first language dialogical factors in toddlers’ © the ......(1) young children’s use of pronouns...

28
FIRST LANGUAGE Article Corresponding author: Anne Salazar Orvig, Institut de Linguistique et Phonétique Générales et Appliquées, Université Sorbonne Nouvelle – Paris 3, 19 rue des Bernardins, 75005 Paris, France. Email: [email protected] First Language 30(3-4) 375–402 © The Author(s) 2010 Reprints and permission: sagepub. co.uk/journalsPermissions.nav DOI: 10.1177/0142723710379957 http://fla.sagepub.com Dialogical factors in toddlers’ use of clitic pronouns Anne Salazar Orvig Université Sorbonne Nouvelle – Sorbonne Paris Cité Haydée Marcos CNRS, France Aliyah Morgenstern Université Sorbonne Nouvelle – Sorbonne Paris Cité Rouba Hassan Université de Lille 3, France Jocelyne Leber-Marin and Jacques Parès EA 1483 - RFC Abstract Young (1;9–2;4) children’s use of third person clitic subject pronouns in natural dialogues was examined in both longitudinal and cross-sectional data. Considering that young children mainly use pronouns in the context of referential continuity, this study aims at identifying some of the factors that affect this use. Two possible dialogical factors are examined: (1) the use of clitic pronouns can be interpreted as a reproduction of the adult’s discourse, either by taking up whole utterances containing a pronoun or by taking up only the clitic pronouns without reproducing the adult’s utterance. (2) The use of pronouns could be driven by pragmatic-discursive factors. In order to assess this hypothesis the use of clitic pronouns was observed in the context of dialogical continuity. Three kinds of links were considered: children repeat or reformulate the adult’s utterances, add a new predication on the same topic, or establish a contrast. The results suggest that the reproduction of the adult’s utterance does not significantly

Upload: others

Post on 29-Jan-2021

1 views

Category:

Documents


0 download

TRANSCRIPT

  • FIRSTLANGUAGEArticle

    Corresponding author:Anne Salazar Orvig, Institut de Linguistique et Phonétique Générales et Appliquées, Université Sorbonne Nouvelle – Paris 3, 19 rue des Bernardins, 75005 Paris, France.Email: [email protected]

    First Language30(3-4) 375–402

    © The Author(s) 2010Reprints and permission: sagepub.

    co.uk/journalsPermissions.navDOI: 10.1177/0142723710379957

    http://fla.sagepub.com

    Dialogical factors in toddlers’ use of clitic pronouns

    Anne Salazar OrvigUniversité Sorbonne Nouvelle – Sorbonne Paris Cité

    Haydée MarcosCNRS, France

    Aliyah Morgenstern Université Sorbonne Nouvelle – Sorbonne Paris Cité

    Rouba Hassan Université de Lille 3, France

    Jocelyne Leber-Marin and Jacques ParèsEA 1483 - RFC

    Abstract

    Young (1;9–2;4) children’s use of third person clitic subject pronouns in natural dialogues was examined in both longitudinal and cross-sectional data. Considering that young children mainly use pronouns in the context of referential continuity, this study aims at identifying some of the factors that affect this use. Two possible dialogical factors are examined: (1) the use of clitic pronouns can be interpreted as a reproduction of the adult’s discourse, either by taking up whole utterances containing a pronoun or by taking up only the clitic pronouns without reproducing the adult’s utterance. (2) The use of pronouns could be driven by pragmatic-discursive factors. In order to assess this hypothesis the use of clitic pronouns was observed in the context of dialogical continuity. Three kinds of links were considered: children repeat or reformulate the adult’s utterances, add a new predication on the same topic, or establish a contrast. The results suggest that the reproduction of the adult’s utterance does not significantly

  • 376 First Language 30(3-4)

    influence children’s use of pronouns, whereas pragmatic-discursive factors are found to affect their choice of referential expressions.

    Keywords

    anaphora, clitic pronouns, dialogue, pragmatic-discursive factors, toddlers’ language, topic maintenance

    The study of children’s early uses of third person clitic pronouns is a domain in which the morpho-syntactic and pragmatic-discursive levels are closely intertwined. Various studies on the use of referential expressions highlight the sensitivity of young children to pragmatic features such as referent accessibility and its mention in previous discourse, showing that the pragmatic-discursive level is constitutive of emergent grammar.

    The purpose of the present research is to examine two competing factors that can be proposed to explain the use of third person clitic pronouns by French-speaking children in a dialogical context. Two main questions are addressed: (1) is the adult’s discourse the model for the child’s utterance and/or (2) does the use of pronouns respond to a pragmatic-discursive choice by the child?

    The acquisition of clitic pronounsFrench presents a double series of personal pronouns: clitic pronouns and tonic pro-nouns. Clitic pronouns, which are characterized by their prefixal and unstressed position, are used as subjects (je, tu, il, elle, on, nous, vous, ils, elles), or direct and indirect objects (me, te, le, lui, nous, vous, les, leur). Tonic pronouns (moi, toi, lui, elle, nous, vous, eux, elles) are characterized by an autonomous and stressed position. They can reduplicate the subject pronoun (lui, il court, ‘as for him, he runs’) and are used for all other functions. Since it presents almost no personal verbal inflexion, spoken French is a non null subject language. In spoken French, the subject position is preferentially filled by a clitic pro-noun (Jeanjean, 1980) which can coexist with a noun (Blanche-Benveniste, Bilger, Rouget, & Van den Eyden, 1990; François, 1981). This characteristic leads some authors to consider clitic pronouns as flexional morphemes (Jakubowicz & Rigaut, 1997; but see De Cat, 2005 for a discussion).

    Previous research on French shows that clitic pronouns are acquired quite early. Parisse and Le Normand (2000), for example, note that at around 2 years old, pronouns already represent 8% of the linguistic units used by children (the average is about 18% for adults). The sequences pronoun + verb are among the most frequent sequences of syntactic categories in Parisse and Le Normand’s data. In a more specific study on the acquisition of nominative clitics, Jakubowicz and Rigaut (1997) note that, in children aged from 2;0 to 2;7, a majority of utterances include a nominative clitic with a prefer-ence for third person pronouns. However, they identify an important rate of omissions: younger children’s use of clitic pronouns seems to be optional. Authors also stress (Hamann, Rizzi, & Frauenfelder, 1996; Jakubowicz & Rigaut, 2000) the asymmetry between the earlier acquisition of subject clitics and the later acquisition of object clitics.

  • Salazar et al. 377

    Pronouns in discourse and dialogue

    Studies of the acquisition of referential expressions such as determiners or pronouns have concentrated predominantly on their use in monologue (Bennett-Kastor, 1983; Hickmann, 2002; Karmiloff-Smith, 1985; Peterson & Dodsworth, 1991), emphasizing both their precocious acquisition and the errors children make, especially in using pro-nouns to introduce new referents. In this perspective, for young children, reference might be a deictic bottom-up process that does not suppose a discursive construction or repre-sentation independent of the situational context.

    But long before children can master monological discursive genres, they experience adults’ use of pronouns and determiners in various dialogic contexts. In particular, involve-ment in dialogue supposes the sharing of an intersubjective representation of discourse (Salazar Orvig, 2002), which is constitutive of the anaphoric relation (Cornish, 1999; Givón, 1995). The possibility for children to share such an intersubjective representation has been prepared by their participation in joint attention episodes (Tomasello, 1999). Attentional continuity – which according to Bruner (1978) is at the root of the topic-comment structure – is constructed through joint action and non-verbal communication. On the other hand, studies of children’s very first utterances, during the one word stage, show that the first verbal exchanges correspond to the ‘beginning of shared linguistic meaning’ (Veneziano, 2000, p. 239, our translation) and that children can chain two or more turns within the same communicative intent and, most often, in continuity with the adult’s discourse (Scollon, 1979). More generally, dialogue can be considered as the framework within which children experience, acquire and use linguistic units. The first values of referential expressions might therefore be constructed in this type of context.

    Various studies, based on experimental or natural data, have explored the use of referential expressions in dialogue. Experimental studies (Campbell, Brooks, & Tomasello, 2000; Wittek & Tomasello, 2005) on children between 2;6 and 3;6 show that the type of question asked by the interlocutor determines which referring expression is used by the child: when the question mentions the discourse object (What did X do?), English-speaking children from both age groups tend to respond with a null reference, whereas nouns and pronouns tend to be used more frequently when preceded by open questions (What happened?). In German, children mostly use a pronoun or a null form when the question specifically bears on the referent (Where is the broom?) and more general questions (What do we need?) elicit nominal responses.

    Studies on non Indo-European languages such as Inuktitut (Allen, 2000; Allen & Schöder, 2003) and Korean (Clancy, 2003) show the impact of the discursive and prag-matic dimensions on children’s use of referential expressions. In Inuktitut, in which arguments are either made explicit by using nouns or are omitted, children ‘produce overt arguments significantly more often if the argument they wish to represent is informative than if it is not’ (Allen, 2000, p. 511). Moreover, Clancy considers that ‘the clusters of semantic, discourse-pragmatic and formal properties of referents organized into A, S and O roles are available to young children as a source of hypotheses about the functional foundations of the morphosyntactically marked categories in the language being acquired’ (Clancy, 2003, p. 104). The results have been confirmed for null subject Indo-European languages such as Italian (Serratice, 2005) and for non null subject

  • 378 First Language 30(3-4)

    languages such as English (Hughes & Allen, 2006). In this last case, whereas null arguments appear in the context of great accessibility, it is not the case for overt arguments (which include pronouns) for which the distribution is more variable. However, factors such as newness and absence of the referent induce the coding of overt arguments. These studies confirm the fact that the acquisition of grammar is intertwined with pragmatic and discursive aspects of language.

