The Structure and Time-Course of Information Interaction During Speech Comprehension: Lexical Segmentation, Access, and Interpretation

The comprehension of speech is clearly an integrative process. In fact, such a statement seems so self-evident as to be nearly tautologous; to understand an utterance we must, in some fashion, retrieve information about the words in that utterance, discover the structural relationship and semantic properties of those words, and interpret these in the light of the various pragmatic and discourse constraints operating at the time. Further, all of this takes place at a remarkably rapid pace, a fact (among several others) that has led a number of theorists to characterize the comprehension system in general as being a contextually determined or "top down" process, a system that allows all temporally previous information to affect the analysis of new information at any point in its procesing, thus providing maximally rapid processing and interpretation of new sensory information. Some impressive empirical work demonstrating the very rapid effects of certain "higher" order information sources upon the processing of "lower" order information has recently provided some important substantiation for this position (see, in particular, Marslen-Wilson & Tyler, 1980). To a great extent, such interactionist points of view have developed in response to the limitations of a particular model of speech understanding, one derived from a rigid mapping of Transformational Generative Grammar onto a putative comprehension system. Under this particular system, comprehension is viewed as being composed of a series of serially encountered, hierarchically organized autonomous processing modules (see Marslen-Wilson, 1976, and Tyler, 1982, for excellent detailed discussions of these approaches). The goal of this paper is to examine both the interactionist and