The role of modality and typology in the acquisition of

1

The role of modality and typology in the acquisition of verb agreement morphology in British Sign Language∗ Gary Morgan1, Isabelle Barrière2&3 & Bencie Woll1

1

Department of Language and Communication Science, City University, London

2

Department of Cognitive Science, Johns Hopkins University

3

Linguistics, Department of Humanities, University of Hertfordshire

Corresponding author: Bencie Woll Department of Language and Communication Science City University London Northampton Square London EC1V 0HB e-mail: [email protected]

∗

Earlier versions of this paper were presented at the Child Language Seminar, University of Hertfordshire July 2001; the 9th International Congress for the Study of Child Language, University of Wisconsin-Madison July 2002; at the Department of Cognitive Science, Johns Hopkins University in 2001-2002; at BU 27th conference on Language Development, University of Boston November 2002; at the Psycholinguistics seminar series, Department of Psychology, University of Connecticut, USA, November 2002; and at the Verbs: Properties, Processes and Problems conference, UCL-City University, London, April 2003. We thank the audiences at these venues for their generous feedback. Thanks to Elena Lieven and Christian Rathmann for very helpful comments and to Géraldine Legendre for discussions on object-verb agreement. Without the support provided by Mark’s family this research would have been impossible. The work described was funded by a City University pump priming grant to Gary Morgan, Bencie Woll and Isabelle Barrière. Our Deaf colleagues Toby Burton, Frank Thomson and Sami Salo assisted us with parts of the data recording and transcription.

2

The role of modality and typology in the acquisition of verb agreement morphology in British Sign Language Abstract In the field of language acquisition, the acquisition of subject-verb agreement and of lexical sub-categorization are typically thought of as two different phenomena. Given the typological characteristics of British Sign Language (henceforth BSL), the acquisition of verb morphology in this language pertains to these two issues: a) it is a pro-drop language and b) the marking of both subject and object-verb agreement interacts with syntactic and semantic sub-lexical categorization features. This study reports the outcome of a case study of a young child acquiring BSL as a first language in light of current issues raised in the literature on the acquisition of agreement, argument structure and aspect, and on the role of the input.

3 The role of modality and typology in the acquisition of verb agreement morphology in British Sign Language Abstract In the field of language acquisition, the acquisition of subject-verb agreement and of lexical sub-categorization are typically thought of as two different phenomena. Given the typological characteristics of British Sign Language (henceforth BSL), the acquisition of verb morphology in this language pertains to these two issues: a) it is a pro-drop language and b) the marking of both subject and object-verb agreement interacts with syntactic and semantic sub-lexical categorization features. This study reports the outcome of a case study on a young child acquiring BSL as a first language in light of current issues raised in the literature on the acquisition of agreement, argument structure and aspect, and on the role of the input. 0. Introduction Once linguists began to seriously study sign languages they were faced with the inevitable conclusion that language was not synonymous with speech. Sign language research so far has demonstrated that in the organisation of grammar, in the constraints on articulation and processing, in the breakdown through brain lesions or developmental impairments and lastly in children’s first language acquisition, broadly similar patterns appear across the spoken and sign language modalities (Emmorey, 2002; Meier, Cormier & Quinto- Pozos, 2003). In one particular area, language acquisition, researchers have recently begun to go further and ask questions concerning some of the observed differences between sign and spoken language (Lillo-Martin, 1997; Lieven, 2002). In what ways and for what reasons may modality and typological characteristics affect typical first language acquisition? We focus on one particular grammatical deviceagreement - as a window onto this question. The paper is organised as follows: Section 1 reviews current work on the acquisition of subject and object-verb agreement marking. Section 2 includes a description of verb agreement in BSL as well as an analysis of subcategorisation based on syntactic and semantic cues, which is shown to interact with the marking of agreement. Section 3

4 describes the previous literature on the development of verb agreement across different signed languages. This section concludes with an examination of the main differences between child directed sign and spoken language. This leads us onto Section 4 and the current study which looks in detail at the emergence of BSL verb agreement morphology in a child between 1;10 and 3;0. In Section 5 we discuss the particular factors that may play a role in the development of morphological verb agreement in signing children. The final section discusses these results in the light of current research on spoken and signed languages. 1. The acquisition of subject- and object-verb agreement in spoken languages: findings and debates Agreement is best defined as a morphosyntactic phenomenon by which “the appearance of one item in a sentence in a particular form requires a second item which is grammatically linked with it to appear in a particular form” (Trask, 1993: 12). From a semantic perspective, subject- and object-verb agreement, in combination with word order indicate who is doing what to whom. At least two reasons have motivated the prolific study of subject- (and to a lesser extent, object-) verb agreement in the literature on language acquisition: Accounting for subject and object verb agreement in the adult grammar requires sophisticated linguistic specific-concepts, including that of functional categories; From the early days of studies of language acquisition, researchers noticed that the first word-combinations produced by young children acquiring English did not exhibit systematic marking of subject-verb agreement (Brown, 1973). This topic is probably the most controversial in language acquisition, as generative and constructivist approaches provide drastically different accounts and rely on different research strategies. We first review the findings that emerge on the acquisition of subject-verb agreement from studies of typologically-different languages. Secondly the debates and the limitation of both the generative and constructivist approaches are highlighted. The last section summarizes the literature on the acquisition of object-verb agreement which has been the focus of a much less extensive literature in language acquisition.

5 1.1. The acquisition of subject-verb agreement: findings Phillips (1995) provides a thorough synthesis of the cross-linguistic literature on the acquisition of subject-verb agreement from a generativist perspective. In doing so, he appeals to the concept of root-infinitive first proposed by Wexler (1994) that refers to the use of non-finite forms as a default in children’s utterances that contain verbs. Phillips’ synthesis relies on a number of detailed case studies (for which at least seven samples per child were available) as well as some additional corpora for which at least two phases of development were documented. Collapsing the data obtained at a maximum of three month intervals, he outlines the developmental pattern of the use of root-infinitives in contexts that require subject-verb agreement in a number of languages: •

English: For both Adam and Eve, a very high rate (more than 50%) of root infinitives are produced up till 2 years. Around age 2, this proportion dramatically goes down for Eve (around 40%, no data analysed after this age) while it remains high for Adam (above 60%, across samples) until at least 3 (no data analysed after that age);

•

Swedish (data from Platzack, 1990): After 2, 40%-50% use of root-infinitives (no data analysed after that age);

•

German (data from Behrens, 1993): between 1;8 and 2;5: 40%-70% use; with a drop to 30% between 2;5 and 3;0; 10%-20% for ages 3;0 to 3;5; and to less than 10% between 3;5 and 4;

•

Dutch (data from Haegeman, 1995): between 2;3 (no data analysed before this age) and 2;5: 20-30% use; after 2;5 a drop to 15%; and to 10% around 3 (no data analysed after this);

•

French (data from Champaud (MacWhinney, 1995) for Grégoire and Suppes; Smith & Leveillé, 1973, for Philippe) in which the subject clitic was treated as the agreement marker: Grégoire used around 30% root infinitives between 1;7/8 and 2; after 2 this percentage drops to 20% at 2;3; then to less than 15% around 2;3. Philippe used 10%-40% between 2;2 and 2;3 and about 10% from 2;3 (no data analysed after this);

•

Hebrew (data from Rhee &Wexler, 1995): 30%-60% between 1;6 and 2 and a sudden drop to 0% after 2 (data analysed until 2;5) in the corpus collected on one

6 child, Naama; in other Hebrew-speaking children: 20%-40% before 2; and a drop to 0% around 2;5 (data available until age 3); •

Spanish (data from Grinstead, 1993): about 12% use at 1;8, dropping to less than 5% around 2;5;

•

Italian (data from Guasti, 1992): Martina produces 7%-25% of root infinites between 1;7 and 1;8; 5%-15% until age 2;0; after 2 the percentage drops to 3% at 2;1 and 0% at 2;3 (no data analysed after this age).

The data analyses reported above document the omission rather than the substitution of agreement inflections. According to Phillips (1995), reports of substitutions are very rare in the literature. For instance, in the data on the acquisition of German collected by Clahsen & Penke (1992), the inappropriate use of subject-verb agreement marking is almost non-existent. The same applies to the analysis of Italian by Guasti (1992) (cited in Phillips, 1995: 4). However more recent studies of the acquisition of French (Ferdinand, 1996), Catalan (Grinstead, 1998) and Spanish (Grinstead, 1998) report the inappropriate use of 3rd person inflection as a default form. Gathercole, Sebastián & Soto (1999, 2000) also report substantial proportions of person errors in the samples produced by two monolingual Spanish-speaking subjects. Phillips (1995) also describes the interaction between subject-verb agreement marking and word-order. This is important in adult grammar as finite and non-finite forms appear in different syntactic contexts. For instance, in German, a V2 language, in the speech samples produced by a child aged 2;1 (Wagner, 1985) almost all verbs that carry agreement markers appear in second position while most root-infinitives appear at the end of a clause (Poeppel & Wexler, 1993). In French, Pierce (1990, 1992) found that root-infinitives follow the negation marker, while verbs that are marked for agreement through the use of the subject clitic are located before negation markers. The distribution of null subjects in children’s speech has also been shown to be sensitive to the presence of subject-verb agreement marking (Pierce, 1990, 1992): root-infinitives typically appear in null subject contexts. In most of the languages mentioned above, including English and French, tense and agreement markers are fused; and in the analyses presented above they are not systematically distinguished in children’s speech samples. The originality of the

7 contributions of Davidson & Goldrick (2001), Legendre et al (2002) and Davidson and Legendre (in press) lie in distinguishing tense and agreement marking in their analysis of the acquisition of French (Legendre et al, 2002; Davidson & Legendre, in press) and Catalan (Davidson & Goldrick, 2001; Davidson & Legendre, in press). In Catalan, the acquisition of both tense and agreement markers occur in parallel. In contrast, in French, a 3-stage model emerges from the reanalyses of three corpora: •

Phase I, during which tense-markers are used productively while agreement markers hardly ever appear: between 1;9 and 1;10 for Grégoire; and 2;2 and 2;3 for Stéphane (Rondal, 1985).

•

Phase II, during which agreement markers compete with tense markers, which results in a higher proportion of omitted tense-markers, compared to Phase I: between 2;0 and 2;3 for Grégoire; 2;6 and 2;8 for Stéphane; and 2;1 and 2;2 for Philippe.

•

Phase III, characterized by the mastery of both tense and agreement markers: at 2;5 for Grégoire; 3;3 for Stéphane; and 2;6 for Philippe.

These phases in the acquisition of tense and agreement are also distinguished on the basis of the Preferred Length of Utterance (Vainikka, Legendre & Todorova ,1999), an alternative measure to MLU that considers two dimensions: the number of words per utterance and the percentage of utterances that contain verbs. Crago & Allen (2001) contribute additional cross-linguistic data as their study focuses on the acquisition of finiteness by four typically developing Inuktitut-speaking children. The data analysed were collected at the ages of 2;0.11, 2;0.20, 1;11.7, and 2;1.11. The results demonstrate that the proportion of inflected forms in these children’s language was very high: between 91% and 100%, with an average of 96% across children (Crago & Allen, 2001: 79). Their analysis included the consideration of the productivity of morpheme-affixation - i.e. whether the relevant morpheme appears with more than one verb root - and of the root - i.e. whether the same root appears with different morphemes. A common feature characterizes the studies by Crago & Allen (2001), Davidson & Goldrick (2001), Legendre et al (2002) and Davidson & Legendre (in press): they consider both the typological characteristics of the language involved and draw

8 quantitative comparisons between children’s production of finite and non-finite forms and that found in adult language, although their approaches are not solely input-driven. Although the findings reported above are based on a small number of children for each language, the examination of the acquisition of subject-verb agreement by bilingual French-English children has confirmed distinct developmental patterns across languages: for example, finiteness has been found to be acquired first in French (Paradis & Genesee, 1997). Garman, Schelletter & Sinka (2000) who compare German-English and LatvianEnglish bilingual development also report within-individual cross-linguistically distinct acquisition patterns of functional categories. On the basis of the findings briefly outlined above, the factors that seem to play a role in the acquisition of subject-verb agreement include: a) Whether the language allows null subjects: around age 2, children acquiring these languages seem to produce mostly inflected verbs; b) Whether the verbal paradigm exhibits a rich morphology: in English, the systematic use of inflections emerges much later than in Italian and Inuktitut. It has proved difficult to disentangle these two factors, as languages that exhibit a rich morphological verbal paradigm are typically pro-drop (but not the reverse, cf Chinese). The data on the acquisition of Icelandic (Sigurjónsdóttir, 1992), (a language with rich agreement marking but which does not allow null subjects) suggest that it is the lack of subjects that triggers an extended stage of use of non-finite forms. Studies of the acquisition of Greek (Varlokosta, Vainikka & Rohrbacher, 1998) and Spanish (Grinstead, 1998) appear to provide contrastive evidence, as in these languages which allow null subjects and exhibit rich morphology, children go through a (short, compared to English) stage during which they use default forms. 1.2 The acquisition of subject-verb agreement: debates Two approaches have attempted to account for the acquisition of subject-verb agreement: the constructivist approach and the generative approach. The merits and the limitations of these approaches are briefly reviewed below. The constructivist approach (Tomasello, 1992, Lieven, Pine & Baldwin, 1997) adopts a lexically-driven perspective on the learning of verbs forms and frames (argument structure): specific verbs occur in specific inflected forms and are associated with specific

