The Role of Dynamic Frequency Synchrony in Syntactic Processing
Currently, the most widely-used method in electrophysiological linguistic research involves grand-averaging of brain responses across trials and subjects -- a technique designed to overcome the low signal-to-noise ratio of the brain's electrical response to a stimulus. Results of this technique (event-related potentials, or ERPs) have uncovered several reliable responses to linguistic variables, including responses to anomalous syntactic (left anterior negativity, or LAN; P600) and semantic items (N400). Psycholinguistic and neurolinguistic researchers often infer the functional significance of ERPs from their eliciting conditions, however in many cases a single category of representation cannot explain all occurrences. This work is an examination of several linguistically-distinct conditions which traditionally elicit only a limited number of syntax-related ERPs (LAN, P600), from the perspective of event-related changes in induced frequency synchrony. Induced activity represents part of the multidimensional EEG signal that is removed in the ERP grand-averaging process, namely oscillatory activity which is not phase-locked to the stimulus. Modulations in phase asynchronous oscillations reflect local changes in neural activity which control the frequency components of ongoing EEG (Pfurtscheller & da Silva, 1999), and are a critical component of the characterization of dynamics in the frequency signal. Frequency synchrony is of interest to theories of sentence processing given its long-established association with the binding of elements into complex representations (Gray et al., 1989a) and given that it has been argued to facilitate activity of spatially-distant, functionally-connected networks (e.g., Singer, 1993). Furthermore, recent work by (Bastiaansen et al., 2010) has revealed the role of certain frequency activity in syntactic (Beta) and more general sentence processing (Theta). Results of the work presented here indicate that a variety of grammatical processes occur within the time windows of the LAN and P600, including top-down creation of filler-gap relations (primarily reflected in early increases in Theta band activity), processing of complex syntactic representations (primarily reflected in increases in Beta band activity) and evaluative processes which reflect the probability of a (syntactic) event (reflected in corresponding decreases in Alpha and Beta activity in late processing windows). Overall, this work supports a theory in which the LAN and P600 are not single events in grammatical processing, and provides hypotheses about the role of certain frequency activity during sentence processing which can be examined in subsequent confirmatory research.