An acoustic study of the /i:/-/ɪ/ contrast in the 431 French learners of English in the PARAAF corpus

Abstract

The PARAAF (Perception de l’Anglais et Reconnaissance Automatique d’Accents à la Fac) corpuswas created at the Université Paris Cité in November and December 2024. It consists in read speechrecorded by 431 students from the Licence LLCER ´Etudes anglophones.The corpus includes sentences and word lists in English and in French that were chosen tostudy different variables. The sentences were created to focus on the differences between BritishEnglish and American English. The word lists include the /hVd/ words already present in theexisting literature on English vowels. Each word list was repeated three times while the sentenceswere only recorded once. Half the participants started with the English part of the corpus and theothers read the French part first.The metadata contains the participants’ sexes, their birthdates, their regions of origin, their agewhen they started learning English, the languages they speak, their university level, the languagein which they started the recording, which recording booth they were in, and the names of theexperimenters who took care of them.After giving a summary of the metadata, we shall present a study of how French learnerspronounce the non-native contrast /i:/-/ɪ/. We measured vowel duration as well as vowel overlapwith Pillai scores in order to analyse their acquisition of L2 phonological categories. Results showedthat the participants did not produce a significant duration difference between /i:/ and /ɪ/, and thattheir realisations of these vowels tended to overlap in the acoustic space. We found varying degreesof acquisition as Pillai scores range from nearly complete overlap to nearly perfect separation,and duration differences from low to very high (nearly exaggerated when compared to nativemeasurements). The implications of these results will be discussed further in our presentation.