9 frames with respect to the distribution of subject and object. The main factor that predicts the forms and frames of specific verbs is the quantitative aspect of the input. Adopting this approach, Lieven, Pine & Baldwin (1997) account for 60% of children’s early multi-word utterances. This figure contrasts with Crago & Allen (2001), Legendre et al (2002) and Davidson & Legendre (2003) whose explanations attempt to account for 100% of the children’s relevant utterances. The second methodological problem with the constructivist approach has been recently emphasized by Naigles (2002, 2003): lack of evidence of productivity in a speech sample does not in and by itself constitute valid evidence against productivity. Finally, although the vast majority of studies on the acquisition of subject-verb agreement are based on production data, a recent study has explored English-speaking infants’ sensitivity to grammatical versus ungrammatical subject-verb agreement, using the Head Turning Procedure (Soderstrom, 2002, Soderstrom, Wexler & Jusczyk, 2003): the results show that 19 month old infants prefer grammatical passages over ungrammatical ones, including samples where nonce-verbs appear in the stimuli. It is difficult to see how the constructivist approach would account for these findings although their work is important in reminding researchers to examine the productivity of the root verb and inflections (as Crago & Allen, 2001 did for instance). Given that for constructivists the early stages of multi-word combinations are not underpinned by grammatical principles, they do not consider the interactions mentioned above with respect to finiteness and other linguistic phenomena, such as wordorder, the status of null subjects etc. Constructivist claims are disputed by authors who view early multi-word utterances as a first stage of morphosyntactic development and who rely on generative and Optimality Theory frameworks. The generative approach has given rise to two groups of hypotheses: a) the Weak Continuity Hypothesis, according to which phrasal structure associated with functional categories (e.g. Agreement, Tense, etc) undergoes a maturational process: these structures do not emerge until later stages of language development (e.g. Aldridge, 1989; Radford, 1990 1996; Rizzi, 1993,4; Vainikka, 1993,4, Wijnen, 1995); and b) the Full Competence Hypothesis, according to which early multiword utterances provide evidence for the availability of functional categories. In this view, adult-like phrase structure is available to the child from the beginning of syntactic

10 acquisition (e.g. Hyams, 1992; Pierce, 1992; Poeppel & Wexler, 1993, Phillips, 1995; Wexler, 1998), but performance factors such as morphological processing (Phillips, 1995) and/or pragmatic factors (Wexler, 1998) explain children’s use of nonfinite/default forms. There are limitations to the generative approach: from a methodological perspective the criteria for assessing the productivity of certain forms are not always clear; and the fluctuations from one stage to the next are not systematically accounted for (in that the accounts typically attempt to explain the first and last data points or general increases and decreases). In the Optimality Theory framework, Legendre et al (2002) and Davidson & Legendre (in press) adopt an approach that combines assumptions from both the Weak Continuity Hypothesis and the Full Competence Hypothesis and provide a stage by stage account of the fluctuations of tense and agreement markers in children’s speech samples. Phonological Bootstrapping accounts (Morgan & Demuth, 1996; Weissenborn & Höhle, 2000) that have come out of the generative approach consider the interface between phonology and morphosyntax: for instance Demuth (2000) argues that the distinct prosodic features of Spanish and English explain the different patterns of grammatical development that children acquiring these two languages exhibit. The acquisition of object-verb agreements: findings and accounts The acquisition of object-verb agreement has been the focus of a much less extensive literature, as it is not found in most of the Indo-European languages that have been the focus of language acquisition studies. The results that emerge from Nairobi Swahili (Deen, 2002, 2003), Basque (Meisel & Ezeizabarrena, 1996) and Georgian (Imedadze & Tuite, 1992) suggest that language-specific typological characteristics may play a role in the pattern of acquisition of subject- and object-verb agreement and the age at which the full paradigm of these markers is mastered. Object agreement marking in Nairobi Swahili is constrained to specific objects (Deen, 2002, 2003) while subject agreement is obligatory except in specific discourse-marked contexts. The analyses of four longitudinal corpora of children aged between 1;8 and 3 reveal that:

11 From 1;8, their pattern of object-verb agreement marking is close to that of adults in two ways: object agreement hardly ever occurs in the context of intransitive verbs (0.4%) and in 93% of the contexts of names that are object-specific, object agreement is marked; In contrast, subject-verb agreement marking is unadultlike in that it is omitted (in 61%72% of relevant utterances) in the earliest stages (that are defined according to MLU, verbs-per-utterance and filler syllables); omission drops to 44% at stage 3 and to 32% at Stage 4 (compared to adult omission of subject-verb agreement in discourse-appropriate contexts in only 5% of the relevant utterances). A different developmental pattern emerges from the acquisition of object-verb agreement in Basque (Meisel & Ezeizabarrena, 1996), based on the analysis of the data collected on three children Jurgi, Mikel and Peru, aged between 1;06 and 5;03: Phase IA: use of non-finite verbs in contexts in which finite verbs are required. Phase IB (reached at 1;07 by Mikel): emergence of finite verbs, no evidence of productive use of inflections; Phase II: productive use of subject-verb agreement marking (between 1;09-1;11 for Mikel, 2;04-2;07 for Jurgi) The productivity criterion for assigning a child’s sample to phase 2 consisted of a significant increase in number of finite forms compared to the previous time point. In some of the subjects’ speech samples, periphrastic forms, with inflected auxiliaries, emerge. Aspectual markers also emerge. In all cases, subject agreement only exhibits 3rd person singular marking. Phase III: distinction between grammatical person in subject-verb agreement (between 2;0 and 2;03 for Mikel, 2;08-3;0 in Jurgi). More than one form for agreement occurs (including unaccusative and transitive constructions that require different third person morphemes) and errors in person agreement disappear. Phase IV: Object-Verb Agreement (between 2;04 and 2;06 for Mikel and 3;01 and 3;02 for Jurgi, from 2;09 for Peru). Two typological features impede the identification of the stage at which object-verb agreement emerges: a) null objects are allowed and b) in the data samples, most contexts in which object-verb agreement are obligatory bear the 3rd person singular feature and the

12 marking corresponding to this person is a zero-morpheme. In all children, contexts for 3rd person singular marked with zero morphemes are identified at 1;10 (Mikel), 2;04 (Peru) and 2;08 (Jurgi). This is followed by the emergence of the overt 2nd person plural object-verb agreement marking -z at 2;04 (Mikel), 2;09 (Peru) and 3;01 (Jurgi) and the overt 3rd person plural object-verb agreement marking –it at 2;10 (Mikel), 3;04 (Jurgi) and 3;05 (Peru). Evidence of 1st singular n- is found at 3;09 (Peru) and 3;11 (Jurgi). Thus evidence of acquisition of object-verb agreement is found four to five months after productive use of subject-verb agreement marking. Although different types of overt object-agreement marking morphemes are used at this stage, children still produce unadultlike constructions, including inappropriate use of subject-, direct object- and indirect object-agreement with verbs that require only subject- and object-verb agreement marking. Omissions are still identified (between 2;04 and 2;10 for Mikel and 3;01 and 3;04, for Jurgi). Phase V: Indirect Object-Verb-Object agreement (2;07 for Mikel, 3 for Peru and 3;03 for Jurgi) As in the case of direct object-verb agreement marking it is difficult to identify the emergence of indirect object markers. When they start using these forms, children produce unadultlike utterances in that they use object-verb agreement in appropriate contexts (transitive verbs) as well as inappropriate contexts (unaccusatives). The authors also identified the emergence of dative agreement: although very few examples are identified in the child’s data, no evidence of inappropriate use or substitution is found. To summarize the findings on the acquisition of Basque, subject-agreement appears to be acquired before object-agreement; and children’s unadultlike constructions: a) occur in contexts that require plural markings and b) involve changes, typically increases, in valency. Although the latter phenomenon may be related to the overgeneralizations of argument structure alternations found in typologically different languages from around 2;6-3;0, including Brazilian Portuguese (Figueira, 1984), English (Bowerman, 1974, 1982a, 1982b, 1990; Lord, 1979; Pinker, 1989 among others) Hebrew (Berman, 1982, 1993, 1994) , Inuktitut (Allen, 1996), Kiche (Pye, 1994) and French (Barrière, Lorch & le Normand, 2000), evidence for Basque-speaking children’s distinction between transitive and intransitive verbs involves the absence of use of a transitive affix with intransitive

13 verbs (Meisel & Ezeizabarrena, 1996: 234). It is interesting to note that the authors mention that such distinctions might be made by children on the basis of the verb semantics rather than morphological marking: this seems to contradict the lack of semantic factors involved in the overgeneralizations of argument structure alternation found in Inuktitut and Kiche (see Barrière, Lorch and Le Normand, 2000 for discussion of the data). The findings on the acquisition of Basque are partly similar to those reported for the acquisition of subject and object clitics in French. The findings which emerge from the identification of the clitics in a child’s speech samples show that subject clitics first emerge at 2;09.2 while the first object clitic is produced at 2;2.23 (only 1 occurrence) (Hamann, Rizzi & Frauenfelder, 1995). Three more examples are produced at 2;6.16 while 10 are identified when the child reaches 2;9.30. Unfortunately, Hamann, Rizzi & Frauenfelder (1995) provide no information with respect to the verb hosts with which the object clitics appear. Except for the earliest stages, where two samples contain only 4 subject clitics, there is a very high number of subject clitics (in total 278 out of 771 verbal utterances, as opposed to 36 object clitics). It is therefore unlikely that subject clitics only appear with specific verbs. Kaiser (1990) reports a similar pattern of delay in the productive use of object clitics compared to subject clitics. Barrière, Lorch & Le Normand (2000) used more conservative productivity criteria to assess the acquisition of the (Reflexive and Reciprocal) clitic-se and report similar ages. On the basis of these findings the gap between the acquisition of subject and object clitics resembles that between subject and verb-object agreement marking in Basque: about 5 months; but an earlier and higher number of object clitics produced by French-speaking children constitutes more robust evidence of acquisition. The emergence of object clitics in French, including the se-clitic, has been shown to correspond to the age at which children use argument structure alternation productively (around 3, the same age as in English, Hebrew, Inuktitut and Kiche) (Barrière, Lorch & Le Normand, 1999, 2000, 2001). Instances of omissions and substitutions of French object clitics involve unadultlike casemarking (Jacubowicz, 1991) and overgeneralizations of arguments structure alternation (Barrière, Lorch & Le Normand, 1999, 2000, 2001), but at a quantitative level these

14 unadultlike utterances seem to be much rarer than in the Basque data analysed by Meisel and Ezeizabarrenna (1996). The acquisition of subject and object verb agreement has also been documented for Georgian (Imedadze & Tuite, 1992), an ergative language exhibiting typological characteristics that differ from both Basque and Nairobi Swahili. Georgian is best described as an agglutinating language: lexical meanings and most grammatical functions are expressed with different morphemes that can be separated from one another (Spencer, 1991: 38). A Georgian verb is composed of a number of morphemes, including: At least one preverb; One or two agreement prefixes; A preradical vowel; A verb root; A passive/inchoative suffix; A series marker; causative suffix; An imperfect stem suffix; A tense/mood vowel; One or two person/number suffixes. (based on Imedadze & Tuite, 1992: 41) Agreement marking is subject to three constraints: 1

It is obligatory with 1st and 2nd person subjects and objects;

2

With respect to 3rd person agreement marking: a) animate subjects control

agreement; b) agreement with 3rd person plural direct or indirect object rarely occurs; and c) agreement with inanimate subjects rarely occurs; 3 A 3rd person NP can control agreement only if it is formally marked for plurals, i.e. if it carries a plural suffix that does not apply to nouns modified by quantifiers. (Adapted from Imedadze & Tuite, 1992: 94)

