Catégorie d'événements : Séminaire

Cet évènement fait partie des séminaires du LPL ayant lieu tous les 1ers vendredis du mois 2025/2026.

[Détails à venir]

Vendredi 7 novembre 2025

Séminaire de l'équipe VESPA

Mélanie Lancien

(Laboratoire ATILF Nancy)

Le rôle de la réduction phonétique dans l'expression de la proximité sociale

A 10h30 en salle de conférences B011 au LPL, Aix-en-Provence

Evènement ouvert à tout le monde

Résumé :
Que révèle la manière dont nous articulons nos voyelles sur la relation que nous entretenons avec notre interlocuteur ? Cette recherche s’attache à répondre à cette question à partir de productions de locuteurs et locutrices du français québécois, en mobilisant les cadres théoriques de l’hyper-hypoarticulation et de l’accommodation. Notre corpus comprend 140 436 occurrences vocaliques issues de 20 participants, enregistrés dans des contextes variés : lecture de mots isolés, lecture en phrase porteuse, et réalisation d’une tâche de discrimination d’images, soit en solitaire, soit en interaction avec leur partenaire, un enquêteur québécois inconnu ou une enquêtrice française inconnue.
L’analyse de quatre mesures spectrales et d’une mesure temporelle révèle une hiérarchie claire du degré d’hypoarticulation : interaction < tâche en solitaire < lecture, ainsi que couple < interlocuteur inconnu local ≤ enquêtrice française. Par ailleurs, l’examen de quatre indices rythmiques indique un ralentissement de la parole lorsque la distance sociale entre interlocuteurs augmente et dans les situations les moins interactives.
Ces résultats suggèrent que l’hypoarticulation constitue un marqueur potentiel de la distance sociale, reflétant une adaptation fine au type d’interlocuteur et au contexte communicatif.

Séminaire de l’équipe LSD

Mardi 16 septembre 2025 de 12h à 14h
Laboratoire Parole et Langage, salle de cours A003

Elise Lefèvre

Application of network analyses to reveal the cognitive architecture underlying reading ability

Résumé :
This presentation illustrates an application of network analyses to explore the complex structure of cognitive abilities related to reading acquisition and dyslexia, combining confirmatory and exploratory approaches.
We will present studies using structural equation modelling (SEM), including mediation models and longitudinal analyses, to test predefined hypotheses about causal relationships between key cognitive skills. Additionally, we will introduce an innovative approach based on graph theory to explore the structure of the cognitive networks without à priori hypothesis. The latter method complements the SEM by revealing complex and unexplored interconnections between different networks. Together, the combination of the rigour of the confirmatory approach and the flexibility of the exploratory approach allows us to provide a richer and more nuanced understanding of typical and atypical reading development.
Following the presentation, further discussions on the application of the network analyses on other research topics will be welcome.

Séminaire inter-équipe S2S-LSD

Vendredi 12 septembre 2025 de 12h à 14h
Laboratoire Parole et Langage, salle de conférences B011 & en ligne

Accès en ligne par Zoom

Sheng-Fu Wang

Modeling durational variability in Taiwanese with surprisal from neural language models

Résumé :
Using variability in syllable duration in Taiwanese spontaneous speech as a case study, this study aims to demonstrate that probabilistic measurements from neural language models can be used not only to improve the modeling of phonetic variability but also configured in different ways to test hypotheses about speech planning. The focus on probability (surprisal) and phonetic variability arises from decades of research showing that acoustic properties of speech sounds have been shown to correlate with their contextual probability — the likelihood of occurring in specific linguistic contexts (Seyfarth, 2014; Van Son & Van Santen, 2005; Jurafsky et al., 2001), which is often interpreted as listener-oriented speech planning that aims to balance information distribution within and across utterances (e.g., Aylett & Turk, 2004). Phonetic studies in this area often have the implicit assumption that it is local contexts that are relevant, based on their employment of simple n-gram (and mostly bigram) models, despite evidence of nonlocal speech planning (e.g., Fuchs et al., 2013; Krivokapic, 2012). Neural language models, with their ability to easily take longer contexts into account when estimating probability, open up opportunities to test this locality assumption that was potentially a byproduct of methodological restriction. Employing bidirectional Transformer-based masked language models (RoBERTa, Liu et al., 2019), this study seeks to answer questions on the relevance of longer contexts and the potential role of linguistic chunking.

