Introduction
This study examines the acquisition of possessive constructions in Northern East Cree, a polysynthetic language indigenous to Québec, Canada. Transcripts from recorded naturalistic interactions between one adult and one child (age 2;01.12 to 3;08.24) reveal that the child produces possessive constructions using a morphosyntactic strategy that circumvents the robust inflection found in adult input. The child's production is explained in light of contextual factors as well as the fact that she is expressing possession before having productively acquired the target inflectional system. Her reliance on such a strategy remains unaffected by the eventual appearance of English elements in her possessive constructions. The child's production also resembles cross-linguistic findings, where children commonly express possession without obligatory grammatical marking.
Northern East Cree
Northern East Cree (NEC) is a variety in the Cree–Innu–Naskapi dialect continuum of the Algonquian language family (MacKenzie, Reference MacKenzie1980). East Cree is the first language of 12,000–13,000 speakers in Québec (Burnaby & MacKenzie, Reference Burnaby and MacKenzie2001; Junker, Reference Junker and Goddard2008). The Northern dialect of East Cree (ISO 639-3 code crl) is spoken in four communities near James Bay: Wemindji, Chisasibi, Whapmagoostui, and Eastmain (Junker, MacKenzie, & Brittain, Reference Junker, MacKenzie and Brittain2012).
NEC is a polysynthetic language, where an utterance can consist of a single word constituted by multiple grammatical morphemes, as in (1):
The acquisition of polysynthetic languages
As others have noted (e.g., Forshaw, Reference Forshaw2016; Kelly, Forshaw, Nordlinger, & Wigglesworth, Reference Kelly, Forshaw, Nordlinger and Wigglesworth2015; Kelly, Wigglesworth, Nordlinger, & Blythe, Reference Kelly, Wigglesworth, Nordlinger and Blythe2014), polysynthetic languages are under-represented in language acquisition research. This gap entails problems for the scientific understanding of first-language acquisition, because one cannot assume what holds for acquiring languages such as English also holds for languages with much different morphosyntax.
This is not to say that the acquisition of polysynthesis is uncharted territory. Over the years a growing and insightful body of work devoted to the issue has developed (e.g., Allen, Reference Allen1996; Fortescue & Olsen, Reference Fortescue, Olsen and Slobin1992; Mithun, Reference Mithun1989; Pye, Reference Pye1983), several recent projects have contributed to the ongoing enrichment of the literature (e.g., Chee, Reference Chee2017; Forshaw, Davidson, Kelly, Nordlinger, Wigglesworth, & Blythe, Reference Forshaw, Davidson, Kelly, Nordlinger, Wigglesworth, Blythe, Fortescue, Mithun and Evans2017; Pedro, Reference Pedro2015; Stoll et al., Reference Stoll, Bickel, Lieven, Paudyal, Banjade, Bhatta and Rai2012), and efforts such as the ACQDIV project (www.acqdiv.uzh.ch/en.html) explicitly include polysynthetic languages. See Kelly et al. (Reference Kelly, Wigglesworth, Nordlinger and Blythe2014) and Forshaw (Reference Forshaw2016, pp. 45–115) for lengthier general overviews of this kind of research. The present study does not aim to solve the question of which factors drive polysynthetic acquisition. Instead, this is an investigation of how a particular polysynthetic construction develops.
The present study relies upon data from the Chisasibi Child Language Acquisition Study (CCLAS), a longitudinal exploration of the first-language acquisition of NEC in the community of Chisasibi (Brittain, Dyck, Rose, & MacKenzie, Reference Brittain, Dyck, Rose, MacKenzie and Wolfart2007). With the support of the local community and the Cree School Board, CCLAS collected almost three years of data (between November 2004 and May 2007) from two small cohorts of children. At the time of data collection, Chisasibi had a population of approximately 4,000 people who had predominantly acquired NEC as their mother tongue (2007). In more than a decade of work, CCLAS has investigated a variety of phenomena in NEC. This includes the acquisition of stress (Rose, Brittain, Dyck, & Swain, Reference Rose, Brittain, Dyck, Swain, Franich, Iserman and Keil2010; Swain, Reference Swain2008), syllable structure and consonantal segments (Thorburn, Reference Thorburn2010), and verb types and verbal inflection (Johansson, Reference Johansson2012; Terry, Reference Terry2010). The focus of the present study is possessive nominal inflection in NEC, a subject not yet approached by the CCLAS project.
