Quick viewing(Text Mode)

Morphological Variation in Cree Inflection*

Morphological Variation in Cree Inflection*

Morphological variation in *

Eun Bi Kang University of Manitoba

This paper discusses two cases of morphological variation found in the conjunct order verb inflection of Cree dialects. The first half of the paper examines the distribution of the reflexes of the Pre-Cree markers *-ik and *-wa:w among the Western Cree dialects. Syntactic and crosslinguistic evidence suggests that the distribution of these markers is influenced by morphological dissimilation. The second half of the paper examines the patterning of three obviative markers, the reflexes of Pre-Cree *-ah, *-im and *-ri. The reflexes of *-ah are concluded to be general obviative markers while *-im and *-ri are found to serve more specific functions that vary depending on the dialect.

1 Introduction

This paper is concerned with two instances of variation in the conjunct order of verb inflection across Cree dialects: (1) the alternation of two different third person plural markers and (2) the existence of three distinct obviative markers. It will be shown that the patterning of the plural markers results from morphological dissimilation while each of the three obviative markers serves a slightly different function. After some background on important concepts in Algonquian linguistics (§2), section 3 presents an analysis of the distribution of reflexes of the Pre-Cree plural markers *-ik and *-wa:w, specifically focusing on the Plains, Swampy, Moose and Atikamekw dialects of Cree (henceforth collectively referred to as the Western Cree dialects). I will argue that the factors conditioning the distribution of these involve their syntactic positions along the clausal spine (§3.1) and the grammatical features they carry (§3.2). In section 3.2, crosslinguistic data from Spanish and Basque will be used to argue that the inability of the reflexes of *-ik and *-wa:w in the Western Cree dialects to co-occur is a result of morphological dissimilation implemented through a process of “obliteration” (Arregi & Nevins, 2006). Section 3.3 discusses the nature of and number in Cree and the implications highlighted by the proposed analysis of the reflexes of *-ik and *-wa:w. Section 4 turns to the obviative markers. Section 4.1 addresses the lack of explicit obviative verb in configurations in which one argument is proximate and the other is obviative. I will propose that Cree treats such configurations as “default” obviative configurations due to the presence of a proximate argument. In light of this, I will then propose that the reflexes of the Proto-Algonquian obviative suffix *- ah can be seen as a non-default configuration obviative marker. In section 4.2, I will propose that the reflexes of the Proto-Algonquian obviative markers *-im and *-ri serve as participant obviative markers, that is, that they agree with specific features possessed by the obviative argument. Although the distribution of these reflexes varies by dialect, I will approach the data by grouping the dialects according to what specific features their reflexes of *-im and *-ri agree for in section 4.3. Two groups emerge through this approach: one where the usage of the obviative markers is dependent on what appears to be a nominative- accusative alignment unique to obviative participants, while the other employs the obviative markers based on the obviative participant’s position on the Person Hierarchy relative to the other participant in the alignment.

* I wish to thank Will Oxford for his invaluable guidance and feedback.

Toronto Working Papers in Linguistics (TWPL), Volume 39 © 2017 Eun Bi Kang EUN BI KANG

In section 5, I will discuss the markedness of obviation by observing how obviative morphology seems to be exempt from rules that govern non-obviative morphology. In addition, I will compare the grammatical treatment of the obviative morphology discussed in section 4 and the pluralisers in section 3, showing that the former defy expectations given their low ranking on the Person Hierarchy. Finally, in section 6, I will conclude my paper by summarising my findings.

2 Background on Algonquian morphosyntax

This section provides background on Algonquian morphosyntax that is relevant to the discussion in this paper: the distinction between independent and conjunct inflection (§2.1), the central and peripheral layers of agreement inflection (§2.2), the person hierarchy (§2.3), and obviation (§2.4).

2.1 Independent and conjunct orders

Most Algonquian have two distinct yet parallel sets of verbal inflection known as the independent and conjunct orders. While both orders mark the same features, the shapes of the morphemes used in the two orders are generally different (Oxford, 2014: 8). The use of the two sets of inflection is determined by clause type: independent order morphology is generally found in main clauses while conjunct order morphology is generally used in subordinate clauses (Oxford, 2014: 8-9). This paper focuses on inflectional forms from the conjunct order.

2.2 Central and peripheral inflection

Some of the morphology that appear on verbs (as well as ) can be identified as belonging to one of two slots: central inflection and peripheral inflection. Central inflection on Cree verbs in the independent order consists of a person and a “central suffix”, which marks person and number, whereas peripheral inflection is composed of a single ending that is lexically specified for third person and can inflect for gender (animate/inanimate), number, and obviation for third persons only (Goddard, 1974: 318). Illustrative examples are given for Plains Cree in (1) below (Wolfart, 1973). Central morphemes are indicated by italicised text whereas peripheral inflection is underlined; morphemes that are neither italicised nor underlined do not belong to either category of inflection. All examples henceforth will be given in Plains Cree unless otherwise noted. A list of abbreviations can be found in the appendix at the end of the paper.

(1) a. ni- pahpi -n 1- laugh -NON3 ‘I laugh’

b. ni- pahpi -na:n 1- laugh -1P ‘We (excl.) laugh’

c. nipa: -w -ak sleep -3 -3P ‘They sleep’

d. ni- wa:pam -a: -na:n -ak 1- see -DIR -1P -3P ‘We (excl.) see them’

2 MORPHOLOGICAL VARIATION IN THE CONJUNCT ORDER AMONG THE CREE DIALECTS

Since (1a) and (1b) are intransitive sentences with only first person arguments, no peripheral inflection occurs. In (1c), the sole argument is a third person, so peripheral inflection appears, as well as a reduced form of third person central agreement (-w). In (1d) both central and peripheral inflection are present. In the conjunct order, peripheral inflection patterns the same as in the independent order, albeit with different phonological shapes. Central inflection in the conjunct order, on the other hand, consists solely of a single suffix that marks both person and number, illustrated in (2), entirely different from the prefix-suffix combinations that occur in the independent order:

(2) wa:pam -a: -ya:hk -ik see -DIR -1P -3P ‘(that) we (excl.) see them’

The patterning of central and peripheral inflection can be roughly described as follows. When a verb has two arguments, central inflection agrees with the argument that is ranked higher on the Person Hierarchy while peripheral inflection agrees with the lower ranked participant (Oxford, 2014: 38). (The Person Hierarchy will be explained further in section 2.3 below.) This description holds for most Cree verb forms, but there are a few exceptional cases in which central inflection indexes both participants using a portmanteau agreement , such as the -ak suffix in (3):

(3) wa:pam -Ø -ak -ik see -DIR -1S:3 -3P ‘(that) I see them’

The participants indexed by central and peripheral inflection are known as the “central participant” and “peripheral participant”, respectively (Goddard, 1974). It is important to note that the central and peripheral participants are not equivalent to the logical subject and object of any given sentence.1 Subject and object roles are instead indicated by a separate morpheme known as the theme sign: the direct theme sign aligns the role of subject with the central participant while the inverse theme sign aligns this role with the peripheral participant (Goddard, 1974: 319). The direct-inverse pattern is suspended, however, in many forms in which the logical object is first or second person, in which case the theme sign marks the person of the object rather than “direct” or “inverse”:

(4) a. wa:pam -i -yan see -1OBJ -2S ‘(that) you (sg) see me’

b. wa:pam -is -k -ik see -2OBJ -3 -3P ‘(that) they see you (sg)’

Lastly, when discussing transitive verbs, it is convention among Algonquianists to represent the configuration of participants in the following format: φ1—φ2, where φ1 represents the person and number of the logical subject and φ2 represents the person and number of the logical object. For example, 3P—1S indicates a transitive configuration wherein the subject is animate third person plural and the object is first person singular. In this paper, I will be using φ1↔φ2 as a shorthand to indicate all transitive configurations that include the participants φ1 and φ2 regardless of their role. For example, 3↔3′ refers to all configurations

1 I use the terms “(logical) subject” and “(logical) object” to indicate the external and internal arguments of a transitive verb, respectively.

3 EUN BI KANG that include an animate third person participant as well as an animate obviative third person participant, such as the configurations 3S—3′, 3P—3′, 3′—3S and 3′—3P.

2.3 Person Hierarchy

The direct-inverse alignment system is governed by a Person Hierarchy which ranks a verb’s arguments according to centrality in the discourse; the more central a participant is, the higher its rank. The Person Hierarchy is normally formulated as follows (Pentland, 1999: 235):

(5) 1st/2nd > indefinite/passive > 3rd animate proximate > 3rd animate obviative > 3rd inanimate

The Person Hierarchy is central to the proper functioning of the direct-inverse alignment system as it governs the designation of the central and peripheral participants as well as the selection of direct/inverse theme signs. As mentioned above, central and peripheral inflection agree with the higher ranked and the lower ranked participants, respectively.

2.4 Obviation

When a sentence involves more than one third person, Cree, as with the other , distinguishes the third persons in accordance to their centrality to the discourse. The “proximate” is the third person that is most central to the discourse. Proximates are morphologically unmarked. All other third persons are overtly marked as “obviative” (Dahlstrom, 1991: 91). Obviative inflection in Cree is number- neutral, so obviative participants are grammatically ambiguous between singular and plural.

3 Reflexes of *-ik vs. *-wa:w

Cree dialects employ two different third person pluralisers in the conjunct order, the reflexes of Pre- Cree *-ik and *-wa:w. (Pre-Cree is an intermediate stage of the between Proto-Algonquian and the modern Cree dialects (Oxford, 2014: 200-201)). In this section I will offer an analysis of the conditions accompanying their usage. Table 1 presents a cross-dialectal of the usage of third person pluralisers in conjunct indicative forms. Bolded text indicates a reflex of *-wa:w while italicised text indicates a reflex of *-ik. Dashes indicate that no relevant morpheme is present in the form while crossed cells indicate forms which are unattested or were not available in the sources consulted. Some dialects may use both suffixes depending on the type of verb stem, such instances are represented by a forward slash. As Table 1 shows, with the exception of , the western dialects consistently only use one pluraliser, -ik, in the indicative mood whereas in the eastern dialects, the distribution of the pluralisers appears somewhat haphazard. It would seem that Plains, Swampy, Moose and Atikamekw have phased out -wa:w. However, a comparison of the conjunct subjunctive forms in the western dialects reveals a different picture, as illustrated by Table 2. Here the western Cree dialects, again with the exception of Woods Cree, consistently utilise - wa:w as the pluraliser, excluding -ik almost entirely from the subjunctive paradigm. While it may be tempting to simply conclude that -ik and -wa:w are third person plural fusional morphemes that also mark indicative and subjunctive mood respectively, this cannot be the case, as allomorphs of -wa:w are found in indicative preterite forms as well, as shown in (6).

(6) Moose Cree a. wa:pam -Ø -ak -wa: -pan see -DIR -1S:3 -3P -PRET ‘(that) I saw them’ (Ellis, 1971: 91)

4 MORPHOLOGICAL VARIATION IN THE CONJUNCT ORDER AMONG THE CREE DIALECTS

Table 1: Conjunct indicative forms including a plural 3rd person2 Verb Plains Woods Swampy Moose Atikamekw SE Cree NE Sheshatsiu Form Cree Innu 3P -ik -ik/-wa:w -ik -ik -ik -wa:w -c -t 0P — —/-wa:w — — — -wa:w -c —/-a:w 1S—3P -ik -wa:w -ik -ik -ik -wa:w -wa:w -a:w 2S—3P -ik -ik -ik -ik -ik -wa:w -wa:w -a:w 1P—3P -ik -ik -ik -ik -ik -wa:w — — 12—3P -ik -wa:w -ik -ik -ik -wa:w -c -t 2P—3P -ik -ik -ik -ik -ik -c -c -t X—3P -ik -ik -ik -ik -wa:w -c -t 3P—1S -ik -ik -ik -ik -ik -wa:w -c -:t 3P—2S -ik -wa:w -ik -ik -ik -wa:w -c -a:w 3P—1P -ik -ik -ik -ik -ik -wa:w — — 3P—12 -ik -wa:w -ik -ik -ik -c -c -t 3P—2P -ik -wa:w -ik -ik -ik -c -c -t 3P—0 -ik -wa:w -ik -ik -ik -wa:w -c -(a:w)

