Michèle Kail

French children's ability to introduce referents in narratives as a function of mutual knowledge

This study examines referent introductions in narratives produced by French children of 6, 9, and 11 years in two situations where they either could or could not assume mutual knowledge. In one situation (MK) the children and their interlocutor were looking at a picture book together; in the other (NMK) the interlocutor was blindfolded. Children in all age groups used more indefinite determiners in Situation NMK than in MK. However, some clear developmental changes were also found: the 6-year-olds used both definite and indefinite determiners in NMK; the 9-year-olds selected indefinite determiners in NMK and definite ones in MK; the 11-year-olds produced indefinite determiners not only in NMK, but also in MK. These results show that children acquire the rules governing referent introductions in three steps: (a) they have no systematic rule in the absence of mutual knowledge; (b) they acquire a rule associating appropriately different linguistic devices with the presence versus absence of mutual knowledge; (c) they acquire an additional rule that generalizes indefinite forms to all narrative situations, regardless of mutual knowledge conditions.

Local and Topological Processing in Sentence Comprehension by French and Spanish Children.

This cross-linguistic study investigates the relative importance of validity in terms of the strengths of syntactic cues and cue processing cost in sentence comprehension by French and Spanish children (4; 6–6; 6). The notion of cue cost refers to the distinction between local and topological processing types. Choices of the agent (cue strength) and latencies (cue cost) were collected through the acting out of sentences containing different syntactic cues. These cues (word order, clitic pronoun, verbal agreement plus accusative preposition a in Spanish) are ordered on a continuum from the most topological (word order) to the most local (preposition a ). The analysis of cue strengths reveals that, while for French children a linguistic cue is all the stronger the more topological it is (verbal agreement a ). The fact that Spanish childrens latencies are always shorter (2150 msec) than those of French children (3110 msec) must be related to the effect of the preposition a which permits efficient role assignments with minimal cost. These results stress the importance of locality in sentence processing. On the other hand, a comparison with our similar adult cross-linguistic data demonstrates that the impact of cue cost changes over time.

Cohesive anaphoric relations in French children's narratives as a function of mutual knowledge

This study examines how French children of 6, 9, and 11 years use referring expressions for reference maintenance in narratives elicited in two situations: children and their interlocutor were looking at a picture book together (mutual knowledge) or the inter locutor was blindfolded (no mutual knowledge). Local coreference has a strong impact on the selection of pronouns (coreferential) vs. nominals (non-coreferential) at all ages and in both situations. However, children from 9 years on produce more pronouns in the absence of mutual knowledge and the extent to which children mark story structure varies as a function of age and situation. Regardless of situation, 6-year-olds mark boundaries across successive pictures (external structure) and episodes (internal structure) by means of nominals. Although a similar pattern can be observed at other ages in the mutual knowledge condition, it gradually disappears with increasing age in the absence of mutual knowledge. It is concluded that discourse-internal functions of referring expressions are a late development characterized by the increasing impact of coreference, which gradually overrides other factors, as children learn to rely maximally on discourse cohesive relations in the absence of mutual knowledge.

Verb agreement processing in French : A study of on-line grammaticality judgments

This study investigates the implementation of the syntactic constraint of verb agreement in French sentence processing. An experimental on-line error detection paradigm is used to examine how variations in linguistic material interposed between the subject and the verb affected the detection of number-verb agreement violations. Detection times mainly vary as a function of the linguistic structure of the interposed material: detection times are clearly longer for noun complements than for relative clauses and adverbial complements. Effect of length is shown to depend marginally on the structure type. Number mismatch between local and head nouns increases detection times, which also varies as a function of the morphology of the head noun. Results are analyzed with regard to processes found in agreement-error production studies. The role of structural, morphological, and semantic information in syntactic verb agreement processing is discussed within a crosslinguistic perspective.

