Contenu du sommaire : Processus d'écriture et marques linguistiques, sous la direction de Irène Fenoglio et Sabine Boucheron-Pétillon

This paper sets out the linguistically relevant features of a particular kind of writing, the outline, or ébauche as Zola calls it, which is not yet a text but which is supposed to prepare for it: this function implies many self-injunctions, evaluations, interrogations. The example that is being analysed, the outline of Zola's Bête humaine, suggests several explanatory hypotheses: an "externalised inner speech", an interlocution between a writing-I and a reading-I, and a speech act through which the "I" generates a program for a world about to be created through writing. The so-called "performative" hypothesis subsumes these three approaches. According to it, the outline is a performative act, giving a tangible form to an inner discourse, involving two agencies (the writer and the reader) mirroring the splitting of the subject. At the end of the paper, this analysis of the outline is extended to cover every kind of avant-texte.

Michel Leiris often refers to the "parasitic formations" wich "proliferate" in his texts and wich, according to him, play only "a secondary role". This article studies the genesis of one of them in an excerpt from À cor et à cri. It shows that once the central anecdote has been told, Michel Leiris' writing work consists mainly in elaborating on the long digression situated in the first part of the passage. Leiris systematically introduces new elements as if this anecdote was only the basis of a digression that would constitute the real narrative in the text - the exhilarating search for all possible cross-words and junctions between words.

The study of Ponge's drafts for the poems collected in Cinq Sapates confirms my initial hypothesis, which considers the tittle - as a rule a noun phrase whose kernel is the name of the objet topicalized - as the initial cell of the text. Drafts tittles besides usual functions of tittle program genetic process and punctuate its continuity. Studying the selection of specifiers assigned to nouns naming the objet topicalized I find evidence for an alternation between singularity and genericity, realism and fiction, referenciation and autotelism. Eponym name is a text matrix invested by cratylean mimography, mimophony and what we would call mimotextualization.

The paper aims to find suitable basic theoretical assumptions which could underly the analysis of successive manuscripts as traces of text genesis. The major linguistic theories in the twentieth century fail to account simulaneously for verbal production, for relevant temporal properties and for specific textual dimensions. They are unable to propound dynamical forms for the succession of writing and rewriting stages and, for each of these stages, to separate (and link) mental and verbal processes, prospecctive and executive moments, as well as anticipating and memorizing ones. The complexity of the relations between these activities can lead to conflicts between them and to lacks of synchronization. Neuropsycholinguistics can offer some interesting parameters to deal with this problem.