    Studies by Salazar Orvig et al. (2004, 2006, 2010) lead to similar conclusions. The authors analyse spontaneous dialogues with children between 1;9 and 3;0 in various situ-ations. Their results show that the children are sensitive to the discursive and attentional context of verbal exchanges : in the majority of cases, predications about an implicit referent concern a referent that has is fact been previously mentioned in the dialogue. Children also use third person clitic pronouns in the context of shared attention and as second or subsequent mentions in the dialogue in the context of discursive continuity. An experimental study also supports this interpretation: 2-year-old children ‘respond to prior mention as an indicator of referent accessibility before they do to perceptual availability’ (Matthews, Lieven, Theakston, & Tomasello, 2006, p. 408).

    What do these performances correspond to? Several possibilities can be considered. The first, proposed by Levy (1989, 1999), suggests that the relation between the pro-noun and its antecedent corresponds to the repetition of noun–pronoun strings pro-duced by adults. In this case, since the child does not establish the co-referential relation existing in his or her discourse autonomously, the pronoun does not have an anaphoric value. The second possibility presented by Matthews et al. (2006) – drawing on Pickering and Garrod’s (2004) model – suggests that there is a dissociation between the co-referential links with the interlocutor’s discourse, which would be of a mechan-ical nature, and the child’s representation of the interlocutor’s informational needs. Thus, it could be considered that the co-referential links are not anaphoric (see also Karmiloff-Smith, 1985). Finally, a third alternative (De Cat, 2004a; Salazar Orvig et al., 2010) differentiates the values of referential expressions from the difficulties children may have in the management of discursive sequences, joint attention and common ground. In this case, we can consider that children acquire pronouns with a proto-anaphoric value (maintaining continuity in the context of a shared discursive representation), while they present performance errors, in particular when they have to manage com-plex interlocutive situations.

    Aim of this studyThe present research studies aspects of the dialogical context of the use of third person clitic pronouns. In view of the questions raised above, two possibilities are considered: (1) young children’s use of pronouns is prompted by the form of the adult’s utterances; (2) children’s choice of a pronoun as a referential expression could be explained by pragmatic-discursive factors. To address the first option, we analyse how the form of adults’ utterances influences children’s use of referential expressions. The second possi-bility will be tackled through the analysis of the ways in which children’s utterances are anchored in the dynamics of dialogue and, more specifically, the type of continuity between the child’s utterance and the previous utterance of the adult.

  • Salazar et al. 379

    The analysis of these factors will allow us to evaluate the type of relation existing between the pronoun and its antecedent. If the children reproduce the whole utterance (utterance reproduction), we may consider that the coreferential relation is only a side effect of repetition. If the children reproduce only the referential expression (item repro-duction), we may think they take up a model from the adult and establish a basic type of discursive relation, as the reference of the expression they use is grounded in the dia-logical context. Finally, only if the use of pronouns can be related to other pragmatic-discursive factors, we may conclude that these early uses are anaphoric.

    We concentrate on the subject function, since, on one side, it is the privileged slot to express topics and, on the other, in French, subject clitic pronouns are acquired long before object clitic pronouns. To determine whether our results are specific to this grammatical category, pronouns are compared to other referential expressions in subject position.

    Method

    Participants

    The data presented here are part of a larger corpus (Salazar Orvig et al., 2004, 2006). For the purpose of the present study we considered only the children (in the cross-sectional corpora) or the sessions (for the longitudinal follow-up) for which there was at least one clitic pronoun in subject position in the data. Three videotaped corpora were used (Table 1):

    Corpus 1: A longitudinal corpus of two boys (Daniel and Léonard). For six sessions per child – about one session per month between 1;9 (for Daniel) or 1;11 (for Léonard) and 2;4 – were analysed. The children were observed in different situations at home (play, snack time, picture book reading, bath). Corpus 2: A cross-sectional corpus at 1;11 based on five dialogues of five children (four girls, Alice1, Cécile, Lisa, Pauline1 and one boy, Thibault) with their father or mother during a standardized play session at home.Corpus 3: A cross-sectional corpus at 2;3 based on 10 other children (six girls, Alice2, Chloé, Elodie, Léa, Margaux, Pauline2 and four boys, Arnaud, Maxime, Rémi, Théo) with their mother in various activities (play, snack). For five of those children (Chloé,

    Table 1. Population: number of children, age, MLU, number of sessions, number of turns

    Corpus Age MLU MLU level

    No. of children

    No. of sessions

    Total no. of turns

    Mean no. of turns

    Cross-sectional Per child Per child1;11 mean: 1;11

    (1;11.3–1;11.26)1.56–2.83 mean: 1.98

    1, 2, 3 5 1 306 61.20

    2;3 mean: 2;3.5 (2;2.5–2;3.29)

    1.32–3.01 mean: 2.07

    1, 2, 3 10 2–4 2620 262

    Longitudinal (home)

    2 Per child Per session

    Daniel 1;9–2;4 1.36–2.50 1, 2, 3 6 1603 229Léonard 1;11–2;4 2.04–2.85 2, 3 6 384 647

  • 380 First Language 30(3-4)

    Elodie, Margaux, Rémi and Théo), play and snack sessions were also recorded at their nursery school.

    The utterances of all participants were transcribed and contextual information necessary for the interpretation of the verbal production was noted.

    At each step, we first present the results from the longitudinal data and then from the cross-sectional data. The longitudinal data are divided into two-months periods 1;9–1;10, 1;11–2;0, 2;1–2;2 and 2;3–2;4.

    In the cross-sectional data, the two corpora present a similar range of mean length of utterance (MLU) (computed in number of words out of 100 utterances per child). In order to draw a developmental trend, it is thus better to contrast children’s productions accord-ing to their MLU rather than their age. Even though there is not necessarily a direct link between MLU and pragmatic development, this method is also appropriate because chil-dren’s use of referential expressions cannot be considered independently of morphological development. The cross-sectional data have thus been divided according to three groups of children, roughly corresponding to Brown’s I, II and III stages (Brown, 1973):

    Level 1: Five children whose MLU is between 1.32 and 1.73. Fewer than 17% of their utterances include a verb, with or without arguments; the utterances are either bare nouns (mean 34%), or one- to two-word pre-syntactic predications (mean 14%).Level 2: Five children whose MLU is between 2.04 and 2.44. 30% of their utterances have a verbal predicate (with or without arguments). Nominal utterances are less prevalent (19%) but 15% of their utterances can still be described as pre-syntactic predications.Level 3: Five children whose MLU is between 2.50 and 3.01. The mean number of verbal predicate utterances corresponds to 40%, nominal utterances to 22% and pre-syntactic predications to 9% of their utterances.

    The two children from the longitudinal corpus follow a similar trend: at 1;10 they mostly produce non-verbal utterances and a very low percentage of utterances with ver-bal predicates. This proportion is inverted at the end of the recordings: there are between 40% and 50% utterances with verbal predicates.

    All children belong to French-speaking middle-class families.

    Data analysisIn this section we first present the types of referential expressions and the criteria used to categorize referential continuity between the children’s utterances and their interlocutors’ utterances. Then, we turn to the categories concerning the influence (of the form) of the adult’s discourse and the type of continuity between the speech turns of the adults and of the children.

    Referential expressions. Considering any linguistic unit that can refer to an entity or to events processed as entities as a referential expression,1 we contrast clitic pronouns with other referential expressions.

  • Salazar et al. 381

    Clitic pronouns: For this study we have considered all adult-like forms as clitic pronouns ([il], [εl]). Some authors (Jakubowicz & Rigaut, 1997; Veneziano, 2003) con-sider that the forms [i] and [ε] (e.g., [εtɔb̃] ‘e(lle) tombe’ ‘she’s falling’) do not clearly correspond to pronouns. However, in oral French, adults frequently omit the lateral [l] in front of a consonant; we have thus included these forms in the pronominal category. Forms in [l] (e.g., [leasi] ‘(i)l est assis’ ‘he’s sitting’) frequently used by children, were also included.

    Other referential expressions: • Nouns: Both proper names and nouns are included in this category. • Demonstrative pronouns: ‘Ça’ ([savala] ‘ça va là’ ‘it goes there’) is the most fre-

    quent form of demonstrative pronoun in our data. Other forms such as ‘celui-ci’ (‘this one’) and ‘celui-là’ (‘that one’) are occasionally used.

    • ‘C’est’ is considered separately from other demonstratives because it can correspond either to the association of the demonstrative and the copula or to a presentative (Martinet, 1979) and because it is difficult to determine whether children are produc-ing a creative combination or a frozen expression for a given utterance. Besides, ‘c’est’ is used very often (e.g., [seamwa?] ‘c’est à moi?’ ‘is it mine?’] by young children.

    • Dislocation constructions: In oral French, pronouns are often used in co-occurrence with a noun (e.g., [anaεlepati] ‘Anna elle est partie’ ‘Anna she’s gone’ or [ifedodoləʃa], ‘il fait dodo le chat’ ‘it is sleeping, the cat’). Dislocations also involve demonstratives (e.g., [sepabõsa] ‘c’est pas bon ça’ ‘it’s not good, that’).2

    • Other pronouns: There also are a few occurrences of indefinite and interrogative pronouns.

    This study is focused on referential expressions in subject function. As the period under study covers the emergence of syntax, position was not the sole criterion to iden-tify ‘subjects’; semantic criteria have also been used as in (1) where [lapupe], the agent, has been considered as the ‘subject’:

    (1) Alice 1;11 – MLU 2.2 Alice: bwa atas lapupe3

    ‘boit F 4 tasse la poupée’ ‘drinks F cup the doll’

    We also include the category of unmarked reference: during the period under study, the subject can be absent.5 This can be the case in ‘pre-syntactic’ productions, such as (2):

    (2) Daniel 1;9 – MLU 1.59 Daniel has just put a piece of something in his shirt. He is looking for it.

