An Introduction to Woods : The Case of South Indian Lake

DONNA STARKS University of

Introduction Cree is usually divided into four dialect areas referred to as Plains, Woods, Swampy and . The four dialect areas differ in respect to the reflexes of one Proto-Algonquian , *1. The reflex of PA *1 is /6/, in contrast to Plains Cree /y/, /n/, and Moose Cree /l/. This paper presents a brief description of the least known of these dialects, Woods Cree. Woods Cree is spoken in northwestern Manitoba and northeastern . Data presented in this paper were collected in South Indian Lake, a Woods Cree community in northwestern Manitoba, whose dialect has been referred to as Eastern Woods Cree, in contrast to Western Woods Cree spoken in Saskatchewan. In the sections to follow, I outline the major phonological and mor­ phological features found in the Woods Cree community of South Indian Lake. The features discussed are not to be construed as distinctive Woods Cree features since little comparative work has been completed in the sur­ rounding area. These features may very well turn out to be: 1) features of Woods Cree in general; 2) general Cree features that have not yet been documented in any detail; 3) areal features found in the other variants of Cree in the area, particularly, Northern Plains Cree and Western Swampy Cree; or even 4) features unique to this community.1 The features are meant to be compared with other published sources such as Wolfart (1973), Ellis (1983) providing a background against which the question, "Is Woods Cree a dialect?" can be raised. The material

1 Preliminary research in the neighboring community of Pukatawagan indicates many of the features found in South Indian Lake occur in both communities.

335 336 DONNA STARKS presented, unfortunately, is insufficent to answer this question completely since the data are drawn from only one community. This paper then is a first attempt to provide comparative data on Woods Cree.

Phonology The of Woods Cree are as follows: Vowels Consonants i I 06 p t c k 5 1 s h a a m n w y The phoneme /l/ is marginal. It occurs only in speech directed towards children e.g., kala 'don't', English caiques apalakoscocin 'hat with pompon [apple] on top', and English based nicknames like malcin 'Mary Jane'. Phonemically, there are three major differences between Woods Cree and the Moose and Plains Cree dialects as described by Ellis and Wolfart. First, Woods Cree has the unique phoneme /6/. Second, Woods Cree had merged *e with *I,2 with a resulting symmetric vowel system consisting of three long and three short vowels. Finally, Woods Cree lacks the contrast between /s/ and /s/ that occurs in Moose Cree and certain Swampy Cree dialects (cf. Pentland 1978). Woods Cree like other Cree dialects also has a number of consonant clusters: /hp, ht, he, hk/ and /h6/. The cluster /h5/ is rare, occurring in only a few items, e.g., wah5aw 'far'. The phonemic structure of Woods Cree is obscured however by a number of phonological processes.

Phonological Processes The term "phonological processes" refers to optional phonetic rules that operate upon the underlying phonological system. Since this is not a sociolinguistic study, a statistical analysis of the extent and distribution of these processes has not been undertaken. However, many of the processes appear to be dependent on sociological factors like age and sex. In this paper, I examine nine of the most common phonological processes.

Fricative plus Stop The first phonological process is the phonetic loss of /k/ in /hk/ and /sk/ clusters, which occur only in word medial and word finalposition . The

2 A few older speakers at South Indian Lake still pronounce the word [kekwan] 'some­ thing' with [e]. Wolfart notes the merger of the two long high front vowels in northern Plains Cree. WOODS CREE 337 weakening of /hk/ occurs in both positions, as is illustrated in the following words, wisakicahk — wisakicdh 'Wlsaklcah', a culture hero, sdkahikanihk -> sakahikanih 'on the lake', pahkikin —• pahikin 'hide', tahkonam —• tahonam 'carry it', /hk/ also weakens in /hkw/ clusters e.g., nahkwan —»• nahwdn 'heel' (cf. Greensmith 1986). The weakening is more pronounced word finally, to the extent that some younger speakers have reanalyzed final /hk/ as /h/. This phonological process in respect to /sk/ clusters is more complex. In final position, /sk/ weakens to /s/ e.g., oki5askisk —* okiddskis 'liar'. Medially /sk/ only weakens to [s] when an /s/ occurs in the following syllable, e.g., miskwami-kiskisikan —• miskwami-kisisikan 'ice pick'. Many younger speakers hyper-correct final /s/ to [sk], e.g., kayas —• kaydsk 'a long time ago'. Clearly, there is a tendency for clusters to weaken in both medial and final position. The weakening is however more common in word final position. This weakening is extending to other fricative clusters in Woods Cree. In Pukatawagan, final /hp/ optionally weakens to [h] among younger speakers in at least one lexical item, tipakohp —• tipakoh 'seven'. In both South Indian Lake and Pukatawagan /h6/ is also in the process of weakening (cf. entry for hd for further details). k- Weakening Co-existing with the weakening of /hk/ is the weakening of /k/. This is most prevalent in word-initial position, where /hk/ clusters do not occur, especially with the conjunct marker /ka-/. Younger speakers almost aways delete /k/ in this , /k/ deletion also occurs medially, maka —• mda, 'but', or word finally, atoskihkak —* atoskihka 'let's work later', /k/ deletion can also occur in a word with an /hk/ cluster, pakahkam —• paham 'maybe'. hd In addition to /hk/ another cluster /hd/ is also weakening. At South Indian Lake, /hd/ weakens to /6/. The weakening process, however, appears to be lexically based. Of eight vocabulary items that Pentland (1979:208) indicates should, for historical reasons occur with the cluster /hd/ in Woods Cree, three retain the /hd/: wahbaw 'far off'; nohdiw 'suckle him'; and ahbiw 'place him'. Two vary between /hd/ and /6/: wihdiw 'name him'; and 6ih5iw 'breathe'. The final three occur only in /6/: abapiy 'net', ata&ohkiw 'tell a legend', and pimi6dw 'fly'.3