The acquisition of possession
Children younger than 2;0 grasp and linguistically express the concept of possession, well before they acquire a full grammatical system of encoding possession (e.g., Golinkoff & Markessini, Reference Golinkoff and Markessini1980; Tomasello, Reference Tomasello and Newman1998). The present study concerns the development of the linguistic expression of possession, a matter which has received much cross-linguistic treatment. Such research tends to rely on small samples, where inter-child variability makes it difficult to associate particular grammatical milestones with age-points (Marinis, Reference Marinis, Lidz, Snyder and Pater2016), but the literature nonetheless evinces some important cross-linguistic patterns.
Research on languages such as English, French, German, Greek, and Japanese (e.g., Brown, Reference Brown1973; Clancy, Reference Clancy and Slobin1985; Eisenbeiß, Matsuo, & Sonnenstuhl, Reference Eisenbeiß, Matsuo, Sonnenstuhl and McGregor2009; Leroy-Collombel & Morgenstern, Reference Leroy-Collombel and Morgenstern2012) provides evidence that children go through similar stages in expressing possession. Young children will commonly begin by producing possessor-only constructions, followed by combinations of a possessor and possessee but lacking grammatical marking such as case inflections or adpositions, until they eventually acquire the target system of their language (see, e.g., the review by Marinis, Reference Marinis, Lidz, Snyder and Pater2016). In contrast to these languages, polysynthetic languages often encode possession by inflecting the possessee with affixal elements (Mithun, Reference Mithun1999). The acquisition of possession in polysynthetic systems has not received much express attention, but some familiar patterns emerge from existing literature.
In some polysynthetic languages, possessive inflection appears quite early. For example, children acquiring Inuit languages productively use grammatical marking for possession around two years of age, earlier than English-acquiring children (Allen, Reference Allen, Fortescue, Mithun and Evans2017; Fortescue & Olsen, Reference Fortescue, Olsen and Slobin1992). This early emergence of inflection is likely a function of inflection-heavy language-specific characteristics as well as adult input (Allen, Reference Allen, Fortescue, Mithun and Evans2017). However, data from other polysynthetic languages more closely resemble cross-linguistic trends in the acquisition of possessive marking.
Two studies of Mohawk offer somewhat equivocal results, but these differences could be due to inter-child variability between two small datasets. Mithun (Reference Mithun1989, p. 294) describes one child productively using possessive inflection at age 2;9. On the other hand, Feurer (Reference Feurer1980) describes a different child at age 2;10 producing constructions that resemble those found in non-polysynthetic languages: This child “quite frequently” expressed possession in two-word combinations of an independent personal pronoun as possessor with an uninflected common noun as possessee (p. 31). In fact, Feurer reports that the child acquired possessive inflection “late”, although no age is given (p. 35).
Children acquiring Mayan languages have also exhibited patterns where possession is expressed early but without obligatory grammatical marking (e.g., Pye, Reference Pye and Slobin1992). For example, Brown (Reference Brown1998, p. 738) reports that two children acquiring Tzeltal – at an unspecified age before 2;5 – produced two-word combinations of possessor (often an independent personal pronoun) and possessee but lacking adult-like inflection. For K'iche, Pye (Reference Pye1979, p. 460) also describes a child approximately aged 2;9 producing the same kinds of constructions without obligatory inflection of the possessee.
