Bozic, M. 1, 2 , Tyler, L. K. 1, 2 & Marslen-Wilson, W. 1, 2
1 MRC Cognition and Brain Sciences Unit, Cambridge
2 Department of Experimental Psychology, University of Cambridge
Existing neuroimaging and neuropsychological evidence suggests that human speech comprehension engages two functionally distinct neurocognitive systems. One system, distributed over bilateral perisylvian regions, supports general processing demands associated with mapping whole words onto their lexical representations. The other is a specialised left hemisphere perisylvian system, which supports core decompositional and combinatorial processing of linguistically complex words. In two separate fMRI experiments and using a combination of univariate and multivariate analysis techniques, we tested how inflected and derived words (e.g. played, agreed vs bravely, darkness) engage these mechanisms. In both experiments we co-varied increases in perceptual, non-linguistic, complexity of spoken words (presence of an onsetembedded stem, e.g. claim/clay), with variations in their linguistic complexity (presence of a morphological suffix, e.g. play+ed). Perceptual processing complexity, generated by the on-line competition between the full word and its onset-embedded stem, was found to activate left and right perisylvian brain regions for both inflections and derivations. In contrast, linguistic complexity activated left-lateralised areas only. Furthermore, this left-lateralised effect was only observed for inflectional combinations (play+ed), and not for derivational combinations (brave+ly). The results suggest that only words that are inherently decompositional in nature (i.e., regular inflected forms)engage specialised grammatical computations in the left fronto-temporal system. In contrast, the processing of derived words, that are likely to be stored as whole forms, primarily engages the bilateral system which underpins whole-word, stem-based lexical access. These results imply that inflectional and derivational complexity in English present different challenges for the processing system, and underline the importance of a neural framework for understanding these mechanisms.