3Pentland also notes one other example /kindiw/ 'golden eagle'. To date, I have not 338 DONNA STARKS ki -* 0 Initial /ki/ often deletes when the following syllable contains /s/, kisipikinam -* sipikinam 'he washes it', kikiskam -» kiskam 'he wears it'. (This process is similar to the weakening of /sk/ word medially). s —* h Another change is the weakening of /s/ to /h/ when it is followed by a diminutive, a morpheme with /s/, ndpisis —• ndpihis 'boy'. This process is marginal and appears to be restricted to younger speakers. w —> y The semi-vowels /w/ and /y/ and to some extent /h/ are extremely variable, /w/ and /y/ regularly alternate in the same word, piyapisk ~ piwapisk 'metal', tdniwa ~ taniya 'where'. The alternation even exists across morpheme boundaries, /ki-wdpamiw/ —• ki-yapamiw 'he sees him', Iotoniwawaj —• otoniyawa 'their lips'. Although the phonological process appears to be conditioned by a surrounding /i/, the change is not regular. Many examples of /w/ remain when they are preceded or followed by /i/. Part of the irregularity is due to language change. This change is very widespread among younger speakers.

V -» 0/C C Short vowels optionally delete between homoganic consonants in Woods Cree: nicanis —• ncans 'daughter'. This feature has been observed in other dialects of Cree (cf. Wolfart and Carroll 1981:13; Ellis 1983:19). In Woods Cree, however, short vowels also optionally delete between nasals, anima —• anma 'that', tanima —+ tanma 'which thing'. hi - /// Short /a/ occasionally merges with short /i/ realized phonetically as [i]. The following examples illustrate: asamak —• isdmak 'snowshoes', abapiy —* i6ipiy 'net'. This process is especially prevalent in word initial position.

/a/ + semi-vowel -f /i/ -* [e]/ or [t\ Among younger speakers the sequence /a/ -f semi-vowel + /i/ often merges with [e] or [T]: kawisimow -* kesimow 'he goes to bed', nitawiftimaw -* nitebimaw 'I want it', masinahikan —• masinekan 'book'. been able to find anyone at South Indian Lake even that can identify, let alone name this bird. WOODS CREE 339

To summarize, there are a number of phonological processes affecting both consonants and vowels in Woods Cree. The distribution of these phonological processes within the community is as yet unclear. There is also a lack of comparative material, since few sources mention variability in Cree. Ahenakew (1987) and Voorhis (1981) however suggest that many of the vowel processes discussed above may be pan-Cree. Other possible pan-Cree processes include short vowel deletion and w/y alternation. Material on consonant processes is less available. The exact relation­ ship between the consonantal processes in Woods Cree and consonantal pro­ cesses in other Cree dialects will have to wait until further cross-linguistic material is available.

Morphology

The verbal structure of Woods Cree is less complex morphologically than that described by Ellis for Moose Cree, or Wolfart for Plains Cree. There are no dubitative paradigms, no functioning initial change, the future preverb Are does not occur, few relational forms exist, the TA -im- suffix has only been attested in the 1st person singular, most delayed imperative singular and obviative forms have merged, and finally preterits are used only in conditional constructions. However the function of these structures is still maintained in Woods Cree, albeit by different grammatical means. For example, dubitative con­ structions are marked in three ways in Woods Cree: either by one of two dubitative particles: itoki 'it must be' or paham 'it might be', or by a dubi­ tative preverb wah-, occuring with both Independent and Conjunct verbs. The following conditional structure will illustrate the use of this preverb:

1) wdA-itohtlyani, ka-waA-wanisininaw. 'If I had gone, we would have been lost'. Also, the function marked by productive initial change in Moose Cree is indicated by two frozen changed conjunct markers a- and i- in Woods Cree. The following two WH-questions will be used to illustrate:

2) tantT a-takosih? 'Where did he arrive?'