    Mother: où il est? ‘where is it?’Daniel takes out the object and looks at it.Mother: le voilà! ‘here it is!’Daniel: a!6 kaçe ‘ah! cassé’ ‘oh! broken!’

  • 382 First Language 30(3-4)

    It can also be the case in utterances that are already syntactically fairly advanced, such as (3):

    (3) Arnaud 2;3 – MLU 3.01Mother: qu’est ce qu’elle fait la caméra? ‘what does the camera do?’

    Arnaud: feε̃pətipəobwi ‘fait un petit peu F bruit’ ‘makes a little bit F noise’

    During this period, (proto)verbs can be associated to proto-forms ([atu]: ‘F tou(rne)’ ‘F turn{/s}’) called ‘fillers’ (Peters, 2000; Veneziano & Sinclair, 2000). The morphological status of these units is uncertain: it is particularly difficult to differentiate a proto-pronoun and a proto-auxiliary in preverbal position which implies that it would also be difficult to specify whether these items are referential. We thus include them in the ‘unmarked reference’ category.

    Relation to the interlocutor’s discourse. This analysis7 concerns the referential continuity between the children and their interlocutor’s previous utterances. We consider as inter-locutors not only the mother, the father, or the caretaker but also other children (siblings in home recordings, other children in the nursery) and the observer who is recording.8 The referential expression used by the child can be linked to these previous utterances in four different ways:

    Immediate explicit mention (IM): the interlocutor has mentioned the referent (with any referential expression, including pronouns) in the immediately preceding utterances:

    (4) Léonard 2;4 – MLU 2.45 Mother and child are looking at a book.Mother: Et oui, et Adèle? ‘yes, and Adèle?’

    Léonard: [εletõbela əgad] ‘elle est tombée là regarde’ ‘she fell over there look’

    Non-immediate explicit mention (NIM): this groups three different categories:• Distant explicit mention: the interlocutor has mentioned the referent in another

    topical sequence, which means that there are at least two shifts in the ‘discourse topic’ (Dik, 1997) between the target utterance and the one produced by the inter-locutor:

    (5) Léa 2;2 – MLU 2.4 Léa and her mother are playing in the room. They are looking for stuffed animals.

    Léa: ijeu ‘il est où’ ‘where is it?’

    Mother: je sais pas. tu sais où il est? hum? ‘I don’t know. do you know where it is? hum?’

    . . . eight turns involving Léa’s name and age.

  • Salazar et al. 383

    Léa points to something behind her mother’s back.Léa: ejeja . ʃwimbɔmatata. ɔʃõ ‘elle est là. Simba Matata. (c)ochon’ ‘it’s here. Simba Matata. piggy’

    In this case, Léa reintroduces the reference to an object that both interlocutors were looking for previously.

    • Inferable reference: the referent has not been explicitly mentioned by the inter-locutor but can be inferred from her/his utterance:

    (6) Daniel 2;4 – MLU 2.39 Daniel sits on his train. Daniel: vəvwa mɔ̃ {ə} tuʃu {bi bije}. vwεj

    vwa {ə} vaʃ vəjvaʃ ‘veux voir mon F tchou tchou {???}. ‘want to see my F tchu

    {veux} voi(r) F {vache/va}. veux tchu. {want} see vache’{vache/va} F {cow/go}, want {cow/go}’

    Observer: tu t’en vas? ‘are you leaving?’ Daniel: wi ‘oui’ ‘yes’ Daniel: kɔsa isãva ‘comme ça i(l) s’en va’ ‘like that it’s leaving’

    The attention of both participants (the observer and the child) is focused on the little train, which has been mentioned only by the child. Even though we can consider that the observer’s utterance ‘tu t’en vas’ is alluding to the train it does not explicitly mention it.

    • Absent referent: the referent mentioned by the child is absent from the previous interlocutor’s discourse:

    (7) Arnaud 2;3 – MLU 3.01 Arnaud is having his snack with his mother. Mother: tu vas manger une tarte ‘you are going to eat an apple pie’

    aux pommes Arnaud: wε ‘ouais’ ‘yeah’ Mother: ouais ‘yeah’ Arnaud: {uwe} klεmari?

    ‘{ou est} Claire-Marie?’ ‘where is Claire-Marie?’

    Influence of the adult’s utterance. In the cases of immediate referential continuity, chil-dren’s use of referential expressions can be influenced by the form of the interlocutor’s utterance. Four cases are considered:

  • 384 First Language 30(3-4)

    Utterance reproduction: Children reproduce the lexical and syntactic form of the adult’s utterance:

    (8) Margaux 2;3 – MLU 2.62Margaux is having her afternoon snack at the nursery school.Adult: non elle est pas là ‘no she’s not here’Margaux: {selasepala{xx}la++ palasalɔt} ‘c’est là c’est pas là {xx} ++ ‘she’s not here, not here.’ pas là Cha(r)lotte’ Adult: Charlotte? ‘Charlotte?’Margaux: εlepala ‘elle est pas là’ ‘she’s not here’

    The reproduction may not immediately follow the interlocutor’s utterance, Margaux replicates the adult’s utterance word for word in her second turn.

    Item reproduction: The child takes up the referential expression used by the adult but constructs a different utterance. The child’s utterance shares with his or her interlocutor’s utterance the referential expression (e.g., the noun, the pronoun) for the same referent, without it being a replication of the whole interlocutor’s utterance:

    (9) Arnaud 2;3 – MLU 3.01Arnaud and his mother are talking about the observer. Mother: elle va revenir dans un petit ‘she’s coming back soon,

    moment, tu finis ta tarte finish your apple pie’ aux pommes

    Arnaud: lepati ‘(e)lle est partie’ ‘she’s gone’

    The cases when the verb is also taken up are included in this category:

    (10) Pauline2 2;3 – MLU 2.52 Mother: elle est où la brioche? ‘where is the cake?’Pauline: εledødã) ‘elle est dedans’ ‘it’s inside’

    As illustrated in this example, cases where a clitic pronoun follows an adult’s disloca-tion are also included in this category: the child is taking up a form available in the adult’s discourse.

    Elicited: A question by the adult can elicit the form used by the child:

    (11) Alice 2;3 – MLU 2.44Mother: qu’est ce que tu veux maintenant? ‘what do you want now?’Max: sε ‘(de)ssert’ ‘desert’

  • Salazar et al. 385

    Absence of direct influence: The form of the adult’s utterance is not directly reflected in the referential expression used by the child:

    (12) Léa 2;2 – MLU 2.4Mother: tu veux qu’on fasse une ‘you want us to build a house with lego?’

    maison avec les legos?Léa: wi {isɔ̃dədã} ‘oui {i(ls) sont dedans}’ ‘yes they’re inside’

    Unmarked references are, by definition, not taken up from the interlocutor’s utterances. Nevertheless we have considered the cases where a filler follows a clitic pronoun (alone or in a dislocation construction) produced by the adult, and cases where they appear after an adult’s question.

    Types of dialogical continuity. This analysis deals with a pragmatic-discursive factor that can account for the use of referring expressions in the dynamics of dialogue. In narra-tives, the choice of a referential expression depends on the dynamics of the topics. In dialogue, it can also correspond to the maintenance or the change of perspective towards the discourse of the interlocutor. Three categories can be distinguished:

    Repetitions and reformulations of a preceding utterance without adding any new predi-cation to the topic:

    (13) Arnaud 2;3 – MLU 3.01Mother: ben c’est Nanaud qui ‘well. It’s Nanaud who is hitting

    tape sur les boutons the buttons.’Arnaud: nanoitaibutõ ? ‘Nanaud i(l) ta(pe) F boutons?’ ‘Nanaud is hitting F buttons?’

    Contrastive sequences: despite referential continuity there is a shift in the way the topic is tackled – with opposition, a change of thematic focusing (14), perspective shifting, explanations, clarifications or requests for clarification:

    (14) Chloé 2;3 – MLU 2.71 Adult: c’est Morgane qui l’a ‘Is it Morgane who has the giraffe?’

    la + girafe?Chloé: morgan’nenosapεl +

    morgan’nenosapεl ‘Morgane [neno] s’appelle ‘Morgane [neno] is called

    Morgane [neno] s’appelle.’ Morgane [neno] is called’

    Plain continuity: only a new predication is added on the same topic without any change of perspective or genre shifting:

    (15) Daniel 2;0 – MLU 2.02Daniel: pupa (noise for fire men trucks)

  • 386 First Language 30(3-4)

    Observer: ah papoupa oui! ‘oh! pa pou pa, yes.’Daniel: a ka katje pupa ‘{arête} poupa’ ‘{stopped} pupa’ Observer: ça s’est arrêté oui ‘it stopped, yes’

    This category also includes answers to questions about the same discourse object, as in (10).

    ReliabilityIntercoder agreement for each of the categories under study was tested for 10% of the data from children randomly selected. The rates of agreement were 98% for grammatical categories and 87% for the presence of the discourse object in the interlocutor’s dis-course. The coding of the two last sets of categories was done separately by different coders (the co-authors) and afterwards discussed by the whole group of coders in order to resolve divergences.

    A descriptive approach was adopted for the data analysis. This approach seemed preferable for two main reasons: the first is that children strongly differ as to the amount of occurrences in each category; the second is that, even for adults, the use of a given referential expression depends on various factors, and there are no sharp contrasts for the use of referential expressions for each functional category. We present the total in absolute values and the percentages for each group, as well as the mean and standard deviation for each functional category. Tables A1–A6 in the Appendix9 give the data for each child.

    Results

    Referential expressions

    Table 2 presents the distribution of referential expressions and unmarked reference in subject position.

