Lexical access in American Sign Language: an ERP investigation of effects of semantics and phonology

That language forms (phonology) are arbitrarily related to their meanings (semantics) is often considered a basic property of human languages. Naturally occurring sign languages, however, often appear to conflate form and meaning. In this paper we examine whether this close coupling has processing consequences for lexical access. We examine the electrophysiological correlates of on-line sentence processing in an attempt to clarify the time-course of lexical access in American Sign Language. EEG was recorded while 17 native signers watched ASL sentences for comprehension. Participants were presented with sentences in which semantic expectancy and phonological form were systematically manipulated to create four types of violations. These four conditions of interest are contrasted to a baseline sentence with a preferred semantic ending. Two different effects were observed in early time windows. Evidence for an early effect of semantic pre-activation of plausible candidates (150-250 ms) was found, followed by a negativity associated with lexical selection (350-450 ms) for only phonologically related (-S, +P) and for only semantically related (+S, -P) signs. These findings provide evidence for a novel mapping of signal form and meaning that may be a unique signature of sign language. In the 450 to 600 ms window, all conditions showed an increased N400 with respect to the expected ending, suggesting greater difficulty in semantic integration with the established context. Overall, these findings provide important insights into the on-line processing of visual-manual language.