15 According to Imedadze and Tuite (1992), as soon as they start producing agreement markers, 2 year old children “grasp the relationship between animacy and number agreement” (94). However, for a long time, until at least 5, children’s agreement marking is not quantitatively like that of adults, in that they frequently omit agreement markers including those in contexts in which the subject is plural and refers to an animate. Given that in some dialects of Georgian, such omissions are permitted in the adult language under specific discourse conditions, Imedadze and Tuite (1992) hypothesize that children’s protracted use of unadult-like forms can be accounted for on the basis of pragmatic factors. A factor that also seems to play a part in the acquisition of verb agreement morphology is whether the person marking morpheme constitutes a portmanteau morph: in one of the conjugation paradims, information about 3rd person subject marking, tense and mood are fused (Imedadze and Tuite, 1992: 64). This particular morpheme seems to trigger unadultike forms until at least children’s fifth birthday. A second typological feature of Georgian that is particularly relevant to this study (see section 2 below on BSL) is that Georgian verb classes depend on both syntactic (transitivity) and semantic (aspectual) features that are briefly outlined below: The vast majority of 1st conjugation verbs are transitive; Most of the verbs in the 3 other conjugations are intransitive but differ with respect to their aspectual class: a) 3rd conjugation verbs refer to activities, in the sense of [-punctual], [- telic] and [+dynamic], as defined by Vendler (1967) and Dowty (1979); b) most 4th conjugation verbs refer to states, in the sense of [punctual], [- telic], [- dynamic] as defined Vendler (1967) and Dowty (1979) c) 2nd conjugation verbs divide into 3 groups: prefixals that correspond to English passives and have a corresponding transitive in the 1st conjugation paradigm; suffixals that refer to changes of states and most of them to the beginning of a change of states, i.e. inchoatives; root second

16 conjugation verbs that refer to achievement, that is [+ punctual], [+telic] and [+ dynamic] in Vendler’s (1967) classification. (Adapted from Imedadze & Tuite, 1992: 41-43). It is important to note that the mapping between aspect and verb class is not systematic: the 2nd conjugation also includes a few activity verbs and the 1st conjugation includes some stative verbs (Imedadze & Tuite, 1992). The assignment of a verb to one or another class has very important consequences for crucial aspects of the morphosyntax as it interacts with case-assignment: 1st and 3rd conjugation can assign ergative case while 2nd and 4th conjugation cannot. With respect to the acquisition of these classes, Imedadze & Tuite (1992) note that it is during the end of their second year or the first half of their third year that children start employing adult-like aspectual and verb class contrasts that also give rise to adult-like case-marking. It is at the same time that the basic agreement rules (with subject, direct and indirect object) start to be mastered (Imedadze & Tuite, 1992: 57-58) although some aspects of person marking (see above) remain unadultlike until after the 4th year.

Unlike Deen (2003), for Nairobi Swahili, and Meisel and Ezeizabarrena (1996) for Basque, Imedadze & Tuite (1992) do not draw an order of acquisition for subject versus object agreement. They do mention the fact that specific person marking seems to remain unadultlike for a relatively long time, such as the 3rd person marking described above and with 1st and 2nd person object with indirect verbs (Imedadze & Tuite, 1992: 58). There are several limitations to this study: a) the authors do not provide any criteria regarding their definition of early mastery; b) many of the features that seem to be mastered late by children also seem subject to dialectal variation but little information is provided regarding the dialect to which the children are exposed; and c) no quantitative comparison is drawn with the adult language to which children are exposed. On the basis of this brief review of the literature on the acquisition of object-verb agreement, it seems that a number of factors may explain the order of acquisition of subject and object verb agreement. These include at least the following typological characteristics:

17 the morphosyntactic constraints for obligatory object verb agreement: only with specific objects for Nairobi Swahili versus a complex range of semantic (based on animacy features, person number and thematic role) and syntactic (subject, direct object and indirect object) features for Georgian; the complexity of the sublexical features that interact with agreement: Georgian appears to be more complex than both Basque and Nairobi Swahili with respect to the interaction between the nature of the agreement marking and the verb class; the realization of the markers: the fact that 3rd person singular object-marking is not overt in Basque impedes the identification of object agreement marking in children’s speech samples, and a portmanteau morph that fuses person, mood and tense marking in Georgian is mastered late by children. The authors also used different research strategies: while Deen (2003) clearly defined obligatory contexts and drew quantitative comparisons with the adult input (as also in Crago & Allen, 2001; Legendre et al, 2002; Davidson & Legendre, 2003; and Goldrick & Davidson, 2001 when investigating the acquisition of subject-verb agreement), neither Imedadze & Tuite (1992) nor Meisel & Ezeizabarrena (1996) considered quantitative aspects of the adult language data as a point of comparison. 2. Agreement in BSL In the literature on the acquisition of verb agreement, distinctions have typically been drawn between languages that exhibit: subject-only or both subject- and object-agreement marking; the obligatory versus non-obligatory use of overt subjects; a rich or poor verbal agreement paradigm; distinct prosodic patterns. One of the limitations of the generative accounts mentioned above, including those that propose that children’s lack of access to morphology explains their production of default forms (such as Phillips, 1995), is that they consider only the richness of the verbal paradigm (English versus Italian, for instance) and have little to say about the possible effects of specific morphological processes.

18 Legendre et al (2002) point indirectly to this issue when drawing our attention to the fusional character of tense and agreement marking. Authors who have questioned whether children acquiring English (Ingham, 1998) and Dutch (Hoektstra & Hyams, 1998) use tense-markers as temporal or aspectual markers also indirectly refer to this phenomenon. Although Meisel & Ezeizabarrena (1996) provide very little information on the morphological processes that characterize subject- and object-verb agreement marking in Basque, they do mention the fact that subject-verb agreement (as opposed to object-verb agreement) marking, at least for some persons, is fused with tense-marking. Imedadze & Tuite (1992) also explain the protracted acquisition of one agreement marker on the basis of its fusional character. Although the Phonological Boostrapping literature considers the phonological realization of the relevant morphosyntactic features involved, accounts such as the Prosodic Constraint Hypothesis (Demuth, 2000) focus on phonological features and their interaction with the morphosyntax rather than on morphological processes (listed below) per se. Traditional typology classifies the morphological processes involved into four groups: isolating, agglutinating/ve, (in)flectional and polysynthetic (Spencer, 1991): Isolating applies to languages with very little overt morphology (except for compounding) and in which, therefore, distinct meanings are expressed using different words (Spencer, 1991: 38). Thus words in these languages (e.g.,Vietnamese and Chinese) tend to be monomorphemic. Agglutinating/ive morphology characterizes languages, such as Hungarian and Turkish, in which lexical meanings and grammatical functions are conveyed using different morphemes that can be separated from one another (Spencer, 1991: 38). Inflectional applies to languages in which morphemes are best characterised as portmanteau morphs in that they carry the meaning of more than one morpheme (Trask, 1993: 211): the term fusional is sometimes used to characterize this property. Examples of such processes include the English morpheme –s which expresses both 3rd person singular and present tense. In languages such as Latin, Spanish, French, Italian etc, such processes are common , especially with the verb endings that typically fuse person, tense and mood marking.

19 Polysynthetic applies to languages in which words carry many bound morphemes, the meaning of which would typically be expressed by a word in another language. In these languages a word may stand for what would be expressed through a whole sentence in other languages (Spencer, 1991: 38). These languages may also exhibit the use of portmanteau morphs (see for instance Inuktitut (Allen, 1996). Although the use of these concepts to classify languages has been questioned, researchers in language acquisition often employ them to describe the language they are investigating (for instance Allen (1996) characterizes Inuktitut as polysynthetic). Another concept - Non-Concatenative Morphology- has been proposed more recently to capture the non-linear nature of the morphological processes involved in e.g. Semitic languages, in which the base of the word may be altered and reduplication (repetition of some morphological material within a single form (Trask, 1993: 231)), conversion (alteration of a word lexical category without any affixing (Trask, 1993: 309; Spencer, 1991: 20)) and suppletion (change of stem (Spencer, 1991: 8)) occur. The description of BSL below discusses the characteristics that have been considered to be relevant to the acquisition of verb agreement, as well as to the typological characteristics of the morphological processes it employs. With respect to the number of bound morphemes that can be affixed to a sign, sign languages are best characterised as polysynthetic, as a single sign may be polymorphemic. In (1) YOU-HAND-ROUND-OBJECT-ME, the direction of the sign’s movement and the embedded classifier1 for ‘round entity’ allow several meaning components to be articulated simultaneously. This same meaning in English requires 4 content words and two function words i.e. 6 morphemes to translate (‘give the round thing to me’). Sign languages have been typically compared to agglutinating languages in the acquisition literature (Meier, 2002: 126), (i.e., languages which do not typically exhibit portmanteau morphs and in which morphemes are linearly ordered). Zwitserlood (2002, 1

In all sign languages described to date, handshapes in verbs of motion and location represent object class

(e.g. long thin object, small cylindrical object, human, animate non-human, etc) (Emmorey, 2003)

20 2003) convincingly argues that morphophonology in sign languages is best characterized as non-concatenative, in that the phonological features involved are sequential but not adjacent. From this morphological perspective, sign languages are closer to Semitic and Bantu languages than to Turkish. Three types of verbs are distinguished in BSL (Sutton-Spence & Woll, 1999): a)

Plain verbs: marked for manner and aspect;

b)

Agreement verbs: marked for manner, aspect, person and number

(for both subject and object); c)

Spatial verbs: (cf. Fig. 1 above) which make use of topographic (as

opposed to syntactic) space, and are marked for manner, aspect, location and movement.

It is important to note that compared to the spoken languages that have been investigated in the literature on language acquisition, BSL & other signed languages exhibit a distinctive feature: morphological subject and object verb agreement only apply to one class of verbs, namely agreement verbs. In (2) the meaning of ‘ask’ requires two thematic roles: the agent and the patient. Word order and morphological agreement provide the intended interpretation. (2) The girl asks the boy The same meaning in BSL requires a different linguistic mapping. In signed languages, morphological agreement is realised by the movement of the verb stem between locations which have been previously indexed as subject (agent) and object (patient).

21 This is shown in an English gloss in (3). The movement of the sign between locations in sign space is shown in figure 1 2. (3) BOYj

IXj

GIRLk

IXk

kASKj

‘The girl asks the boy’

[Insert figure 1 here)

kASKj

‘(she) asks (him)’

In (3) the object of the sentence is first indexed to a location in front of the signer (BOY IX). Then the signer indexes a second location with the NP GIRL IX at the location of her own body (the IX is directed to her own chest). This location is understood as the subject index because the movement of the verb starts from the location of the signer

2

Signed sentences that appear in the text follow standard notation conventions. Signs are

represented by upper-case English glosses. IX is a point to an area of sign space, which acts as a syntactic index for referring to an argument in the sentence. In some examples the direction of the verbs movement is indicated by an arrow moving within a semicircle. Subscripted lower-case letters indicate coindexation. Above the glosses, nonmanual markers such as eye closes (∅∅) and upper face and head movements are indicated by a horizontal line across the affected segment(s).

22 (GIRL) and moves towards the location previously indexed for BOY (shown in figure 1 through an arrow indicating the direction of the movement).

Thus in BSL, as in other sign languages, spatial locations act as referential indexes (either the spatial location of the present referent or an arbitrary location assigned to a nonpresent referent). The referential index of a noun phrase is normally unexpressed in spoken languages; while in sign languages the referential index is overtly established through an index point to a location in sign space (Lillo-Martin, 2002). Agreement verbs make use of these locations. Some sign languages (for example SLN: Sign Language of the Netherlands or DGS: German Sign Language) have auxiliary verbs with a meaning of ‘act on’ (Bos, 1994; Rathmann & Mathur, 2002). In these sign languages a plain verb (one that does not move between locations) is coupled with the auxiliary verb which indicates the subject and object. In contrast in BSL, plain verbs, while remaining uninflected in sign space, can either be coupled with overt noun phrases and use word order to indicate who is doing what to whom or points towards indexed spatial locations to illustrate arguments. This is shown in figure 2. [Insert figure 2 here]

In spatial verbs (cf. example 1 above), to indicate who is doing what to whom is the location(s) - rather than the roles of agent and theme- of the entities involved are expressed. So in ‘you hand a round object to me’, the initial location of the verb HAND is at the location assigned to the object, not to the locations of either YOU or I. The development of verb agreement morphology is particularly interesting in BSL because the presence of inflectional morphology on verbs is constrained by the interaction of the semantic distinction between eventive and stative meanings and the transitive / intransitive verb frame pattern. Transitive eventive verbs (e.g. ASK, GIVE,

23 PUSH, BITE, HIT) can be inflected for subject / object agreement, while transitive stative verbs (KNOW, LIKE, WANT, BELIEVE) cannot. Intransitive frames can be filled with both eventive and stative verbs. Therefore BSL agreement verbs exhibit both specific semantic and syntactic subcategorisation features: Syntactic constraints: transitivity Semantic constraints: eventive encompasses activities, accomplishments and achievement, according to classical categorizations (Vendler, 1967 and Dowty, 1979), and excludes statives (i.e. events that do not express change over time, that do not involve an agent and do not have any recognizable goal) (Vendler, 1967, Dowty, 1979). Cross-linguistically the process of morphological verb agreement is similar in signed languages but the verbs that can and cannot take agreement may differ. The transitive stative verb HATE in ASL can be inflected for morphological verb agreement while in BSL it cannot, conversely the transitive eventive verb SUPERVISE can in BSL but in ASL cannot (Janis, 1995). These differences have led some ASL researchers to conclude that in ASL ‘there is a basic dichotomy between verbs that agree and those that do not. This dichotomy is not related to the semantics or transitivity of the verbs in any obvious way.’ (Janis, 1995 p198). Regardless of whether it is morphologically or syntactically (i.e through word order) expressed, the nature, number and distribution of participants play a role in the licensing of null arguments (Lillo-Marin, 1986). The signer’s own body is normally associated with the agentive role in the event being described (see Kegl, 1990). As a consequence subjects are less overtly marked than objects in BSL sentences. When there is one agreement slot available it is with the object, not the subject. If there are two slots available then the object is obligatory and the subject is optional (Rathmann & Mathur, 2002). Agreement verbs normally appear in sentence final position while plain verbs normally appear in SVO word order as shown in (4) (4a) JOHNj IXj MARYk IXk jASKk ‘John and Mary (he) asks (her)’

