Words and voices: Episodic traces in spoken word identification and recognition memory

Abstract

Most theories of spoken word identification assume that variable speech signals are matched to canonical representations in memory. To achieve this, idiosyncratic voice details are first normalized, allowing direct comparison of the input to the lexicon. This investigation assessed both explicit and implicit memory for spoken words as a function of speakers' voices, delays between study and test, and levels of processing. In 2 experiments, voice attributes of spoken words were clearly retained in memory. Moreover, listeners were sensitive to fine-grained similarity between 1st and 2nd presentations of different-voice words, but only when words were initially encoded at relatively shallow levels of processing. The results suggest that episodic memory traces of spoken words retain the surface details typically considered as noise in perceptual systems. In a now-classic article, Oidfield (1966) first described the mental lexicon, a collection of words in long-term memory that mediates perceptual access to lexical knowledge. The lexicon has since been a focus of extensive investigation and theoriz-ing. Painting in broad strokes, there are two basic views on lexical representation: Abstractionist theories view the lexicon as a set of ideal, modality-free units, and episodic theories assume that groups of detailed traces collectively represent individual words (see Roediger & McDermott, 1993; Ten-penny, 1995). The abstractionist view is prominent in current theories; perception is typically assumed to involve informa-tion reduction, which is the decoding of specific episodes (tokens) into canonical representations (types; Morton, 1969; Posner, 1964). However, some theories posit episodic represen-tations and perception, bypassing such decoding. Global memory models (Eich, 1982; Gillund & Shiffrin, 1984; Hintz-man, 1986; Underwood, 1969), exemplar categorization mod-els (Nosofsky, 1991), and distributed memory models (McClel-land & Rumeihart, 1985) all assume episodic traces (although processing assumptions clearly differ).