Memory-search and rehearsal processes and the word length effect in immediate recall: a synthesis in reply to service.

Abstract

Wewould like to begin by offering a new theoretical account of the duration-based part of the word length effect, inspired partly by the recent literature and partly by Service’s comments (this issue). Baddeley (1986) suggested that recall depends on the speed of a covert verbal rehearsal process. Cowan and colleagues have argued that it may, but that recall also depends on the speed of some sort of short-termmemory search process. Both of these processes could serve to refresh a short-termmemory representation of list items. The memory search process would differ from rehearsal in that search would not require covert articulation and would proceed at the same rate regardless of the word length, perhaps based on a lexical representation. The main evidence for this is that the duration of silent pauses within correct spoken responses in memory span tasks depends upon the number of verbal items in the list but not upon the durations of those items (Cowan et al., 1994; Hulme, Newton, Cowan, Stuart, & Brown, 1999). Subjects may have to search through the list after each word in order to determine which word to recall next, and that search process would take longer when the list includes more words. This pattern of pauses in spoken recall resembles what is found in memory search tasks with a recognition probe, in which the reaction time depends upon the list length but not upon the word length (Chase, 1977; Clifton & Tash, 1973). Cowan et al. (1998) found that memory search and rapid speaking rates both were moderately related to memory span but were unrelated to each other. We would suggest that either memory search or rehearsal processes (or both) can be used to refresh items in a list. Search processes would tend to be used when there is a strong lexical representation of each word, so that recall of the word is an all-or-none event. In contrast, when the lexical representation is weak, as when the items are pseudowords, verbal rehearsal would have to be used because recall of one part of a pseudoword does not ensure recall of its remainder and rehearsal activates each part of the item separately, in turn. This type of rehearsal requires covert articulation of the list items and takes longer when the items are longer. These suggestions are supported to some