24

(4b) JOHN KNOW MARY ‘John knows Mary’ According to the brief outline presented above, the key characteristics of BSL agreement morphology are as follows: It is a pro-drop language in which subjects are more frequently omitted than objects; The morphological paradigm used to mark agreement is rich; Indication of who is doing what to whom with other types of verbs makes use of word order; The morphological processes which agreement marking involves are best described as non-concatenative and polysynthetic; Subject and object morphological agreement only apply to a subclass of verbs that are semantically (according to the type of events they refer to) and syntactically (transitivity) defined. In spoken languages such as Spanish, Italian and Hebrew mentioned in the review of the acquisition of subject-verb agreement, different verbs belong to different inflectional paradigms, and in this sense require sub-lexical knowledge. In this respect, the distinction between verb classes in those languages is less drastic than in BSL, in which two systems co-exist: one that involves morphological agreement marking and one that does not. It also differs from most ergative languages in which agreement marking is typically distinct for transitive and intransitive verbs but no difference is found to depend on the verbs aspectual dimension, as far as agreement marking is concerned. One of the spoken language that is most similar to BSL in this respect is Georgian. In the latter, as mentioned above, the verb conjugation paradigms and the nature of agreement marking are constrained by semantic (aspectual) and syntactic (transitivity) features. However the aspectual distinctions relevant to agreement marking in Georgian (activities versus statives versus inchoatives versus achievement) are more subtle than in BSL (eventiveincluding activity, achievement, and accomplishment- versus statives). The morphological realization is also more complex in Georgian in that it can be marked for a) subject, direct and indirect object versus subject and direct object in BSL, b) the nature

25 of the agreement marking in Georgian depends on animacy and person features assigned to the subject and object. The literature review on the acquisition of signed languages below considers in turn the roles that each of these characteristics may play in the acquisition of agreement.

3. Verb agreement in signed languages: developmental patterns and modalityspecific input 3.1 Developmental patterns Although there have been several studies of the acquisition of verb agreement morphology in sign languages (see Meier, 2002 for a review) it still remains an understudied area compared to spoken language development. The majority of studies carried out so far have been on American Sign Language (ASL) although data has been reported on other sign languages e.g. LIBRAS (Brazilian Sign Language: De Quadros, 1997), BSL (Morgan, Herman & Woll, 2002), SLN (Van den Bogaerde & Mills; 1996) and LIS (Italian Sign Language: Pizzuto, 2002). Many of these studies are on a small number of children (1-30) and use varying methodologies (e.g. spontaneous or elicited samples). There are two important issues to consider when looking at the development of morphological verb agreement. As described above, in the adult language, the inflection of verbs from the signer’s own location is normally accompanied with subject and object NPs (referential indexes). Verbs can be used without NPs if subject and object can be retrieved from the surrounding discourse (either as antecedents or when physically present); they indicate the identity of the arguments through movement to either the real world location of the referent or to an assigned location in sign space. Studies of development have largely focused on the use of agreement inflections for present referents (see Meier, 2002 for a review). The small number of studies of the development of agreement to non-present referents have used analysis of narrative as a

26 methodology (e.g. Loew, 1983; Morgan 2000). However mastery of narrative involves an additional cognitive load that may influence the age at which inflections are used. The use of agreement in narratives with non-present participant roles is a late development, with children showing a prolonged period of acquisition that continues past age 5;0 (Loew, 1983; Morgan 2000), marked by the use of appropriate directions of movements in agreement verbs but without identification of their arguments. Thus these studies show that signing children initially use verb agreement to indicate present referents before it is extended to use with absent referents. It is not clear whether this problem is to do with linking a word to a non-present referent (a general conceptual issue) or more to do with the particular linguistic devices used to refer to a non-present referent (abstract locations in sign space). In their review, Newport and Meier (1985: 905) concluded that ‘we see no reason to implicate morphology of verb agreement per se as the source of these errors. Rather, the data seem to us to suggest that the errors arise from difficulties in establishing and maintaining spatial loci’. There are few studies of spoken language acquisition to address the issue of inflection for non-present referents. However, in her discussion of the development of the notion of agency, Budwig (1990) demonstrates that in the early stages of acquisition, children only refer to themselves as agents. In the first studies of this feature of sign language grammar it was found that children master morphological verb agreement in present referent contexts at around 3;0 (Fischer, 1973; Hoffmeister, 1978; Kantor; 1982). Before this age children use word order without inflections e.g. POINT2 ASK POINT3, ‘you ask him’ where the arguments are instantiated through pointing to them instead of 2ASK3 where the arguments are encoded in the inflection. The movement of the verb is grammatically uninflected where the syntactic context as well as the meaning of the verb makes the inflection obligatory. In Meier (1981) there are errors of this type reported for 3 children. Between 2;8 and 3;8, one child (Shirley) omitted the inflection in 30 obligatory contexts, all involving an

27 inflection between a second and third person present referent. In general, children begin to use ASL agreement morphology from 2;0 – 2;6 with many verbs remaining uninflected up till 3;6. In their study of the acquisition of SLN, Van den Bogaerde & Mills (1996) report that in a 10 minute interaction at 2;6, one child produced 15 utterances containing a verb (27% of the total utterances) but none of them were inflected. One piece of evidence for the acquisition of a grammatical feature is overgeneralizations. There are some reports of cases of children overgeneralizing agreement to verbs that do not require agreement e.g. the class of BSL plain verbs: EAT, DRINK, SLEEP. Fischer, (1973) and Casey (2000) have reported on overgeneralisation of agreement morphology to plain verbs in ASL. Casey (2000) also reports errors of misagreement between the ages of 2;7 and 2;11, where verbs are moved towards the location of their subjects rather than towards the object location as was intended by the context. This last set of errors provided compelling evidence for the young child’s morphological rather than mimetic analysis of the movement of the verb (Meier, 2002). In summary, children master agreement for present referents in several different sign language around the age of three years. Non-first person referents continue to cause problems up to and beyond 4 years but this may be because of factors other than linguistic ones for example in the establishment and maintenance of spatial loci (Newport & Meier, 1985). 3.2 Modality-specific input One surprising common finding from several studies of early child/parent sign language interaction is that the number of signs addressed to young children is considerably less than that in hearing parent/hearing child dyads in the same community and also less than in adult/adult sign interaction (e.g. Kantor, 1982; Kyle & Ackerman, 1990; Gallaway & Woll, 1994; van den Bogaerde & Mills, 1996; Harris, 2001). Hearing mothers produce more utterances when talking to their hearing children than deaf mothers when signing to their deaf children. Even when the number of morphemes in child-directed signing

28 (CDS) is counted rather than number of signs the findings are the same (van den Bogaerde & Mills, 1996). Why is this? CDS is slower and produced at a different rate to adult directed signing because it must be linked to the child’s gaze at the signer (Kyle and Ackerman, 1990). In Harris (2001) interaction between seven deaf mother/children dyads was evaluated. Across the group 55-100% of the language addressed by mothers to 18-month-old children were single-signs utterances. Between 0 and 12% were 3+ sign utterances. Although the children were 18 months old mothers were rarely using multi-sign utterances with them. Harris (2001) compared sign and spoken language interaction explicitly. In a 10-minute play situation with 18 month old deaf children, she calculated that deaf mothers used between 40 and 84 utterances (average 62), with a majority being single signs compared with hearing mothers with hearing children of the same age who produced on average 166 utterances, the majority being multi-word utterances. Signing mothers provide fewer signs overall and also fewer syntactic contexts.3 Van den Bogaerde & Mills (1996) reported that the mean length of utterance in morphemes (MLU) of Dutch Sign Language (SLN) addressed to a deaf child aged 2;6 in a 10 minute interaction was 1.97 and that there were 12 verbs (21% of utterances). They suggested that the small number of verbs in CDS was linked to there being no copula in SLN. This difference between spoken and signed child addressed language is linked to fewer opportunities for using sign with a child in interaction. Put simply, you can only sign to a 3

The modality appears to be the important factor rather than the deafness of the child, as

hearing mothers produce a greater number of multi-word utterances to their deaf 18month olds than deaf mothers do in sign, even through the hearing mothers are aware that their children are not accessing all of this language through the spoken modality (Harris, 2001).

29 child when she is looking at you. As deaf children get older they better manage the requirements of shifting their visual attention to their signing parents (Harris, 2000) and to objects in the environment, but they are also more mobile and so more likely to want to explore and play with toys and thus move out of the ‘eye-shot’ of the ambient language. Although hearing children do not look at their parents all of the time, the language provided is accessible, since spoken languages can be heard while the child is looking at a book or overheard, even when it is not addressed to them directly. It should be noted, however, that it is not clear how much ambient language is taken up by the child. For instance, Crago & Allen (2001) extend their consideration of the input to the speech overheard by children and the utterances produced by their siblings while most studies that consider the input in other languages tend to pay attention to the speech of parents (Pine et al., 1997). Recent experimental research has shown that children can profit from ambient language that is overheard when later asked to generalise something from the overheard language to a new context (REFERENCE). Undoubtedly children can see sign not addressed to them in the ambient language but they have less access to this, as it cannot be ‘overseen’ if they are not looking towards the adult signers.

Despite there being fewer tokens in the sign language input, deaf children prefer sign motherese to adult/adult signing (Masataka, 2000). Hearing children not exposed at all to sign prefer signed motherese to random gesture (Masataka, 2000). The fact that language development occurs at generally the same rate across spoken and sign languages suggests that the amount of language addressed to children is not an important measure, as long as it is accessible (c.f. Goldin-Meadow’s research on deaf children exposed only to spoken language – Goldin-Meadow 2003). The differences between signed and spoken motherese should be seen to be more of type than quality. Differences between modalities are more comparable to those crosslinguistic differences that have been shown to play a role in language acquisition (e.g. Allen, 1996). Although adults may be using only single or two-sign combinations with children it is likely that more information is being supplied than just vocabulary (e.g. sign

30 language prosody, use of space, use of correct facial information and use of eye-gaze, etc). A more crucial difference between modalities on the input side is how morphological processes with signs are used. Deaf mothers make extensive use of modifications to a sign’s movement in order to make it more visible or to attract the child’s attention (e.g. articulating the sign in the child’s line of sight) (Harris, 2001).

This affects how much

morphological information can appear on the sign. Although there have been very few studies looking at verb agreement morphology in CDS, there exists the possibility that mothers may adapt their language by omitting inflections in certain contexts. Van den Bogaerde & Mills, (1996) reported that in a 10minute sample of motherese containing a total of 47 verbs, only 4 carried inflections. It should be noted that they conflated inflections for person agreement and spatial agreement and crucially they do not report on the relative frequency of omissions in obligatory contexts (see Pizzuto, 2002) but with less than 10% of the verbs being inflected it may be that morphological inflections for verb agreement were omitted. Inconsistency in the use of inflections in sign input may trigger variability in the patterns the child is attempting to analyse. While the features of CDS make communication successful they may have the effect of making important grammatical processes harder to identify.4 In adult-adult interaction, morphological inflections for verb agreement are obligatory in possible contexts. The possibility of grammatical optionality in child directed signing has received very little attention in the sign acquisition literature to date (Pizzuto, 2002). 4

There have been similar claims made about competition between grammar and non-

linguistic communication with the use of ASL ‘furrowed brows’ facial expression for marking questions and negations and the motivation to show a ‘baby friendly’ face to young children (Reilly & Bellugi, 1996).