Table 2: Conjunct subjunctive forms including a plural 3rd person Verb Form Plains Woods Swampy Moose Atikamekw 3P -wa:w -wa:w -wa:w -wa:w -wa:w 0P -wa:w -wa:w -wa:w -wa:w 1S—3P -wa:w -wa:w -wa:w -wa:w -wa:w 2S—3P -wa:w -wa:w -wa:w -wa:w -wa:w 1P—3P -wa:w — -wa:w -wa:w -wa:w 12—3P -wa:w — -wa:w -wa:w -wa:w 2P—3P -wa:w — -wa:w -wa:w -wa:w X—3P -wa:w -wa:w -wa:w 3P—1S -wa:w -wa:w -wa:w -wa:w -wa:w 3P—2S -wa:w -wa:w -wa:w -wa:w 3P—1P -wa:w -wa:w -wa:w -wa:w 3P—12 -wa:w -wa:w -wa:w -wa:w -wa:w 3P—2P -wa:w — -wa:w -wa:w -wa:w 3P—0 -wa:w -wa:w -wa:w -ik -wa:w

As such, it cannot be said that grammatical mood is the factor determining which pluraliser is used. Thus, the mechanism enabling this alternation in the Plains, Swampy, Moose and Atikamekw dialects of Cree must lie elsewhere.

3.1 The syntax of -ik and -wa:w

We can gain insight into the distribution of the pluralisers by considering the structure of the verb complex. As shown in the examples below, the pluraliser -ik always occurs in -final position:

2 Data collected from Wolfart 1973 (Plains), Starks 1992 (Woods), Ellis 1983 (Swampy), Ellis 1971 (Moose), Beland 1978 (Atikamekw), Junker 2000-2014 (), Clarke 1982 (Sheshatsiu Innu)

5 EUN BI KANG

(7) Plains Cree a. wa:pam -Ø -ak -ik see -DIR -1S:3 -3P ‘(that) I see them’ (Wolfart, 1973: 42)

Woods Cree b. wa:pam -i -c -ik see -1OBJ -3 -3P ‘(that) they see me’ (Starks, 1992: 285)

Moose Cree c. wa:pam -Ø -akihc -ik see -DIR -1P:3 -3P ‘(that) we (excl.) see them’ (Ellis, 1971: 90)

Atikamekw d. wa:pam -Ø -ac -ik see -DIR -2S:3 -3P ‘(that) you (sg) see them’ (Beland, 1978: 65)

In contrast, -wa:w is always followed by another suffix:

(8) Plains Cree a. wa:pam -Ø -ak -wa:w -i see -DIR -1S:3 -3P -SJV ‘(that) if I see them’ (Wolfart, 1973: 42)

Woods Cree b. wa:pam -i -t -wa:w -i see -1OBJ -3 -3P -SJV ‘(that) if they see me’ (Starks, 1992: 287)

Moose Cree c. wa:pam -Ø -akiht -wa:w -e: see -DIR -1P:3 -3P -SJV ‘(that) if we (excl.) see them’ (Ellis, 1971: 93)

Atikamekw d. wa:pam -Ø -at -wa:w -e: see -DIR -2S:3 -3P -SJV ‘(that) if you (sg) see them’ (Beland, 1978: 73)

From the data presented above, it is inconclusive as to whether -ik and -wa:w both realise the same syntactic position. While -ik always appears at the end of the verb complex and -wa:w is always the penultimate morpheme, we could imagine that a null indicative mood marker -Ø is present after -ik in the examples in (7), thus putting it in the same morphological slot (and thus, assuming the Mirror Principle, the same syntactic position) as -wa:w. However, a case can be made for the pluralisers occupying different syntactic positions once data from other conjunct paradigms are taken into account. Consider the conjunct dubitative preterite forms in Moose Cree and Atikamekw shown in (9):

6 MORPHOLOGICAL VARIATION IN THE CONJUNCT ORDER AMONG THE CREE DIALECTS

(9) Moose Cree a. wa:pam -a: -w -ak -epan -e:n -ak see -DIR -DUB -1S:3 -PRET -DUB -3P ‘(that) maybe I saw them’ (Ellis, 1971: 92-93)

Atikamekw b. wa:pam -a: -w -ak -ipan -e:n -ak see -DIR -DUB -1S:3 -PRET -DUB -3P ‘(that) maybe I saw them’ (Beland, 1978: 82-85)

These forms contain the peripheral suffix -ak, which is an allomorph of third person plural -ik that is also found on nouns and independent order verbs. The dubitative preterite forms show that -ak/-ik appears to the right of the preterite morpheme -epan, whereas, as shown above in (6), the pluraliser -wa:(w) appears to the left of -epan. If we assume the Mirror Principle, this difference in morpheme order strongly suggests that -ik and -wa:w do occupy different syntactic positions along the clausal spine. The syntactic structure that underlies the Algonquian verb complex may be schematised as in (10), modified from Oxford (2014: 179) (cf. Halle & Marantz, 1993):

(10)

The order of morphemes in the Cree dubitative preterite forms in (9) maps to this sequence of heads straightforwardly, as shown in (11) (assuming iterated head movement from to C).

