Le statut grammatical du mot de dans ses emplois partitifs (article ou préposition) est un sujet fort discuté en grammaire française.

L'objectif de cette étude est de montrer que le problème du statut grammatical du mot de dépasse le cadre de ses emplois partitifs, que ce sont en fait tous les emplois de de qui sont concernés.

Un rapide inventaire des différentes valeurs sémantiques possibles pour de, dans toute l'histoire du français, débouche sur une répartition des emplois répertoriés dans différentes catégories grammaticales.

Cette répartition révèle les limites du système traditionnel des parties du discours. Des propositions récentes, dues tant à la grammaire générative (catégorie du complémenteur) qu'à la psychomécanique du langage (catégorie de l'inverseur) sont alors discutées.

An experiment is described in which 74 French subjects were asked to list colour terms within a limited time period. Results are compared with those from a similar experiment with Russian subjects. The resulting data are used as a measure of psychological salience to establish which French colour terms are basic. Particular attention is paid to the two possible French terms for brown (marron and brun), and to beige ‘beige’, which may be emerging as an additional basic term. As Russian also has an additional term for blue, both languages are particularly interesting in the light of Berlin and Kay's 1969 model of basic colour term development.

There is currently no generally accepted theory of metre in French such as the one available for English. Students of French verse-texts look to linguists in vain for a clear decision-procedure for the metrical description of French verse. This essay attempts to derive such a decision-procedure from a historical testing of the hypothesis that the word in French has a prosodic identity and that metre in French can be seen as a structure which orders that identity.

Changing patterns of assonance in early Old French texts have traditionally provided the basis for the claim that vowel height has a universally predictable effect on the development of distinctive nasalization over time. However, the results of recent studies cannot be accounted for by such a hypothesis. Furthermore, an alternative hypothesis suggests that the same patterns of assonance reflect modifications in vowel quality rather than nasality. Additional cross-linguistic and phonetic data examined here further undermine the purportedly universal character of the development of distinctive nasalization in Old French.

This article reports the results of three studies dealing with dislocated sequences in French. A text count of written and spoken French reveals that word order is not entirely free in these sequences but instead obeys certain constraints. Two studies then investigate the use of cues (word order, clitic pronoun type, clitic pronoun agreement and animacy) in the interpretation of dislocated sequences by native speakers of French. Results indicate that the constraints found in the text count, with minor modifications, are also operative in processing.

We consider recent Government-Binding work on sentential negation, e.g. by Pollock, and evaluate a fundamental assumption made about the syntax of negative clauses. While accepting that ne is generated as the head of NegP, we reject the dual claim that pas is characteristically: (a) a maximal projection, and (b) base-generated as the specifier of ne. We offer a three-sided argument against such an analysis, invoking: (a) the incompatibility of the proposal with the status of pas as a nominal; (b) the interaction between pas, etc, and indefinite direct objects; and (c) the syntax of ‘adverbials’. We go on to consider Obenauer's work on ‘quantification at a distance’ and Battye's work on ‘nominal quantification’. On the basis of this work, we posit that pas is generated lower in clause structure, either VP-adjoined or as the head of a determiner-less direct object DP.

Old and Middle French interrogative sentences of the type sez tu pas? have often been interpreted, erroneously, as providing evidence for the early dropping of ne as in modern je sais pas. The construction occurs most widely in Old and early Middle French with point (occasionally pas, mie) and the particle had a positive or ‘potential’ value (e.g. Quenois le tu point? ‘Do you know him [at all]?’). It was only in late Middle French that, influenced by the increasing use of pas and point in negative clauses, the interrogative clauses in question came to have a negative value.

In the diachronic development of the modern Romance languages, reflexes of the Latin verb habere have replaced reflexes of esse, to a greater or lesser extent in particular languages. In this article it will be argued that the distribution of esse-reflexes is determined by a hierarchy of unaccusativity based on the semantic distinctions concreteness/abstracteness and movement/staticity, and that habere-reflexes have been spreading systematically from the periphery of this hierarchy towards the core. This process has affected Italian and French to different degrees: Italian has largely retained a syntactically and semantically consistent auxiliary system, whereas the French system, at a more advanced stage of the evolution, shows greater variation and inconsistency.

Evidence is presented from the linguistic intuitions of very advanced Italian non-native speakers of French and French non-native speakers of Italian about equivalent unaccusative verbs, and it is argued that the unaccusative hierarchy conditions both the degree and the directionality of difficulty in second language acquisition. The findings reveal an asymmetric pattern: it is easier for the French learners to fully acquire the facts of essere-selection than for the Italian learners to internalize the facts of être-selection, and the degree of difficulty experienced by the Italian learners with individual unaccusative verbs is correlated to their position along the hierarchy.

Les langues offrent souvent des exemples d'évolution phonique qui semblent, à tort, contredire le principe des changements phonétiques ‘vrais’, principe selon lequel ‘toutes les réalisations d'un phonème donné, dans un contexte donné, changent dans le même sens et à la même allure’. De plus, un même locuteur peut, d'un moment à l'autre, prononcer le même mot avec des phonèmes différents. Ce phénomène, connu sous le nom de fluctuations, n'affecte pas tous les monèmes dans lesquels figurent les phonèmes en question et on ne l'observe que lorsque se trouvent en contact constant des locuteurs aux systèmes partiellement différents, chaque locuteur adaptant inconsciemment sa prononciation, pour certains mots, à celle de ses interlocuteurs. Ce type de changement est en fait propagé par emprunt de formes lexicales à un système voisin. Lorsqu'on peut étudier ces changements sur le vif au moment même où ils se produisent, il s'avère particulièrement important de distinguer entre les évolutions que l'on peut décrire uniquement en termes phoniques, indépendamment de toute référence grammaticale ou lexicale, et les changements qui se font ‘mot par mot’, qui touchent donc au lexique et à la grammaire et qui connaissent des périodes de fluctuations, avant de se fixer par convergence des usages.

Le but de cet article est de donner un panorama synthétique d'un certain nombre de concepts et d'idées qui caractérisent la phonologie générative moderne et d'examiner quelques applications à l'étude du français. Derrière le foisonnement de la terminologie et des modèles, il y a quelques idées clefs qui sont partagées par de nombreux chercheurs qui travaillent dans ce qu'on appelle souvent ‘phonologie non-linéaire’, ‘phonologie multilinéaire’ ou ‘phonologie multidimensionnelle’. On partira ici de l'approche de Chomsky et Halle dans le Sound Pattern of English (1968), illustré dans de nombreux travaux sur le français, pour déboucher sur des cadres tels que la Phonologie du Gouvernement, la Phonologie de Dépendance ou la Phonologie Moraïque. On verra alors qu'à un système bâti sur des structures relativement pauvres (séquences de traits simultanés et de frontières) que les règles phonologiques manipulent pour produire des formes de surface, on a substitué des approches plus ‘configurationnelles’ où les règles sont écartées en faveur de ‘principes’ et de ‘paramètres’.