31

A possible motivation for mother’s using less morphological verb agreement with young children is if they believe it simpler for the child if all sign movements take place in the same line of sight (See figure 3). [Insert figure 3 here]

In 1 the adult moves the sign ASK across the child’s line of sign towards the third person’s location. To process this sentence the child needs to move her eyes with the sign processing the meaning of the sign and the morphology simultaneously. In 2 the adult breaks the sign down into pronouns and an uninflected verb. It should be noted that these features of CDS become less common after 24 months (Erting, Prezioso & Hynes, 1990). This is just the time that children exposed to a spoken language are starting to use of inflectional morphology productively, to a larger or lesser extent depending on the typology of the language. Before describing the present study it may be useful to summarise the main issues raised so far. In our review of the acquisition of agreement morphology in spoken languages we have illustrated that the typological characteristics of the morphological processes a given language employs have an impact on the developmental pattern (i.e. proportions of default forms versus productive and appropriate use of inflected forms) and mastery of the agreement system. In BSL the morphological processes for marking agreement are polysynthetic and non-concatenative. While in BSL the morphological paradigm to mark agreement is rich, subject and object agreement are only applied to a sub-class of verbs based on syntactic and semantic grounds. In terms of the input we have described the particular differences between sign and spoken language addressed to children. The major effects of modality are on the child’s differential ability to ‘over-see’ grammatically rich adult/adult sign language and the

32 possibility that signing mothers may choose to omit obligatory agreement morphology in certain situations. While the first aspect has meant that signing deaf children do not acquire language differently to speaking hearing children in any general sense, the second issue may play a role in the pattern of acquisition of verb agreement morphology. The aim of the study reported here was to investigate the development of morphological verb agreement in BSL. It assesses the contribution of each of the linguistic phenomena and the environmental factors outlined above to the pattern of acquisition of BSL verb morphology. 4. Methodology

4.1 Subject The subject is a deaf boy referred to by the pseudonym ‘Mark’. He has been exposed to BSL from his parents and older siblings (all native signers) since infancy. He was filmed in naturalistic interaction in the home from 1;10 to 3;0. 4.2 Data collection and transcription Deaf and hearing investigators filmed Mark in 2-3 hour sessions at least once a month. The data are described in Table 1.

[Insert table 1 here] Subsequently the videos were transcribed and coded by trained Deaf and hearing investigators. For the purposes of the current analysis all Mark’s signed utterances that contained a verb were transcribed. Across the time period he was studied, we recorded 287 verb phrase tokens and 166 verb phrase types. Of these there were 75 agreement verb tokens, 128 plain verb tokens and 84 spatial verb tokens. All the following analyses involve the 75 agreement verb tokens.

33 Analyses and results 5.1 Introduction We calculated the total number of verb tokens (plain, agreement and spatial) as a percentage of total utterances of all types (noun phrases, questions prepositional phrase etc) for each session. We collapsed the data across sessions as follows: 1;10 – 1;11 - 3 sessions 2;1- 2;2 - 2 sessions 2;3, 2;3 – 2 sessions 2;4-2;5 - 2 sessions 2;6-2;8 – 3 sessions 2;9-2;9 - 2 sessions 2;11-3;0 - 2 sessions This is presented in Figure 4.

[Insert figure 4 here]

5.2 Emergence of verb agreement Next we counted the morphological inflections for verb agreement as tokens across the sessions. This is presented in Figure 5.

[Insert figure 5 here]

At the same time as the inflections were emerging there were many cases where the child omitted the inflection in a context where it was obligatory. The child used points to indicate arguments instead of the morphological inflection on the verb which was semantically and syntactically licensed. These omissions became increasingly rare. This is shown in figure 6.

34 [Insert figure 6]

The omissions are shown as a percentage of the total number of verbs. We also looked at the prevalence of the type of agreement relation (e.g. first to second or second to third) across the omissions. All the possible types of inflection appeared as omissions across the data (e.g. first to third, second to first, third to first, third to third etc) but there were many cases of omissions with third person arguments, either as subject or object.

The productivity of the inflections was also assessed and three degrees were defined: 1.

Emergence of an inflection: no evidence of the clear

analysis of a morpheme; 2.

Weak evidence of productivity of productivity: the same

verb appears with and without inflection; 3.

Strong evidence of productivity: more than one verb

appears with and without inflection. On the basis of the analysis, the pattern outlined in Table 2 emerges:

[Insert table 2 here]

5.3 Emergence and mastery of agreement in BSL At the start of two-sign combinations Mark treated nouns and verbs differently. Between the ages of 1;10 to 2;1, although two and three nouns were being combined, almost all verbs were produced in single sign utterances. From a total of 60 verb tokens recorded during this period, there was one combination of a verb with another sign: PUSH ME (1;10,19).

35 All the verbs first appeared in citation form i.e. there was no use of sign space for grammatical purposes. Descriptions of actions included verbs classified as intransitive / eventive in the adult language and so unable to inflect for verb agreement: FLY, CRY, SLEEP and JUMP; and also transitive/ eventive verbs which were produced without morphological markers (for example, BITE, CLOSE, EAT and THROW). All verbs were produced without an overt subject or object. At this early age the child used himself as the subject and agent in keeping with the BSL preference for using the signer to represent the subject location.

At 2;1 there was one attempt to provide information about how the subject and object were related in a transitive eventive context. When describing a picture of a boy biting a girl, the child signed BITE in the uninflected form, followed by a depiction of the bitten girl’s reaction (a shudder of the body and a startled facial expression). Through this combination of a sign with a whole-body gesture the child went some way in mapping out the concept of cause and effect through quasi argument structure. Thus the utterance used the ‘own body’ option for the subject but a non-linguistic attempt to indicate the object.

At 2;2 an adult signed to Mark 1BITE3 ‘(I) bite (it). Immediately after Mark signed BITE 1IX3 ‘Bite

me on it’. The verb BITE was uninflected but the index point moved between

himself and the object location. This resembled the auxiliary verb in SLN ‘act on’ described earlier. At 2;2 subject and object verb agreement emerged with a small number of verbs. At this age Mark began to introduce uninflected verbs into multi-sign combinations. After 2;2 there was an absence of inflections in the data until 2;5 when Mark signed DOG MAN DOG 3WASH3 ‘The dog/man washed him’. The verb was inflected towards the assumed subject location of the signer’s own body but because there was no index assignment for the arguments DOG or MAN it was difficult to interpret the intended meaning. In the adult language this construction would be understood as ‘the dog/man washes it/himself’ Up to this point subject and object agreement inflections were appearing with different grammatical person (1st to 3rd, 3rd

36 to 3rd etc). At 2;6 he used a first to second person inflection 1COMB-HAIR2 ‘I comb your hair’. From 2;9 onwards morphological agreement appeared on a more verbs especially those expressing visual perception, object transfer and causality e.g. BOY GIVES-FOOD, BOY PUSH-OUT,

SNAKE SEE,

3BITE3

The verb LOOK was used in more diverse constructions, including the sentence ‘both of them look at the kangaroo’ which is a third person plural to third person single inflection.5 e.g. LOOK-AROUND, LOOK-AT-ME (2;9,21), BOTH-LOOK IX KANGAROO (2;10)

Although verb agreement morphology was being used consistently at this age, the majority of new transitive / eventive verbs that entered the lexicon around this time continued to be produced as citation forms e.g. MAN KICK, DUCK BITE, ICE-CREAM POUR, MUMMY BREAK or in single sign utterances: BREAK, SCRATCH, CUT, SEE. Some intransitive/eventive verbs already in Mark’s lexicon appeared with subject NPs e.g. DRIVE MUMMY, BOY CRY, BIRD FALL; however new verbs entering his vocabulary were signed in one-sign utterances e.g. DRAW, VOMIT, WEE-WEE, SLEEP, BUMP-OWN-HEAD, CYCLE, WAKE-UP.

5.4 The input data The first issue to be raised about the input data relates to whether Mark could ‘oversee’ inflected forms in the signing of his mother when she signed to other adults. In general, there were hundreds of utterances with verb agreement morphology in our adult-adult 5

LOOK is unusual. It appears very frequently in the corpus, and can be interpreted as either an agreement

verb or a spatial verb, since the handshape ‘V’ in LOOK can be interpreted as a classifier representing an object with two long thin extensions (e.g. ‘two persons’ or ‘eye gaze’).

37 data. Whether the child was attending to all of these is very difficult to assess. Here we present a 15 minute sample of adult-adult conversation in the presence of Mark when he is 2;4, an age at which he has started to produce agreement inflections, although he omits them in 50% of obligatory contexts (see figure 6) and there is not evidence of productivity (see table 2). Mark is sitting on his mother’s lap while she signs to another deaf adult; this was a common place for him to sit. Her arms are around Mark’s body while she signs and so it would be difficult for Mark not to see his mother’s signing. In this short time span the mother used 8 morphological inflections with subject and object agreement (the referents were all recoverable from the discourse context). These included 2nd to 1st person (YOU-EXPLAIN-ME, YOU-TAKE-ME), 3rd person to 3rd person (HE-SAY-SHE, SHE-LOOK-SHE, HE-PUSH-SHE, SHE-LOOK-HE, SHE-GIVE-HE), and 3rd person to 1st person (HE-ASK-ME). In this sample, all transitive / eventive verbs carried verb agreement morphology, i.e. there were no omissions.

The second part of our examination of the input was the identification of the occurrence of inflections in the language addressed to Mark by his mother. The first thing to point out was that throughout our data as a whole, there were much less verbs in CDS than in adult-adult signing. We analysed 10-minute CDS samples from four recorded sessions. In all the samples sessions there were 15 agreement verbs overall used by the mother in interaction with the child. The percentage of verbs that appeared with omission is shown in table 3.

[Insert table 3 here]

As the figures in table three illustrate 27% (4/11) of the mother’s transitive / eventive verbs were signed to the child without inflections in obligatory contexts. These appeared as citation forms, with indexing towards the referent subject or object. Some verbs were

38 signed with and without agreement morphology in the same sampled session (e.g. ASK in 2;8). Although the proportion of omission in Mark’s mother signing is not negligible, Mark’s percentage of omission is much higher during that time as it amounts to 50%.

5.5 Summary of the results The results indicate that the percentage of verb tokens (plain, agreement and spatial) in the child’s production (figure 4) was relatively low up until age 2;4. Coupled with this, it was not until after 2;2 that Mark combined verbs with other signs. This scarcity of verbs and verb combinations in the early sessions is a major factor in the late onset of verb agreement morphology in the data. To underline this point we observed no verbs at all in the child’s language at 1;10;5.6 As figure 5 illustrates morphological agreement is used as early as 2;1-2;2 but because of the many co-occurring omissions at this stage (78%) and the lack evidence of productivity of the agreement inflections, these examples are probably un-analysed wholes. The agreement marker may be said to have emerged at this age without being productive. An interesting case at this age (2;2) illustrates the child’s reluctance to use agreement morphology. The adult signs ‘I bit him’ with subject and object agreement but the child copying what he had just seen marks indicates the identity, roles and number of participants through a moving index point which traces the line between the subject (himself) and the object. This error is revealing in that it is an unusual substitution by the child. Although in other sign languages (but not BSL) there exist auxiliary verbs (Bos, 1994; Rathmann & Mathur, 2002) that allow arguments to be expressed through this type of moving point, they are used with plain verbs rather than agreement verbs. If at this age the child is not distinguishing between agreement and non-agreement verbs systematically it could explain why he uses a point rather than morphological agreement on the verb as modelled in the adult’s utterance. 6

Of course this is only in the two hours of recorded data from 2 weeks of unrecorded

data. We did not ask the mother to keep a diary.

39

Between the ages of 2;3 and 2;6 the number of inflections decreased in Mark’s data as did the number of all types of verbs in general. From 2;6 to 2;9 we saw a re-emergence of the inflection but with a similar increase in omissions. It was at 2;9 that the child produced the most number of verb constructions made up of all verb types (plain, agreement and spatial) and agreement verbs with and without morphological markers. This contrasts with the period 2;3 – 2;6 where very few verb constructions meant fewer opportunities to try out new linguistic devices. During this period we also saw that the child was using a small number of verbs in a lot of different types of constructions. The verb LOOK became test model for attempting to express more complex morpho-syntactic constructions. It was not until 2;11 - 3;0 that the child began to produce consistently more than 80% of inflections in obligatory contexts. 6. Discussion 6.1 Introduction In this section the findings that emerge from this study are compared with those reported on spoken and signed languages. The next section discusses factors that may play a role in the acquisition of agreement morphology in BSL. Finally, the contribution of this study on agreement constrained to transitive and eventive verb is shown to bring an interesting contribution to current debates on the acquisition of aspect. It is important to bear in mind that the discussion below focuses on findings that emerge from the data collected on a single child. Given the individual variation reported in the literature and the relevance of this issue to the validity of accounts of morphosyntactic development (Bates et al, 1995, Shore, 1995), it is hoped that the hypotheses outlined below will be tested on other children acquiring BSL as an L1.

40 6.2 Comparison with the acquisition of spoken languages Compared to all spoken languages mentioned in section 1.1 and 1.3 (including Georgian), one of the findings that emerge from this study is a late onset of the production of verbs. In previous studies of morphologically rich (including languages that exhibit nonconcatenative morphology such as Hebrew) and pro-drop languages, subject agreement is acquired early at around 2 years (see section 1.1). This is considerably earlier than in the BSL data where we evaluated the productive stage of agreement marking at 2;9. In fact this age of mastery is more like non-pro-drop languages such as German (Behrens, 1993) and languages such as English with poor morphological verb paradigms (Phillips, 1995). In contrast to studies of the acquisition of Basque and Nairobi Swahili which report the non-simultaneous acquisition of subject and object agreement and of Romance languages in which subject clitics are acquired before object clitics, we found very little difference in the age of mastery between the child’s use of BSL agreement morphology to mark subject or object. We observed weak evidence for mastery of object agreement at 2;5 and subject agreement at 2;6. It is interesting to note that although subject-drop is less marked than object-drop in BSL (Rathmann & Mathur, 2002), object agreement appears to precede subject-agreement only by one month in our data. There are differences in age of acquisition of object agreement across spoken languages. In Basque object-verb agreement emerges around 3 years (Meisel & Ezeizabarrena, 1996) while in Nairobi Swahili, object verb agreement is produced at 1;8 (Deen, 2003). The optionality of object marking and the existence of non-overt markers seem to contribute to the protracted development reported in Basque.