Page Web : https://sheng-fu.github.io/

Séminaire de

Shelley Xiuli Tong

Speech, Language, and Reading Lab, Faculty of Education, The University of Hong Kong

On the Primacy of Prosody: Perception, Production, and Their Connections with Reading Comprehension Across Languages

Jeudi 12 juin 2025 de 10h à 11h, en salle de conférences B011 au LPL et à distance par Zoom

Lien Zoom : https://univ-amu-fr.zoom.us/j/91072651899?pwd=mxadrKlb11b4gdxRkLEB2QIAHaRaam.1

Suivi d’une table ronde (11h-12h) avec la participation de Cristel Portes et Chotiga Pattamadilok du LPL.

Résumé du séminaire :
When reading stories to children, adults often employ prosodic reading, using animated and expressive tones to act out the characters. This technique is not only entertaining, but plays a crucial role in enhancing comprehension. In this talk, I will discuss the findings from a decade of research conducted by my lab on speech prosody and its impacts on reading comprehension by addressing three questions. First, what are the acoustic-phonetic and lexical factors influencing linguistic pitch perception across languages? Second, how does prosodic sensitivity, or perceptual acuity to acoustic–phonetic cues associated with prosodic structure, contribute to reading comprehension? And third, to what extent does prosodic reading, characterized by pitch and pause patterns when reading aloud a connected text, predict reading comprehension? The answers to these questions support a new theoretical framework, the prosodic catalyzing hypothesis (Tong et al., 2024), which suggests that prosodic reading integrates lexical, syntactic and semantic representations of texts to facilitate comprehension.

Evènement ouvert à tout le monde.

Bio de S. X. Tong :
Shelley Xiuli Tong, Ph.D., is a Full Professor at the University of Hong Kong’s Faculty of Education where she directs the Speech, Language, and Reading Lab. Recognized as an RGC Research Fellow and Fulbright Senior Scholar, her research, which has been funded by the U.S. National Academy of Education, the UK ESRC- HK RGC joint grant scheme and the Hong Kong Research Grants Council, focuses on utilizing cognitive-behavioural, neurophysiological, and machine learning approaches to investigate the roles of prosodic reading in bilingual reading comprehension difficulties; the cognitive and neural mechanisms underlying statistical learning in children with dyslexia; and optimal solutions for classifying dyslexia, autism, and hearing-impairment. She has developed an intelligent dyslexic interface design (I-DID) that capitalizes on individual strengths of children with dyslexia and reflects her life-long commitment to transform scientific evidence into public policy and practice. Her work has resulted in over 80 publications in journals such as Child Development, Cognition, Neuroimage, Autism Research, Journal of Educational Psychology, Learning and Instruction, and Educational Psychology Review. She was an Associate Editor of Applied Psycholinguistic and currently serves as an Associate Editor for Scientific Studies of Reading.

LSD seminars on prediction and learning

Monday 12/05/2025, 9.30 to 17.00

LPL, room A003

Presentation

EN During this first event of a series of scientific days, the LSD team meets to present new findings and discuss hot topics. This first event will focus on the role of prediction in phonetic convergence and learning. Specifically, two talks will be followed by round tables in the morning and evening sessions.

All the lab members are welcome to attend and participate in the round tables.

FR Pendant ce premier évènement d’une série de journées scientifiques, l’équipe LSD se réunit pour présenter des résultats récents et discuter de la thématique prédiction et apprentissage. Cette première journée ce focalisera sur le role des prédictions dans la convergence phonétique et l’apprentissage. Deux exposés seront suivis de tables rondes dans les sessions du matin et de l’après-midi.

Tous les membres du laboratoire sont invités à assister et à participer aux tables rondes.

 Program

9.30-9.45 Elin Runnqvist & Clément François : Introduction / coffee
9.50-10.30 Leonardo Lancia : Disentangling the contribution of top-down and bottom-up processes to phonetic convergence
10.35-11.15 Noël Nguyen : Perceptual assessment of phonetic convergence between speakers: A meta-analysis
11.20-12.30 Round-table Predictions and predictive coding
12.30-14.00 Lunch
14.00-14.40 Clément François : Sensitivity to statistical regularities as a basic principle for learning
14.45-15.25 Antoni Rodriguez-Fornells : Electrophysiological evidence for the role of prediction during novel word learning
15.30-17.00 Round-table Linking predictions and learning