Event Date: Thursday, 31 May, 2018, 11 a.m.
Location: Via Santa Maria, 36, Pisa, PI, Italia [2nd floor seminar room]
Speaker: Dr. Francesca Carota (Max Planck Institute for Psycholinguistics, Nijmegen)
Title: Neurocognitive representations of lexical complexity
Abstract: Morphological derivation offers a cross-linguistically relevant strategy to generate novel lexical items based on existing lexical material. How are the resulting complex lexical forms neurocognitively represented? Influential accounts posit that language comprehension engages a bihemispheric system supporting the semantic/pragmatic interpretation of whole-word forms and a left-lateralised fronto-temporal sub-system (especially the left inferior frontal gyrus, LIFG), specialised for decomposing combinatorial grammatical sequences like inflections. Previous cross-linguistic evidence showed that morphological derivation engages the bilateral fronto-temporal network, but does not selectively involve the LIFG, suggesting that derived words are stored as whole-word forms. However, cohort competition-based bilateral effects driven by semantic opacity (but absent in transparent forms) indicate that the word internal structure still affects neural processing. In recent work using functional magnetic neuroimaging (fMRI), I re-examined these positions in the context of the combinatorially rich Italian lexicon, asking whether different types of morphological complexity along transparency/opacity, productivity and ‘storedness’ dimensions affected the representation of complex word forms in the language systems. Only opaque forms triggered cohort competition effects in the bihemispheric system, consistent with earlier evidence. However, behavioural and corpus-based models of lexical complexity identified differences in the multivoxel representational patterns for opaquely and transparently derived words. Lexical competition and semantic relatedness correlated significantly with the bihemispheric system, whilst suffix productivity in transparent words selectively activated the LIFG, pointing to a degree of decompositionality. Combined univariate and novel multivoxel-pattern analyses, as performed here, offer a promising method for detecting the fine-grained representational properties of lexical complexity, which may vary across languages depending on their typological and historical characteristics.
Francesca Carota is a senior research associate at the MPI and DCCN in Nijmegen. Her research investigates the neurobiological bases and neural dynamics of language comprehension and production combining behavioural, computational linguistic and neuroimaging methods. After graduating in Computational Linguistics at the University of Pisa and ILC-CNR, she has received postdoctoral training in neuroimaging at ISC-CNRS Lyon. She worked as a staff member of the speech and language team at the MRC-Cognition and Brain Sciences Unit in Cambridge, studying language comprehension and semantic decoding using fMRI and MEG – with Prof. Y. Shtyrov, F. Pulvermueller and N. Kriegeskorte. She then joined the Neurolex group at the Dept. of Psychology, Cambridge University, with Prof. William Marslen-Wilson, to study the neurocognitive representations of lexical complexity across different languages. She has been a senior fellow at the Freie Universität and Humboldt Universität zu Berlin and is a visiting scientist at the Dept. of Psychology of Cambridge University.