7 EUN BI KANG

(11) Moose Cree

As the mapping in (11) suggests, I propose that -ak/-ik is found in the CP layer of the clausal spine. Locating peripheral agreement in such a high position is not a novel proposal: Branigan and MacKenzie (1999: 478) propose a similar analysis of peripheral agreement in Innu, noting the similarity with agreement morphology that appears on appear on Dutch complementisers in embedded clauses:

(12) da-n-ze zunder komen that-3P they come (Branigan & MacKenzie, 1999: 478)

If -ik is found in C, then what is to be made of -wa:w? In Oxford’s analysis of Proto-Algonquian verb structure, it is proposed that the “inner suffix” portion of central inflection, that is, the portion of central inflection which does not include person , in the independent order forms a part of the TP, or as he calls it, the InflP (Oxford, 2014: 176-228). Hamilton supports this model in his analysis of verb morphology in Mi’gmaq, another Algonquian language (Hamilton, 2013). Since conjunct order central inflection only consists of the suffix, this “inner suffix” is automatically assumed to be T. Referring back to form outlined in (11), the T in wa:pama:wakepane:nak ‘(that) maybe I saw them’ would then be -ak ‘1S:3’. I propose that the pluraliser -wa:w is actually also a component of central inflection that occurs alongside morphemes like -ak ‘1S:3’ and co-resides in the TP. This may appear strange since -wa:w is a third person pluraliser and the third person is a peripheral participant in all the relevant forms; how can a morpheme agreeing for a peripheral participant constitute a part of central inflection? However, as noted in section 2.2 above, the functional separation of central and peripheral inflection is not as clear-cut in the conjunct order as it is in the independent, as there are portmanteau central suffixes such as -ak ‘1S:3’ that index features of both arguments. From a morphological standpoint, it would appear as though both arguments are being treated as central participants, since portmanteau suffixes occur where one would normally expect the central suffix in an independent order clause to appear. As such, -wa:w can then be considered to be a kind of central pluraliser, one that is utilised when the peripheral ending slot is inaccessible such as in a conjunct subjunctive form like wa:pamakwa:we: ‘if I see them’, as illustrated by (13). Unlike other moods such as the dubitative, I propose that the subjunctive marker occupies C instead of Mod, since, despite its name, the Cree subjunctive has the function of a complementiser (‘if’ in the form in (13)) rather than a modal. Given the proximity of -wa:w to the central portmanteau suffix -ak ‘1S:3’ in (13) as opposed to that of -ik in (11), I analyze -wa:w as a second component of central inflection (i.e., a second component of the realization of T) expressed through morphological fission.

8 MORPHOLOGICAL VARIATION IN THE CONJUNCT ORDER AMONG THE CREE DIALECTS

(13) Moose Cree

While analyzing the pluralisers -wa:w and -ik as realizing different heads (T and C) captures their different positions in the inflectional template, it does not explain why they do not co-occur. I suggested above that the pluraliser -wa:w surfaces in T when the peripheral suffix slot (i.e., C) is inaccessible to number marking. However, there is nothing inherent within the syntax that forbids one morpheme from surfacing alongside another except in a scenario where they would both be competing to realise the same terminal node—a scenario which is precisely the opposite of what my analysis proposes, since -ik and -wa:w are identified with different terminal nodes. Furthermore, since multiple third person markers are fully acceptable in forms like Plains/Moose/Atikamekw wa:pamahkik ‘(that) they see it’, where the central suffix -k and peripheral suffix -ik both agree for third person, what is the reason for the ungrammaticality of forms like *wa:pamakwa:wik ‘(that) I see them’, where the central suffix -wa:w and the peripheral suffix -ik both agree for third person plural? We must conclude that syntax alone is not sufficient in explaining the distribution of -ik and -wa:w and other factors must be examined.

3.2 Morphological dissimilation

Although -ik and -wa:w do not share the same syntactic position, they do presumably share the same grammatical features. A deeper understanding of the alternation of -ik and -wa:w can be gleaned from a consideration of crosslinguistic evidence. Cases analogous to that of -ik and -wa:w—in which two morphemes with similar or identical features should theoretically be able to co-occur but in reality cannot— have been documented in Spanish and Basque. Spanish possesses a set of pronominal inflected for case and although it permits the occurrence of multiple neighbouring pronominal clitics, it does not allow two adjacent clitics to have identical person features when one of them is dative. Instead, as Nevins (2007) demonstrates, the relevant must be substituted with the impersonal/reflexive pronoun se:

(14) a. El premio, lo dieron a Pedro ayer the prize 3-ACC gave-P to Pedro yesterday ‘The prize, they gave to Pedro yesterday.’

9 EUN BI KANG

b. A Pedro, le dieron el premio ayer to Pedro 3-DAT gave-P the prize yesterday ‘To Pedro, they gave the prize yesterday.’

c. *A Pedro, el premio, le lo dieron ayer to Pedro the prize 3-DAT 3-ACC gave-P yesterday ‘To Pedro, the prize, they gave yesterday.’

d. A Pedro, el premio se lo dieron ayer to Pedro the prize se 3-ACC gave-P yesterday ‘To Pedro, the prize, they gave yesterday.’ (Nevins, 2007: 275)

In other work, Arregi and Nevins (2006) and Nevins (2010) observe that Basque does not tolerate configurations in which multiple [+PARTICIPANT] features (i.e., first and/or second persons) are present; offending configurations are repaired by deleting one of the first/second person morphemes:

(15) a. (Suk guri emon) d- o- sku- su → d- o- su you us gave 3S.A TR 1P.DAT 2S.E 3S.A INT 2S.E ‘You gave it to us.’ (Arregi and Nevins, 2006: 9)

b. (Guk hiri emon) d- o- tzu- u → d- a- tzu we you gave 3S.A TR 2S.DAT 1P.E 3S.A INT 2S.DAT ‘We gave you it.’ (Nevins, 2010: 14)