To date there has been no modern scientific study of possession in the first language acquisition of an Algonquian language, but a few studies offer important clues. Leonard (Reference Leonard2007) reports that two children aged approximately 4;0 and 6;0 had not acquired possessive inflection in Miami. However, this represents a case study in language reclamation rather than typical first language acquisition. In the 1980s, a project from Upper and McKay (Reference Upper and McKay1987, Reference Upper and McKay1988) pioneered the investigation of the L1 acquisition of an Algonquian language, Anishininiimowin (Oji-Cree). Although Upper and McKay themselves explain that their limited data were not phonetically transcribed in a precise and consistent scientific format (Reference Upper and McKay1987, p. 27), their tentative findings point to some familiar cross-linguistic patterns: Without a clear specification for age or MLU, their earliest reported productive pattern for expressing possession is pairing an independent personal pronoun as possessor with an uninflected common noun as possessee (Reference Upper and McKay1987, p. 108). Furthermore, Upper (Reference Upper1993) reports that by the age of 3;0 one child had begun to produce some obligatory inflection on possessees, but not in a clearly productive manner.
Taken together, these findings point to some important cross-linguistic similarities in the acquisition of possessive constructions. Children understand and express possession at an early age, but mastery of possessive marking can take a few years. Instead, children commonly produce possessor-only utterances as well as two-word possessor–possessee combinations that omit obligatory grammatical marking for possession. The present study sheds light on a similar phenomenon in NEC.
Possession in NEC: two grammatical strategies
This study introduces terminology to distinguish between two strategies for creating possessive constructions in NEC: the morphological possessive strategy (MPS) and the equational possessive strategy (EPS). Cree linguistic literature does not distinguish between these two strategies, but instead focuses almost exclusively on the MPS.
The MPS involves inflecting a common noun as possessee. NEC has polysynthetic morphology, where verbs carry the majority of the grammatical load, and an entire utterance can consist of just one inflected verb (Junker, Reference Junker2004). Although they undergo more limited inflection than verbs, nouns inflect via affixation for a variety of categories. This includes grammatical animacy, number, diminutive, locative, obviation, and possession (Junker et al., Reference Junker, MacKenzie and Brittain2012, p. 22). Inflection for possession is most important for the present study, and a child acquiring this system must master the interaction of a range of factors to produce the target morphology.
First, all possessees must inflect with a prefix marking the person of the possessor. This is demonstrated in examples (2–4), where the noun misinihîkin ‘book’ is possessed by different persons indicated by a personal prefix.
This required marking involves the distinction between alienable and inalienable possession, as nouns pertaining to kinship, body parts, and some close personal belongings must carry the personal prefix to indicate a possessor (Junker, Reference Junker2003a).
Many possessees take an obligatory possessive suffix -im. This is required for most animate nouns but also occurs with some inanimate nouns, without a clear pattern for the suffix's distribution (Junker, Reference Junker2003a).
Number affects possessive inflection as well. Possessees with plural referents take a plural suffix, whose form is determined by grammatical animacy (Bloomfield, Reference Bloomfield1946; Goddard, Reference Goddard and Wolfart2002). Plural possessors are marked with an additional suffix, and marking for first person plural possessors includes an inclusive–exclusive distinction.
Finally, NEC also marks possessees with a suffix to indicate obviation, which distinguishes between third persons in a particular discourse or syntactic environment (Rhodes, Reference Rhodes1990). Animate possessees with a third person possessor require an obviative suffix, but inanimate nouns with a third person possessor take an obviative suffix only in certain grammatical contexts (Junker, Reference Junker2003b).
Adapted from Junker et al. (Reference Junker, MacKenzie and Brittain2012, pp. 25–6), examples (5–7) illustrate several of these inflectional elements with the stem shîshîp ‘duck’. The stem in (5) is accompanied by four affixes: the prefix indicating the second person possessor, the possessive suffix -im, a plural suffix for the possessor, and a plural marker for the possessee.
Similarly, (6) shows a third person prefix, the possessive suffix, and an obviative suffix, while (7) demonstrates the optional presence of a free-standing possessor with the same inflected possessee.
Linguistic descriptions of possession in Cree–Innu–Naskapi (e.g., Wolfart, Reference Wolfart1973) almost invariably focus on the morphology of possessive inflection. However, the CCLAS corpus abounds with a different method for creating a possessive construction, the EPS.