It seems that productive use of subject and object agreement marking in both Georgian and BSL emerges around 3: in both languages the expression of agreement marking depends on syntactic and semantic factors. This issue is further discussed in section 6.4.4 below.

41

6.3 Comparison with the acquisition of other signed languages According to findings published in the literature and reviewed in sections 3.1, subject and object agreement marking in other sign languages, including ASL in which only a subset of verbs that are neither syntactically nor semantically constrained exhibits morphological agreement, seem to emerge around 2-2;6 and is not used in an anadutlike way until about 3;6. As far as the emergence and the mastery are concerned our findings are in line with those based on other sign languages. Some qualitative aspects of the data collected on mark provide an interesting contrast to the ASL data, namely a) the fact that no specific person marking seems to attract unadultlike use in BSL and b) the lack of affixation of agreement markers with verbs that are not agreement verbs. In previous studies of ASL, Meier (1981) reported that before children master morphological agreement they omit inflections of which a high number of omissions were reported for 2nd to 3rd person contexts. That is the verb moves between locations in front of the signer. We found no sole type of inflection was omitted over others in Mark’s data but we did see that many of the contexts that required the movement of the verb for third person agreement (either as subject or object) caused errors of omission. It may be that moving the verb between locations some distance away from the body is phonetically more difficult than other types of inflections. Fischer (1973) and Casey (2000) both report overgeneralization of agreement morphology to plain verbs in ASL between the ages of 2;7 and 2;11. We did not observe these types of overgeneralizations. This issue is further discussed in sections 6.4. and 6.5 below.

42 6.4 Accounts 6.4.1 Introduction The cross-linguistic approach adopted above reveals that with respect to the age at which productive use of agreement markers emerge, BSL is not only similar to languages that exhibit the same modality but also to a languages, Georgian, with it shares some typological features. In sections 1, a number of accounts that are attempting to explain cross-linguistic findings on the acquisition of agreement were presented. The factors that they consider include: the role of the input, children’s morphophonological processing and morphosyntactic features. We assess the merit of each of these in light of our data. Next we turn to the contribution that this study brings to current hypotheses on the acquisition of a currently hotly debated topic in the acquisition of spoken languages, i.e that of aspect. Input account Sign languages are very different to spoken languages when the child’s attention to the ambient language is considered (Harris, 2000). It is still an unresolved question in the general language acquisition literature how much do children need to see or hear to ensure fluent native acquisition (Lieven, 2002; Goldin-Meadow, 2003). Rather than attempting to find a direct correlation between certain structures in the input and the appearance of the same structures in the child’s output we have been concerned in this study with the general difference between amount and consistency of child directed signing and in specific verb agreement inflections. Despite being exposed to less child directed language (due the different constraints of visual attention and visual saliency) deaf children exposed to sign from infancy go through the same general series of stages at the same ages as children developing spoken languages (see Chamberlain, Morford & Mayberry, 2000, Morgan & Woll, 2002). This finding has provided some important data for the how much question.

43 In our analysis of the input we see that the ambient language contains a high proportion of verbs marked for subject and object agreement. Although adult-adult dialogue is morphologically rich it is not transparent. Subject and object referents are not always overtly expressed, in many cases the inflection refers anaphorically to a preceding discourse topic or directly topresent referents out of the immediate field of vision of the child. We do not know how much attention the child pays to this language especially faced with the choice between television or language, picture book or language or as the child gets older - bike riding or language. This choice is not an issue in hearing children’s access to spoken language. The mother’s signing to the child was concerned with the child’s successful comprehension rather than correct BSL grammar. Within the same conversation we observed her using and omitting obligatory morphological inflections with the same verb. This optionality – although the mother’s proportion of omission is half that of her son’scoupled with the typological issues described previously may contribute to the late onset of verb agreement in BSL. We have suggested that one motivation for splitting apart polymorphemic signs into a sequence of elements by the mother is her concern for visual saliency for the child. We are left with the problem of deciding which comes first however – does the child omit inflections because of performance limitations in perception or production or because they observe omissions in the input? In addition at a quantitative level, the proportion of the child’s and the mother’s production of agreement marked in CDS, at a given stage are very different. Although more research with more children and adults will directly looking at this question will be needed to assess the role of the optional agreement marking in CDS, our results suggest that it may be limited.

Morphophonological processing account The main findings from our data revolve around the late onset of verb agreement morphology. As is described in section two (p21) the morphological processes in BSL may be different to those in other polysynthetic languages such as Inuktitut and nonconcatenative languages such as Hebrew or Bantu where subject verb agreement is acquired from around two years. In sign languages morphemes appear to be sequentially

44 ordered but importantly non-adjacent. In sign languages the nature of verb agreement may involve even more overlap of simultaneous morphemes than in spoken languages with non-concatenative morphology (Zwisterlood, 2002, 2003), which would explain why agreement is acquired later in BSL than in Hebrew. Children acquiring sign languages tend to produce the uninflected forms of the verbs with pronominal points produced sequentially or by marking agreement morphologically but on the pronoun rather than the verb. This finding has been reported several times in studies of deaf children’s development of different sign languages (e.g. De Quadros, 1997; Meier, 2002; Pizzuto, 2002; Van den Bogaerde & Mills; 1996). It suggests that , as far as production is concerned, children find it easier to produce markers sequentially. This may be related to the issue discussed in the review on the acquisition of Basque and Georgian in which portmanteau morphs seem to be acquired alter than separate morphemes conveying different information. However the lack of sequentiality may not be the only factor that explains the late production of agreement in BSL: in Hebrew in which the morphological processes are non-concatenative and children have to able to extract roots and insert infixes, children seem to master the agreement system between 2 and 2;5. What may delay the acquisition of agreement in BSL is the added layer of complexity invoked by Zwisterlood (2002, 2003). Although, from a perceptual perspective, the movement of a verb across sign space from subject to object arguments however may be perceptually very salient, the identification of each morpheme may be complex. One of the limitations of this study is the focus on language production: it is difficult to capture what the child’s underlying representation of the grammar of BSL between 2 and 3. One difficulty in evaluating this explanation is that we know very little about how the sign-stream is segmented by young infants compared to the explosion of findings in the last 30 years on infant speech perception (e.g. Jusczyk, 1997) that have shed light on early discriminative capacities that may enable them to bootstrap into crucial aspects of the morphosyntax of their language (Morgan & Demuth, 1996, Jusczyk, 1997, Soderstrom et al, 2003).

45 One piece of evidence which suggests sign language agreement is harder to process is that children in production attempt to separate out these overlapped morphemes (i.e. substitute inflections for points) even past 3 years of age (Morgan, Herman & Woll 2002). In this view performance factors such as morphological processing partly explain the late onset. Comprehension data testing 2 year old children’s understanding of agreement making use for instance of the Visual Preferential Paradigm (Golinkoff et al, 1987), would also help assess whether they are able to map the relevant inflections and person marking.

Typological characteristic accounts: pro-drop, richness of verb morphological paradigm and sub-lexicalization features As mentioned in the literature review on the acquisition of spoken languages, pro-drop and rich verbal morphological paradigms typically trigger early productive use and mastery of the subject verb agreement system. Although BSL is pro-drop and exhibits a morphologically rich verbal paradigm, the emergence and mastery of subject agreement occur relatively late (comparable to non-pro drop and morphologically poor languages). With respect to object agreement, given the few languages investigated to date with respect to this phenomenon, although a few possible factors were suggested in section 1.3, there is no consensus with respect to these factors in the literature Two aspects of our finding suggest that the syntactic and semantic sublexical features tat interact with agreement marking in BSL play an important role in the acquisition of this aspect of the morphosyntax. First with respect to the first productive use of agreement markers, the spoken language the closest to BSL was Georgian in which , like in BSL, syntactic and semantic features of the verbs interact with the nature and expression of agreement marking. Secondly, the consideration of these sublexical features explain a qualititiave difference between the findings that emerge from ASL and from our study, namely:

46 in ASL, agreement verbs are an ‘arbitrary class’ in that it is not constrained by semantic and syntactic factors and around 3 years of age, children use agreement morphology with verbs that do not allow it; in BSL, when Mark reaches 3, his subject and agreement marking is productive and he omits less than 20% of agreement inflections in obligatory contexts and he does not overgeneralize agreement to verbs that are intransitive and/or statives. The literature on the acquisition of spoken languages has revealed that 3 is an age at which a) children’s use of argument structure alternation is productive and b) some aspectual distinctions are understood. The characteristics of BSL enable the child to make use of these syntactic and semantic cues in order to identify agreement verbs and restrict subject and object agreement to this verb class. In contrast, given that none of these cues is relevant to agreement marking in ASL, children overgeneralize agreement marking. Relevance of this study to accounts on the acquisition of aspect in spoken languages This section further discusses the acquisition of aspect and demonstrates that this study casts an interesting light on current accounts of the acquisition of aspect. Agreement marking in BSL interacts with both the transitivity of the verbs and their aspectual features. In ‘classic’ categorizations (Vendler, 1967, as described in Dowty, 1979), these include: [Insert Table 4 here] It is important to note that compared to the verbal conjugation in Georgian for instance in which four classes- state, activity, accomplishment and achievement- which depend on the sensitivity to three semantic features- punctuality, telicity and dynamics- are relevant to agreement marking, in BSL the semantic distinction stative versus eventive (in addition to the frame) enables the identification of the agreement verb class.

47 Accounts of the acquisition of aspect, like those of agreement marking, can be divided into two broad categories: those for which the child extracts patterns based on the input versus those that assume that children at the onset or at some point in their development represent relevant semantic features in order to distinguish aspectual classes. The prototype account proposed by Shirai & Andersen (1995) belongs to the former categories. The authors compare the aspectual features of the verbs used with the past and progressive inflections in the children’s and the mother’s speech and conclude that there is a correlation between them that support the view that children’s acquisition of aspect can be accounted for on the basis of a distributional learning account. There are many problems raised by the data analysis of the authors that deserve further attention but which are beyond the scope of this article. One aspect of our findings does not seem to be compatible with the prototype account which would predict a correlation between the morphological pattern assigned to (transitive) eventive verbs, in the case of BSL, in the child and the mother’s speech. The findings reveal that at 2;4, the mother produces much more agreement marking with (transitive) eventive verbs than Mark does. Alternative accounts of the acquisition of aspect include Bickerton (1981) hypothesis according to which the Stative/Eventive distinction is part of the bioprogram. Three sources of data support this hypothesis: the fact that in English children do not overgeralize –ing to stative verbs; Turkish acquisition data analysed by Slobin & Aksu (1980) (cited in Bickerton, 1981) reveal that 2 morphemes used in the adult language to distinguish between direct versus indirect experience are first used- as early as 1;9- to distinguish between dynamic and static events by children (161); In Creole languages present and past tense-marking and the stative/eventive dimension interact (160). This hypothesis would predict that the stative/eventive distinctions would enable children to draw the appropriate semantic distinction between agreement and non-agreement verbs as early as 2. Either this is not the case, or the transitive/intransitive dimension add

48 another factor that impedes the use of aspectual features by the child acquiring BSL as an L1. This study does not bring a precise contribution to accounts that attempt to assess the role of each semantic feature (van Hout, 1997, Olsen & Weinberg, 1999, van Hout, in press) punctuality, telicity and dynamics- since the distinction relevant to the classification of verb agreement in BSL does not distinguish between activity and processes. Instead these recent studies on the comprehension and production of aspectual features inform our study in that most of them reveal that while there is little consensus on the systematicity of associations between aspectual features of verbs and tense markers (see Weist, 2002 for an exhaustive review), comprehension experiments on a number of languages including Slavic languages in which aspectual distinctions are morphologically marked (see Wesit, 2002 and van Hout, in press for exhaustive reviews) show that from 3 years of age children start to be sensitive to aspectual features. Given that a) subject and verb agreement in BSL depends on sensitivity to transitivity and to the aspectual distinction between stative and eventive, b) that sensitivity to these cues has been shown to emerge around 3 in typologically different languages, and c) Mark is found to be adult-like in its use of subject and object verb inflections around 3, it seems reasonable to assume that one of the factors that may explain the acquisition of agreement in BSL lies in the sub-lexical categorization features that apply to agreement verbs. Experimental data, for instance testing chidren’s willingness to inflect transitive and intrasitive and eventive and stative nonce verbs, would enable us to test this hypothesis.

