Incrementality and beyond: What ERPs tell us about utterance comprehension

We've known for a long time, both from real-life language use and laboratory experiments, that utterance comprehension is incremental. In my talk, I will use ERP data to illustrate just how pervasive this incrementality really is. Readers and listeners grab every word as it comes in, and relate it to the wider discourse context (e.g., what has been said before, the identity of the speaker) in a split second, in terms of its semantic, referential and syntactic implications. Listeners also relate words that are unintentionally embedded in other words (e.g., "pain" in "champagne") to the preceding context. All this reveals that we are dealing with a rather eager information processing system, one that doesn't hesitate to commit early. After reviewing ERP evidence for incrementality, I will briefly discuss two lines of research that go beyond standard incrementality. First, we have converging evidence to suggest that people don't just quickly act upon bits of the unfolding input -- they also look ahead, and anticipate upcoming words, referents and concepts. Second, we have initial evidence that, at least in the context of an attitude survey, people rapidly relate the unfolding utterance to their personal value system. Of course, all this makes sense if we consider language in the bigger picture of biological life, where anticipation can prevent you from getting killed, and where rapid valuation may well be the prime motivator for incremental processin