A distinguishing characteristic of the cue-based retrieval architecture is a highly restricted memory capacity for maintaining active information. This is consistent with research in the memory domain suggesting that retrieval speeds are constant for all information except for a few recent items in attentional focus. This means that normal (skilled) language comprehension requires only minimal memory capacity, with a fast cue-based retrieval process providing the computational power necessary to create dependencies in real time. This contrasts sharply with the predominant view of the memory system that supports language comprehension, which assumes an active memory capacity sizeable enough to be a source of individual differences (e.g., Just & Carpenter, 1992). It also provides a means of resolving the puzzling findings that patients with severely limited working memory spans nevertheless show preserved comprehension of quite complex grammatical constructions (e.g., Caplan & Waters, 1999). Shifting the focus away from capacity and towards retrieval suggests that sources of individual variation in comprehension relate either to differences in the nature or efficiency of the retrieval mechanism and/or in the quality of representations. Examining the first possibility is especially important because evidence supporting cue-based retrieval is entirely based on a population of highly skilled (undergraduate and graduate-level) readers. Our primary goal in the current study is to assess the presence of cue-based retrieval in poor readers.
展开▼