    Our data relate to the acquisition process of pronouns in its very beginning. The lon-gitudinal data enable us to observe how clitic pronouns in subject position emerge. Even though the two children under study very quickly use more than two words per utterance on average, the proportion of clitic pronouns does not grow at the same pace (with the exception of Léonard at 2;3–2;4). Nouns and dislocations are (together) the most fre-quent subjects. Unmarked reference, which appears to be prevalent at 1;9–1;10 in Daniel’s corpus, corresponds to a third of both children’s utterances at 1;11–2;0 but still around 11–15% at 2;1–2;2 and 2;3–2;4. The cross-sectional data present a slightly different distribution as there is no predominant category for the subject position: neither clitic pronouns nor nouns (with dislocations), which are more frequent than clitic pronouns for subject function at any stage. Unmarked reference diminishes with the MLU level, from 39% of the utterances at level 1 to 10% at level 3.

  • Salazar et al. 387

    Tabl

    e 2.

    Ref

    eren

    tial e

    xpre

    ssio

    ns a

    nd u

    nmar

    ked

    refe

    renc

    e in

    sub

    ject

    pos

    ition

    : num

    ber

    and

    (per

    cent

    age)

    Cor

    pus

    Clit

    ic

    pron

    ouns

    N

    o. (

    %)

    Nou

    ns

    No.

    (%

    )

    Dis

    loca

    tions

    N

    o. (

    %)

    C’e

    st

    No.

    (%

    )

    Dem

    . pr

    onou

    ns

    No.

    (%

    )

    Oth

    er

    pron

    ouns

    N

    o. (

    %)

    Unm

    arke

    d re

    fere

    nce

    No.

    (%

    )

    Tota

    l N

    o. (

    %)

    Age

    M

    LU le

    vel

    Long

    itudi

    nal

    D

    anie

    l

    1;9–

    1;10

    Le

    vel 1

    5 (1

    1)0

    2 (4

    )1

    (2)

    0 0

    37 (

    82)

    45 (

    100)

    1;

    11–2

    ;0

    Leve

    l 25

    (6)

    26 (

    31)

    16 (

    19)

    8 (1

    0)5

    (6)

    024

    (29

    )84

    (10

    0)

    2;1–

    2;2

    Leve

    l 24

    (11)

    6 (1

    7)12

    (33

    )5

    (14)

    5 (1

    4)0

    4 (1

    1)36

    (10

    0)

    2;3–

    2;4

    Leve

    l 314

    (11

    )17

    (13

    )33

    (25

    )28

    (22

    )18

    (14

    )1

    (1)

    37 (

    15)

    130

    (100

    )

    Léon

    ard

    1;

    11–2

    ;0

    Leve

    l 22

    (13)

    5 (3

    1)2

    (13)

    0 1

    (6)

    06

    (38)

    16 (

    100)

    2;

    1–2;

    2 Le

    vel 2

    4 (1

    1)6

    (16)

    9 (2

    4)9

    (24)

    1 (3

    )4

    (11)

    4 (1

    1)37

    (10

    0)

    2;3–

    2;4

    Leve

    l 315

    (30

    )2

    (4)

    17 (

    34)

    8 (1

    6)0

    1 (2

    )7

    (14)

    50 (

    100)

    Cros

    s-se

    ctio

    nal

    1;

    11/2

    ;3Le

    vel 1

    15 (

    20)

    5 (7

    )12

    (16

    )8

    (11)

    5 (7

    )0

    29 (

    39)

    74 (

    100)

    1;

    11/2

    ;3Le

    vel 2

    32 (

    16)

    21 (

    11)

    32 (

    16)

    56 (

    29)

    9 (5

    )3

    (2)

    41 (

    21)

    194

    (100

    )

    1;11

    /2;3

    Leve

    l 345

    (15

    )36

    (12

    )52

    (18

    )11

    3 (3

    8)16

    (5)

    4 (1

    )29

    (10

    )29

    5 (1

    00)

  • 388 First Language 30(3-4)

    Presence of the referent in the interlocutor’s discourse

    Tables 3a and 3b present the distribution of the referential expressions used by the children in subject position according to the previous mention of the referent by their interlocutor. In order to simplify the tables, immediate explicit mention (IM) is contrasted with all other categories (NIM), on the one hand, and clitic pronouns (CP) with other referential expres-sions (ORE) and unmarked reference (UR), on the other hand. The last column indicates the distribution of IM and NIM for each age or MLU level. Tables A1 and A2 in the Appendix9 present, in raw numbers (and percentages), the detail for each child and for each referential expression.

    The size of the standard deviations indicates that there are considerable individual dif-ferences. However, (see table A2 in the Appendix9 for full data) for the majority of children, the proportion of occurrences in each category follows the same trend as mean percentages. This is also the case for the results we present in Tables 4a and 4b and 5a and 5b. Given that we deal with dialogic data and that we have focused the analyses on the subject position, topical continuity is predominant for all referential expressions and unmarked reference, even if the distribution is less contrasted for the longitudinal data. Tables 3a and 3b also show that clitic pronouns are not used in the same way as the other referential expressions. Longitudinal data show that, even when clitic pronouns are not very frequent in the corpus (before 2;3), children use them more often in immediate continuity with a referent explicitly mentioned by the interlocutor than when the referent has not been immediately mentioned by the interlocutor (with the exception of Daniel at 2;3–2;4, for whom there is just a slight tendency). The opposite is true for the other referential expressions, especially for disloca-tions, nouns and demonstratives. With the same exception of Daniel at 2;3–2;4, they appear fairly less often in the context of immediate explicit mention than in the context of no immediate mention. On the contrary, unmarked reference follows the same trend as clitic pronouns, being more frequent in the IM than in the NIM condition.

    In the cross-sectional data the effect of the dialogical context is clear: for all three lev-els, clitic pronouns are more frequently used when the discourse object is immediately present in the interlocutor’s utterance than when this is not the case, the other referential expressions, and more specifically nouns, dislocation constructions and demonstratives, appear more often in the non-immediate explicit mention condition. The examination of individual data (see also Table A2) shows that nearly all the children at the three levels present this pattern. The exceptions are: Lisa (whose referential expressions other than pronouns are very scarce) and Alice2 (who, on the contrary presents a much higher pro-portion of ORE in subject function). Unmarked reference does not present the same con-figuration as clitic pronouns; it appears in level 2 and level 3 as often in IM condition as in NIM condition. Individual data show that for the majority of children unmarked refer-ences are more frequent in the NIM context than in the IM context.

    Influence of the adult’s utteranceTables 4a (longitudinal corpus) and 4b (cross-sectional corpus) present the distribution of formal links with the preceding adult utterance for each type of referential expression produced by the children. In order to grasp the prevalence of the direct influence of the

  • Salazar et al. 389

    Tabl

    e 3a

    . D

    istr

    ibut

    ion

    of r

    efer

    entia

    l exp

    ress

    ions

    acc

    ordi

    ng t

    o th

    e m

    entio

    n of

    the

    ref

    eren

    t in

    the

    inte

    rloc

    utor

    ’s di

    scou

    rse

    – lo

    ngitu

    dina

    l dat

    a (n

    umbe

    r an

    d pe

    rcen

    tage

    )

    Chi

    ldA

    geM

    entio

    nC

    litic

    pro

    noun

    s%

    CP

    Oth

    er

    refe

    rent

    ial

    expr

    essi

    ons

    %

    OR

    EU

    nmar

    ked

    refe

    renc

    e%

    UR

    Tota

    l%

    Tot

    al

    Dan

    iel

    1;9–

    1;10

    IM 4

    16 1

    420

    80 2

    5 5

    6N

    IM 1

    5 2

    1017

    85 2

    0 4

    4To

    tal 1

    ;9–1

    ;10

    511

    3 7

    3782

    45

    100

    1;11

    –2;0

    IM 5

    1122

    5017

    39 4

    4 5

    3N

    IM 0

    032

    82 7

    18 3

    9 4

    7To

    tal 1

    ;11–

    2;0

    5 6

    5465

    2429

    83

    100

    2;1–

    2;2

    IM 4

    2013

    65 3

    15 2

    0 5

    6N

    IM 0

    015

    94 1

    6 1

    6 4

    4To

    tal 2

    ;1–2

    ;2 4

    1128

    78 4

    11 3

    610

    02;

    3–2;

    4IM

    911

    5973

    1316

    81

    62

    NIM

    510

    3878

    612

    49

    38

    Tota

    l 2;3

    –2;4

    1411

    9774

    1915

    130

    100

    Léon

    ard

    1;11

    –2;0

    IM 2

    22 2

    22 5

    56

    9 5

    6N

    IM 0

    0 6

    86 1

    14

    7 4

    4To

    tal 1

    ;11–

    2;0

    213

    850

    637

    16

    100

    2;1–

    2;2

    IM 3

    1714

    78 1

    5 1

    8 4

    8N

    IM 1

    515

    79 3

    16 1

    9 5

    1To

    tal 2

    ;1–2

    ;2 4

    1129

    78 4

    11 3

    710

    02;

    3–2;

    4IM

    1339

    1340

    721

    33

    66

    NIM

    212

    1588

    0 0

    17

    34

    Tota

    l 2;3

    –2;4

    1530

    2856

    714

    50

    100

    IM: e

    xplic

    it im

    med

    iate

    ly m

    entio

    ned;

    NIM

    : non

    -imm

    edia

    tely

    men

    tione

    d (d

    ista

    nt, i

    nfer

    able

    , or

    abse

    nt).

