Our results strongly support Integration Models, which posit that word representations are shared across language modalities. /end

A fitting salute to my friend - a giant on the neurobiology of language 🫡

@cnrs.fr @univ-amu.fr @ilcb.bsky.social #LPL @agencerecherche.bsky.social @erc.europa.eu

Results were clear: The same brain networks were activated in production and comprehension.

Topographically in the motor cortex (with stronger lip activity for bilabial words and stronger tongue activity for alveolar words), and distributed in the temporal cortex /4

To test this, we scanned (fMRI) participants (n=37) while they both named aloud object names and passively listened to the same words. All words were minimal pairs only differing in their first phoneme (bilabial words like ‘Monkey’ vs. alveolar words like ‘Donkey’). /3

We asked whether the brain uses the same phonological representations when speaking vs. understanding words.

While most brain language models predict asymmetries, a model as that of Friedemann suggests they’re shared across modalities. /2

🫶

Thanks Riccardo! That means a lot coming from you 🙏

🙏

Yes! We hope to finish those analyses in the coming months 🤞

And myself 😉
(And wine posts, though that’s more about me having a conversation with myself 😂)

Oh wow, that’s awesome! Thanks so much Laurel!

👋 Divulgative highlights of this study by Kudos are also available: www.growkudos.com/publications...

🙏Emilia Kerr, @bnmorillon.bsky.social

@cnrs.fr @cnrs-dr12.bsky.social @univ-amu.fr @ilcb.bsky.social #LPL
@erc.europa.eu
@apajournals.bsky.social /end

💡This study confirms hypotheses of predictive dialog models and theories of joint action

More generally, this work demonstrates that language in interaction can rely on processing dynamics which are different or even absent in the dominant individualistic research tradition! /7

💡Our data show that during interactions we not only predict what we are about to say, but also the words our partner may utter.

This dyadic prediction effect is not linear, but ‘incremental’ (perhaps advanced planning like in chess?), and requires ‘interpersonal synergy’ /6

⏱️Importantly, in an individual control experiment where everything was identical except that one of the interlocutor’s was replaced with a loudspeaker, the typical prediction effect for picture naming was the same, but the prediction effect for replying with a related word disappeared completely! /5

⏱️In predictable contexts picture naming was faster (typical prediction effect), but interestingly, the reply given by the partner was even faster… twice (!) as fast compared to the interlocutor who named the picture to which the actual predictable content applied (a dyadic prediction effect) /4

💬 In a simple semantic association game, one interlocutor names a picture (eg dog), and the partner needs to reply with a semantically related word (eg cat).

Prior to this interaction, contex could be predictable for the upcoming picture (´man’s best friend is a…’) or not (‘outside there is a…)./3

❔Prediction helps us process language faster. We use it to anticipate words, understand meaning, and plan what to say next. But nearly all studies have tested this within individuals. What happens when we actually talk to each other?

We explored this question with a new dyadic paradigm: /2