Now you see it, now you don’t: On semantic P600s and cross-linguistic variation in online verb– argument linking
The cognitive neuroscience of language has recently seen a very prominent debate about so-called “semantic P600” effects. Semantic P600s, i.e. observations of an ERP component that was long considered syntactic in nature in response to semantic violations, have been claimed to challenge the dominant role of syntax in determining online sentence interpretation and to rather demonstrate a greater independence of combinatory semantic processing. Hence, these effects have initiated a discussion about fundamental properties of the language processing architecture. In this talk, I will present a different perspective: based on the architectural assumptions of a cross-linguistic neurocognitive model of language comprehension (the eADM), I will argue that semantic P600s are best viewed as correlates of the syntax-semantics interface. Crucially, these effects occur only under certain circumstances and only in languages of a certain type. In support of this hypothesis, I will present electrophysiological data from German, Turkish, Chinese and Icelandic, all of which show qualitatively different patterns from English and Dutch for typical semantic P600 manipulations. From these results, I conclude that the presence or absence of semantic P600s can be used to typologise languages with respect to their verb–argument linking characteristics.