Catégorie d'actualités: Release

Learning to read: marked inequalities in overseas France

We are delighted to announce the latest article published today by Stéphanie Ducrot in collaboration with Laurie Persia-Leibnitz, Marie Vernet, Bruce Brossette, Chloé Prugnières and Jonathan Grainger in the International Journal of Educational Development.

This study reveals that “nearly half of the children in Martinique are at risk of reading delay, with errors three times more frequent than those of their peers in France.”

 The CNRS wanted to highlight these societal issues, and sent out a press alert to journalists in its network (in French): https://www.cnrs.fr/fr/presse/apprentissage-de-la-lecture-des-inegalites-marquees-en-outre-mer

Article reference: Stéphanie Ducrot, Laurie Persia-Leibnitz, Marie Vernet, Brice Brossette, Chloé Prugnières, Jonathan Grainger. Children in French overseas departments are at a 3-fold increased risk of developing reading problems. International Journal of Educational Development, 2025, 115, pp.103277. ⟨hal-05043884v1⟩

Full-text article:https://www.sciencedirect.com/science/article/pii/S0738059325000756

Credits: L. Persia-Leibnitz

Learning to read starts early: The benefits of premature exposure to the written word

We are pleased to share with you the latest article published by Stéphanie Ducrot (LPL) in collaboration with Jonathan Grainger (CRPN) in the Journal of Experimental Child Psychology:

Reference: Stéphanie Ducrot, Jonathan Grainger. The development of letter representations in preschool children is affected by visuomotor integration skills and visual field asymmetries. Journal of Experimental Child Psychology, 2025, 257, pp.106277. ⟨hal-05042015⟩

Abstract & full-text article: https://doi.org/10.1016/j.jecp.2025.106277

Credits: The authors

How does our brain process prosody?: Shedding light on the involvement of Heschl’s gyrus

A long-standing international collaboration between James S. German (LPL/amU) and Barath Chandrasekaran (University of Pittsburgh) has taken shape in a multidisciplinary study just published in Nature Communications:

Reference: G. Nike Gnanateja, Kyle Rupp, Fernando Llanos, Jasmine Hect, James S German, Tobias Teichert, Taylor J Abel, Bharath Chandrasekaran. Cortical processing of discrete prosodic patterns in continuous speech. Nature Communications, 2025, 16 (1), pp.1947.

Full-text article: https://www.nature.com/articles/s41467-025-56779-w
Link to HAL database: https://hal.science/hal-04973949v1
EurekAlert news release: https://www.eurekalert.org/news-releases/1075245

Abstract:
For years, scientists thought that all aspects of prosody were essentially processed in the superior temporal gyrus, a brain region known for speech perception. Through analysis of intra-cortical recordings in humans and non-human primates (Macaque), this study reveals that Heschl's Gyrus is a cortical region crucial to prosody perception, processing melodic accents as abstract phonological units. These findings inform neurolinguistic models of prosody processing by expanding the role of Heschl's Gyrus in speech processing beyond the low-level representations suggested so far. They also have important theoretical implications for linguistics since, in line with what has been proposed by metrical self-segmental theory, they show that melodic accents are linguistic categories of a discrete nature.

 

Description of figures 1 and 2
Credits: Authors

The nature of speech representations in the “literate brain”

We are pleased to announce the latest article published by Chotiga Pattamadilok (LPL) , Shuai Wang (LPL, ILCB) , Deirdre Bolger (ILCB) and Anne-Sophie Dubarry (CRPN). This study was carried out using the EGI recording system of the Laboratoire Parole et Langage (LPL):

Reference: Chotiga Pattamadilok, Shuai Wang, Deirdre Bolger, Anne-Sophie Dubarry. Learning to read transforms phonological into phonographic representations. Scientific Reports, 2025, 15, ⟨10.1038/s41598-025-88650-9⟩.

Full-text article: https://hal.science/hal-04975350v1

Abstract:
The ability to understand speech in one's native language is considered universal. This assumption implies that literate and illiterate people recognize spoken words in the same way. Our study provides new neurophysiological evidence against this assertion, showing that the speech representations stored in the “phonological” lexicon of literate people are contaminated by orthographic features. This observation calls into question the universality of spoken word recognition models that have been built exclusively on the basis of results obtained from literate participants.

 

Photo credits:
1) lil_foot_ / Pixabay
2) EGI system / LPL

How to assess the risk of virus transmission through speech?

As early as spring 2020, several members of the LPL and La Conception Hospital were working on the risks of spreading the SARS-Cov-2 virus, particularly when dealing with swallowing disorders or dysphonia. Based on this initial work, the SpeedVel study was financed and carried out in collaboration with colleagues from the Institut Universitaire des Systèmes Thermiques et Industriels in Marseille, to analyze droplets emitted in oral communication situations, in order to better assess the risk of contamination.

We are pleased to announce the publication of detailed results from the SpeedVel project in the journal Language Resources et Evaluation:

Reference: Francesca Carbone, Gilles Bouchet, Alain Ghio, Thierry Legou, Carine André, Muriel Lalain, Caterina Petrone, Antoine Giovanni. Investigating Droplet Emission During Speech Interaction. Language Resources and Evaluation, 2024, ⟨10.1007/s10579-024-09789-x⟩. ⟨hal-04385770v2⟩

Abstract & full-text article: https://hal.science/hal-04385770

Credits: SpeedVel / LPL

The texture of sounds: do you hear rough or smooth sounds?

On 20 November, the journal JASA published a new article on the association of sounds with shapes and textures, by a group of international scientists including Caterina Petrone (LPL-CNRS) and Susanne Fuchs (ZAS Berlin, currently Iméra/ILCB Chair).

