Automatic contribution of long-term knowledge to verbal short-term memory

Introduction: The contribution of long-term memory (LTM) knowledge to verbal short-term memory (VSTM) is explained by language-based models assuming fast and automatic interactions between STM and LTM. However, direct evidence for automatic STM-LTM interactions is scant. The purpose of the present study was to test the degree of automaticity of LTM-STM interactions using fast running span procedures minimizing the intervention of strategic processes during verbal STM encoding and maintenance.

Method: We assessed the influence of several types of long-term linguistic knowledge (lexicality, lexical frequency, semantic-similarity and imageability) on performance in a running-span procedure in which subjects had to encode and recall auditory lists of unpredictable length, with verbal items being presented at a very fast rate (2.5 items/s).

Results: We observed reliable effects of linguistic variables in all conditions: VSTM span was higher for words vs. nonwords (p < .001), high vs. low frequency words (p < .001), related vs. unrelated words (p < .001) and high vs. low imageability words (p < .001).

Discussion: These results provide the most direct evidence as so far for a direct and automatic interaction between verbal STM and linguistic knowledge stored in verbal LTM. Importantly, this was also the case for semantic knowledge, disconfirming claims that semantic LTM effects would depend on slower controlled activation processes.