Event-related potentials reflecting the frequency of unattended spoken words: A neuronal index of connection strength in lexical memory circuits?

How are words represented in the human brain and can these representations be qualitatively assessed with respect to their structure and properties? Recent research demonstrates that neurophysiological signatures of individual words can be measured when subjects do not focus their attention on speech input. These automatic activations, which take the form of negative deflections of event-related potentials, can appear surprisingly early (within similar to 200 ms) and are based on robust connections within neuronal memory circuits encoding individual words that ignite even when attentional resources are scarce. A new and critical prediction of this framework is that words with high frequency of occurrence have especially strong connections of their underlying memory circuits and should thus yield more negative ERPs compared with rarer words. We tested this prediction by presenting our subjects, in passive non-attend conditions, with acoustically matched high- and low-frequency words along with pseudo-words. Using factorial and correlation analyses, we found that already at similar to 120 ms after the spoken stimulus information was available, amplitude of brain responses was modulated by the words' lexical frequency. Topographic mapping and source analysis suggested that this early automatic frequency effect originates from the left inferior-frontal cortices. While, at this early latency, lexical differences between words and pseudo-words (more negative-going potential for meaningful words) could be seen only for the most frequent word stimuli, later-on (similar to 270 ms), a more global lexicality effect with bilateral perisylvian sources was found for all stimuli, suggesting faster access to more frequent lexical entries. Our results support the account of word memory traces as interconnected neuronal circuits, and suggest that speed and magnitude of their activation are determined by their internal connection strength, which, in turn, is determined by the everyday language use. (C) 2010 Published by Elsevier Inc.