As (14) and (15) show, both Spanish and Basque use dissimilation strategies to resolve instances in which two featurally identical morphemes are adjacent to each other. In Spanish, the pronominal clitics le and lo are third person, meaning they both possess a [–PARTICIPANT] feature. In order to avoid identical adjacent features, one of the clitics is replaced with an impersonal reflexive (Nevins, 2007: 275). On the other hand, Basque removes, or “obliterates”, one of the offending morphemes entirely (Arregi and Nevins, 2006; Nevins, 2010: 14). As third person pluralisers, Cree -ik and -wa:w both share the features [–PARTICIPANT, +PLURAL]. I propose that, similar to how Spanish and Basque do not tolerate adjacent [–PARTICIPANT] features, the Western Cree dialects do not tolerate multiple, though not necessarily adjacent, [–PARTICIPANT, +PLURAL] features, and that the absence of -wa:w in the conjunct indicative forms is a result of the same dissimilation strategy employed by Basque: obliteration. Thus, a form containing both -ik and -wa:w such as *wa:pamakwa:wik is ungrammatical and must be repaired; the structurally lower of the two featurally identical suffixes, -wa:w, is deleted, and the realised surface form becomes wa:pamakik ‘(that) I see them’. When the subjunctive marker occupies the C position, thus precluding the realization -ik in this position, - wa:w automatically surfaces in T, as the absence of -ik in C means that there is no longer a conditioning environment for the obliteration of -wa:w in T.

3.3 The nature of person and number in Cree

A morphological dissimilation analysis of -ik and -wa:w provides insight into the distribution of these markers and helps to highlight the differential treatment of number and person in Cree morphosyntax. Just as the Person Hierarchy designates certain persons as “more important” than others, there appears to be a more general designation of person itself as “more important” than number. For example, while the person features of the two arguments of a transitive are always marked, the number features are not; inanimate and obviative participants in transitive configurations are number-neutral. The absence of number marking for inanimate and obviative participants may be tied to their low ranking on the Person Hierarchy. Meanwhile,

10 MORPHOLOGICAL VARIATION IN THE CONJUNCT ORDER AMONG THE CREE DIALECTS for proximate third persons, number is marked, but it cannot be double-marked: in a form where plural -ik and -wa:w ought to both appear, only -ik is permitted.

4 Obviative markers

In this section, I will present and discuss my observations concerning the distribution of the various obviative morphemes found in the Cree dialects. As with the third person pluralisers, the Cree dialects possess multiple morphemes that mark obviation in conjunct indicative forms, though only two are in complementary distribution. These morphemes are the reflexes of Pre-Cree *-ah, *-im and *-ri.

4.1 Default and non-default configurations

Before discussing the distribution and usage of the obviative morphemes, it is important to note in all the Cree dialects, transitive forms in which one argument is proximate and the other is obviative lack any obviative marking at all despite including an obviative participant. This is illustrated in Table 3, where any obviative morphology is indicated in bold. These forms include these specific configurations: 3S—3′, 3P—3′, 3′—3S and 3′—3P. The only exception is the Southern East Cree dialect, which uses obviative -h, a reflex of *-ah, when the proximate participant is plural (Junker, 2000-2014).

Table 3: Configurations lacking obviative marking across the Cree dialects Verb Plains Woods Swampy Moose Atikamekw SE Cree NE Sheshatsiu Form Cree Innu 3S—3′ -a:t -a:t -a:t -a:t -a:t -a:t -a:t -a:t 3P—3′ -a:cik -a:cik -a:cik -a:cik -a:cik -a:twa:wh -a:c -a:t 3′—3S -ikot -ikot -ikot -ekot -ikot -ikoth -ikot -ikot 3′—3P -ikocik -ikocik -ikocik -ekocik -ikocik -ikotwa:wh -ikoc -ikot

What is striking about these forms is that, although the usage of obviative markers varies between dialects as will be discussed below, all dialects but Southern East Cree consistently lack obviative marking in these configurations. I propose that the absence of obviative marking from these forms arises from factors that are analogous to the factors that motivate ellipsis. The key point is that in the forms in Table 3, it is fully predictable that one of the arguments must be obviative, since, as stated earlier, obviation is obligatory whenever a transitive clause has two animate third person arguments. In other , when both arguments are animate third persons, the presence of a proximate entails the presence of an obviative. Since the existence of an obviative is fully recoverable from the existence of a proximate, overt obviative morphology can be omitted, just as ellipsis allows syntactic material to be omitted when its content is recoverable from the context. An analysis along these lines would be consistent with Pentland’s (1999) use of the terms “predictable obviative” and “unpredictable obviative”: a predictable obviative, as in 3↔3′ configurations, need not be marked by overt obviative morphology, but an unpredictable obviative, as in 1↔3′ configurations, does require overt marking. From a formal perspective, however, Pentland’s description presupposes a feature [UNPREDICTABLE], which would be unconventional and unprecedented. I suggest that an analysis drawing on parallels with ellipsis, as suggested above, is preferable as it allows for better theoretical integration. Regardless of whether or not it is the presence of a proximate argument that is the cause behind the lack of obviative morphology in the 3↔3′ forms, it is evident that the Cree dialects treat these forms separately from the rest. As such, I shall treat these forms as comprising a “default” obviative configuration; any configuration that departs from the parameters of this default requires explicit obviative marking.

11 EUN BI KANG

4.2 Reflexes of *-ah as non-default obviation marker

If non-default obviative configurations are marked through explicit morphology then the difference between reflexes of *-ah, *-im and *-ri quickly becomes apparent. Table 4 compares non-default obviative configurations across the Cree dialects. As can be seen from the table, reflexes of *-ah appear in nearly every non-default obviative configuration across the , with the exception of Plains Cree, which Wolfart reports as using a reflex of *-ah only in the independent order (Wolfart, 1973: 49). The most straightforward generalisation of *-ah given its prevalence and its ability to co-occur with the other obviative markers would be that reflexes of *-ah simply mark non-default configurations. Unfortunately, there are forms that systematically resist this view: the singular inanimate intransitive and 3↔3′′ configurations. Again, the 3↔3′′ forms merit special attention here. Although explicit obviative marking is still expressed in these forms through the reflexes of *-im in all the dialects, *-ah is absent in all dialects but South East Cree and Sheshatsiu Innu, where it seems to be conditioned by number. Incidentally, South East Cree expresses obviative marking in the plural form but not the singular, a pattern exhibited in the default 3↔3′ forms. It is no coincidence that the 3↔3′′ forms appear to systematically omit the non- default obviative marker, since these are the only non-default configurations that possess a proximate third person argument. In this light, an alternative approach to *-ah emerges: *-ah generally indicates obviation in contexts where no proximate argument is present. I believe this latter perspective is preferable as it reduces the recalcitrant forms to just the singular inanimate intransitive. Moreover, the recalcitrant forms do not appear to be arbitrary as all Cree dialects lack obviation on their singular inanimate intransitive obviative forms, leaving potential for theoretical integration in the future.

