.

3 The term "relational noun" is due to Nicklas (1974).

385



INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

However, for the most part these prefixes are used to denote ownership when marking possessors.

(3a) am-ofi 1Poss-dog

'my dog'

(3b) alla --towa child 3Poss-ball

'the child's ball'

(3c) chi-chokka 2Poss-house

'your house'

2. Verb agreement. Subjects, direct objects, indirect objects, and beneficiaries are marked by affixes (differentiated for case) on the predicate of the clause in which they occur.4 For the most part, this can be described as follows:

(4) Verb Agreement Rules5 a. Subjects determine nominative agreement. b. Direct objects determine accusative agreement. c. Indirect objects determine dative agreement. d. Beneficiaries determine benefactive agreement.

This agreement is illustrated in the clauses in (5)-(7).

(5) Chi-bashli-li-tok. 2Acc-cut-lNom-Pst

'I cut you'.

(6) Holisso chim-a:-li-tok. book 2Dat-give-1 Nom-Pst

'I gave the book to you'.

(7) Alla off imi Ihioli-li-tok. child dog 3Ben chase-lNom-Pst

'I chased the dog for the child'.

4 Choctaw verb agreement has been referred to as inseparable pronouns (Byington 1870), echoes (Nicklas 1974 and Jacob, Nicklas, and Spencer 1977), and case affixes (Heath 1977). Additionally, some have used semantic notions such as actor (Jacob, Nicklas, and Spencer 1977), agentive (Heath 1977), and patient (Heath 1977 and Jacob, Nicklas, and Spencer 1977) rather than the notions used here. See Davies (1981a) for arguments against a purely semantic characterization of Choctaw verb agreement.

5 In Davies (1981a) I argue that the rules in (4) must apply to each NP in a clause disjunctively in the order c, b, a, d. See that discussion for further details and examples of these rules.

386




In (5), the first-person singular subject is marked by the first-person singular nominative morpheme li and the second-person singular direct object is marked by the second-person accusative morpheme chi.6 In (6), a first-person singular nominative morpheme again marks the subject; also a 0-morpheme third-person accusative agreement marker is determined by holisso 'book'.7 In addition, the indirect object is marked by a second-person singular dative morpheme chim. The clause in (7) illus- trates benefactive agreement; the third-person beneficiary, alla 'child', determines the third-person benefactive morpheme imi. Verb agreement is obligatory.

The situation is more complex in clauses with a possessed body part as subject or direct object.

(8a) Sa-yyi:-t basha-h.8 IPoss-foot=Dt-Nom cut-Pred

'My foot is cut'.

(8b) Sa-yyi:-t sa-basha-h. 1 Poss-foot=Dt-Nom 1 Acc-cut-Pred

'My foot is cut/ I am cut on my foot'.

(9a) Chi-nishkin-a-t hottopa-h-o? 2Poss-eye-Dt-Nom hurt-Pred-Q 'Do your eyes hurt?'

(9b) Chi-nishkin-a-t chi-hottopa-h-o? 2Poss-eye-Dt-Nom 2Acc-hurt-Pred-Q 'Do your eyes hurt?/ Do you hurt in your eyes?'

(IOa) Ofi':-t sa-bbak kopoli-tok. dog=Dt-Nom 1Poss-hand bite-Pst

'The dog bit my hand'.

(10b) Ofi:-t sa-bbak sa-kopoli-tok. dog= Dt-Nom 1Poss-hand 1 Acc-bite-Pst

'The dog bit my hand/The dog bit me on the hand'.

6 As in many other Amerindian languages, unemphatic pronouns do not occur in Choctaw utterances for the most part.

7 Third-person nominative and accusative agreement morphemes are 0-morphemes. 8 Although most of the consultants who supplied the data for this study often use the h

predicate or tense marker only when predicates and nouns might be confused due to word order, full or partial homophony, or other conditions, I use the suffix in the present study to conform to the general practice of Choctaw scholars.

With vowel-final nouns the determiner a is optionally realized as lengthening of the final vowel, as indicated in (8) and other data. Alternatively, a rule of y-epenthesis inserts a y before a when immediately preceded by the final vowel of a noun. Thus, there are alterna- tions such as ofi:t - ofyvat 'dog=Dt=Nom', alla:t - allayat 'child=Dt=Nom', and so on.