  • 390 First Language 30(3-4)

    Tabl

    e 3b

    . D

    istr

    ibut

    ion

    of r

    efer

    entia

    l exp

    ress

    ions

    acc

    ordi

    ng t

    o th

    e m

    entio

    n of

    the

    ref

    eren

    t in

    the

    inte

    rloc

    utor

    ’s di

    scou

    rse

    – cr

    oss-

    sect

    iona

    l dat

    a (n

    umbe

    r an

    d pe

    rcen

    tage

    )

    MLU

    gro

    upM

    entio

    nC

    litic

    pro

    noun

    sO

    ther

    ref

    eren

    tial e

    xpre

    ssio

    nsU

    nmar

    ked

    refe

    renc

    eTo

    tal (

    %)

    No.

    %

    Mea

    n %

    SDN

    o.%

    Mea

    n %

    SDN

    o.%

    Mea

    n %

    SD

    Leve

    l 1IM

    1223

    2915

    .1 1

    631

    3514

    .124

    4635

    1952

    (70

    )N

    IM 3

    1415

    33.5

    14

    6365

    38.7

    523

    2012

    .822

    (30

    )To

    tal L

    115

    2022

    21.1

    30

    4146

    22.8

    2939

    3217

    .774

    (10

    0)Le

    vel 2

    IM28

    2219

    21.6

    76

    5859

    27.9

    2620

    2224

    .113

    0 (6

    7)N

    IM 4

    6 4

    8.1

    45

    7073

    16.9

    1524

    2218

    .564

    (33

    )To

    tal L

    232

    1615

    16.9

    121

    6264

    15.9

    4121

    2115

    .819

    4 (1

    00)

    Leve

    l 3IM

    3820

    2424

    .413

    670

    6622

    .919

    1010

    2.5

    193

    (65)

    NIM

    7 7

    1217

    .6 8

    583

    7518

    .810

    1012

    6.3

    102

    (35)

    Tota

    l L3

    4515

    2123

    .522

    175

    6922

    .829

    1010

    2.2

    295

    (100

    )

    IM: e

    xplic

    it im

    med

    iate

    ly m

    entio

    ned;

    NIM

    : non

    -imm

    edia

    tely

    men

    tione

    d (d

    ista

    nt, i

    nfer

    able

    , or

    abse

    nt).

  • Salazar et al. 391

    Tabl

    e 4a

    . In

    fluen

    ce o

    f the

    inte

    rloc

    utor

    ’s ut

    tera

    nce

    on c

    hild

    ren’

    s us

    e of

    ref

    eren

    tial e

    xpre

    ssio

    ns –

    long

    itudi

    nal d

    ata

    (num

    ber

    and

    perc

    enta

    ge)

    Chi

    ldA

    geIn

    fluen

    ceC

    litic

    pr

    onou

    ns%

    CP

    Oth

    er

    refe

    rent

    ial

    expr

    essi

    ons

    %

    OR

    EU

    nmar

    ked

    refe

    renc

    e%

    UR

    Tota

    l%

    To

    tal

    Dan

    iel

    1;9–

    1;10

    Item

    rep

    rodu

    ctio

    n 2

    100

    00

    00

    28

    No

    influ

    ence

    2

    9 1

    420

    8723

    92To

    tal 1

    ;9–1

    ;10

    4 1

    6 1

    420

    8025

    100

    1;11

    –2;0

    Utt

    eran

    ce

    repr

    oduc

    tion

    1 3

    3 2

    670

    03

    7It

    em r

    epro

    duct

    ion

    3 3

    3 2

    224

    459

    20N

    o in

    fluen

    ce 1

    3

    1856

    1341

    3273

    Tota

    l 1;1

    1–2;

    0 5

    11

    2250

    1739

    4410

    0

    2;1–

    2;2

    Utt

    eran

    ce

    repr

    oduc

    tion

    0

    0 3

    100

    00

    315

    Item

    rep

    rodu

    ctio

    n 2

    40

    360

    00

    525

    No

    influ

    ence

    2 1

    7 7

    583

    2512

    60To

    tal 2

    ;1–2

    ;2 4

    20

    1365

    315

    2010

    02;

    3–2;

    4U

    tter

    ance

    re

    prod

    uctio

    n 0

    0

    810

    00

    08

    9It

    em r

    epro

    duct

    ion

    4

    0 0

    00

    100

    2430

    Elic

    ited

    0 1

    720

    831

    01

    1N

    o in

    fluen

    ce 5

    10

    2165

    1225

    4859

    Tota

    l 2;3

    –2;4

    9 1

    131

    7213

    1681

    100

    Léon

    ard

    1;11

    –2;0

    Item

    rep

    rodu

    ctio

    n 0

    0

    00

    310

    03

    33N

    o in

    fluen

    ce 2

    33

    233

    233

    667

    Tota

    l 1;1

    1–2;

    0 2

    22

    222

    556

    910

    02;

    1–2;

    2U

    tter

    ance

    re

    prod

    uctio

    n 1

    50

    150

    00

    211

    Item

    rep

    rodu

    ctio

    n 2

    50

    125

    125

    422

    No

    influ

    ence

    0

    012

    100

    10

    1267

    Tota

    l 2;1

    –2;2

    3 1

    714

    781

    618

    100

    2;3–

    2;4

    Utt

    eran

    ce

    repr

    oduc

    tion

    0

    0 1

    100

    00

    13

    Item

    rep

    rodu

    ctio

    n10

    83

    18

    18

    1239

    No

    influ

    ence

    3 1

    710

    565

    2818

    58To

    tal 2

    ;3–2

    ;413

    42

    1239

    619

    3110

    0

  • 392 First Language 30(3-4)

    Tabl

    e 4b

    . In

    fluen

    ce o

    f the

    inte

    rloc

    utor

    ’s ut

    tera

    nce

    on c

    hild

    ren’

    s ch

    oice

    of r

    efer

    entia

    l exp

    ress

    ion

    – cr

    oss-

    sect

    iona

    l dat

    a (n

    umbe

    r an

    d pe

    rcen

    tage

    )

    MLU

    gr

    oup

    Influ

    ence

    Clit

    ic p

    rono

    uns

    Oth

    er r

    efer

    entia

    l exp

    ress

    ions

    Unm

    arke

    d re

    fere

    nce

    Tota

    l (%

    )

    No.

    %M

    ean

    %SD

    %N

    o.%

    Mea

    n %

    SD %

    No.

    %M

    ean

    %SD

    %

    Leve

    l 1El

    icite

    d0

    00

    310

    040

    54.8

    00

    03

    (6)

    Item

    rep

    rodu

    ctio

    n4

    5741

    .750

    343

    46.7

    50.6

    00

    07

    (13)

    Utt

    eran

    ce

    repr

    oduc

    tion

    00

    01

    100

    2044

    .70

    00

    1 (2

    )N

    o in

    fluen

    ce8

    19.5

    29.3

    16.9

    922

    24.3

    20.3

    2458

    .546

    .317

    .941

    (79

    )To

    tal L

    112

    2329

    15.1

    1631

    3514

    .124

    4635

    18.9

    52 (

    100)

    Leve

    l 2El

    icite

    d0

    00

    110

    020

    44.7

    00

    01

    (1)

    Item

    rep

    rodu

    ctio

    n14

    5039

    37.1

    1346

    58.5

    34.1

    10

    00.

    128

    (22

    )U

    tter

    ance

    re

    prod

    uctio

    n3

    1925

    36.3

    1381

    7536

    .30

    00

    16 (

    12)

    No

    influ

    ence

    1113

    13.5

    13.5

    4958

    58.9

    24.3

    2529

    .427

    .627

    .285

    (65

    )To

    tal L

    228

    2219

    21.6

    7658

    5926

    .926

    2022

    24.1

    130

    (100

    )Le

    vel 3

    Elic

    ited

    00

    03

    100

    6054

    .80

    00

    3 (2

    )It

    em r

    epro

    duct

    ion

    2649

    45.1

    32.3

    2445

    49.8

    293

    00.

    10.

    153

    (27

    )U

    tter

    ance

    re

    prod

    uctio

    n1

    525

    5021

    9575

    500

    00

    22 (

    11)

    No

    influ

    ence

    1110

    10.3

    7.3

    8877

    75.1

    7.4

    1613

    .914

    .713

    .511

    5 (6

    0)To

    tal L

    338

    2024

    24.5

    136

    7066

    22.9

    1910

    9.9

    2.49

    100

  • Salazar et al. 393

    interlocutor’s utterance form in the children’s use of referential expressions, the last column presents the global percentage of each functional category for each age or MLU level.

    Tables 4a and 4b show that direct influence of the interlocutor’s utterance is not a homoge-neous phenomenon across children and ages or MLU levels. For both the longitudinal and cross-sectional data, direct influence of the form of the interlocutor’s discourse ranges from 0% to 60% (see also and Tables A3 and A4 in the Appendix9 for full data). In the longitudinal data, item reproduction ranges from 8% to 40% of the children’s utterances, utterance reproduction from 0% to 15% and elicitation is a marginal phenomenon present only in Daniel’s data. These data show that the two children therefore do not behave in the same way as far as this direct influence is concerned. Daniel tends to produce clitic pronouns after the adult has also pro-duced one (item reproduction) in the first months and then use them autonomously, whereas Léonard presents the opposite pattern with item reproduction for clitic pronouns only after 2;1. But for both children, utterance reproduction does not seem to be a privileged context for the production of clitic pronouns. It therefore seems difficult to consider that there is a systematic trend for the influence of the form of the interlocutor’s utterance on children’s utterances.

    The cross-sectional data show that utterance reproduction grows from 2% to 12% and that item reproduction grows from 13% at level 1 to 27% of the children’s utterances at level 3. Elicitation by questions appears as a minor process in any case and, as it could be expected, unmarked references are more often associated with an absence of influence. In item reproduction contexts, clitic pronouns seem to be as frequent as the other referential expressions. Individual data confirm that children behave differently with respect to the way they anchor their productions in their interlocutor’s utterance. At each MLU level, some children (two at level 1, two at level 2 and four at level 3) proportionally take up clitic pronouns more often than other referential expressions from their interlocutors’ utterances. The other children either behave equally for clitic pronouns and other referen-tial expressions, or take up the other referential expressions more often.