4.3 Reflexes of *-im and *-ri as narrow obviation markers

Unlike the reflexes of *-ah, the reflexes of *-im and *-ri are interesting in that they do not co-occur with one another and therefore are in complementary distribution in virtually all dialects. However, the actual distribution of these obviative markers varies depending on the dialect, as shown in Table 5. The most prominent variation appears to revolve around the inverse forms containing a non-third animate argument, that is, the 3′—1/2 forms. Based on their treatment of these forms, the dialects can be classified into one of two groups: those that employ *-im regardless of whether the configuration is direct or inverse (Swampy, Moose, Atikamekw, and Sheshatsiu Innu) and those that refrain from using *-im in the inverse configurations (Plains and the East Cree dialects). From a cursory glance, I would suggest that the group of dialects which uses -im in both the direct and inverse sets of the 1/2—3′ configurations (Swampy, Moose, Atikamekw, and Sheshatsiu Innu) is the more innovative of the two as it appears that they have levelled -im to include both the direct and the inverse paradigms. Conversely, the other group can be labeled as being more conservative in this respect. Whether this is actually the case is beyond the scope of this paper; the terms “innovative” and “conservative” dialects are simply used for the sake of convenience for the remainder of this section.

4.3.1 Conservative dialects

In this section, I consider the obviative markers in the conservative dialects, which use *-im in direct forms and *-ri in inverse forms. While the conservative dialects seem to have more irregularities such as forms that do not exhibit either of the obviative morphemes in question, if one were to consider only the forms that do possess either reflexes of *-im and *-ri, a rather simple but intriguing observation arises. In all forms where reflexes of *-ri appear in Plains and East Cree, the obviative participant is the one performs the action. Conversely, in all but one form (NE Cree 3′—2P), -im only appears in configurations where the obviative participant is the recipient of the action. In other words, the reflexes of *-ri and the reflexes of *-im in the conservative dialects can be seen as comprising a crude nominative-accusative case system exclusive to obviative configurations, in which *-ri is a nominative obviative marker and *-im is an accusative obviative marker.

12 MORPHOLOGICAL VARIATION IN THE CONJUNCT ORDER AMONG THE CREE DIALECTS

Table 4: Non-default obviative configuration forms across the Cree dialects Verb Plains Swampy Moose Atikamekw SE Cree NE Cree Sheshatsiu Form Innu 3′ -yit -ilici -ilicih -irici -yich -yich -nci: 0′S -yik -ilik -ilik -irik -yic -yic -nt 0′P -yiki -iliki -ilikih -iriki -yich -yich -nci:/- nika:w 1S—3′ -imak -imaki -imakih -imaki -imach -imich -imaki: 2S—3′ -imat -imaci -imacih -imaci -imath -imith -imaci: 1p—3′ - -imakihci -imakihcih -imakihci -imacihth -imicihch -imaci:ci: ima:ya:hk 12—3′ -ima:yahk -imahko -imahkoh -imahko -imihkwh -imihkwh -imakwi: 2P—3′ -ima:ye:k -ime:ko -ime:koh -ime:ko -ime:kwh -ima:kwh -ime:kwi: X—3′ -imiht -imihci -imihcih -imihci - - -a:kanit / a:kano:yic a:kiniwi:yic -a:kanci: h h 3S— -ima:t -ima:t -ima:t -ima:t -ima:t -ima:t 3′′ 3P— -ima:cik -ima:cik -ima:cik - -ima:c -ima:ci: 3′′ ima:twa:w h 3′—1S -iyit -imici -imicih -imici -iyich -iyich -imi(ni)ci: 3′—2S -iyisk -imiski -imiskih -imiski -iskh -isciyich -imiski: 3′—1P - - - - -iyamihth -iyimihth -iminami:t ikowa:ya: (im)iyam (im)iyamihc imiyamihci hk ihci ih 3′—12 - - -imitahkoh -imitahko -itahkwh -iyitihkwh -imita:kw ikowa:yah imitahko k 3′—2P - -imita:ko -imita:koh -imita:ko -ita:kwh -imita:kwh -imita:ko ikowa:ye: k 3′—3′′ -a:yit -a:lici -a:licih -a:rici -a:yich -a:yich -a:nci: 3′′—3′ -ikoyit -ikolici -ikolicih -ikorici -ikoyich -ikoyich -ikonci: 3′—0 -amiyit -amilici -amilicih -amirici -amiyich -imiyich -aminci: 0—3′ -ikolici -ikolicih -ikorici

Bolded text indicates the obviative marker *-ah or its allomorph -ih; italicised text indicates other obviative markers. Woods Cree is omitted due to insufficient data.

13 EUN BI KANG

Table 5: Distribution of the reflexes of *-im and *-ri across the Cree dialects Verb Plains Swampy Moose Atikamekw SE Cree NE Cree Sheshatsiu Form Innu 3′ -yi -li -li -ri -yi -yi -n 0′S -yi -li -li -ri -yi -yi -n 0′P -yi -li -li -ri -yi -yi -n 1S—3′ -im -im -im -im -im -im -im 2S—3′ -im -im -im -im -im -im -im 1P—3′ -im -im -im -im -im -im -im 12—3′ -im -im -im -im -im -im -im 2p—3′ -im -im -im -im -im -im -im X—3′ -im -im -im -im -yi -yi -im/-ni 3S—3′′ -im -im -im -im -im -im 3P—3′′ -im -im -im -im -im -im 3′—1S -yi -im -im -im -yi -yi -im (-ni) 3′—2S -yi -im -im -im — -yi -im 3′—1P — -im -im -im — — -im 3′—12 — -im -im -im — — -im 3′—2P — -im -im -im — -im -im 3′—3′′ -yi -li -li -ri -yi -yi -n 3′′—3′ -yi -li -li -ri -yi -yi -n 3′—0 -yi -li -li -ri -yi -yi -n 0—3′ -li -li -ri

Reflexes of *-im are indicated in bold text and reflexes of *-ri are italicised. Again, Woods Cree has been omitted due to insufficient data concerning non-default configurations.