387




The (a) clauses in (8)-(10) have the expected agreement morphemes. In each case the head noun of the NP is third-person singular and determines agreement, which is realized as a 0-morpheme. However, in the (b) clauses there is an accusative morpheme agreeing in person and number with the possessor of each body part. Since the agreement rules generally reference the person and number of the head noun of an NP, the problem is to account for the accusative morphemes in the verbal complexes in (8b), (9b), and (1Ob), which reference the possessors.

2.1. Are the accusative morphemes in the (b) clauses agreement mor-

phemes? One might hypothesize that the possessive prefixes have simply been copied onto the predicates of these clauses. However, a prefix- copying analysis is ruled out by the data in (1l) and (12).

( la) Issoba:-t am-ashaka habli-tok. horse=Dt-Nom 1Poss-behind kick-Pst

'The horse kicked my behind'.

(1 lb) Issoba:-t am-ashaka sa-habli-tok. horse=Dt-Nom 1Poss-behind 1Acc-kick-Pst

'The horse kicked my behind/The horse kicked me in my behind'.

(12a) Am-anokfila:-t okpolo-h. 1Poss-think=Dt-Nom broken-Pred

'My mind is troubled'.

(12b) Am-anokfila:-t si-okpolo-h. 1 Poss-think=Dt-Nom 1 Acc-broken-Pred

'My mind is troubled/ I am troubled in my mind'.

In (11), the direct object is a possessed body part, am-ashaka 'my behind'. Here the first-person possessive prefix is am, the prefix generally used to show alienable possession. The hypothesis claiming that the accusative

morphemes in (8b), (9b), and (1Ob) are copies of the possessive prefixes predicts that the morpheme marking the possessor in the verbal complex in both (llb) and (12b) will be am. However, the morpheme which in fact occurs is the first-person singular accusative agreement morpheme. This fact (in addition to switch-reference data considered in 3) casts doubt on the copying analysis.

Additional evidence that these accusative morphemes are agreement markers is the fact that the agreement normally determined by NPs in these environments is accusative agreement. Given the rule in (4), one

expects accusative agreement with direct objects, as in (13) and (14).

388




(13) Ofi:-t sa-kopoli-tok. dog=Dt-Nom 1Acc-bite-Pst

'The dog bit me'.

(14) Issoba:-t sa-habli-tok. horse=Dt-Nom 1 Acc-kick-Pst

'The horse kicked me'.

In both (13) and (14), the direct object is marked by the first-person singular accusative morpheme sa. Compare these to (1Ob) and (1 lb), in which a body part functioning as direct object has a first-person singular possessor and there is a first-person singular accusative agreement mor-

pheme in the verbal complex. Since direct objects determine accusative

agreement, the accusative morphemes in (10b) and (1 lb) are arguably agreement markers determined by possessors, which have optionally superseded their head nouns in the determination of agreement.

Likewise, subjects of clauses containing the predicates basha 'cut' and

hottopa 'hurt' determine accusative agreement.

(15) Sa-basha-h. 1 Acc-cut-Pred

'I am cut'.

(16) Chi-hottopa-h-o? 2Acc-hurt-Pred-Q 'Are you hurt?'

In both (15) and (16), the subject of the clause is marked by an accusative agreement morpheme, sa 'lAcc' in (15) and chi '2Acc' in (16). At first this seems to contradict the agreement generalization in (4) that

subjects determine nominative agreement. However, I have argued elsewhere (Davies 1981a) that the subjects of clauses such as (15) and (16) are actually direct objects at the initial level of structure and subjects at the final level of structure-what is referred to as an unaccusative structure in the relational grammar literature (Perlmutter 1978 and Perlmut- ter and Postal 1984). The accusative agreement marking is consistent with the unaccusative analysis since the NPs in (15) and (16) are initial direct objects. Since the subjects of basha and hottopa determine accusative agreement, the accusative morphemes in (8b) and (9b) are consistent with an analysis in which possessors optionally supersede the head in determining agreement. The fact that these accusative morphemes occur only in environments in which one expects accusative agreement indicates that the morphemes may well be agreement morphemes.

389




Further evidence that the morpheme is an agreement marker comes from a particular Choctaw construction. For one class of verbs it is possible for what appears to be a direct object to determine dative agreement. This is illustrated by the clause pairs in (17) and (18).