    Types of dialogical continuityTables 5a (longitudinal) and 5b (cross-sectional) present the type of dialogical continuity existing between the children’s utterances and their interlocutor’s immediately preceding utterance (see also Tables A5 and A6 in the Appendix9).

    Globally, plain continuity is the most frequent link between the children’s utterances and their interlocutor’s turns. The longitudinal data show that pronouns appear more often, at each age period, in this context whereas the other referential expressions tend to be acti-vated more frequently when there is a contrast, a repetition or a reformulation. With the exception of Daniel at 1;9–1;10 and Léonard at 2;1–2;2, unmarked reference appears more frequently in the context of plain continuity than in the context of contrast sequencing. But it does not show the same pattern as clitic pronouns for repetition – reformulation.

    As for the cross-sectional corpus, the data show an evolution between level 1 and level 3. At level 1, there are no differences between plain continuity and contrast for clitic pronouns, but the other referential expressions appear slightly more frequently in the context of con-trast and repetition or reformulation. Two types of profile can be set on the basis of the individual data (Table A6 in the Appendix9): at level 1, three children use clitic pronouns more often in the context of plain continuity and other referential expressions in the context

  • 394 First Language 30(3-4)

    Tabl

    e 5a

    . D

    istr

    ibut

    ion

    of r

    efer

    entia

    l exp

    ress

    ions

    acc

    ordi

    ng t

    o th

    e ty

    pe o

    f dia

    logi

    cal c

    ontin

    uity

    – lo

    ngitu

    dina

    l dat

    a (n

    umbe

    r an

    d pe

    rcen

    tage

    )

    Chi

    ldA

    geD

    ialo

    gica

    l co

    ntin

    uity

    Clit

    ic

    pron

    ouns

    % C

    PO

    ther

    ref

    eren

    tial

    expr

    essi

    ons

    % O

    RE

    Unm

    arke

    d re

    fere

    nce

    % U

    RTo

    tal

    % T

    otal

    Dan

    iel

    1;9–

    1;10

    Plai

    n co

    nt.

    327

    00

    8 73

    11

    44C

    ontr

    ast

    00

    113

    7 87

    832

    Rep

    ./ref

    1 17

    00

    5 83

    624

    Tota

    l4

    161

    420

    80

    2510

    01;

    11–2

    ;0Pl

    ain

    cont

    .2

    204

    404

    4010

    24C

    ontr

    ast

    2 12

    953

    635

    1742

    Rep

    ./ref

    .1

    76

    437

    5014

    34To

    tal

    5 12

    1946

    1741

    4110

    02;

    1–2;

    2Pl

    ain

    cont

    .3

    503

    500

    06

    30C

    ontr

    ast

    117

    583

    00

    6 30

    Rep

    ./ref

    .0

    05

    633

    378

    40To

    tal

    4 20

    1365

    315

    2010

    02;

    3– 2

    ;4Pl

    ain

    cont

    .5

    249

    437

    3321

    27C

    ontr

    ast

    312

    2285

    14

    2633

    Rep

    ./ref

    .1

    326

    815

    1632

    40To

    tal

    9 11

    5772

    1316

    7910

    0Lé

    onar

    d1;

    11–2

    ;0Pl

    ain

    cont

    .2

    501

    25 1

    25

    4 4

    4C

    ontr

    ast

    00

    150

    150

    222

    Rep

    ./ref

    .0

    00

    03

    100

    333

    Tota

    l2

    222

    225

    569

    100

    2;1–

    2;2

    Plai

    n co

    nt.

    2 50

    250

    00

    422

    Con

    tras

    t1

    253

    750

    04

    22R

    ep./r

    ef.

    00

    990

    110

    10

    56To

    tal

    3 17

    1478

    15

    1810

    02;

    3–2;

    4Pl

    ain

    cont

    .10

    55

    528

    317

    1855

    Con

    tras

    t0

    05

    712

    297

    21R

    ep./r

    ef.

    338

    338

    225

    824

    Tota

    l13

    3913

    397

    2133

    100

  • Salazar et al. 395

    Tabl

    e 5b

    . D

    istr

    ibut

    ion

    of r

    efer

    entia

    l exp

    ress

    ions

    acc

    ordi

    ng t

    o th

    e ty

    pe o

    f dia

    logi

    cal c

    ontin

    uity

    – c

    ross

    -sec

    tiona

    l dat

    a (n

    umbe

    r an

    d pe

    rcen

    tage

    )

    MLU

    gro

    upD

    ialo

    gica

    l co

    ntin

    uity

    Clit

    ic p

    rono

    uns

    Oth

    er r

    efer

    entia

    l exp

    ress

    ions

    Unm

    arke

    d re

    fere

    nce

    Tota

    l (%

    )

    No.

    %M

    ean

    %SD

    %N

    o.%

    Mea

    n %

    SD %

    No.

    %M

    ean

    %SD

    %

    Leve

    l 1Pl

    ain

    cont

    inui

    ty8

    2824

    26.8

    724

    3841

    .514

    4837

    28.3

    29 (

    57)

    Con

    tras

    t3

    2719

    37.5

    436

    4142

    .54

    3640

    49.0

    11 (

    22)

    Rep

    ./ref

    .1

    911

    19.2

    436

    5439

    .96

    5535

    35.7

    11 (

    22)

    Tota

    l L1

    1223

    2915

    .116

    3135

    14.1

    2446

    3518

    .951

    (10

    0)Le

    vel 2

    Plai

    n co

    ntin

    uity

    1833

    2433

    2342

    4930

    .314

    2527

    29.1

    55 (

    42)

    Con

    tras

    t4

    86

    8.1

    3672

    7616

    .410

    2018

    14.4

    50 (

    38)

    Rep

    ./ref

    .6

    2438

    46.5

    1768

    5241

    .52

    810

    22.4

    25 (

    19)

    Tota

    l L2

    2822

    1921

    .676

    5859

    26.9

    2620

    2224

    .113

    0 (1

    00)

    Leve

    l 3Pl

    ain

    cont

    inui

    ty33

    4138

    27.3

    3644

    4424

    .312

    1518

    13.7

    81 (

    42)

    Con

    tras

    t2

    3 2

    .96.

    452

    8888

    14.7

    5 8

    99.

    359

    (31

    )R

    ep./r

    ef.

    3 6

    5.1

    7.0

    4891

    937.

    92

    4 2

    2.9

    53 (

    27)

    Tota

    l L3

    3820

    2424

    .513

    670

    6622

    .919

    10 9

    .92.

    519

    3 (1

    00)

  • 396 First Language 30(3-4)

    of contrasts or reformulations, whereas two other children invert this pattern. At level 2, clitic pronouns are clearly more frequent for plain continuity than for contrast and this dif-ference increases at level 3. The pattern is inverted for the other referential expressions. At level 2, one child does not reproduce this pattern and two of them also present pronouns more often in the context of repetition or reformulation. A level 3, there is only one child for whom there is no real contrast between the use of clitic pronouns and other referential expressions.

    Unmarked reference is, at each level, more frequent in the context of plain continuity than in the context of contrast sequencing. At level 1 it is a prevalent feature of utterances in repetition or reformulation; this characteristic diminishes dramatically at levels 2 and 3. For these two levels, the unmarked reference pattern is similar to that of the clitic pronouns.

    Discussion The data presented in the Results section concern the very first periods in the acquisition of clitic pronouns. In this research, several issues related to children’s use of clitic pro-nouns as opposed to other referential expressions were explored. The analysis was restricted to the subject position. In the first place, the results confirm that most referen-tial expressions tend to code a referent previously mentioned by the interlocutor, which is natural in dialogue. It can be assumed from previous studies (Matthews et al., 2006; Salazar Orvig et al., 2010) that children draw on intersubjective sharing and dialogue to construct their discourse rather than relying on their sole apprehension of the world.

    On these bases, two main issues were investigated. The first concerns the links between the child’s production of a clitic pronoun and the forms used by his or her inter-locutor in the immediate context. Four possibilities were considered: the child’s utter-ance is merely a repetition, the child takes up the referential expression (clitic pronoun or other) used by the adult in the immediate context, the referential expression (clitic pronoun or other) is elicited by a question, or the referential expression (clitic pronoun or other) mainly stands for a co-referential relation without directly depending on the form used by the interlocutor. The second main issue concerns the role of a pragmatic-discursive factor in the use of pronouns in contrast to the use of other referential expressions.

    The results for the first issue show that this relation does not always correspond to the mere uptake of the interlocutor’s construction. This is true for elicitation by a question, as well as for the exact reproduction of the interlocutor’s utterances. Concerning the use of pronouns by the children just after the adults have used one themselves, our definition of the category ‘item reproduction’ includes several cases of possible direct influence by the form of the adult’s utterance, including when a subject–verb construction produced by the adult (be it assertive or interrogative) shapes the child’s response (example 10). The results form a complex picture. The ‘item reproduction’ phenomenon is not perva-sive throughout the whole corpus and there are considerable differences among the chil-dren in the use of clitic pronouns in this context. The hypothesis of a mere reproduction of an anaphoric link established by the adult (Levy, 1989, 1999) should thus be dis-missed. Even if children anchor their production in the adults’ utterances, they do not always preferentially take up the referential expression used by their interlocutor.