3) tantT T-takosih? 'Where is he going to arrive?' Although the relational and the TA -im- infix are used infrequently, the structures are marked where possible on the verb root by diminutive symbolism. The following example illustrates:

4) kl-sipwlhclw opapawa. 'He left with her father". 340 DONNA STARKS

In cases where the relational or the TA -im- infix optionally occur, diminutive symbolism is still marked on the verb stem: 5) nikl-sipwlhclmawa opapawa. 'I left her father'. 6) T-kT-sipwlhcTwakiht opapawa. 'We left with her father'. Finally, although the preterit does not mark past events, the past is indicated by the past tense preverb ki. 7) ki-plkotahah waslnamawin clkayikan ohci. 'He/she broke the window with an axe'. The only clear loss of a grammatical function is the loss of the delayed imperative obviative: 8) atoskihkan 'work later' (3 or 3') A similar situation exists for the nominal and pronominal paradigms of nouns. Woods Cree does not mark unpossessed inanimate obviatives: 9) T-pikotahah waslnamawin. 'He broke the window'. There are also no distinct indefinite paradigms for 'someone' and 'some­ thing' in Woods Cree. Also, there is an optional loss of the singular/plural distinction within these paradigms. Again, the functional distinctions are still maintained. The indefinite is distinct from the definite in two respects. First, the forms never occur in the same environment. WH-words occur in clause initial-position while indefinite pronouns usually occur in non-initial posi­ tion. Second, the verb in a question that contains awina 'who' or kikway 'what' is usually marked with an a-conjunct verb, whereas the verb in a main clause with an indefinite pronoun usually occurs with either an i"-conjunct or an independent verb. In other dialects the singular/plural pronominal distinction is marked on the verb and the pronoun. The loss of the grammatical distinction on the pronoun is therefore not a loss of a functional distinction. The above features might suggest that Woods Cree is overall less com­ plex than other Cree dialects and that Woods Cree is characterized by the loss of grammatical categories. Although many categories have been lost, Woods Cree is not alone (as a comparison of Wolfart or Ahenakew with El­ lis will show). It is apparent that there are fewer grammatical distinctions in the west than in the east. There are however grammatical forms that occur in Woods Cree that are absent in at least one of Moose or Plains Cree. For example, in Plains and Woods Cree there is a three-way number contrast in the delayed imper­ ative. In Moose Cree, the plural forms have merged. In the other direction, WOODS CREE 341 there is a disjunctive set of pronouns (nista, kista, wista etc.) that occur in Moose and Woods Cree that is absent in Plains Cree.

Morphological Forms

Another major difference between the published sources and my data from South Indian Lake pertains to morphological forms. There are two questions that must be raised. First, when the other dialects differ among themselves, will my data agree with the western or the eastern data. Sec­ ond, are there any morphological forms in Woods Cree that are distinct from the other dialects. The answer to the first question appears to be that Woods Cree agrees more with the western data than the eastern data. For example, the sub­ junctive morpheme in Woods Cree is e, but in both Plains and Woods Cree it is -i, kisaspin wapamaki 'if I see him'. Also, the AI subjunctive 1st person inclusive is yahki at South Indian Lake, but in Moose Cree and in the variant of Plains Cree cited by Wolfart (1973:48), it is yahko. Wolfart also cites the form yahki as a variant of the Plains Cree form which corresponds with the form in the data from South Indian Lake. Most of the inflectional forms that vary between Plains and Moose Cree also vary within the South Indian Lake community. It is therefore impossible to add any conclusive details here. In many cases, any variability marks a change toward a western feature. To illustrate, I will describe the plural conjunct -hkwaw. This morpheme occurs optionally on the following conjunct verbs: most TA plural objects (with the optional exception of the 2sg, and the third person singular and plural), the 33 II form, the TI plural 3rd person plural subject, and the plural forms of the indefinite -katik and naniwan. There is however a considerable amount of variability. Plural forms evident in Moose Cree (e.g., 33 II -i, Indef pi. katiki) optionally occur among some individuals in South Indian Lake. The two variants often occur in the same text. Another area of variability is the AI verb stem. AI verb stems ending in -i- are being reanalzed as a/i stems in Woods Cree (e.g., nimacin —» nimacan 'I hunt'). The change is dependent largely on age. Many older speakers retain the distinction, while many younger speakers have merged the two paradigms. The merger is however not totally complete.4 Yvonne Carifelle (personal communication) has noted that this merger is complete