(17a) Chi-alikchi-li-tok. 2Acc-doctor-l Nom-Pst

'I doctored you'. (I 7b) Chim-alikchi-li-tok.

2Dat-doctor-l Nom-Pst

'I doctored you'.

(18a) Issoba shilli-li-tok. horse comb-I Nom-Pst

'I combed the horse'.

(18b) Issoba r-shilli-li-tok. horse 3Dat-comb-l Nom-Pst

'I combed the horse'.

In (17a) and (18a), as expected, the direct objects are marked in the verbal complex by accusative agreement morphemes, chi '2Acc' and a 0-morpheme, respectively. However, in the (b) clauses the same NPs are marked by the dative agreement morphemes chim '2Dat' and T '3Dat'. Elsewhere (Davies 1981a; 1982) I have presented evidence that clauses such as (17b) and (18b) have a structure in which an initial direct object is demoted to indirect object. This accounts for the dative agreement morphemes and other facts which need not concern us here.

The data in (19) illustrate an interaction between possessed body parts and this clausal construction.

(19a) Oho:yo:-t sa-ppashi shilli-tok. woman=Dt-Nom IPoss-hair comb-Pst

'The woman combed my hair'.

(19b) Oho:yo:-t sa-ppashi T-shilli-tok. woman=Dt-Nom 1Poss-hair 3Dat-comb-Pst


(19c) Oho:yo:-t sa-ppashi a-shilli-tok. woman=Dt-Nom 1Poss-hair 1Dat-comb-Pst


In (19a), sa-ppishi 'my hair' is a final direct object and determines the expected third-person accusative agreement 0-morpheme. In (19b), sa-

390




ppashi determines a third-person dative agreement morpheme, T, signaling that it has been demoted to indirect object. However, the verbal

complex in (19c) contains a first-person singular dative agreement morpheme, j, which agrees in person and number with the possessor of the body part. This can be accounted for as a case in which sa-ppashi 'my hair' has become an indirect object and its possessor determines agreement in the verbal complex. As we have seen in (17b), (18b), and (19b), this environment manifests dative agreement. The clause in (19c) therefore provides evidence that this morpheme is an agreement morpheme.9

The indirect object demotion data provide an argument against an analysis in which both the head noun and its possessor determine agreement. Based on the (b) clauses in (8)-(12), one might claim that the possessor determines agreement in addition to the head noun. Since the head noun is third person and the third-person accusative agreement marker is a 0-morpheme, there would be no overt manifestation of that agreement. However, (19c) falsifies such an analysis since the head noun ippashi 'hair' does not determine agreement; if it did, there would be a third-person dative agreement marker in the verbal complex, which there is not.

The lack of third-person agreement in (19c) provides striking evidence that the dative morpheme agreeing in person and number with the possessor is an agreement marker. Some type of agreement with the NP sa-ppashi is necessary for grammaticality. The first-person singular dative marker fills this role and is therefore an agreement marker. Here the possessor assumes the place of the head noun for agreement purposes.

2.2. Not all possessors may optionally supersede their head nouns and determine accusative agreement, however.?1 Possessors of alienably possessed nouns cannot determine accusative agreement.

9 Note that a possessive prefix copying analysis fails to account for (19c) since the

possessive prefix and the morpheme in the verbal complex are distinct. 10 The construction under consideration is distinct from the possessive construction

exemplified in (ib). (ia) Chim-alla:-t cha:ha-h.

2Poss-child=Dt-Nom tall-Pred 'Your child is tall'.

(ib) Alla:-t chT-cha:ha-h. child=Dt-Nom 2Dat-tall-Pred 'Your child is tall'.

In (ib), the possessor is marked by a dative agreement marker (and the possessive marking fails to occur on the possessed noun). Elsewhere (Davies 1981a; 1981b) I have analyzed this as a Possessor Ascension construction. That construction differs from the phenomenon being considered here in a number of ways. Most crucial is the fact that the raised

possessor invariably determines dative agreement rather than the agreement one would

391




(20a) Issoba:-t am-ofi habli-tok. horse=Dt-Nom 1Poss-dog kick-Pst

'The horse kicked my dog'.

(20b) *Issoba:-t am-ofi sa-habli-tok.

(21a) Am-ofi:-t hohchafo-h. 1Poss-dog=Dt-Nom hungry-Pred

'My dog is hungry'.