4.3.2 Innovative dialects

The innovative dialects require a slightly more complicated explanation in order to account for the usage of *-im in both direct and inverse configurations. The fact that reflexes *-im appears in both direct and inverse configurations rules out the possibility of a nominative-accusative distinction such as the one apparent in the conservative dialects. The variation here instead seems to be dependent on the person feature of the other participant. As outlined in Table 6, all the configurations in which reflexes of *-im appear include an animate non-obviative participant whereas reflexes of *-ri only appear when the other argument is second obviative or inanimate, or when there is no other argument at all (intransitive). These forms seem to neatly correspond to two halves of the Person Hierarchy: reflexes of obviative *-im appear when the other participant belongs to the upper proximate-animate echelon (i.e., 1, 2, impersonal X, and proximate 3) while reflexes of obviative *-ri appear when the other participant belongs to the lower obviative-inanimate echelon of the hierarchy (i.e., 3′ or 0) or is absent entirely. We can thus understand *-im and *-ri as marking obviation in the presence of a particular type of central participant: the reflexes of *-im indicate a peripheral obviative argument when the central participant is proximate or a speaker/hearer while the reflexes of *-ri appear in configurations where the central participant is necessarily obviative, due to there either being only one participant, which happens to be obviative, or due to the partner participant ranking even lower on the Person Hierarchy. One of the strengths of this proposal is that it fits well into the established literature of the Algonquian languages and reinforces the concept of central and peripheral participants as well as the direct-inverse alignment system. If it is assumed that the reflexes of *-ri in these dialects mark obviation for central participants, then perhaps it is no surprise that they appear beside the central agreement suffix (-c/t). I propose that, similar to how central inflection can be composed of -wa:w plus a central suffix (§3.1), the

14 MORPHOLOGICAL VARIATION IN THE CONJUNCT ORDER AMONG THE CREE DIALECTS reflexes of obviative *-ri together with the adjacent third-person suffix -c/t also comprise two components of central inflection, which is realised through morphological fission just as it is in the examples with - wa:w.

Table 6: Distribution of the reflexes of *-im and *-ri across the “innovative” dialects Verb Form Swampy Moose Atikamekw Sheshatsiu Innu *-im 1S—3′ -imaki -imakih -imaki -imaki: 2S—3′ -imaci -imacih -imaci -imaci: 1P—3′ -imakihci -imakihcih -imakihci -imaci:ci: 12—3′ -imahko -imahkoh -imahko -imakw(i:) 2p—3′ -ime:ko -ime:koh -ime:ko -ime:kw(i:) X—3′ -imihci -imihcih -imihci -a:kanit/-a:kanci: 3S—3′′ -ima:t -ima:t -ima:t 3P—3′′ -ima:cik -ima:cik -ima:ci: 3′—1S -imici -imicih -imici -imi(ni)ci: 3′—2S -imiski -imiski -imiski -imiski: 3′—1P -(im)iyamihci -(im)iyamihci -imiyamihci -iminami:t 3′—12 -imitahko -imitahko -imitahko -imita:kw 3′—2P -imita:ko -imita:ko -imita:ko -imita:ko *-ri 3′ -ilici -ilicih -irici -nci: 0′S -ilik -ilik -irik -nt 0′P -iliki -ilikih -iriki -nci:/nika:w 3′—3′′ -a:lici -a:licih -a:rici -a:nci: 3′′—3′ -ikolici -ikolicih -ikorici -ikonci: 3′—0 -amilici -amilicih -amirici -aminci: 0—3′ -ikolici -ikolicih -ikorici

5 The markedness of obviation

Having seen how the pluralisers and the obviative markers vary in the Cree dialects and the conditions determining both variations, it is safe to say that obviation is a marked category. At least in the conjunct order, obviation is treated differently in that it appears to be exempt from the same rules that characterise less marked inflectional categories such as person. For example, the central agreement suffixes are not concerned with the grammatical function of their argument as they are assigned through Cree’s direct-inverse alignment system. Therefore, one would reasonably expect obviative forms to display similar behaviour. Despite this, the narrow obviative markers, the reflexes of *-im and *-ri, in the conservative dialects seemingly operate under a nominative-accusative distinction. Another example of the exceptional treatment of obviatives is the simultaneous occurrence of multiple obviative markers. Earlier, the variation of the third person pluralisers among the Western dialects were explained as a product of dissimilation to eliminate multiple identical features; since -ik and -wa:w occupy different positions on the clausal spine, it is syntactically possible for them to co-occur, but a morphological filter on multiple instances of [–PARTICIPANT, +PLURAL] result in only one pluraliser surfacing in any given form. When it comes to obviation, essentially the same situation results in the opposite outcome. As discussed in 3.3, it seems that Cree treats grammatical person as the most valuable piece of information and is willing to sacrifice distinctions in number. From this perspective, obviative persons seem to be treated as being low in importance as they are not marked for number. Yet while the Western Cree dialects permit only one plural marker to surface for proximate participants, they seemingly have no qualms with obviation taking up two slots in the verb complex.

15 EUN BI KANG

Clearly, obviation occupies a special place in the Cree inflectional system, and while the theoretical reasons for its exceptional status are beyond the scope of this paper, I believe it is worth consideration in future work.

