This article explores an unexpected extraction pattern out of adjuncts, as in What did John die whistling? and What did John hurt himself trying to fix? in English. This type of extraction is possible if the adjuncts are predicative and the matrix verbs are either ergative or (inherent) reflexive. The predicativity of the adjuncts is essential in getting these adjuncts interpreted as arguments; this requires coindexation with a thematic role of the verb, which is shown to obtain. Once transparent adjuncts are integrated into the thematic structure of the clause, the resulting representation exhibits the hallmarks of reflexivity. Reflexive predicates must receive appropriate marking. This explains the second requirement on transparent adjuncts: the matrix verbs are precisely of the type capable of licensing reflexivity.

This article argues that the Impersonal Contruction (IC) of French, which displays unaccusative syntax but allows unergative verbs, refutes the Unaccusative Hypothesis (UH) and the principles of lexically driven mapping. It is demonstrated that three types of argument intended to explain unaccusative mismatches are untenable while maintaining the UH. An account of the verbs appearing in the IC is proposed, based on the principles of free linking of arguments and postsyntactic compositional interpretation, taking into account meaning contributed by diverse sources, including lexical, syntactic, constructional, and morphological meaning, which must be compatible. It is argued that the shape of IC sentences requires that a state of affairs be predicated of a locative; various consequences for verb and argument selection and tense arise from this requirement. By removing certain semantico-syntactic functions from the lexicon, economy is achieved in both the lexicon and the syntactic component.

The present study compares relative clause production in French learners and in mature speakers. Previous studies on French postulate deficiencies in children’s grammar of relatives in order to explain their defective production (Labelle 1990; Guasti and Shlonsky 1995). However, these studies do not strictly compare child and adult data. A new production task was administered to 25 French learners ages 4 to 7, and to 25 adults. Three types of stimuli were presented in order to elicit subject, direct object and indirect object relatives. Results show that children’s performance varies according to the type of stimuli presented, and that the performance of adults is qualitatively similar. These patterns undermine the notion that a given component of UG is either inoperative or unspecified in the developing grammar. Such unsteady lacunae point more naturally to peripheral processing limitations in learners rather than core deficiencies within their grammar proper (cf. Goodluck and Tavakolian 1982).

We report a study that tests children’s knowledge of an effect of Principle C of the binding theory: In the adult grammar of English and French, coreference between a main clause object pronoun and a non-pronominal subject of a sentence-final temporal clause is permitted, whereas coreference between a subject pronoun and the subject of a temporal clause is blocked. In an act-out task, both French-speaking adults and children aged 3–7 were found to be sensitive to the position of a main clause pronoun (subject vs object) in selecting a referent for the subject of a temporal clause, permitting coreference more frequently when the pronoun was in object position. This result replicates earlier work done on English. A sentence judgement task produced clear results only for adults. Results from the act-out suggest that children are relatively inept at integrating non-mentioned participants into their interpretation of sentences. We suggest that children’s knowledge of the principle C effect we tested constitutes a “poverty of the stimulus” argument for innateness.

In this article, a system of categorizing the c’esf-cleft into different types is developed, based on their pragmatic, syntactic and prosodic properties. This goal has already been accomplished in English linguistic studies for a similar construction, the it-cleft. The c’est-cleft, however, is found more frequently than the it-cleft, and in many contexts, it is obligatory. In general, in the relative clause of the c’est-cleft, there is an open proposition that is saturated through replacing the missing variable by the element that is found in post-copular position. In most cases, the material found in the relative clause is presupposed; however, there are some cleft types for which this is not necessarily the case. This study is based on Lambrecht’s study of Information Structure.

This article concerns the distribution of clitics in the child root infinitives in French and in Dutch. In the material studied, subject clitics are absent both from French and Dutch root infinitives; object clitics are present in French root infinitives while remaining absent from Dutch root infinitives. It is proposed that this cross-linguistic difference is related to the difference in the licensing site of the object clitics, which is taken to be AgrO in French and AgrS in Dutch. The hypothesis thus casts doubt on proposals according to which clitics are universally base-generated in specialised functional positions in the structure (Sportiche 1996; Schaeffer 1997).

Many questions about the expressive or distinctive character of focal accent and emphatic prominence remain unanswered in studies concerning natural language prosody. Recent works show that fundamental frequency variation underlies a categorial perception that permits a differenciation in the representation of these two types of prominence. Using data from French, the contribution of each type of prominence to the intonation contour is analyzed. This study reconciles the difference in expressive character and phonological description associated with each prominence type through the use of association rules linking tones and text in different prosodic domains. An instrumental study of 280 utterances documents the phonetic particularities in the production of both prominences, and in neutral utterances.

In this article, the acquisition of French subject and object pronouns by Anouk, a French-Dutch bilingual girl is studied and compared with the acquisition by monolingual French children. At first sight there are no major differences between Anouk and the children discussed in the literature. There are however some indications that the status of the pronouns in Anouk’s data is not the same as that in the monolingual data. In Anouk’s data, there are no utterances where a quantified nominal subject is doubled by a subject clitic. Consequently, it is impossible to argue that her subject pronouns have the status of agreement markers which they are claimed to have in the case of monolingual children. Moreover, Anouk acquires both subject and object pronouns at roughly the same moment—to what is found for the monolinguals. Finally, Anouk makes position errors with object pronouns which are similar to those found in French L2 acquisition data. It is proposed that this may be due to the (indirect) influence of Dutch, her other language.

This article presents and discusses data on nominative and object clitics used by twelve monolingual French-speaking children aged 2;0 to 2;7 years in a spontaneous interaction setting and in an elicited production task. It is shown that nominative clitics surpass object clitics, and that reflexive clitics fare better than accusative clitics. It is argued that these two dissociations are compatible with the computational complexity hypothesis put forth by Jakubowicz and Nash (to appear), applied to the analysis of third person Romance pronominal clitics proposed by Jakubowicz, Nash, Rigaut, and Gérard (1998).

This article is mostly devoted to a critical review of the main analyses of root infinitives. Confronting these analyses to acquisition data from French, it is shown that an analysis in terms of a “null auxiliary” is more compatible with the data than the alternative hypotheses. However, an analysis in terms of a null or deleted auxiliary raises some difficult questions with respect to the existence and the nature of null elements in child language. Rather than appealing to null or deleted elements, it seems more appropriate to assume that children utterances do not include an auxiliary or a modal and that the corresponding node is simply not projected. An analysis along these lines is developed in the last section of the article.