(21b) *Am-ofi:-t sa-hohchafo-h. The agreement option is not merely restricted to inalienably possessed

nouns. Although most kin terms belong to the class of inalienably possessed nouns in Choctaw (cf. 1), possessors of kins terms cannot determine accusative agreement.

(22a) Ofi:-t si-oshi kopoli-tok. dog=Dt-Nom 1Poss-son bite-Pst

'The dog bit my son'.

(22b) * Ofi:-t si-oshi sa-kopoli-tok.

(23a) Sa-shki:-t nayokpa-h. 1Poss-mother=Dt-Nom happy-Pred

'My mother is happy'.

(23b) *Sa-shki:-t sa-nayokpa-h. In addition, possessors of alienably possessed body parts cannot

determine accusative agreement. In (24), am-iyyi talhlha.pi'my five legs' refers to legs that I own, not to legs attached to my body. When the possessor of an alienably possessed body part determines accusative agreement, as in (24b), the clause is ungrammatical.

(24a) Am-iyyi talhlha.pi arpa asha:chi-li-tok. 1Poss-leg five table put-1Nom-Pst 'I put my five legs on the table'.

expect if the possessor were superseding the head noun for purposes of verb agreement. Subjects of the predicate cha:ha 'tall' determine accusative agreement, as in (ii).

(ii) Sa-cha:ha-h. Acc-tall-Pred

'I am tall'. Were the raised possessor of the subject of (i) to determine accusative agreement, the clause would be ungrammatical.

(iii) *Alla:-t chi-cha:ha-h. child=Dt-Nom 2Acc-tall-Pred ('Your child is tall'.)

These facts establish that the Possessor Ascension construction is distinct from the construction under consideration in this work.

392




(24b) *Am-iyyi talhlha:pi aTpa si-asha:chi-li-tok.

Thus, in Choctaw, only possessors of inalienably possessed body parts may take the place of the head noun for agreement purposes. It is therefore necessary to include a statement such as the following in the grammar of Choctaw:

(25) The possessor of an inalienably possessed body part may optionally supersede the body part in the determination of verb

agreement. In the sections which follow, I show that statements such as (25) must also be formulated for switch-reference and interclausal reflexives and that the phenomenon is subject to a further restriction.

3. Switch-reference. Another form of referential coding in Choctaw is the switch-reference system, which morphologically distinguishes whether or not clause pairs have coreferential subjects.

(26) Switch-Reference Marking" a. Same-subject (SS) marking occurs if the subject of clause A is

coreferential with the subject of clause B.

b. Different-subject (DS) marking occurs if the subject of clause A is not coreferential with the subject of clause B.

Switch-reference marking is illustrated in (27) and (28).

(27a) Ofi pashohli-li-cha tamaha ia-li-tok. dog rub-lNom-SS town go-lNom-Pst

'I patted the dog and went to town'.

(27b) * Ofipashohli-li-na tamaha ia-li-tok. DS

(28a) Tobi apa-li-na tachi ish-pa-tok. bean eat-lNom-DS corn 2Nom-eat-Pst

'I ate beans and you ate corn'.

(28b) *Tobi apa-li-cha tachi ish-pa-tok. SS

In (27), the subject of each clause is 'I' and coordination can be marked with the SS suffix cha (27a) but not with the DS suffix na (27b). Con- versely, in (28), the subject of the first clause is 'I' and the subject of the second is 'you'; here DS marking occurs (28a) but SS marking cannot (28b).

l See Davies (198 la; 1984) for more detailed discussion of switch-reference in Choctaw.

393




However, when a body part is the subject of a clause and the subject of the other clause is the possessor of the body part, either SS or DS marking is possible.

(29a) Sa-nishkin-a-t hottopa-na okhTsh chopa-li-tok. 1Poss-eye-Dt-Nom hurt-DS medicine buy-lNom-Pst

'My eyes hurt, so I bought some medicine'.

(29b) Sa-nishkin-a-t hottopa-cha okh?sh chopa-li-tok. 1Poss-eye-Dt-Nom hurt-SS medicine buy-lNom-Ost


(30a) Oka lawa tahli-li-na sa-ttakoba:-t water much finish-INom-DS Poss-stomach=Dt-Nom

hottopa-h. hurt-Pred

'I had a lot of water, and my stomach hurts'.