6 Conclusion

This paper has examined two variations in the patterning of agreement morphology in the Cree conjunct order inflection: the distribution of the third person pluralisers -ik and -wa:w in the Western Cree dialects and the use of the three obviative markers which are the reflexes of Proto-Algonquian *-ah, *-im and *-ri. It was argued that the third person pluralisers -ik and -wa:w realise different heads along the clausal spine and thus are able to co-occur syntactically. It was suggested that the Western Cree dialects opted toward -wa:w when the syntactic position which -ik occupies, C, was inaccessible, such as in subjunctive forms where C is occupied by a subjunctive marker -i/-e:. However, in order to explain the pluralisers’ inability to co-occur in forms which they theoretically could, crosslinguistic evidence from Spanish and Basque was examined to propose that -ik and -wa:w are mutually exclusive because the Western Cree dialects do not tolerate multiple morphemes possessing the same [–PARTICIPANT, +PLURAL] features. For the obviative markers, it was proposed that the 3↔3′ configuration of participants is a “default” in which no explicit obviative morphology was required, as the presence of a proximate argument automatically requires the other animate third person argument to be obviative. The reflexes of *-ah were suggested to mark non-default obviative configurations while the reflexes of *-im and *-ri were proposed to be dependent on the features of the arguments. The precise features differ by dialect: conservative dialects showed a nominative-accusative pattern, with *-ri marking obviative subjects and *-im marking obviative objects, while innovative dialects employed *-im in all 1/2↔3′ forms regardless of the role of the obviative argument. In these dialects, *-im and *-ri were shown to be conditioned by the features of the central participant. The innovative dialects have thus begun to incorporate the obviative markers into the direct- inverse alignment system, thereby abandoning the isolated nominative-accusative pattern shown by the conservative dialects. Lastly, attention was drawn to the markedness of obviative inflection after the treatment of number and obviation in the Cree dialects was compared in light of the variations discussed. Among the peculiarities exhibited by obviation are the Cree dialects’ apparent tolerance for multiple obviative markers despite restrictions on multiple proximate plural markers. In addition, there are what appear to be nominative and accusative obviative markers in the conservative group of dialects, despite Algonquian languages being well-known for their direct-inverse alignment system. It was suggested that obviation occupies an exceptional place in Cree morphology and should be considered accordingly in the future.

References

Arregi, K. & Nevins, A. (2006). Obliteration and impoverishment in the Basque g-/z- constraint. University of Pennsylvania working papers in linguistics, 13, 1–14. Philadelphia: Penn Linguistics Club. Beland, J. P. (1978). Atikamekw morphology and lexicon. Berkeley: University of California. Branigan, P & MacKenzie, M. (1999). Binding relations and the nature of pro in Innu-aimun. In P. Tamanji, M. Hirotani, & N. Hall (Eds.) Proceedings of the north east linguistic society, 29, (pp. 475-485). Amherst: University of Massachusetts. Clarke, S. (1982). North-west river (Sheshatshit) Montagnais: A grammatical sketch. National Museum of Man Mercury Series, Canadian Ethnology Service Paper no. 80, Ottawa. Dahlstrom, A. (1991). Plains Cree morphosyntax. New York: Garland. Ellis, C. D. (1971). Cree verb paradigms. International journal of American linguistics, 37, 76–95. Ellis, C. D. (1983). Spoken Cree. Edmonton: Pica Pica Press.

16 MORPHOLOGICAL VARIATION IN THE CONJUNCT ORDER AMONG THE CREE DIALECTS

Goddard, I. (1967). The Algonquian independent indicative. In A. D. DeBlois (Ed.), Contributions to anthropology, linguistics I (Algonquian), (pp. 66–106). Ottawa: National Museum of Canada Bulletin 214. Goddard, I. (1974). Remarks on the Algonquian independent indicative. International journal of American linguistics, 40, 317–327. Halle, M & Alec Marantz. (1993). Distributed morphology and the pieces of inflection. In K. Hale & S. J. Keyser (Eds.), The view from building 20: Essays in linguistics in honor of Sylvain Bromberger (pp. 111-176). Cambridge: MIT Press. Hamilton, M. D. (2013). An account of verbal person suffixes. Presented at the 45th Algonquian Conference, Mohegan Tribal Nation, Uncasville CT. October 19. Junker, M (Ed.). (2000–2014). The interactive East Cree reference grammar. Retrieved from Nevins, A. (2010). Morphophonological dissimilation, morphosyntactic dissimilation, and the architecture of exponence. In J. Trommer (Ed.), The Handbook of Exponence. Nevins, A. (2007). The representation of third person and its consequences for person-case effects. National language and linguistic theory, 25, 273-313. Oxford, W. R. (2014). Microparameters of agreement: A diachronic perspective on Algonquian verb inflection (Doctoral dissertation). Toronto: University of Toronto. Pentland, D. H. (1999). The morphology of the Algonquian independent order. In D. H. Pentland (Ed.), Papers of the 30th Algonquian Conference (pp. 222–266). Winnipeg: University of Manitoba. Starks, D. J. (1992). Aspects of Woods Cree syntax (Doctoral dissertation). Winnipeg: University of Manitoba. Wolfart, H. C. (1973). Plains Cree: A grammatical study. Transactions of the American Philosophical Society, 63, 1–90. Philadelphia: American Philosophical Society.

Appendix A: List of Abbreviations

0 inanimate third person IND indicative mood 1 first person Infl(P) inflection (phrase) 12 inclusive first person plural INT intransitive auxiliary 2 second person Mod(P) mode (phrase) 3 third person NE Northeast (Cree) X indefinite person Neg(P) negative (phrase) ′ obviative OBJ object ′′ double obviative P plural NON3 non-third person formative PRET preterite aspect A absolutive case S singular ACC accusative case SE Southeast (Cree) Asp(P) aspect (phrase) SJV subjunctive mood C(P) complementiser (phrase) T(P) tense (phrase) DAT TA transitive animate verb DIR direct alignment TI transitive inanimate verb DUB dubitative mood TR transitive auxiliary E ergative case (P) voice (phrase) GNO gnomic or neutral aspect v(P) verb (phrase)

17