Formes réduites en conversation: caractéristiques des séquences et des locuteurs

Abstract

Dans le langage parlé, une partie non négligeable de mots sont produits avec des formes phonétiques lacunaires ou sous-spécifiées. Ce phénomène s'appelle la réduction phonétique. Cette étude vise à mieux comprendre son fonctionnement. Pour cela nous analysons 13 formes réduites (mots et combinaisons de mots) dans un corpus de conversations en français. En s'appuyant sur une Transcription Orthographique Enrichie pour l'extraction des items réduits, nous avons observé leur distribution et leur durée dans un corpus de parole conversationnelle. Les résultats montrent que la fréquence de réduction des items est variable et dépend de l'item lui-même mais aussi du locuteur. Les analyses post-hoc ont révélé une faible corrélation entre le taux de réduction de chaque locuteur et son débit dans le corpus. Enfin, une analyse sur 3 items nous montre que leur réduction est, là encore, dépendante des items et des locuteurs.

Vowel Reduction in Conversational Speech: the case of reductions in French

“…Other forms we detected in a previous study involve /fɛ / for "enfin" (finally), /ʃɥi/ for "je suis" (I am), and /tse/ for "tu sais" (you know) among others (Bodur et al, 2022).…”

“…In a previous study (Bodur et al, submitted) 1 , we established a distinction between two types of reductions in spontaneous speech: lexicalized and nonlexicalized reductions. Lexicalized reductions can be defined as reduced variants of a same lexicalized representation, affecting mostly the words/sequences that occur frequently in casual, informal speech (Bodur et al, 2022). We assume that these reductions are stored within the lexicon as reduced variants of lexical representations.…”

“…In addition to studying vowel reduction, we aim to establish a relationship between speaker-specific vowel metrics and the frequency of other types of reductions the speakers produced in their conversations which are defined in the Introduction. For this purpose, selecting 13 forms that are frequently encountered/perceived in reduced form in casual conversations (Bodur et al, 2022), we manually detected 2575 lexicalized reductions with a top-down approach. Regarding the extraction of non-lexicalized reductions, the collection process can be considered more challenging, as they do not regularly affect the same lexical forms and are the result of interplays between several linguistic levels (e.g., prosody, phonetics, pragmatics).…”

Exploring the Nuances of Reduction in Conversational Speech: Lexicalized and Non-Lexicalized Reductions

“…For this reason, we have established a distinction between two types of reductions occurring in casual speech which we will call lexicalized and non-lexicalized reductions from now on. Lexicalized forms can be described as different reduced examples of a same lexicalized representation, relating to words or sequences that occur frequently in everyday language (Bodur, Fredouille, & Meunier, 2022). These forms are more commonly studied in literature and are often recognizable by listeners and transcribers who have perceptible representations of several examples of these sequences (see Figure 1).…”