Language-particular or language-independent factors in acquisition? Children's comprehension of object pronouns in Dutch, French and German

Childrens comprehension of direct and indirect object pronouns in Dutch, French and German was studied in order to determine the role of language-particular factors in acquisition. The pronoun systems of the three languages differ on a number of dimensions that have been assumed to influence the acquisition process: the phonological proper ties of pronouns (clitic vs. non-clitic); the position of the pronoun in the sentence and the phrasal category in which the pronoun occurs: prepositional ([ppPREP[NP[PRO]]]) vs non-prepositional ([NP[PRO]]). A picture-matching task was used. Three age groups, 4-, 5-, and 6-year-old children, were tested. The results indicate that language- particular properties, at least in the part of the grammar investigated here, do not play the role assumed in some theories of language development, but that more abstract, language-independent proper ties of the pronouns, such as their function as internal verbal argu ments, are relevant for acquisition. Those language-particular factors which were observed are secondary and restricted to limited domains.

On-line sentence processing in Swedish: cross-linguistic developmental comparisons with French

This study examined on-line processing of Swedish sentences in a grammaticality-judgement experiment within the framework of the Competition Model. Three age groups from 6 to 11 and an adult group were asked to detect grammatical violations as quickly as possible. Three factors concerning cue cost were studied: violation position (early vs. late), violation span (intraphrasal vs. interphrasal) and violation type (agreement vs. word order). Developmental results showed that children were always slower at detecting grammatical violations. Irrespective of age, participants were faster at judging sentences with late violations, especially in the younger groups. Intraphrasal violations were more rapidly detected than interphrasal ones, particularly in adults. Finally, agreement violations and word order ones did not differ. The hierarchy of cue cost factors indicated that violation span was the dominant one. A cross-linguistic analysis with French (Kail, 2004) underlines the developmental processing abilities and the interdependence between cue cost and cue validity.

A developmental cross-linguistic study of adversative connectives: French 'mais' and German 'aber/sondern.

This study concerns the acquisition of the meaning of adversative connectives in French and German children from 7; 8 to 9; 11. French mais has both a substitutive and a contrastive use which is expressed by two different connectives in German, i.e. sondern and aber . 36 French and 36 German children were tested in a completion and a judgement task. Two hypotheses are confirmed: ( a ) substitutive but is easier to process and hence is acquired earlier than contrastive but ; ( b ) the interpretation of contrastive but -sentences depends on their inferential complexity relative to a given context. A third assumption about the facilitative effect of lexical differentiation in German was only partially supported by the data. This issue needs further investigation with more languages and younger children.

Developmental Production of Utterances from a Series of Lexemes.

51 children from 4; 11 to 8; 2 were given three words (a triplet made up of two nouns and one verb) and asked to produce an utterance with them. Words in the triplet were presented in every possible order. At all ages, the dominant responses to non-reversible triplets were utterances in which the animate noun served as the subject of the verb and the inanimate noun as its object, whatever the order of words within the triplet. Production of utterances in response to reversible triplets showed a clear differentiation with age (with the exception of responses to symmetrical triplets NVN for which the SVO structure is dominant at all ages). With non-symmetrical triplets (NNV or VNN), the strategy that assigns the function of subject to the first noun becomes progressively established. In the youngest children, a significant number of responses that preserve the order of stimulus elements is observed. This strategy results in attributing an identical grammatical function to the two nouns. As anticipated, isomorphism between stimulus structure and response structure diminishes with age.

Sentence processing studies and linguistic literacy.

Among the various processes that characterize the development of literacy competence presented by Ravid & Tolchinsky in their paper, two are particularly relevant for studies of language development focusing on crosslinguistic sentence processing: (1) the continuity vs. discontinuity between the differential access to oral and written codes within the same language and between languages; (2) the representational status of the conscious access to language provided by the development of literacy, which renders variability more accessible and controllable. In this short note, I shall deal only with the first of these.

Chapter 15. Online sentence processing in simultaneous French/Swedish bilinguals

Within the Competition Model, cue validity and cue cost can serve to make predictions about real-time sentence processing in a cross-linguistic perspective. Previous research with monolingual children and adults in French (Kail 2004) and Swedish (Kail et al. 2012) proposed that cue cost is determined by contextual and structural information, word order and morphology. On the basis of online grammaticality judgments, we investigated whether these cue cost constraints are equally efficient and follow the same hierarchy in simultaneous French/Swedish bilinguals and in their monolingual counterparts. Although bilinguals were slower and less accurate, the weight of each cue cost component was similar for both groups. Bilinguals’ longer detection times are linked to specific interactions between cue cost components not observed in monolinguals. This result is compatible with the cognitive cost implied by the need to inhibit the non-relevant language during bilingual processing.