The EPS is named after null-copula equational constructions described in existing documentation of Cree–Innu–Naskapi varieties, including NEC (8) and Plains Cree (9). The EPS works much the same way by pairing a possessor with a possessee, but this construction has a crucial feature: the possessee is a demonstrative rather than a common noun that would require possessive inflection. This is evident in (10–11) from the CCLAS corpus, where the demonstrative does not take the obligatory person prefix for inflected possessees.
The EPS can also involve a possessor alone, as in (12–13). These constructions again circumvent possessive inflection by omitting a common noun possessee.
The EPS has received little attention in linguistic description – the mechanics of the EPS have been noticed in a few examples from discussions of Plains Cree demonstratives (Wolvengrey, Reference Wolvengrey2011) and nominal predication (Déchaine, Reference Déchaine and Pentland1997; Déchaine, Cardinal, Johnson, & Kidd, Reference Déchaine, Cardinal, Johnson, Kidd, Macaulay and Valentine2015) – but it abounds throughout the CCLAS corpus.
Possession in English
A child acquiring NEC as a first language will also have another crucial factor influencing her morphological development: English is certain to be present to some degree in the linguistic environment. Chisasibi is a Cree-dominant community, but English is heard in most homes at least through television, and NEC has been used as the medium of instruction in schools only up to grade four (Burnaby & MacKenzie, Reference Burnaby and MacKenzie2001; Junker et al., Reference Junker, MacKenzie and Brittain2012). The exact state of diglossia in Chisasibi requires future exploration, but these indications mean that a child could be aware that the English language presents a suite of options for expressing possession. Depending upon English-language input, a child could encounter the English possessive morpheme <’s> (e.g., “Victoria's toy”), the genitive construction (“the toy of Victoria”), a possessive pronoun with a possessee (“her toy”), or just the possessive pronoun alone (“hers”).
All four of these English constructions generally involve fewer morphological components than the MPS, where the decision tree of inflection necessarily concerns navigating (in)alienable possession, person, the possessive suffix, number, grammatical animacy, and obviation. Furthermore, the reduced morphological complexity of the EPS resembles English-language constructions such as “Victoria's toy”, “her toy”, and “hers”.
Altogether then, for expressing possession a child acquiring NEC will need to master one strategy that entails relatively more complex morphological considerations (the MPS), along with one strategy that does not (the EPS). English may also present options that resemble the EPS and generally reduce inflectional complexity in expressing possession.
Research questions
Given this context, the present study poses the following research questions:
RQ1: How (if at all) are the EPS and the MPS employed in child-directed and child speech in NEC?
RQ2: How (if at all) does English appear in possessive constructions in child-directed and child speech?
Method
Source of data
All data come from the CCLAS project (www.mun.ca/cclas/). Data for the present study consist of transcripts from naturalistic recording sessions conducted in a home setting between one child and one adult. These sessions were videotaped and transcribed into IPA and NEC orthography by CCLAS using Phon (www.phon.ca), a software program developed for corpus management and analysis (Rose & MacWhinney, Reference Rose, MacWhinney, Durand, Gut and Kristoffersen2014; Rose et al., Reference Rose, MacWhinney, Byrne, Hedlund, Maddocks, O'Brien, Wareham, Bamman, Magnitskaia and Zaller2006). CCLAS has a strict privacy protocol in place with the participating families in Chisasibi, and so this study relies upon transcripts rather than video from the recording sessions, which were created and processed by members of CCLAS in Chisasibi and at Memorial University in Newfoundland, Canada. See Brittain et al. (Reference Brittain, Dyck, Rose, MacKenzie and Wolfart2007) for more details regarding CCLAS, data collection, and processing.