According to the results of this study, the alveolar trill Is perceived as jagged/rough by speakers of different languages. The researchers believe that this association may be more universal than the famous bouba/kiki effect.

Reference: A. Ćwiek, R. Anselme, D. Dediu, S. Fuchs, S. Kawahara, G. E. Oh, J. Paul, M. Perlman, C. Petrone, S. Reiter, R. Ridouane, J. Zeller, B. Winter. The Alveolar Trill Is Perceived as Jagged/Rough by Speakers of Different Languages. J. Acoust. Soc. Am. 156, 3468–3479 (2024)

Full text: https://pubs.aip.org/asa/jasa/article/156/5/3468/3321514/The-alveolar-trill-is-perceived-as-jagged-rough-by

Abstract:
Typological research shows that across languages, trilled [r] sounds are more common in adjectives describing rough as opposed to smooth surfaces. In this study, this lexical research is built on with an experiment with speakers of 28 different languages from 12 different families. Participants were presented with images of a jagged and a straight line and imagined running their finger along each. They were then played an alveolar trill [r] and an alveolar approximant [l] and matched each sound to one of the lines. Participants showed a strong tendency to match [r] with the jagged line and [l] with the straight line, even more consistently than in a comparable cross-cultural investigation of the bouba/kiki effect. The pattern is strongest for matching [r] to the jagged line, but also very strong for matching [l] to the straight line. While this effect was found with speakers of languages with different phonetic realizations of the rhotic sound, it was weaker when trilled [r] was the primary variant. This suggests that when a sound is used phonologically to make systemic meaning contrasts, its iconic potential may become more limited. These findings extend our understanding of iconic crossmodal correspondences, highlighting deep-rooted connections between auditory perception and touch/vision.

 

Illustration: Fig. 1 The oscillograms and spectrograms for the recording of (a) the alveolar trill [r] and (b) the alveolar lateral approximant [l]. The superimposed red line is the intensity curve with a range between 55 and 85 dB. The jagged line (c) and the flat line (d) were the corresponding visual stimuli presented to participants in the experiment.

Credits: Authors

How can species as far apart as dogs and humans understand each other?

The Institut Pasteur has just issued a press release on the study conducted by researchers from the University of Geneva, the Institut de l'Audition and the CRPN Marseille  in collaboration with Thierry Legou (LPL), published in the journal PLOS Biology in October:

 Reference: Déaux EC, Piette T, Gaunet F, Legou T, Arnal L, Giraud A-L (2024) Dog-human vocal interactions match dogs' sensory-motor tuning. PLoS Biol 22(10): e3002789.

Full text article: https://doi.org/10.1371/journal.pbio.3002789

This study ‘reveals that the pair meet halfway between their differences to communicate together. From their results, the researchers hypothesise that dogs and humans have co-adapted to be able to communicate together. Comparison with other canids that have not been domesticated by humans, such as wolves, would reveal which of the dog's linguistic abilities are due to its genetics and which are due to its socialisation’.

 

Crédits : Thierry Legou, LPL

 

State of the art of ‘graded resources’ for foreign language teaching

We are pleased to announce the publication of the latest article by Núria Gala (LPL-AMU) in collaboration with Thomas François from the Centre for Automatic Language Processing (CENTAL UCL) in the International Journal of Applied Linguistics :

Reference: Thomas François, Núria Gala. Graded resources for learning and teaching foreign languages. ITL - International Journal of Applied Linguistics, 2024, ⟨10.1075/itl.00024.fra⟩. ⟨hal-04697842⟩

Article on the publisher's website (restricted access): https://dx.doi.org/10.1075/itl.00024.fra

Reference on HAL: https://hal.science/hal-04697842

Abstract:
Innovative resources for teaching vocabulary have emerged during the last decades, among which the so-called ‘graded resources’, i.e. lexicons or inventories where linguistic forms have been associated with a difficulty level, having in mind a target reader. The idea of ‘grading’ in Education is not new and has evolved over time: vocabulary or text scales (grades) have been approached as an educational tool with the means available at each period of time. In this survey, our aim is to show how current approaches of graded resources for foreign language learning are rooted in the tradition of building frequency lists for education. We then synthesize the body of work that has been undertaken to design graduated resources, highlighting the different methodologies applied as well as existing resources.

 

Credits: Image by freepik (ed.)

Prediction of events in human language processing and pretrained language models

We are pleased to announce a new paper published by Philippe Blache (LPL) in collaboration with researchers from Hong Kong Polytechnic University (Procore Project) and Purdue University:

Reference: James Britton, Yan Cong, Yu-Yin Hsu, Emmanuele Chersoni, Philippe Blache. On the influence of discourse connectives on the predictions of humans and language models. Frontiers in Human Neuroscience, 2024, 18, pp.1363120. ⟨10.3389/fnhum.2024.1363120⟩. ⟨hal-04717106⟩

Full-text article: https://doi.org/10.3389/fnhum.2024.1363120

Abstract:
This paper investigates how humans and language models process event predictions in sentences, focusing on the role of discourse connectives like and, but, and because. Humans find congruent sentence sequences easier to understand, and these connectives help clarify event relationships, especially when predictions need to be reversed (e.g., concessives and contrastives). The study used Italian and Mandarin Chinese story stimuli to test plausibility and coherence with or without connectives. Results show Mandarin language models are somewhat sensitive to these factors but struggle with prediction reversals, while Italian models showed no significant alignment with human data.

 

Credits: Authors