Links between Short-Term Memory and Word Retrieval in Aphasia
MetadataShow full item record
Effective assessment and treatment of aphasia requires specific and sensitive diagnosis of the linguistic level(s) wherein the linguistic impairment lies (e.g., semantic, lexical, phonological, etc.). Recent work in aphasiology argues that linguistic impairments at both the production and comprehension levels of language are supported, at least in part, by shared underlying processes. These processes aid in the access to and the temporary maintenance of linguistic elements needed to produce and comprehend language. One such process that has been a topic of recent research in aphasia is verbal short-term memory, or the temporary, limited-capacity storage mechanism through which information is briefly retained. A typical measure of verbal short-term memory is an immediate repetition span task: individuals listen to a list of digits or words and repeat them immediately after the list ends. People with aphasia demonstrate digit and word span lengths below that of neurologically healthy individuals. More specifically, resent research demonstrates that individuals with aphasia with greater language comprehension impairments at the semantic level show different patterns of breakdown in serial recall tasks from that of individuals with aphasia with greater language comprehension impairments at the phonological level. Individuals with language comprehension impairments at the level of phonology tend to err on items at the beginning of a list and demonstrate higher repetition accuracy on high frequency and imageability words, while individuals with semantic language comprehension impairments tend to err on items at the end of a list and are not as susceptible to manipulations of frequency and imageability. The current study further explored the link between locus of linguistic impairment in aphasia and immediate serial recall performance by examining the relationship between type of word retrieval impairment (i.e., semantic or phonological) and two aspects of verbal short-term memory performance, 1) location of errors made on a serial recall task (word pair repetition) and 2) susceptibility to word frequency and imageability manipulations during word pair repetition. The extent to which a linguistically-specified short-term memory system, versus a domain-general one, supports the word retrieval process was also examined. The results demonstrated that overall accuracy of word retrieval correlates positively with word span length in the absence of a correlation between word span length and a nonverbal (i.e., spatial) span length, providing support for the existence of a linguistically-specified short-term memory system. Additionally, while a relationship between type of word retrieval impairment and location of errors made on a word pair repetition task was not realized, a significant correlation between type of word retrieval impairment and susceptibility to word frequency and imageability manipulations during word pair repetition was found. As the number of phonological word retrieval errors, relative to semantic errors, increased, the bias towards correct repetition of high imageability over low imageability words increased. These finding are interpreted in the context of an interactive activation model of word retrieval where activation spreads from lexical-semantic to phonological representations. While an error location bias on the word repetition task likely arises due to intricate timing of linguistic activation, a bias towards correct repetition of highly imageable and frequent words likely arises due to the overall activation strength of linguistic representations. Results suggest that word pair repetition, a classic short-term memory task, and picture naming, a classic language production task, may be supported by a shared temporary linguistic activation process dependent on the overall activation strength of linguistic representations. Theoretically, the findings demonstrate the need to rethink the classic distinction between short-term memory and language that views the processes as separable. Clinically, the findings suggest that the analysis of imageability biases in word pair repetition tasks may help determine word retrieval impairment type, an encouraging idea given the much more laborious and time-intensive nature of analyzing word retrieval errors.
- Speech