    On the other hand, if we go back to the results in Tables 4a and 4b, we can observe that the more linguistically advanced children take up the interlocutor referential expression,

  • Salazar et al. 397

    without reproducing his/her utterance, more often than the less advanced children. Thus, the reproduction of the form does not necessarily correspond to an acquisitional move. It can depend on the dynamics of the dialogue. Therefore factors other than the influence of the form used by the interlocutor might play a part in the use of pronouns in continuity with the interlocutor’s discourse. For instance, the shared construction of a topical sequence could partly explain the use of pronouns. As a matter of fact, fewer children in level 1 take up the clitic pronouns from the interlocutors’ utterances. This could be due to the fact that during the first stages of language acquisition, dialogical exchanges are short (McTear, 1985; Ninio & Snow, 1996; Salazar Orvig, 2003). On the contrary, four of the five children in level 3 show a preference for clitic pronoun continuity. This suggests a correlation between pronoun use and children’s ability to maintain a conversation topic in dialogue.

    This leads us to consider pragmatic-discursive factors. The analysis presented here con-cerns the types of continuity between the child’s productions and the previous adult utter-ances in the dynamics of the dialogue. In adult language, the use of pronouns in contrast with other referential expressions depends on referent accessibility (Ariel, 1988; Givón, 1995; Prince, 1981). Similarly, studies on narratives in children (Bamberg, 1986; Hickmann, 2002) show that the use of nouns or dislocations mostly corresponds to shifts and reintro-ductions of topics. However, according to Apothéloz (1995) the choice of referential expressions does not only depend on this factor, it can also be linked to modifications of the discourse genre (for example shifting from description to argumentation). According to our results on dialogical continuity, children’s discourse presents similar phenomena.

    In the case of plain continuity, the children’s discourse might be in line with a discur-sive representation (Cornish, 1999) that was previously elaborated and shared. In the case of repetitions and reformulations children do not add anything to the discursive represen-tation, they simply reiterate it. Finally, when children need to express a contrast, they tend to use nouns or dislocations. Their verbal production functions as a new utterance act, either because children change the discourse focus (request for clarification) or because they change perspective, or finally because they need to assert their own positioning as speaker. Level 2 and 3 children (four out of five, each time) favour the use of clitic pro-nouns when a topic is continued and new predications are added in linear progression. They use nouns and dislocations when there is a contrast with the interlocutor’s utterance. This preference seems to be initiated while the children are in the process of mastering the use of pronouns as it appears if we consider cases where there are five or more pronouns in subject position: Lisa (level 1), Léa (level 2), Cécile, Margaux and Pauline2 (level 3) in the cross-sectional data and Daniel at 1;11–2;0 and 2;3–2;4 and Léonard at 2;3–2;4 in the longitudinal data; only Chloé (level 3) presents an undifferentiated pattern.

    But the data also show that plain continuity can be expressed as well through forms other than pronouns. This might indicate that the use of a particular referential device follows a more complex pattern. Let us consider the case of nouns (alone or in disloca-tions), on the one hand, and the case of unmarked references, on the other hand.

    As far as nouns or dislocations are concerned, they are mostly used in the context of a contrast (shifting topics or points of view) with the preceding discourse. Nevertheless, this can interact with the fact that, in certain cases, children do take up forms previously used by adults (17 out of 29 cases). The following example suggests how these two fac-tors seem to combine:

  • 398 First Language 30(3-4)

    (16) Léa 2;2 – MLU 2.4

    Léa and her mother are talking about a fictional character, a little princess named ‘Poucelina’.Mother 180: Et le prince il est beau le prince? ‘and the prince, he is

    handsome, the prince?’Léa 157: wi ‘oui’ ‘yes’Mother 181: c’est vrai? ‘it’s true?’ Léa 158: əpε̃ijepaʒãti ləpε̃s dəpuselina ‘Fp(r)in(ce) il est pas gentil ‘F prince, he is not nice,

    le p(r)ince de Poucelina’ Poucelina’s prince’Mother 182: pourquoi il est pas gentil? ‘why isn’t he nice?’Léa 159: pasøkø ijemeʃã ‘pa(r)ce que il est méchant’ ‘because he is mean’. . . talking about other topicsLéa 176: jəpε dəpuselina ijemeʃã

    ‘le p(r)in(ce) de poucelina ‘Poucelina’s prince, he is il est méchant’ mean.’

    Mother 205: oh bien ça! on n’a pas réussi ‘well we haven’t found out à savoir pourquoi il était why he was mean. well, méchant mais bon! pourquoi il why is he mean the prince?’ est méchant le prince?

    Léa looks up at the camera and then looks at a toy.Léa 177: pasøkø iiijedijebo meijepabo

    ‘pa(r)ce que {i i} {il est dit/il ‘because {? ?} {it’s said/he dit/j’ai dit} il est beau mais il says/ I said} he is handsome, est pas beau’ but he isn’t handsome’

    In turn Mother180, the mother refers to ‘Poucelina’s prince’ using a double dislocated construction, in turn Léa 158, the child reproduces the double dislocation used by her mother. In the next turns (Mother 182 and Léa 159) referential continuity is marked by the exclusive use of a pronoun, again both by the mother and the child. This could lead us to the conclusion that the child is simply adopting the adult’s model. However, this is not the case in Léa 177 where Léa continues with a pronoun even though her mother did not use it in her question. Returning to the first part of the excerpt, note that in turn Léa 158, there is a slight shift in perspectives. The child adds a predication about the prince, but this predication is in contrast with the one proposed by the mother: as French children usually do, Léa uses here ‘pas beau’ (‘not beautiful, not handsome’) as a synonym of ‘pas gentil’ (‘not nice’). Therefore it can also be considered as being in opposition with the representation the mother was giving of the character. In the third part of the dialogue the referent ‘le prince’ is reintroduced with a dislo-cation (a typical form for reintroductions) by the child and then continued with pronouns both by the child and the mother. The mother’s speech turn (Mother 205) illustrates the dif-ference between plain continuity (we haven’t found out why he was mean) and contrast marked by a new question to the child (why is the prince mean?). The child’s utterance is in continuity with this discursive representation and only the pronoun is used after that turn.

  • Salazar et al. 399

    The results concerning unmarked reference deserve a specific discussion. As far as types of continuity are concerned, unmarked reference is as frequent as clitic pronouns in the dif-ferent contexts, especially in the less advanced children in the cross-sectional data. Therefore before pronouns are mastered, topical continuity seems to be maintained by predications on referents that are implicit in the children’s productions (Tables 5a and 5b) but have been previously mentioned in the discourse, and particularly by the interlocutor. This corrobo-rates the results of the study on English by Hughes and Allen (2006), who consider that pragmatic discursive features of null arguments foreshadow the adult use of pronouns. In a similar vein, De Cat (2004b) suggests that subject omission does not correspond to a lack of pragmatic competence in young children. As a matter of fact, this type of utterance is set in the framework of joint attention episodes, which for Bruner (1978) are the prerequisite for the referential function. When pronouns emerge, children might already master two func-tions: (1) the deictic function, which corresponds in the gestural modality to pointing and in the linguistic modality to devices such as demonstratives; and (2) the topical continuity function, for which children produce predications about unmarked referents (with or with-out fillers) in joint attention episodes. The emergence of clitic pronouns can thus be consid-ered in terms of the well-described process by Slobin: ‘New forms first express old functions, and new functions are first expressed by old forms’ (Slobin, 1973, p. 184). Third person clitic pronouns might not initiate the ‘anaphoric’ function but actually become the privileged tool to convey a function that was already present in the children’s language uses.

    Overall, these findings suggest that children’s uses of pronouns reflect early prag-matic skills, which seem to be acquired during the period under study. Children seem to be able to differentiate moves in dialogue and to choose the adequate linguistic devices to express them. Even though differentiation with the use of other referential expressions does not seem to be established for every child at levels 1 or 2 , there are no cases where a greater number of pronouns are used for other types of continuity.

    The relation between the discourse of the child and the discourse of the interlocutor is not confined to a mere referential coincidence. On the one hand, children master topical continuity before the use of pronouns (Bruner, 1978; Ochs, Schieffelin, & Platt, 1979; Veneziano, 2000) and, on the other hand, the present findings show that when they start to use pronouns, they tend to contrast them with other referential expressions. Hence, we can consider that third person clitic pronouns have the value of marking topical continu-ity in a shared discursive representation. At a more general level, the results suggest that in the process of acquiring morphological devices, children do not first acquire their strict grammatical values and add their discursive value later on. On the contrary, the discursive values are associated with the grammatical level from the onset.

    These findings confirm the relevance of a pragmatic study of the emergence of pro-nouns that takes the dialogical conditions of their acquisition into account. Nevertheless, dialogical conditions must not be understood as mere ‘copying’ of adult productions on the part of the child. Our data suggest that children’s productions are predominantly anchored in an intersubjective space shared in dialogue. Indeed, most referents mentioned by the children were first mentioned by the adult. In a Bakhtinian approach to language (Bakhtin, 1986), this means that children use the interlocutor’s discourse in order to elab-orate their own. This could be the framework for two types of concurrent phenomena that could be explored in further research. First, mediation: adults mediate the construction of

  • 400 First Language 30(3-4)

    longer and more continuous sequences in their dialogues with children; and second, an ‘integration of interactional episodes’ (Veneziano, 2000, p. 254, our translation). Children, in their communicative experience, grasp the contrasts marked by adults in their choice of referential expressions – and in particular the fact that adults use pronouns in the context of plain continuity, and nouns and dislocations for contrastive sequences.

    Acknowledgements

    We would like to thank Edy Veneziano, Kevin Durkin and the three anonymous reviewers for their thorough reading of a first version of this article and their stimulating comments. We would also like to thank Cristina Corlateanu and Gwendoline Fox for their collaboration.

    Notes1 We have excluded all first and second person reference as well as metalinguistic utterances.2 The only three cases of tonic third person pronouns appear in dislocations.3 We present an interpretation of the examples in French and a translation in English along

    with the phonetic transcription. Braces indicate ambiguities and uncertain transcriptions, translations or interpretations. A slash separates alternative interpretations.