Dataset and participants
The present study examines nine transcripts from interactions between one adult and one child, which cover approximately one year and seven months of child development. Each of the recording sessions lasted approximately 30 to 50 minutes, with a total time of 5:57:36 across the nine sessions. The transcripts include 8,234 total utterances, with 4,561 from the adult and 3,673 from the child. The adult is a community member and native speaker of NEC, and she served as the CCLAS project manager in Chisasibi at the time of recording. The child is a young girl aged 2;01.12 for the first recording session and 3;08.24 for the final session. To respect the privacy of participants, the adult is codenamed ‘CD’ (for child-directed speech), and the child is codenamed ‘Ani’. Ani has also been the focus of other existing CCLAS publications (e.g., Rose et al., Reference Rose, Brittain, Dyck, Swain, Franich, Iserman and Keil2010; Swain, Reference Swain2008; Terry, Reference Terry2010; Thorburn, Reference Thorburn2010).
Bilingualism and Ani's linguistic environment
Chisasibi is a Cree-dominant community, where elders are often NEC monolinguals, but most other adults are Cree–English bilinguals (Burnaby & MacKenzie, Reference Burnaby and MacKenzie2001; Pile, Reference Pile2018). CCLAS reports that Ani's parents speak Cree and English, but Ani is not bilingual and NEC is the language used in the home. Ani receives some exposure to English, primarily through television and songs (Swain, Reference Swain2008). The exact state of Ani's bilingualism has not received formal study, but, pending future investigation, existing CCLAS work indicates that Ani is NEC dominant. For example, other analyses of Ani's development (Swain, Reference Swain2008; Thorburn, Reference Thorburn2010) have observed that limited English surfaces in particular topics/domains such as television and toys (especially Dora the Explorer and Barbie) as well as songs (“Happy Birthday” and “Twinkle, Twinkle, Little Star”). Furthermore, Bryant (Reference Bryant2013) has demonstrated that Ani has a single, Cree-governed phonological system that she applies to her production of words from both Cree and English. Recent work has also found overwhelming Cree dominance in the grammar of another child in the CCLAS project, Billy (aged 4;06–6;0) (Pile, Reference Pile2018).
Coding
All 8,234 utterances in the dataset were examined manually using Phon (Rose & MacWhinney, Reference Rose, MacWhinney, Durand, Gut and Kristoffersen2014; Rose et al., Reference Rose, MacWhinney, Byrne, Hedlund, Maddocks, O'Brien, Wareham, Bamman, Magnitskaia and Zaller2006). Each utterance was coded – on the basis of morphology, context, and the translation provided by a native speaker of NEC – as containing a possessive construction or not. Each NEC possessive construction was counted and tagged as MPS or EPS. All EPS constructions were coded as either possessor-only or possessor-plus-demonstrative constructions. In order to capture any NEC–English code-switching, possessive constructions in English were also included. English possessives were coded as either pronoun-only (e.g., “mine”) or pronoun-plus-possessee constructions (e.g., “my doctor”). There were no examples of the English possessive morpheme <’s > or the genitive construction. Any possessive constructions that mixed NEC and English elements were also coded as mixed nec-english. All possessive constructions were then exported using Phon and compiled into a spreadsheet for further annotation and analysis.
Results and discussion
Possession in NEC constructions
To approach RQ1, Table 1 tallies NEC possessive constructions present in child-directed speech for each of the nine transcripts. Overall, CD overwhelmingly uses the MPS (221/272 constructions, or 81.3%). Examples (14–15) provide some typical child-directed morphological possessives, including an overt possessor in (15).
NEC possessive constructions in child-directed speech
Notes. MPS = constructions using the morphological possessive strategy. EPS = constructions using the equational possessive strategy. Possessor + DEM = EPS constructions with a possessor and a demonstrative as the possessee. Possessor only = EPS constructions with only a possessor.
However, CD does model the EPS in her speech to some degree (51/272 constructions, 18.8%). This is exemplified in (16–17).
Turning now to the child, Table 2 provides a count of all the NEC possessive constructions in Ani's speech. Several important patterns emerge from this child data. First, the EPS dominates Ani's total possessive constructions (89/101, 88.1%), which means she usually does not produce nouns requiring possessive inflection.