(30b) Oka lawa tahli-li-cha sa-ttakoba:-t water much finish-lNom-SS 1Poss-stomach=Dt-Nom

hottopa-h. hurt-Pred

'I had a lot of water, and my stomach hurts'.

(31a) Sa-ppashi:-t lakna-ka ikha:na-li-h. 1Poss-hair=Dt-Nom yellow-Comp=DS know-lNom-Pred

'I know my hair is yellow'.

(31b) Sa-ppashi:-t lakna-ka-t ikha:na-li-h. 1Poss-hair=Dt-Nom yellow-Comp-SS know- Nom-Pred

'I know my hair is yellow'.

(29) and (30) show that the body part can occur in either the first or second of two conjoined clauses. In (31), the possessor is the subject of the matrix clause and the body part is the subject of the embedded clause.

The switch-reference data in (29)-(31) exhibit the same pattern as the verb agreement data in 2, that is, possessors of inalienably possessed body parts optionally take the place of the body parts for switch- reference marking. Once again, possessors of alienably possessed nouns (32b) and inalienably possessed kin terms (33b) cannot determine switch- reference marking.

394




(32a) Am-ofi:-t abi:ka-na yaya-li-tok. Poss-dog=Dt-Nom sick-DS cry-lNom-Pred

'My dog was sick, and I cried'.

(32b) *Am-ofi:-t abi:ka-cha yaya-li-tok.

(33a) Sa-shki:-t hottopa-na alikchi pis-t I Poss-mother=Dt-Nom hurt-DS doctor see-SS

ia-li-tok. go- Nom-Pst

'My mother was hurt, and I went to see the doctor'.

(33b) *Sa-shki:-t hottopa-cha alikchi pis-t is-li-tok.

To account for the above switch-reference facts, the following statement can be included in the grammar of Choctaw.

(34) The possessor of an inalienably possessed body part may optionally supersede the body part in the determination of switch- reference marking.

4. Interclausal reflexives. Optional reference of possessors of body parts also occurs in the interclausal reflexive construction. As described in Davies (1981a), in certain environments12 the subject of an embedded intransitive clause can optionally be marked by a reflexive agreement morpheme if coreferent with the matrix subject. This is illustrated in (35b).

(35a) Sa-kapassa-h sa-yimmi-h. 1Acc-cold-Pred 1 Acc-believe-Pred

'I believe I'm cold'.

(35b) Ili-kapassa-h sa-yimmi-h. Refl-cold-Pred 1 Acc-believe-Pred

'I believe myself to be cold'.

When a possessed body part occurs as the subject in a similar environment, a reflexive agreement morpheme may occur on the embedded predicate if the possessor is coreferent with the matrix subject.

(36a) Sa-yyi:-t kobqffa-h sa-yimmi-h. lPoss-foot=Dt-Nom break-Pred 1Acc-believe-Pred

'I believe my foot is broken'.

12 Aside from coreference of the two subjects, the interclausal reflexive construction is

subject to two conditions: (i) the matrix predicate must belong to a small class of verbs whose members include yimmi'believe', banna 'want', ahwa 'think', and others; and (ii) the embedded predicate must belong to the class of unaccusative predicates (cf. 2).

395




(36b) Sa-yyi:-t ili-kobaffa-h sa-yimmi-h. lPoss-foot=Dt-Nom Refl-break-Pred 1Acc-believe-Pred

'I believe my foot is broken'.

(37a) Sa-noshkobo:-t hottopa-h sa-yimmi-h. lPoss-head=Dt-Nom hurt-Pred 1Acc-believe-Pred 'I believe my head hurts'.

(37b) Sa-noshkobo:-t ili-hottopa-h sa-yimmi-h. lPoss-head=Dt-Nom Refl-hurt-Pred 1Acc-believe-Pred

'I believe my head hurts'.

In (36a) and (37a), a third-person accusative 0-morpheme marks agreement with the body-part subjects in the embedded clauses. However, the complement verbal complex in both (36b) and (37b) contains a reflexive agreement morpheme.

Given the coreference condition which holds for all Choctaw reflexivi- zation, such data might seem problematic. However, the parallel between these reflexive data and the verb agreement and switch-reference data is unmistakable. In both (36b) and (37b), an inalienably possessed body part is the subject of an embedded clause whose predicate sanctions interclausal reflexives, and the possessor and the matrix subject are coreferent. Therefore, if the interclausal reflexive rule is sensitive to the possessor, the proper coreference conditions obtain and reflexive morphology in the embedded clause follows. Thus the grammar of Choctaw should contain the following:

(38) The possessor of an inalienably possessed body part may optionally supersede the body part and be referenced by the interclausal reflexive rule.

