SPEECH PERCEPTION AND MEMORY IN AN EPISODIC LEXICON

Arizona State University-Tempe Campus, Tempe AZ

Investigators

Linked publications & trials

Abstract

DESCRIPTION: This research would investigate lexical representations in long-term memory and the processes that both encode and retrieve those representations. Most theories assume that speech waveforms are converted to phoneme strings, which are then compared to the mental lexicon. When comparison yields a match, the word is "recognized." The typical view emphasizes the abstract properties of words: A speech signal containing idiosyncratic information (e.g., speakers' voice, ambient noise) is normalized into a canonical form and then matched to abstract lexical entries. Few theories consider idiosyncrasies of speech as more than "noise" in the signal. However, a multiple-trace memory model posits that such details of perceptual experience are not forgotten; instead, they are integral to later perception. The proposed research and simulations (using Hintzman's, 1986, MINERVA 2 model) have already produced encouraging results (Goldinger, 1998). The proposed research will test the predictive limits of the extant model, with embellished versions as necessary. In Project 1, perception and memory experiments will test the factors that govern episodic content, examining roles of selective attention and working memory. The experiments will also examine the nature and extent of perceptual-conceptual tradeoffs in trace formation. According to the model, these dimensions should be flexible and emergent, depending largely upon processing tasks and working-memory resources. The experiments in Project 2 will examine memory for larger episodic units, using full sentences as stimuli. Memory for sentences will also be assessed using a novel speech-production method, examining the acoustic-phonetic content of subject's spoken responses. As in previous research, subjects are expected to spontaneously imitate the acoustic patterns of the stimulus sentences, in terms of duration, pitch, and intonation. Moreover, the model predicts that degrees of imitation will be affected by "abstract" characteristics of the sentences, such as syntactic and semantic complexity, suggesting that the acoustic manifestations of speech reflect a complex interplay of the linguistic stimuli and their episodic representations.