49

7. Conclusions This study has reported a late onset of verb agreement followed by productive use and mastery at an age comparable to that documented on the acquisition of other sign languages and one spoken language- Georgian. In light of this finding in addition to qualitative difference with the acquisition of ASL, a language that makes use of the same modality as BSL but exhibits distinct characteristics with respect to the lack of syntactic and semantic constraint that applies to agreement verbs, we have proposed three factors that may play a role in the acquisition of agreement in BSL and which are listed below, starting with the most to the least important: 1.

The sublexical categorization features of BSL make identifying the correct

subclass of agreement verbs problematic before 3 but at this age at which children acquiring different languages have been reported to be sensitive to transitive/intransitive alternations and aspectual distinctions, they produce subject and object agreement markers in about 80% of obligatory contexts and they constrain this morphological marking to the appropriate transitive and eventive class; 2.

The complexity of non-concatenative morphology makes the segmentation of

BSL morphology more difficult and 3.

The optionality of the agreement inflections in CDS in comparison to adult/adult

signing delays their mastery. Follow-up studies based on comprehension data and the analysis of production data on additional children acquiring BSL as an L1 will enable us to evaluate each of these proposals.

50 References Adridge, M. (1989) The acquisition of INFL. Bloomington: Indiana University Linguistics Club Publications. Allen, S.E.M. (1996). Aspects of argument structure acquisition in Inuktitut. Amsterdam: John Benjamins. Anderson, D.E. & Reilly, J.S. (1998). The puzzle of negation: How children move from the communicative to grammatical negation in ASL. Applied Psycholinguistics, 18, 411429. Barrière, I.; Perlman Lorch, M. & Le Normand, M.T. (1999) On the overgeneralization of the Intransitive/Transitive alternation in children’s speech: a cross-linguistic account with new evidence from French. The International Journal of Bilingualism, Vol. 3 (4: 351362. Barrière, I.; Lorch, M. & Le Normand, M.T. (2000) Argument Structure Alternation in French Children’s Speech. In M. Perkins & S. Howard (Eds.): New Directions in Language Development and Disorders. New York: Kluwer Academic/Plenum Publishers. pp139-148. Barrière, I; Lorch, M. & Le Normand, M.T. (2001) La cliticisation en SE dans le langage de l’enfant: surgénéralisation des affixations et des omissions. In C. Muller (Ed.): Clitiques et cliticisation. Paris: Honoré Champion.

Bates, E., Dale, P.S. & Thal, D. (1995) Individual differences and their implications for theories of language development. In P. Fletcher & B. MacWhinney (eds) (1995) The handbook of child language. Cambridge: Blackwell. Pp96-153.

51 Behrens, H. (1993) Temporal reference in German child’s language. PhD dissertation. University of Amsterdam. Bellugi, U., van Hoek, L., Lillo-Martin, D. & O’Grady, L. (1989). The acquisition of syntax and space in young deaf signers. In K. Mogford-Bevan & D. Bishop (eds), Language development in exceptional circumstances. Cambridge: L.E.A. Berman, R.A. (1982) Verb-pattern alternation: The interface of morphology, syntax and semantics in Hebrew child language. Journal of Child Language, 9, 169-191. Berman (1993b) Marking of verb transitivity by Hebrew-speaking children. Journal of Child Language, 20, 641-669. Berman, R.A. (1994) Developmental perspectives on transitivity: a confluence of cues. In Y. Levy (Ed.)Other children, other languages: Issues on the theory of language acquisition (pp. 189-242). Hillsdale, NJ: Erlbaum Bickerton, D. (1981) Roots of language. Ann harbor: Karoma. Bogaerde, B. & Baker, A. (1996). Verbs in the language production of one deaf and one hearing child of deaf parents. Poster presented at the 5th International Conference on Theoretical Issues on Sign Language Research, September 1996, Montréal, Canada. Manuscript

Bos, H. (1994). An auxiliary verb in Sign Language of the Netherlands. In: Ahlgren, I., Bergman, B., & Brennan, M (eds): Perspectives on sign language structure, Vol. 1 Durham: ISLA, p37-53. Bowerman, M. (1974) Learning the Structure of Causative verbs: a study in the Relationship of Cognitive, Semantic and Syntactic Development. Papers and Reports on Child Language Development, Vol. 8, 142-178.

52

Bowerman, M. (1982a) Evaluating Competing Linguistic Models with Language Acquisition Data: Implications of Developmental Errors with Causative Errors. Quaderni di Semantica, Vol. III, no. 1. Bowerman, M. (1982b) Starting to talk worse: Clues to language acquisition from children’s late speech errors. In S. Strauss (Ed.) U-shaped behavioral growth. New York: Academic press. Brentari, D. (2002). Modality differences in sign language phonology and morphophonemics. In: Meier, R., Cormier, K. & Quinto-Pozos, D. (eds): Modality and structure in signed and spoken language. Cambridge: Cambridge Univ. Pr. pp. 35-64 Brown, R. (1973) A first language: the early stages. Cambridge: Harvard University Press. Budwig, N. (1990) Do young children linguistically encode the notion of agent? PRCLD, 29: 133-147. Casey, S. (2000). Relationships between gestures and signed languages: Indicating participants in actions. Paper presented at the 7th International Conference on Theoretical Issues in Sign Language Research, Amsterdam, The Netherlands.

Chamberlain, C., Morford, J.P. & Mayberry, R. (2000). Language acquisition by eye. Mahwah, NJ : Erlbaum.

Clahsen, H. & Penke, M. (1992) The acquisition of agreement morphology and its syntactic consequences; new evidence on German child language from the Simonecorpus. In J. Meisel (ed.) The acquisition of verb placement: functional categories and V2 phenomena in language acquisition. Dordrecht: Kluwer. Pp 181-223.

53 Davidson, L. and G. Legendre (in press) “Defaults and Competition in the Acquisition of Functional Categories in Catalan and French,” in Rafael Nuñez-Cedeño, Luis López & Richard Cameron, (eds.), Selected Papers from the 2001 Linguistic Symposium on Romance Languages (LSRL XXVI). Amsterdam: John Benjamins.

Deen, K.U. (2002) The acquisition of Nairobi Swahili: the morphosyntax of inflectional prefixes and subjects. PhD dissertation. UCLA Deen, K.U. (2003) Subject agreement, object agreement and specificity in Nairobi Swahili. Paper presented at GALA 2003, Utrecht. Demuth, K. (2000) Prosodic constraints on morphological development. In J. Weissenborn & B. Höhle (2000) Approaches to bootstrapping, Vol. 2: phonological, lexical, syntactic and neurophysiological aspects of early language acquisition. Amsterdam: John Benjamins. pp.3-21. De Quadros, R. (1997). Aspectos da sintaxe e da aquisicao da Lingua Brasileira de Sinais. [Syntactic Aspects in the Acquisition of Brazilian Sign Language]. In: Letras-deHoje 32: 4 pp. 125-146.

Dowty, D. (1979( Word meaning and Montague Grammar: the semantics of verbs and times in Generative Semantics and Montague’s PTQ. Dordrecht: reidel. Emmorey K (2002). Language, Cognition and The Brain: Insights from Sign Language Research. Mahwah NJ: Lawrence Erlbaum Associates. Emmorey K (ed.) (2003) Perspectives on classifier constructions in sign languages. Mahwah NJ: Lawrence Erlbaum Associates. Erting, C.J., Prezioso, C., & O'Grady Hynes, M. (1990) Mother signs in baby talk. In Edmondson, W.H. & Karlsson, F. (eds): SLR´87: Papers from the Fourth International

54 Symposium on Sign Language Research. Lappeenranta, Finland July 15-19, 1987. Hamburg: Signum Press, 190-199. Ferdinand, A. (1996) The development of functional categories: the acquisition of the subject in French. Holland Intitute of Generative Linguistics. Dordrecht: ICG printing. Figueira, R.A. (1984) On the development of the expression of causativity: a syntactic hypothesis. Journal of Child Language,11: 109-127. Fischer, S. (1973). The deaf child's acquisition of verb inflection in American Sign Language. Paper presented to the Annual Meeting of the Linguistic Society of America. San Diego, Cal 1973 Manuscript (unpubl.) Gallaway, C. Woll, B. (1994). Interaction and childhood deafness. In: Gallaway, C. & Richards, B. (eds). Input and interaction in language acquisition. Cambridge, New York : Cambridge Univ. Pr. pp. 197-218 Gathercole, V.C.M., Sebastián, E. & Soto, P.(1999) The early acquisition of Spanish verbal morphology: across the boeard or piecemeal knowledge? International Journal of Bilingualism, 3: 133-182. Gathercole, V.C.M., Sebastián, E. & Soto, P. (2000) Lexically specified patterns in early verbal morphology in Spanish. In M. Perkins & S. Howard (eds.) New directions in Language development and disorder. New York: Kluwer Academic/Plenum publishers.pp149-168. Goldin-Meadow, S. (2003). The resilience of language : What gesture creation in deaf children can tell us about how all children learn language. Hove : Psychology Press 2003. Golinkoff, Hirsh-Pasek, Cauley & Gordon (1987) The eyes have it : lexical and syntactic comprehension in a new paradigm. Journal of Child Language, 14: 23-45.

55

Guasti, M. (1992) Verb syntax in Italian child grammar. Geneva Generative Papers 1-2: 115-122. Grinstead, J. (1993) Tense, Agree,emt, and Nominative case in Child Catalam and Spanish. and Catalan. M.A.Thesis, UCLA. Grinstead, J. (1998) Subjects, sentential negations and imperatives in child Spanish and Catalan. PhD Dissertation, UCLA. Hamman, C., Rizzi, L. and Frauenfelder, U. H. (1995) On the acquisition of the pronominal system in French. In C. Jacubowicz (Ed.) Recherches linguistiques de Vincennes.Special Issue: Grammaire universelle et acquisition du langage, Vol. 24, 83102. Haegeman, L. (1995) Root infinitives, tense and truncated structure. Language Acquisition, 4 (3): 205-255. Harris, M. (2000). Social interaction and early language development in deaf children. In: Deafness and Education International 2: 1 pp. 1-11 Harris, M. (2001). It's all a matter of timing: sign visibility and sign reference in deaf and hearing mothers of 18-month-old children . In: Journal of Deaf Studies and Deaf Education 6: 3 pp. 177-185.

Hoffmeister 1978 Word order acquisition in ASL. Paper presented at the Third Annual Boston University Conference on Language Development. Boston 1978 Manuscript (unpubl.)

56 Hornstein, N. & Lightfoot, D. (eds.) (1994) Verb movement. Cambridge: Cambridge University Press. van Hout (1999) Event semantics in the lexicon-syntax interface: verb frame alternations in Dutch and their acquisition. Utrecht: Institute of Linguistics. van Hout (in press) Imperfect imperfectives: on the acquisition of aspect in Polish. Hyams, N. (1992) The genesis of clausal structure. In J. Meisel (ed.) The acquisition of verb placement: functional categories and V2 phenomena in Language Acquisition. Amsterdam: Kluwer Imedadze, N. & Tuite, K. (1992) The acquisition of Georgian. In D.I. Slobin (ed.) The cross-linguistic study of language acquisition. Vol.3. Hillsdale: Lawrence Erlbaum. Pp 39-109. Jakubowicz, C. (1991) L’acquisition des anaphores et des pronoms lexicaux en français. In J. Guéron and J. Y. Pollock (Eds.): Grammaire Générative et Syntaxe Comparée (229252). Paris: Édition du C.N.R.S. Janis, W. (1995). A crosslinguistic perspective on ASL verb agreement. In: Emmorey, K. & Reilly, J. (eds). Language, gesture, and space. Hillsdale, N.J. pp. 195-223. Kaiser, G.A. (1990) More about INFL-ection and Agreement. In J. Meisel (Ed.) Two First Languages: Early Grammatical Development in Bilingual Children (47-72). Dordrecht: Foris publications. Kantor, R. (1982). Communicative interactions: Mother modification and child acquisition of American Sign Language. Sign Language Studies 11: 36 pp. 233-282

57 Kegl, J. (1990). Predicate argument structure and verb-class organization in the ASL lexicon. In: Lucas, C. (ed). Sign language research. Theoretical issues. pp. 149-175 Kyle, J. & Ackerman, J. (1990). Signing for infants: Deaf mothers using BSL in the early stages of development. In: Edmondson W. & Karlsson, F. (eds). International Studies on Sign Language and Communication of the Deaf 10. Hamburg: Signum pp. 200-211 Legendre, G., Hagstrom, P., Vainikka, A. & Todorova, M. (2002) Partial Constraint Ordering in Child French Syntax. Language Acquisition, 10 (3): 189-227. Lieven, E. V. M. (2002). Afterword: A view from research on spoken language development. In: Morgan, G. & Woll, B. (eds). Directions in sign language acquisition. (Trends in Language Acquisition Research; 2) Amsterdam, Philadelphia : Benjamins. pp. 277-290 Lieven, E.V.M., Pine, J. & Baldwin, G. (1997) Lexically-based learning and early grammatical development. Journal of child language, 24, 187-219. Lillo-Martin, D. (1986). Two kinds of null arguments in American Sign Language. Natural Language and Linguistic Theory 4 pp. 415-444

Lillo-Martin, D. (1997). The modular effects of sign language acquisition. In: Marschark, Marc et al (eds) Relations of language and thought: the view from sign language and deaf children. Oxford. Oxford University Press. pp. 62-109

Lillo-Martin, D. (2002). Where are all the modality effects? In: Meier, R., Cormier, K. & Quinto-Pozos, D. Modality and structure in signed and spoken language. Cambridge: Cambridge Univ. Pr. pp. 241-262

58 Loew, R. (1983). Roles and reference in American Sign Language: a developmental perspective. Unpublished doctoral dissertation, University of Minnesota.