NEC possessive constructions in Ani's speech
Notes. MPS = constructions using the morphological possessive strategy. An asterisk indicates the number of MPS constructions missing the obligatory person prefix. EPS = constructions using the equational possessive strategy. Possessor + DEM = EPS constructions with a possessor and a demonstrative as the possessee. Possessor only = EPS constructions with only a possessor.
Second, at age 2;07.06 Ani produces her first common nouns that can be reasonably interpreted as possessees, but they are all missing possessive morphology. Example (18) is one such construction. This example includes (i) a representation of the target utterance in NEC orthography, (ii) an IPA transcription of the pronunciation of the target utterance, and (iii) an IPA transcription of Ani's actual production. Note that target pronunciations often do not closely match orthographical representations, primarily due to processes of stress/accent shift and vowel syncope (Knee, Reference Knee2014; Rose et al., Reference Rose, Brittain, Dyck, Swain, Franich, Iserman and Keil2010). In (18), the word ‘pants’ is missing all of the obligatory morphology: the third person prefix, possessive suffix, and animate obviative suffix (required for an animate possessee with a third person possessor).
Possessive inflection does not emerge until age 3;01.18. This is shown in Ani's correct production of an indefinite person prefix on the inalienably possessed noun ‘mouth’ in (19). However, the child is still missing such obligatory morphology in half of her productions at this age. In (20), for instance, the word niwîchâwâkin ‘my friend’ lacks two syllables as well as the prefix indicating a first person possessor.
Ani's final constructions requiring possessive morphology in the dataset occur at age 3;06.23, and the person prefix is present in all three examples: she produces the correct first person possessor morpheme ni- on two tokens of nimischisin ‘my shoe’ and once with a different noun, nipichiwiyân ‘my shirt’. None of these tokens require additional morphology. This is positive evidence of acquisition, but because of the very small sample size, one cannot generalize that Ani has definitively productive knowledge of the role of the obligatory person prefix in possessive inflection by 3;06.23.
Given these findings, RQ1 has clear answers. The vast majority of child-directed constructions employ the MPS and inflect the possessee. Furthermore, possessive inflection is frequent and robust in child-directed speech. However, Ani largely does not produce NEC possessive morphology. For a quantitative assessment of this discrepancy between speakers, a Fisher's exact test reveals that the difference in usage of the MPS and the EPS varies significantly between Ani and CD (p < .001, OR = 31.691), and there is a strong positive association (Rea & Parker, Reference Rea and Parker2014) between each speaker and her reliance upon each NEC possessive strategy (ϕ= 0.64). These results paint very different pictures of possessive strategies in child-directed and child speech, and the present study offers a two-pronged explanation for Ani's reliance on the EPS.
First, context plays a role in this imbalance. Heine (Reference Heine1997, p. 26), for example, suggests that possessive constructions may be categorized into two contexts: (i) where possession is presupposed, as in examples (14, 15, 18); and (ii) where possession is asserted, as in (10, 11, 16). To explore the connection between context and construction, Table 3 compares the frequencies of the two contexts for each speaker. A Fisher's exact test shows that the two speakers differ significantly in how frequently they each express presupposed and asserted possession (p < .001, OR = 18.690), and there is a relatively strong positive association between speaker and each context (ϕ = 0.59). Ani asserts possession significantly more often than CD.
Each speaker's NEC possessive constructions, classified by context
This raises the question of whether each speaker simply uses each type of possessive strategy in a different context. Table 4 classifies all of CD's possessive constructions. A Fisher's exact test shows that CD associates the MPS and EPS each with a different context (p < .001, OR = 18.292), and this association is relatively strong (ϕ= 0.47). In adult input, the MPS overwhelmingly expresses presupposed possession, but the EPS is more split between contexts.
CD's possessive constructions, classified by strategy and context
Notes. MPS = constructions using the morphological possessive strategy. EPS = constructions using the equational possessive strategy.