5. A condition on optional reference. The preceding discussion treats verb agreement, switch-reference, and interclausal reflexive rules optionally referencing a possessor rather than a body part as unrelated phenomena. In 5.1 I formulate a single statement, which includes a syntactic condition, to account for optional reference. In addition, I present evidence against an alternative analysis in 5.2.

5.1. Not all possessors of body parts may determine agreement. In the examples in 2, the body parts are all direct objects at some level of structure.13 When the body part is the subject of an unergative clause (an

13 Recall the discussion of the unaccusative and direct object demotion structures in 2.

396




intransitive clause in which the subject is a subject at all levels of structure; Perlmutter 1978 and Perlmutter and Postal 1984), the possessor cannot determine agreement.

(39a) Sa-yyi:-t hilha-tok. 1Poss-foot=Dt-Nom dance-Pst

'My feet danced'.

(39b) *Sa-yyi:-t hilha-li-tok. 1 Poss-foot= Dt-Nom dance- 1 Nom-Pst

('My feet danced'.)

(39c) *Sa-yyi:-t sa-hilha-tok. 1Poss-foot= Dt-Nom 1 Acc-dance-Pst

('My feet danced'.) In (39b), the first-person singular possessor of 'feet' determines agreement, li 'lNom', and the clause is ungrammatical. If this possessor could supersede its head and determine agreement, we would expect it to determine nominative agreement because, being an unergative clause in which the subject bears the subject relation at all levels of structure, the subject determines nominative agreement (following the rule in 4), as in (40).

(40) Hilha-li-tok. dance-1 Nom-Pst

'I danced'.

(39c), in which the possessor is marked by the accusative agreement morpheme, is also ungrammatical; however, this follows from the fact that accusative agreement is inappropriate for unergative subjects.

Neither can the possessor determine agreement when the body part is the subject of a transitive clause.

(41a) Sa-bbak-a-t okwissoshi Ihopolli-tok. 1 Poss-hand-Dt-Nom window go=through-Pst

'My hand went through the window'.

(41b) *Sa-bbak-a-t okwissoshi Ihopolli-li-tok. 1Poss-hand-Dt-Nom window go=through-lNom-Pst

('My hand went through the window'.) (41 c) * Sa-bbak-a-t okwissoshi sa-lhopolli-tok.

1 Poss-hand-Dt-Nom window 1 Acc-go=through-Pst ('My hand went through the window'.)

Again, if the possessor were able to supersede the body part in the determination of agreement, we would expect it to determine nominative

397




agreement. As in unergative clauses, the subject of an initially and finally transitive clause in Choctaw is a subject at all levels of structure and therefore determines nominative agreement.

(42) Okwissoshi Ihopolli-li-tok. window go=through-l Nom-Pst

'I went through the window'.

The ungrammaticality of (41b) and (41c) parallels the ungrammaticality of (39b) and (39c).

These data point to the fact that possessors can determine agreement only when the body part bears the direct object relation at some level of structure, perhaps restricted to the initial level.14

The same restriction is relevant to the switch-reference phenomena. In those cases in 3 in which the possessor is considered for switch-reference marking, the body part is the subejct of an unaccusative clause, that is, the body part is initially a direct object. When the body part is an unergative or transitive subject, switch-reference marking cannot be sensitive to the possessor.

(43a) Sa-yyi:-t hilha-na sa-nayokpa-tok. Poss-foot=Dt-Nom dance-DS I Acc-happy-Pst

'My feet danced, and I was happy'.

(43b) *Sa-yyi:-t hilha-cha sa-nayokpa-tok.

(44a) Sa-bbak-a-t okwissoshi Ihopolli-na 1Poss-hand-Dt-Nom window go=through-DS

yaya-li-tok. cry-1Nom-Pst

'My hand went through the window and I cried'.

