Investigating the Relationship between Short-Term Memory and Word Retrieval with a Novel Verbal Short-Term Memory Task

• ¹ Temple University, Communication Sciences and Disorders, United States

Introduction and Predictions Individuals with aphasia frequently demonstrate verbal short-term memory (STM) deficits (Minkina, Kalinyak-Fliszar, Rosenberg, & Martin, 2017). These impairments likely coexist due to shared processing mechanisms between language tasks (e.g., picture naming) and verbal STM tasks (e.g., immediate serial recall of words). According to a verbal STM view (Martin & Saffran, 1997) based on an interactive activation language model (Dell & O’Seaghdha, 1992), both task types rely on temporary activation of semantic, lexical, and phonological representations. We investigated the relationship between verbal STM, measured with a novel serial picture name recall task, and word retrieval, measured with a classic picture naming test. Verbal STM is typically tested through immediate serial digit/word recall, during which phonology is activated prior to lexical-semantics. In our novel verbal STM task, lexical-semantic information is activated first (consistent with word retrieval). The interactive activation model of word processing predicts that recall of the first word should depend more on phonological support, as activation has time to spread forward to phonology, while recall of the second word should depend more on semantics, as activation has less time to spread to lower levels before recall is required. Thus, we predicted more semantic (relative to phonological) errors on the picture naming test would be associated with better recall of the first word on the verbal STM task (primacy) and more phonological errors in naming would be associated with better recall of the last word (recency). Methods Three individuals with chronic aphasia as a result of left hemisphere stroke completed this pilot study (Table 1). Participants completed the Philadelphia Naming Test (PNT; Roach, Schwartz, Martin, Grewal, & Brecher, 1996), modified with a 2s picture presentation and a 5s response window, and responses were coded for error type. The number of semantic errors was subtracted from the number of phonological errors (P-S score). Participants also completed a naming test of 139 standardized pictures (Snodgrass & Vanderwart, 1980) three times over three days (same timings as the PNT). Pictures named correctly all three times were grouped into triplets for the verbal STM task. Words in each triplet (72 total) were matched for syllable length, frequency, and phonotactic probability, and controlled for phonological similarity. Three pictures were presented sequentially on the screen (2s per picture), after which the participant was prompted to name the picture sequence in order. Responses were scored for accuracy, and primacy (# first words correct/total words correct) and recency (# last words correct/total words correct) scores were derived. Results Pilot results are listed in Table 1. Consistent with the predictions, participants with negative P-S scores (>semantic impairment; P1 & P2) demonstrated a primacy effect, while P3 (positive P-S score; >phonological impairment) demonstrated a recency effect. Data collection with additional participants is ongoing, and the target number (n=20) will allow for correlational/regression analyses. Conclusion These results point to a possible relationship between word retrieval and verbal STM, and suggest that primacy/recency effects in our novel verbal STM task may provide diagnostic insights into word retrieval impairment in aphasia.