Scrambling and Information Structure: Dynamics of Interpretation

Language is surely the most powerful communicative tool available to humans. Accordingly, languages typically provide flexible means of adapting to a discourse situation, e.g. by partitioning information that is "common ground" between interlocutors from new information. This influence of "information structure" typically leads to a preference for old information to precede new information. However, this type of strategy may conflict with other properties of a particular language, for example with the syntactic requirement that subjects should precede objects whenever possible. The aim of the research proposed here is to shed light on the resolution of this conflict between information structural and syntactic requirements within clause-medial argument order variations in German (`scrambling`). To this end, we will measure the electrical response of listeners` brains (event-related brain potentials, ERPs) elicited during the presentation of object-initial sentences under various contextual influences. ERPs provide a real-time measure of the comprehension process and therefore allow for an investigation of the processing conflict as it occurs and unfolds over time. Additionally, we will examine how information structure influences the acceptability of scrambled sentences, a factor that has been shown to diverge from real-time measures. One of the most important aims of this research is to resolve this apparent paradox and show how the two perspectives are connected. This may be achieved by employing the speed-accuracy trade-off (SAT) technique. The results obtained will be used to develop a model of language architecture that captures the mutual influences of syntax, information structure and prosody in scrambled structures. The model will incorporate both time-sensitive and time-insensitive architectural aspects and may help to determine the locus of the comprehension deficit for object-initial sentences shown by particular patient populations.