(44b) * Sa-bbak-a-t okwissoshi Ihopolli-cha yaya-li-tok. In both (43) and (44), the subject of one clause is a possessed body part that is not a direct object at any level of structure. When DS marking occurs on the first clauses, the sentences are grammatical, (43a) and (44a), because the rule of DS marking (26b) is satisfied. If the switch- reference rule were sensitive to the possessor, the SS marking condition would be satisfied; however, when SS marking occurs on the first clause, the sentences are ungrammatical, (43b) and (44b).'5

14 Given the semantics of such clauses, I have found it impossible to test whether or not

possessors of body parts functioning as initial indirect objects or beneficiaries can determine agreement.

15 The interclausal reflexive construction does not provide a parallel argument that the body part must be a direct object since this construction is possible only when the embedded clause is unaccusative (cf. n. 10).

398




Given the data, the following restriction must be placed on the optional reference of possessors:

(45) A Choctaw referential coding rule may be sensitive to the possessor of a body part if and only if:

i. the body part is the head of the NP, ii. the body part is inalienably possessed, and

iii. the NP bears the direct object relation to its clause.

The term "referential coding rule" in (45) encompasses the verb agreement, switch-reference, and interclausal reflexive phenomena.

5.2. One might hypothesize that the verb agreement, switch-reference, and interclausal reflexive facts would be best accounted for by positing a rule of Possessor Ascension such as (46).

(46) Possessor Ascension

The possessor of an inalienably possessed body part may optionally ascend out of its NP construction and assume the grammatical relation of the NP of which the body part is head.16

The ascension analysis would account for the data considered thus far since the possessor would then be the head noun of the NP which bears the direct object relation and would therefore be referenced by coding rules such as verb agreement, switch-reference, and interclausal reflexivi- zation. Possessor Ascension constructions have been posited for Cebuano (Bell 1983), Chamorro (Crain 1979), Haya (Hyman 1977), and Southern Tiwa (Allen et al., in preparation) to account for data in those languages which parallel the Choctaw data considered here: cases in which the possessor displays grammatical behavior expected of the possessed noun.

The ascension analysis makes an interesting prediction. Under it, the possessor determines verb agreement, switch-reference marking, or reflexive morphology because after raising the possessor bears the grammatical relation previously borne by the NP of which the body part is head. Therefore, there should be a correlation between switch-reference marking and verb agreement; that is, when the possessor determines agreement, it should also determine same-subject marking given the proper environment.

However, there is no correlation between the option to reference a possessor for agreement and switch-reference marking. The data in (29)-(31) show that the possessor can be referenced for switch-reference

16 The Possessor Ascension rule in (46), as well as the proposals cited for other languages, is consistent with the Relational Succession Law proposed by Perlmutter and Postal (1983), which states that a raised nominal assumes the grammatical relation borne by the NP out of which it ascends.

399




in the absence of verb agreement with the possessor. Further evidence of the independence of the two is given in the sentences in (47), which are counterparts of (29) in which the possessor determines agreement in the first clause.l7

(47a) Sa-nishkin-a-t sa-hottopa-na okhish chopa-li-tok. 1Poss-eye-Dt-Nom 1Acc-hurt-DS medicine buy-lNom-Pst


(47b) Sa-nishkin-a-t sa-hottopa-cha okhish chopa-li-tok. 1Poss-eye-Dt-Nom lAcc-hurt-SS medicine buy-lNom-Pst


Even if the possessor determines agreement, DS marking is possible (47a).

(45) correctly makes no prediction about correlations among the specific phenomena. The analysis simply asserts that a referential coding rule may optionally reference a possessor. Since verb agreement and switch-reference are independent, the data in (29)-(31) and (47) are consistent with the proposed analysis.

6. Conclusion. I have presented an account of Choctaw verb agreement, switch-reference, and interclausal reflexives in which a possessor optionally supersedes the head of an NP and is directly referenced by coding rules. I have shown that the Choctaw data are incompatible with a type of Possessor Ascension analysis which has been proposed for other,languages to account for similar data. One may then reasonably

17 I show in Davies (1981a; 1984) that for some speakers of Choctaw either SS or DS

marking is possible whenever an unaccusative clause is conjoined with a clause that has a

subject coreferent to the subject of the unaccusative clause, as in (i). (ia) Sa-hohchafo-cha tobi nonachi-li-tok.

IAcc-hungry-SS bean cook-lNom-Pst 'I was hungry and cooked some beans'.

(ib) Sa-hohchafo-na tobi nonachi-li-tok.

1Acc-hungry-DS bean cook-lNom-Pst 'I was hungry and cooked some beans'.