4There is some residue. The verb cahkiw 'be tired' still maintains an i- stem in the 1st singular. 342 DONNA STARKS in at least one northern variant of Plains Cree. Many western features found in Woods Cree appear to be recent innovations. The answer to the second question is no. There are very few forms which occur only in Woods Cree. One possible example is the preverb wd- 'dubitative' described in the preceding section. Other differences are minimal. For example, one difference occurs in AI indefinite morphemes. In Plains, the indefinite is marked by -ndniwi- and -dniwi-. The former form -naniwi- is used with d and e/d stems while the latter form occurs with all other forms. In Moose Cree, a similar distinction exists but the forms are -naniwan and -dniwan respectively. The Moose forms correspond to the Woods Cree, however due to the merger of i- stems with i/a stems there is a different distribution of these features in Woods Cree than in Plains or Moose Cree.

Summary The situation between Woods Cree and other dialects of Cree is more complex than the presence or absence of /6/. This feature is, however, still the only feature that clearly marks Woods Cree as phonologically dis­ tinct. There is not enough material on phonological processes to consider the dialect split as conclusive. Although many of the vowel phonological processes may be pan-Cree, the consonantal processes may mark Woods Cree as distinct. However little more can be added until more research in the surrounding area is completed. There are a reasonable number of distinctions at the morphological level between various dialects. The evidence suggests that there are more Woods Cree features that agree with western Cree than with eastern Cree. However, a number of morphological features, especially changes due to loss and subsequent syntactic innovation, mark Woods Cree as distinct from both Plains or Moose Cree. It must be kept in mind however it is very likely that the features outlined above for Woods Cree will be found in neighbouring Plains and Swampy Cree dialects, since many of the changes occurring in South Indian Lake seem to be occurring in the surrounding area. Further research is needed before any definitive statement can be made. It is clear however that Woods Cree shares features with both eastern and western dialects. One of the clearest statements about the position of Woods Cree in respect to other Cree dialects comes from the lexicon. Most lexical items that separate the eastern from the western dialects — for example, the numbers 6 to 10 — mark Woods Cree as a western dialect. However a number of lexical items that vary between east and west co-exist in South Indian Lake. For example, the question markers na and ci, the WH-words WOODS CREE 343 klkway and kikwan, and the subjunctive particles kisaspin and kispin 'if'.5 The data for this paper were collected from only one community. These data, however, do indicate that Woods Cree is distinct from the other Cree dialects phonologically, lexically, morphologically and syntactically. The evidence indicates the need of a variability study even within the one Wood Cree community studied here. South Indian Lake is is by no means a homogeneous speech community. There is a high degree of vari­ ability phonologically, morphologically and even lexically. Changes are tak­ ing place that are complicating the concept of a "Woods Cree dialect" and sociolinguistic features such as age and sex are challenging the rigidity of a dialect label.6

REFERENCES Ahenakew, Freda 1984 A Text-Based Grammar in Education. M.A. thesis, Uni­ versity of Manitoba.

1987 waskahikaniwiyiniw-acimowina / Stories of the House People. Winnipeg: University of Manitoba.

Greensmith, Jennifer 1985 Phonological Variants in Pukatawagan Woods Cree. M.A. thesis, Univer­ sity of Manitoba.

Ellis, C. Douglas 1983 Spoken Cree: West Coast of . Winnipeg: Pica Pica Press [rev. ed.].

Pentland, David 1978 A Historical Overview of the Cree Dialects. Pp. 104-126 in Papers of the Ninth Algonquian Conference. William Cowan, ed. Ottawa: Carleton University.

1978 Algonquian Historical Phonology. Ph.D thesis, University of Toronto.

Voorhis, Paul 1981 Varieties of Cree Speech in Manitoba. Paper presented at Linguistics Colloquium, University of Manitoba.

5 All of these pairs of words are semantically and syntactically distinct in Woods Cree. 6 Research was made possible by a grant from the Northern Studies Committee at the University of Manitoba. Special thanks go to my two main informants at South Indian Lake, Ivy Gault and Margaret Moose. 344 DONNA STARKS

Wolfart, H. Christoph 1973 Plains Cree: A Grammatical Study. Transactions of the American Philo­ sophical Society n.s. 63(5). Philadelphia.

Wolfart, H. Christoph, and Janet F. Carroll 1981 Meet Cree: A Guide to the Cree Language. Edmonton: University of Press.