Table 5 shows that the child also associates the MPS more strongly with presupposed possession, but not to the extreme in adult speech. Unlike CD, Ani uses the EPS more frequently in asserted possession. A Fisher's exact test confirms that Ani tends to use each type of strategy in a different context (p < .01, OR = 5.621), but there is only a moderate positive association between strategy and context (ϕ = 0.29). Ani uses the EPS more frequently to express asserted possession, but she also employs the EPS with presupposed possession.
Ani's possessive constructions, classified by strategy and context
Notes. MPS = constructions using the morphological possessive strategy. EPS = constructions using the equational possessive strategy.
A second factor influencing Ani's reliance on the EPS is that she has not yet mastered the inflectional complexities of the MPS. The MPS requires a speaker to mark the head (possessee) of a possessive construction for grammatical properties of both the head and the dependent (possessor). This marking entails interacting categories of person, number, animacy, and obviation. At a minimum, a child must know whether a given noun is inalienably or alienably possessed, and then the possessee takes a prefix indexing the person of the possessor. The acquisitional challenge does not end there. A child must also understand the animacy categorization of that noun, which relates to whether the noun should take a possessive suffix. Animacy also affects whether the noun must be marked as obviative when possessed by a third person, and if so, which obviative suffix the possessee will take. If the possessee or possessor is plural, or if the possessor itself is obviative, then the possessee requires additional affixation again involving these interacting categories. However, the EPS largely circumvents these grammatical complexities. The other languages mentioned in the above review require children to navigate fewer such inflectional considerations. Even with these languages, children can still take until after the age of 3;0 to master the full mechanics of possessive marking. Ani's reliance upon the EPS reflects a child's need to express possession before she has acquired the full system of inflectional encoding.
Possession in constructions with English elements
The discussion so far has considered possessive constructions using only NEC, but at age 3;04.09 Ani begins using English elements in her expression of possession. This raises the question posed in RQ2: Does the presence of English impact Ani's possessive constructions? Perhaps, for example, Ani's reliance on the EPS also hinges on its superficial resemblance to some possessive strategies in English, which also entail more stripped-down marking than the MPS.
The first relevant point for RQ2 is that English is very rare in child-directed possessive constructions: CD uses almost exclusively NEC for expressing possession (272/273 constructions, 99.6%), and she produces only one English-language possessive across the entire sample. However, this particular construction is in direct response to the presence of English in a question from Ani. This exchange is provided in (21–22).
As for child speech, Table 6 provides a count of all the possessive constructions in Ani's production. This includes English-language constructions and those mixing NEC and English. Several important patterns emerge from this data.
All possessive constructions in Ani's speech
Notes. MPS = constructions using the morphological possessive strategy. An asterisk indicates the number of MPS constructions missing the obligatory person prefix. EPS = constructions using the equational possessive strategy. Possessor + DEM = EPS constructions with a possessor and a demonstrative as the possessee. Possessor only = EPS constructions with only a possessor. Pronoun only = constructions using only an English possessive pronoun. Pronoun + possessee = constructions using an English possessive pronoun with a possessee noun. Mixed = possessive constructions mixing NEC and English elements.
First, across the course of the transcripts, Ani tends overall to express possession by using NEC-only (100/142, 70.4%) rather than English-only constructions (38/142, 26.8%). Importantly for RQ2, she produces only NEC possessives from age 2;01.12 to 3;01.18. During this time she has established a clear lack of the MPS, where 69/78 (88.5%) of her total possessives to this point use the EPS. However, a change occurs at age 3;04.09, when Ani begins to produce English-language possessive constructions. At this age she produces “my Dora” and “my cheese”, along with pronoun-only possessives such as “mine”. At 3;06.23 and 3;08.24, she produces constructions consisting of a possessive pronoun with a possessee, such as “my toy”, “my shoes”, and “my doctor”. Once English appears in Ani's possessives, English-only constructions take over a majority of Ani's tokens for the remainder of the sampled sessions (38/64, 59.4%).