Given these facts, one might claim that the apparent independence of the option to reference a possessor for agreement and switch-reference marking is due to the possibility of DS marking in this environment; the Possessor Ascension analysis then would not be a counterexample to the data in (29)-(31) and (47). However, as stated in Davies (1981a; 1984), some speakers do not accept DS marking in this environment and consider (ib) ungrammatical. These same speakers nevertheless consider (29)-(31) and (47) fully grammatical. Therefore, on the basis of these speakers' judgments, the independence of the

option to reference a possessor for agreement and switch-reference marking does provide counterevidence to the Possessor Ascension analysis.

400




wonder whether the type of analysis proposed here could account for Possessor Ascension data in other languages, thus eliminating the Pos- sessor Ascension construction. However, examination of these other data reveals that the devices assumed here cannot adequately handle all such cases. In particular, these Possessor Ascension analyses gain important support from the interaction of Possessor Ascension with grammatical constructions such as passive and other "subject-creating" rules, relative clause formation, subject-to-object raising, and language-particular conditions of one form or another.'8 Thus, the claim here is not that the Choctaw data undermine Possessor Ascension analyses proposed for other languages. Rather, these data provide evidence that coding rules do not always reference the head of an NP specified by a rule, a result which can be checked for other languages.

REFERENCES

ALLEN, B.; ET AL. In preparation. Syntactic levels and possessor ascension in Southern Tiwa.

BELL S. J. 1983. Advancements and ascensions in Cebuano. Studies in Relational Gram- mar 1, ed. D. M. Perlmutter. Chicago: University of Chicago Press.

BOOKER, K. 1980. Comparative Muskogean: aspects of proto-Muskogean verb morphology. Ph.D. dissertation, University of Kansas.

BYINGTON, C. 1870. Grammar of the Choctaw language, ed. D. Brinton. Proceedings of the American Philosophical Society 11:317-67.

CRAIN, C. 1979. Advancements and ascensions in Chamorro. Linguistic Notes from La Jolla 6:3-32.

DAVIES, W. D. 1981a. Choctaw clause structure. Ph.D. dissertation, University of Cali- fornia, San Diego.

1981b. Possessor ascension in Choctaw. Proceedings of the Eleventh Annual Meeting of the North East Linguistic Society, pp. 38-57.

1982. 2-3 retreat, the notion 'absolutive', and levels of grammatical relations. Pro- ceedings of the First Annual West Coast Conference on Formal Linguistics, pp. 307-18.

.1984. Choctaw switch-reference and levels of syntactic representation. Syntax and Semantics, vol. 16: The Syntax of Native American Languages, ed. E. D. Cook and D. Gerdts. New York: Academic Press.

HEATH, J. 1977. Choctaw cases. Berkeley Linguistics Society 3:204-18. HYMAN, L. 1977. The syntax of body parts. Haya Grammatical Structure, ed. E. Byaru-

shengo, A. Duranti, and L. Hyman. Southern California Occasional Papers in Linguis- tics, no. 6.

JACOB, B.; D. NICKLAS; AND B. L. SPENCER. 1977. Introduction to Choctaw. Durant: Choctaw Bilingual Education Program, Southeastern Oklahoma State University.

MUNRO, P., and L. GORDON. 1982. Syntactic relations in Western Muskogean. Language 58:81-115.

18 The reader is referred to the cited works for the specific data supporting a Possessor Ascension analysis in particular languages.

401



402 INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS

NICKLAS, T. D. 1974. The elements of Choctaw. Ph.D. dissertation, University of Michigan. 1975. Choctaw morphophonemics. Studies in Southeastern Linguistics, ed. J. M.

Crawford. Athens: University of Georgia Press. PERLMUTTER, D. M. 1978. Impersonal passives and the Unaccusative Hypothesis. Berkeley

Linguistics Society 4:157-89. ,AND P. M. POSTAL. 1983. The Relational Succession Law. Studies in Relational

Grammar 1, ed. D. M. Perlmutter. Chicago: University of Chicago Press. , AND P. M. POSTAL. 1984. The 1-Advancement Exclusiveness Law. Studies in

Relational Grammar 2, ed. D. M. Perlmutter and C. Rosen. Chicago: University of

Chicago Press.



Inalienable Possession and Referential Coding in Choctaw

Documents