Lord, C. (1979) ‘Dont’you fall me down’: Children’s overgeneralizations regarding cause and transitivity. Papers and Reports on Child Language Development, 17, 81 89. MacWhinney, B. (1995) The CHILDES project: tools for analyzing talk. 2nd edition. Hillsdale: Lawrence Erlbaum. Masataka, N. (2000). The role of modality and input in the earliest stages of language acquisition: Studies of Japanese Sign Language. In: Chamberlain, C., Morford, J.P.& Mayberry, R. (eds). Language acquisition by eye. Mahwah, NJ : Erlbaum. pp. 3-24

Meier, R. P. (1981). Icons and morphemes: models of the acquisition of verb agreement in ASL. Papers and Reports on Child Language Development 20, 92-99.

Meier, R. P. (1991). Language acquisition by deaf children. American Scientist, 79, 6070 Meier, R.P.(2002). The acquisition of verb agreement: Pointing out arguments for the linguistic status of agreement in signed languages. In: Morgan, G. &Woll, B. (eds). Directions in sign language acquisition. (Trends in Language Acquisition Research; 2) Amsterdam, Philadelphia : Benjamins pp. 115-141

Meier RP, Cormier KA, & Quinto-Pozo DG (Eds.) (2002) Modality and Structure in Signed and Spoken Languages. Cambridge: Cambridge University Press.

Meisel, J. (ed.) (1992) The acquisition of verb placement: functional categories and V2 phenomena in language acquisition. Dordrecht: Kluwer.

59

Meisel, J., and M. Ezeizabarrena (1996). Subject-Verb and Object-Verb Agreement in Early Basque in H. Clahsen (ed.) Language Acquisition: Empirical Findings, Theoretical Considerations, Cross-linguistic Comparisons. Amsterdam: John Benjamins. 201–239. Morgan, G. (1998). The development of discourse cohesion in British Sign Language. Unpublished doctoral dissertation, University of Bristol. Morgan, G. (2000). Discourse Cohesion in Sign and Speech. International Journal of Bilingualism, 4, 279-300.

Morgan, G., Herman, R. & Woll, B. (2002). The development of complex verb constructions in BSL. Journal of Child Language, 29, 23 -66

Morgan, G. & Woll, B. (2002). Directions in sign language acquisition. (Trends in Language Acquisition Research; 2) Amsterdam, Philadelphia : Benjamins.

Morgan, G., Barriere, I. & Woll, B. (2003). First verbs in British Sign Language development. City University Working Papers in Language and Communication Science, vol. 2 pp 57-66

Morgan, J.L. & Demuth, K. (1996) Signal to syntax: bootstrapping from speech to grammar in early acquisition. Mahwah: Lawrence Erlbaum.

Naigles, L.R. (2002) Form is easy, meaning is hard: resolving a paradox in early child language. Cognition, 86, 157-199.

60 Naigles, L.R. (2003) Paradox lost? No paradox found! Reply to Tomasello & Akhtar (2003). Cognition, 88, 325-329. Newport, E. L., & Meier, R. P. (1985). The acquisition of American Sign Language. In D. I. Slobin (ed.), The crosslinguistic study of language acquisition. Volume 1: The data. Hillsdale, NJ: Erlbaum, 881-938. Olsen, M.B. & Weinberg, A. (1999) Innateness and the acquisition of grammatical aspect via lexical aspect. In A. Greenhill, H. Littlefield & C. Tano (eds) Proceedings of the 23rd Annual Boston University Conference on Language Development. Somerville: Cascadilla Press. pp529-540. Phillips, C. (1995) Syntax at age two: cross-linguistic differences. MIT Working Papers in Linguistics, 26, pp37-93.

Pierce, A. (1990) On the emergence of syntax: a cross-linguistic study. PhD Thesis, MIT.

Pierce, A. (1992) Language acquisition and syntactic theory: a comparative analysis of French and English grammars. Dordrecht: Kluwer.

Pinker, S. (1989) Learnability and Cognition: the Acquisition of Argument Structure. Cambridge, Massachussetts: the M.I.T Press. Pizzuto, E. (2002). The development of Italian Sign Language (LIS) in deaf preschoolers. In: Morgan, G. &Woll, B. (eds). Directions in sign language acquisition. (Trends in Language Acquisition Research; 2) Amsterdam, Philadelphia : Benjamins pp. 77-114

61 Platzack, C. (1990) A grammar without functional categories: a syntactic study of early Swedish child language. Working Papers in Scandinavian Syntax, 45, 13-34. Lund: University of Lund. Poeppel, D. & Wexler, K. (1993) The Full competence hypothesis of Caluse Structure in Early German. Language, 69: 1-33. Pye, C. (1994) A cross-linguistic approach to the causative alternation. In Y. Levy (Ed.) Other children, other languages: Issues in the theory of language acquisition (pp.243264). Hillsdale, NJ: Erlbaum. Radford, A. (1990) Syntactic theory and the acquisition of English syntax. Oxford: Blackwell. Rathmann, C. & Mathur, G. (2002). Is verb agreement the same cross-modally? In: Meier, R.P., Cormier, K. & Quinto-Pozos, D. (eds). Modality and structure in signed and spoken language. Cambridge: Cambridge Univ. Pr. pp. 370-404

Reilly, J.S. & Bellugi, U. (1996). Competition on the face: affect and language in ASL motherese. Journal of Child Language 23, 219-236 Rhee, J. & Wexler, K. (1995) Optional infinitives in Hebrew. MIT Working Papers in Linguistics 26: 383-402. Rizzi, L. (1993/1994) Some notes in linguistic theory and langauge development: the case of root infinitives. Language Acquisition, 3, 371-393. Rondal, J.A. (1985) Adult-child interaction and the process of language understanding. New York: Prager.

62 Shirai, Y. & Andersen, R.W. (1995) The acquisition of tense-aspect morphology: a prototype account. Shore, C.M. (1995) Individual differences in language development. London: Sage. Sigurjónsdóttor, S. (1992) Binding in Icelandic: Evidence from Language Acquisition. PhD dissertation, UCLA. Soderstrom, M. (2002)The acquisition of inflection morphology in early pereceptual knowledge of syntax. PhD Dissertation, Johns Hopkins University. Soderstrom, M., Seidl, A. Kmeler Nelson, D.G. & Juszyck, P. (2003) The prosodic boostrapping of Phrases: evidence from prelinguistic infants. Journal of Memory and Language, 49: 249-267. Soderstrom, M., Wexler, K. & Jusczyk, P. (2003) English-learning toddlers’sensitivity to agreement morphology in receptive grammar. In B. Skarabela, S. Fish & A.H.J. Do (eds.) Proceedings of the 26th Annual Boston University Conference on language development, vol. 2. Somerville, MA: Cascadilla Press: 643-652. Spencer, A. (1991) Morphological theory. Oxford: Blackwell. Suppes, P., Smith, R., Leveillé, M. (1973) The French syntax of a child’s noun phrases. Archives de Psychologie, 42: 207-269.

Sutton-Spence,. R. & Woll, B. (1999). The linguistics of British Sign Language: an introduction. Cambridge: Cambridge University Press.

Tomasello, M. (1992) First verbs: a case study of early grammatical development. Cambridge: Cambridge University Press.

63 Trask, R.L. (1993) A dictionary of grammatical terms in linguistics. London: Routledge. Vainikka, A. (1993/1994) Case in the development of English Syntax. Language acquisition, 3, 257-324. Vainikka, A., G. Legendre, and M. Todorova. 1999. PLU Stages: An Independent Measure of Early Syntactic Development. Technical Report: JHU-CogSci-99-10.

Varlokosta, S., Vainikka, A. & Rohrbacher, B. (1998) Functional projections, markedness and root infinitives in early child Greek. Linguistic review, 15, 187-207. Vendler, Z. (1967) Linguistics in Philosophy. Pithaca: Cornell University Presss. Wagner, K. (1985) How much do children say in a day? Journal of Child Language, 12: 475-487. Weissenborn, J. & Höhle, B. (2000) Approaches to bootstrapping, Vol. 2: phonological, lexical, syntactic and neurophysiological aspects of early language acquisition. Amsterdam: John Benjamins. Weist, R.M. (2002) The first language acquisition of tense and aspect: a review. In R. Salaberry & Y. Shirail (eds.) The L2 acquisition of Tense-Apsect morphology. Amsterdam: Johns Benjamins. pp21-78. Wexler, K. (1994) Optional infinitives, head movement, and economy of derivation. In N. Hornstein & D. Lightfoot (eds.) Verb movement. Cambridge: Cambridge University Press.pp 305-350. Wexler, K. (1998) Very early parameter setting and the unique checking constraint: a new explanation of the optional infinitive stage. Lingua, 106, 23-79.

64 Wijnen, F. (1995) Incremental acquisition of phrase structure: a longitudinal analysis of verb placement in Dutch Child Language. In J. Beckman (ed.) Proceedings of NELS, Vol. 25, GLSA, University of Massachussetts, Amherst. Zwitserlood, Inge (2002) The complex structure of ‘simple’ signs in NGT. In M. van Koppen, E. Thrift, E.J. van der Torre & M. Zimmermann (eds.) Proceedings of Console IX.1-15. Zwitserlood, Inge (2003) Classifying hand configurations in Nederlandse Gebarentaal (Sign Language of the Netherlands). Utrecht: Utrecht Institute of Linguistics.

65 Figures and tables Figure 1

kASKj

Figure 2

IXa

LIKE

Fig. 3 ‘he likes her’

IXb

66 Figure 3 The sentence ‘you ask him’ (1) with and (2) without morphological verb agreement (1)

YOU-ASK-HE

(2)

HE YOU

ASK

67

Figure 4 Verb tokens as a percentage of all utterances across sessions

100 % of total utterances

90 80 70 60

other

50

verbs

40 30 20 10 0 1;10-1;11

2;1-2;2

2;3

2;4-2;5

2;6-2;8

2;9

2;11-3;0

age (years; months)

number of tokens with agreement

Figure 5 Emergence of agreement 14 12 10 8 6 4 2 0 1;10-1;11

2;1-2;2

2;3

2;4-2;5 age (collapsed)

2;6-2;8

2;9

2;11-3;0

68

% ommission

Figure 6 Percentage of obligatory verb agreement omitted 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1;10 - 1;11

2;1 - 2;2

2;3

2;4 - 2;5 age (years;months)

2;6 - 2;8

2;9

2;11 - 3;0

69

Table 1 Data collection

Recording

Length

Age of child

session

(hours)

(Year; month, day)

1

2

1;10,5

2

3

1;10,19

3

2

1;11,8

4

2

2;1,18

5

3

2;2,24

6

3

2;3,2

7

3

2;3,16

8

2

2;4,13

9

3

2;5,13

10

2

2;6,23

11

3

2;8,13

12

2

2;9,19

13

2

2;10,9

14

2

2;11,0

15

3

3;0,1

70

Table 2 The productivity of verb agreement morphology

Markers

Emergence

Weak evidence of

Strong evidence of

productivity

productivity

Subject agreement

2;2

2;6

2;11

Object agreement

2;2

2;5

2;11

71 Table 3 Number of inflected/uninflected verbs and type of verb agreement morphology in the child directed signing.

Age of child

Mother

(yr; month) No of

V + agreement

V + omission

Agreement type

obligatory

(N =

contexts for V

occurrences)

+ agreement morphology 1;10

4

4

2GIVE3 (N

= 3),

1SEE3

2;4

1

1

2;8

2

1

1SEE3

1

1ASK3,

YOU ASK HIM 3;0

8

5

3

2GIVE3 (N

= 3),

3GIVE2, 3SCRATCH3,

YOU LOOK HIM (N = 3)

Totals %

15

11

4

73%

27%

72 Table 4: Aspectual distinctions (adapted from Shirai & Andersen, 1995) STATES

EVENTIVE PROCESSES

STATE

ACTIVITY

ACCOMPLISHEMENT ACHIEVEMENT

PUNCTUAL -

-

-

+

TELIC

-

-

+

+

DYNAMIC

-

+

+

+