Biography

I am joining the DIETS project (“Automatic diagnosis of errors in end-to-end speech transcription systems based on their reception by users”), led by Richard Dufour (LS2N Nantes), as a postdoctoral fellow for a period of 9 months (March – December 2024). This project has three main strands dedicated to transcription errors in automatic speech recognition (ASR) systems: (i) evaluation and limitation of ASR errors using NLP tools (evaluation metrics, rescoring, neural networks); (ii) psycholinguistic or sociolinguistic evaluation of these errors (survey of speakers on the interpretability and acceptability of these errors); and (iii) categorisation of these errors from a purely linguistic point of view (semantics, morphosyntax). These aspects are linked, with the categorisation of errors being used in particular to draw up surveys on their assessment by speakers, and the assessment of these errors being used in particular to prioritise the improvements to be made on the computing level.

With a background in linguistics, I’m now moving into automatic methods for analysing language and speech. Mentored by Jane Wottawa (LST), I’m mainly involved in parts (ii) and (iii) of the DIETS project. On the research side, I specialise in semantics and prosody. I have specialised in theoretical linguistics (structural and interpretative semantics, analogical linguistics). I am the author of the thesis “Analyse sémantique de la prosodie imitative” (Sorbonne, 2022). As far as teaching is concerned, I taught linguistics at the Sorbonne and at Rennes 2 from 2015 to 2023. My teaching focused mainly on general linguistics, morphosyntax, the approach to enunciation, the remediation of phonetic and morphosyntactic errors in FLE, and grammar for teaching competitions.