Search Abstracts | Symposia | Slide Sessions | Poster Sessions
A semantic locus for iconicity in the brain? Identifying the neural signatures of iconic words in spoken language production
Poster Session E, Friday, October 2, 11:00 am - 1:00 pm, Wangari Maathai
Giulio Massari1, Fanny Meunier1, Raphaël Fargier1; 1Université Côte d'Azur, CNRS, BCL, France
Linguistic units can raise non-arbitrary, perceptual-based associations with their referent: this is iconicity. Present throughout the lexicon, iconic mappings between semantics and phonology have been shown to display reliable consequences upon lexical processing (Perniss & Vigliocco, 2014; Sidhu, 2024). For instance, studies show that, despite differences across language modalities, iconic words and signs are both retrieved faster than arbitrary ones (Massari et al., 2026; Vinson et al., 2015). This suggests that iconicity contributes to structure the cognitive dynamics through which lexical representations are processed for production (Roelofs & Ferreira, 2019). However, the way this translates into brain mechanisms is still unclear. ERP studies on sign language show that the neural patterns underlying the retrieval of iconic and arbitrary signs differ upon both early time windows linked to conceptual-semantic activations, and later time windows linked to lexical-phonological activations (Baus & Costa, 2015; McGarry et al., 2021; 2023). Yet, no similar evidence is available for spoken language. To investigate this, we tested whether iconic and arbitrary words differ in their associated neural signals when retrieved for speech. We designed a picture naming task employing high-density electroencephalography (HD-EEG). Twenty-nine native French speakers named 160 images via iconic or arbitrary words. Iconic items (n=80) consisted of words whose phonology symbolically matched the visual shape, round or spiky, of their associated image (e.g., ballon – ‘balloon’); arbitrary items (n=80) consisted of words whose phonology did not carry an iconic association towards shape (e.g., bouteille – ‘bottle’: see Massari et al., 2026). Behavioral results were analyzed in terms of response times on correct trials. EEG recordings, restricted to correct trials and cleaned for artifacts and outliers, were epoched into baseline-corrected stimulus-aligned ERPs (0 to 500ms post-image). A linear mixed-effects model on RTs revealed a significant effect of iconicity (p=.045), highlighting that iconic words were retrieved reliably faster (-32ms) than arbitrary words. This reproduces the finding of a timing advantage by iconicity in spoken production (Massari et al., 2026). Paired t-tests on ERP amplitudes, FDR-corrected for multiple comparisons, revealed significant differences between conditions (p<.05) in an early time window (139-178ms post-image) on a group of neighboring occipital electrodes, as well as in a later, shorter time window (293-315ms post-image) on a smaller group of non-neighboring occipital electrodes. Inspection of mean amplitudes traced back this pattern to iconic words eliciting greater positivity than arbitrary words in the periods of interest at occipital sites. This is consistent with previous observations of ERP differences according to iconicity in signed production (Baus & Costa, 2015; McGarry et al., 2021). Specifically, the current findings support the view that the increased positivity for iconic units reflects the greater engagement of the conceptual-semantic system upon their retrieval (McGarry et al., 2023). Therefore, our results show that iconicity in spoken language displays distinctive neural signatures upon lexical processing. The timing and topography of such signatures, in turn, argue for a mainly semantic locus of the effect, suggesting that iconicity modulates the dynamics of word retrieval by increasing the accessibility of conceptual networks in the brain.
Topic Areas: Language Production,