However, the emergence of English does not affect the balance of Ani's NEC possessive constructions. In fact, her reliance on the EPS is quite consistent. Before age 3;04.09, her NEC possessives are overwhelmingly EPS constructions (69/78, 88.4%). After this age, the percentage of NEC possessives using the EPS remains nearly the same (20/23, 87.0%). A Fisher's exact test confirms that Ani's usage of the EPS over the MPS did not vary significantly before and after age 3;04.09 (p = 1, OR = 0.884, ϕ = –0.02). Ani continued to rely on the EPS in her NEC constructions to the same extent after English appeared in her overall production of possessives.
Finally, Ani only mixes NEC and English in a possessive three times, and she uses both the EPS and MPS. In (23) she produces a full English noun as the possessee in an EPS construction. In (24) she uses the MPS by producing an English noun inflected with a Cree person prefix and the English plural suffix.
Altogether, these findings offer answers for RQ2. English is such a negligible presence in child-directed possessives that it does not influence Ani's reliance on the EPS. Most importantly, the emergence of English in her possessive constructions at age 3;04.09 has no effect on how much Ani relies on the EPS. The present study can neither account for why English suddenly appears in Ani's possessives, nor why English subsequently occupies a majority of her possessive productions. However, it is clear that Ani establishes a reliance upon the EPS that remains constant before and after English becomes a presence in her possessive constructions.
Conclusion
Ani's reliance upon the EPS, despite the frequency of the MPS in adult speech, is a reflection of two factors. First, Ani tends to express asserted possession, a context in which the EPS is more commonly found than the MPS in both adult and child speech. Second, Ani has not acquired productive mastery of the inflectional complexities of the MPS. The child only produces scant but positive acquisitional evidence of some MPS marking at age 3;06.23, followed by no such evidence at age 3;08.24. Thus, the MPS appears to be a late development. The presence of English in Ani's speech has no bearing here, as Ani depends upon the EPS before and after English appears in her possessives. Instead, Ani has not mastered the encoding required by the MPS, and the EPS generally allows a speaker to express possession without these inflectional elements. In expressing possession without obligatory grammatical marking, Ani's production resembles cross-linguistic findings from some non-polysynthetic and polysynthetic languages.
Implications and future directions
These results are based on a small sample size (one child subject and child-directed input from one adult, in a restricted context). Although this is often the norm for data from understudied languages such as NEC, this does constrain the generalizability of the findings. Nonetheless, the answers to RQ1 and RQ2 raise additional questions for understanding the acquisition of morphologically complex languages and the relationship between child-directed speech and child production.
For example, the role of frequency and morphological regularity in the acquisition of NEC requires further investigation. The MPS, which employs regular and frequent usage of person prefixes, is robust in adult input but accounts for less than one-tenth of Ani's possessives. This is puzzling given the fact that these prefixes are also used with intransitive verbs, which Terry (Reference Terry2010) has demonstrated are the most frequent verb type in Ani's speech. However, Terry (Reference Terry2010, pp. 155–156) concludes that Ani's inflectional morphology for such verbs is only just emerging during ages 3;04.09 to 3;08.24. The present study shows this is also when Ani is showing the first positive evidence of acquiring the prefix morphology for possession.
Additionally, as (11) demonstrates, the EPS does not altogether circumvent inflection. The demonstratives that occupy a central role in the EPS also inflect for number and obviation while contending with grammatical animacy, even though using a demonstrative allows a speaker to eschew the person prefix and possessive suffix. Future research is required to investigate the acquisition of the NEC demonstrative stem and inflectional inventory, as well as other issues. The present study has provided the first step in this direction, shining light on the development of a central construct in early language acquisition.
Acknowledgements
I gratefully acknowledge Dr Kamil U. Deen for his diligent guidance and input on this project. Special thanks go to Dr Julie Brittain and Dr Yvan Rose of the CCLAS project at Memorial University of Newfoundland for sharing their data and insight. I am also grateful for input from Dr Andrea L. Berez-Kroeker, Dr William O'Grady, Dr Geoffrey LaFlair, and the Language Acquisition Group at the University of Hawai‘i at Mānoa, as well as audiences at the 49th Algonquian Conference and the 2018 Winter Meeting of The Society for the Study of the Indigenous Languages of the Americas.