    4 We use F to indicate the probable presence of a filler syllable.5 Contrary to English, German or other languages (Allen, 2000; Givón, 1995; Hickmann,

    2002), in French null forms are not a syntactic choice for subject function (except when two predications are coordinated).

    6 The intonation is clearly exclamatory here, which leads us to interpret this [a] as an interjection and not as a filler.

    7 This issue has been treated for all functions in Salazar Orvig et al. (2010).8 Note that in cross-sectional observations, the observer left the room during the recording.9 Appendix available at http://fla.sagepub.com/content/30/3-4/375/suppl/DC1

    References

    Allen, S. E. M. (2000). A discourse-pragmatic explanation for argument representation in child Inuktitut. Linguistics, 38, 483–521.

    Allen, S. E. M., & Schöder, H. (2003). Preferred argument structure in early Inuktitut spontane-ous speech data. In J. W. Du Bois, L. E. Kumpf, & W. J. Ashby (Eds.), Preferred argument structure: Grammar as architecture for function (pp. 301–338). Amsterdam: John Benjamins.

    Apothéloz, D. (1995). Rôle et fonctionnement de l’anaphore dans la dynamique textuelle. Genève: Droz.

    Ariel, M. (1988). Referring and accessibility. Journal of Linguistics, 24, 65–87.Bakhtin, M. (1986). Speech genres and other late essays. Austin: University of Texas Press.Bamberg, M. (1986). A functional approach to the acquisition of anaphoric relationships.

    Linguistics, 24, 227–284.Bennett-Kastor, T. (1983). Noun phrases and coherence in child narratives. Journal of Child

    Language, 10, 135–149.Blanche-Benveniste, C., Bilger, M., Rouget, C., & Van den Eyden, K. (1990). Le français parlé.

    Etudes grammaticales. Paris: CNRS Editions.Brown, R. (1973). A first language: The early stages. Cambridge, MA: Harvard University Press.

  • Salazar et al. 401

    Bruner, J. S. (1978). From communication to language: A psychological perspective. In I. Marková (Ed.), The social context of language (pp. 17–48). Chichester: John Wiley.

    Campbell, A. L., Brooks, P., & Tomasello, M. (2000). Factors affecting young children’s use of pronouns as referring expressions. Journal of Speech, Language and Hearing Research, 43, 1337–1349.

    Clancy, P. (2003). The lexicon in interaction: Developmental origins of preferred argument structure. In J. W. Du Bois, L. E. Kumpf, & W. J. Ashby (Eds.), Preferred argument structure: Grammar as architecture for function (pp. 81–108). Amsterdam: John Benjamins.

    Cornish, F. (1999). Anaphora, discourse and understanding. New York: Oxford University Press.De Cat, C. (2004a). A fresh look at how young children encode new referents. International Review

    of Applied Linguistics, 42, 111–127.De Cat, C. (2004b). Early ‘pragmatic’ competence and the null subject phenomenon. In R. Bok-Bennema,

    B. Hollebrandse, B. Kampers-Manhe, & P. Sleeman (Eds.), Romance language and linguistic theory 2002. Selected papers from Going Romance 2002 (pp. 17–32). Amsterdam: John Benjamins.

    De Cat, C. (2005). French subject clitics are not agreement markers. Lingua, 115, 1195–1219.Dik, S. C. (1997). The theory of functional grammar. Part 2: Complex and derived constructions.

    Berlin and New York: Mouton de Gruyter.François, D. (1981). Syntaxe du français et pédagogie. Rapport de synthèse du groupe de recher-

    che sur la syntaxe de l’oral. Paris: Université René Descartes.Givón, T. (1995). Functionnalism and grammar. Amsterdam: John Benjamins.Hamann, C., Rizzi, L., & Frauenfelder, U. H. (1996). On the acquisition of subjects and object clitics

    in French. In H. Clahsen (Ed.), Generative perspectives on language acquisition (pp. 309–334). Amsterdam and Philadelphia: John Benjamins.

    Hickmann, M. (2002). Children’s discourse: Person, time and space across languages. Cambridge: Cambridge University Press.

    Hugues, M., & Allen, S. (2006). A discourse-pragmatic analysis of subject omission in child English. In D. Bamman, D. T. Magnitskaia, & C. Zaller (Eds.), Proceedings of the 30th Annual Boston University Conference on Language Development (pp. 293–304). Somerville, MA: Cascadilla Press.

    Jakubowicz, C., & Rigaut, C. (1997). L’acquisition des clitiques nominatifs en français. In A. Zribi-Hertz (Ed.), Les pronoms. Morphologie, syntaxe et typologie (pp. 57–99). Saint-Denis: Presses Universitaires de Vincennes.

    Jakubowicz, C., & Rigaut, C. (2000). L’acquisition des clitiques nominatifs et des clitiques objets en français. Canadian Journal of Linguistics, 45, 119–157.

    Jeanjean, C. (1980). L’organisation des formes sujet en français de conversation: Etude quantita-tive et grammaticale de deux corpus. Recherches sur le français parlé, 3, 99–134.

    Karmiloff-Smith, A. (1985). Language and cognitive processes from a developmental perspective. Language and Cognitive Processes, 1, 61–85.

    Levy, E. T. (1989). Monologue as development of the text-forming function of language. In K. Nelson (Ed.), Narratives from the crib (pp. 123–161). Cambridge, MA and London: Harvard University Press.

    Levy, E. T. (1999). A social-pragmatic account of the development of planned discourse. Human Development, 42, 225–246.

    McTear, M. (1985). Children’s conversation. Oxford and New York: Basil Blackwell.Martinet, A. (1979). Grammaire fonctionnelle du français. Paris: Didier.Matthews, D., Lieven, E. V. M., Theakston, A. L., & Tomasello, M. (2006). The effect of percep-

    tual availability and prior discourse on young children’s use of referring expressions. Applied Psycholinguistics, 27, 403–422.

  • 402 First Language 30(3-4)

    Ninio, A., & Snow, C. (1996). Pragmatic development. Boulder, CO: Westview Press.Ochs, E., Schieffelin, B. B., & Platt, M. L. (1979). Propositions across utterances and speakers.

    In E. Ochs & B. Schieffelin (Eds.), Developmental pragmatics (pp. 251–268). Cambridge: Cambridge University Press.

    Parisse, C., & Le Normand, M.-T. (2000). How children build their morphosyntax: The case of French. Journal of Child Language, 27, 267–292.

    Peters, A. M. (2000). Filler syllables: What is their status in emerging grammar? Journal of Child Language, 28, 229–242.

    Peterson, C., & Dodsworth, P. (1991). A longitudinal analysis of young children’s cohesion and noun specification in narratives. Journal of Child Language, 18, 397–415.

    Pickering, M., & Garrod, S. (2004). Toward a mechanistic psychology of dialogue. Behavioural and Brain Sciences, 27, 169–190.

    Prince, E. (1981). Towards a taxonomy of given-new information. In P. Cole (Ed.), Radical prag-matics (pp. 223–255). New York: Academic Press.

    Salazar Orvig, A. (2002). Remarques sur la continuité et l’engagement dialogique chez le jeune enfant. In L. Danon-Boileau, C. Hudelot, & A. Salazar Orvig (Eds.), Usages du langage chez l’enfant (pp. 61–81). Paris: Ophrys.

    Salazar Orvig, A. (2003). L’inscription dialogique du jeune enfant: Evolution, diversité et hétérogénéité. Tranel, 38–39, 7–24.

    Salazar Orvig, A., Fayolle, V., Hasan, R., Leber-Marin, J., Marcos, H., Morgenstern, A., et al. (2004). Emergence des marqueurs anaphoriques avant 3 ans: Le cas des pronoms de troisième personne. Calap. La Cohésion Chez L’enfant, 24, 57–82.

    Salazar Orvig, A., Hasan, R., Leber-Marin, J., Marcos, H., Morgenstern, A., & Pares, J. (2006). Peut-on parler d’anaphore chez le jeune enfant? Langages, 163, 10–24.

    Salazar Orvig, A., Marcos, H., Morgenstern, A., Hassan, R., Leber-Marin, J., & Parès, J. (2010). Dialogical beginnings of anaphora: The use of third person pronouns before the age of 3. Journal of Pragmatics, 42, 1842–1865.

    Scollon, R. (1979). A real early stage: An unzippered condensation of a dissertation on child language. In E. Ochs & B. Schieffelin (Eds.), Developmental pragmatics (pp. 215–227). New York: Academic Press.

    Serratice, L. (2005). The role of discourse pragmatics in the acquisition of subjects in Italian. Applied Psycholinguistics, 26, 437–462.

    Slobin, D. I. (1973). Cognitive prerequisites for the development of grammar. In C. A. Ferguson & D. I. Slobin (Eds.), Studies of child language development (pp. 175–209). New York: Holt, Rinehart & Winston.

    Tomasello, M. (1999). The cultural origins of human cognition. Cambridge, MA and London: Harvard University Press.

    Veneziano, E. (2000). Interaction, conversation et acquisition du langage dans les trois premières années. In M. Kail & M. Fayol (Eds.), L’acquisition du langage. Vol. I. Le langage en émergence. De la naissance à trois ans (pp. 231–265). Paris: PUF.

    Veneziano, E. (2003). The emergence of noun and verb categories in the acquisition of French. Psychology of Language and Communication, 7, 23–36.

    Veneziano, E., & Sinclair, H. (2000). The changing status of ‘filler syllables’ on the way to grammatical morphemes. Journal of Child Language, 27, 461–500.

    Wittek, A., & Tomasello, M. (2005). Young children’s sensitivity to listener knowledge and per-ceptual context in choosing referring expressions. Applied Psycholinguistics, 26, 541–558.