Slide Slam A6
Influence of item- and individual-level semantic and phonological characteristics on word production in individuals with aphasia
Emily J. Braun1, Swathi Kiran1; 1Boston University
INTRODUCTION: Lexical-semantic and phonological properties of words have been shown to influence naming and word learning in neurotypical individuals. While this has also been investigated in individuals with aphasia, further study is required given inconsistent results across studies and to determine the interaction of these variables with individual impairment profiles. This retrospective study examined whether naming accuracy was predicted by (1) stimulus-level psycholinguistic properties, and (2) the interaction between stimulus-level psycholinguistic properties and individual semantic and phonological skills. METHODS: Participants were 35 individuals with chronic post-stroke aphasia (mean WAB-R AQ 63.5). Assessments evaluated overall aphasia severity using the Western Aphasia Battery – Revised Aphasia Quotient (WAB-R AQ) (Kertesz, 2007); naming via a 180-item picture naming task; and individual impairment profiles via three semantic (e.g., verifying whether a superordinate category applies to a word) and three phonological (e.g., judging whether two words rhyme) processing tasks. A single accuracy score for each domain was generated by calculating the average accuracy of the three tasks in that domain. Nine psycholinguistic variables were selected based on prior literature. Given high correlations among these variables, a principal component analysis (PCA) was completed with the stimuli condensing these variables into three components that represented lexical-semantic (age of acquisition, lexical frequency, typicality, and semantic neighborhood density), phonological (phonological neighborhood density, length in syllables, and length in phonemes), and phonotactic (phonotactic probability of phonemes and biphones) properties. For both questions, mixed effects logistic regression was used to predict binary noun naming accuracy, with WAB-R AQ as a covariate and random intercepts for item and participant. For Question 1, the three stimulus-level PCA scores were predictors with item category as an additional covariate (Model 1). For Question 2, three models were constructed, with predictors being the interaction between individual semantic skills and stimulus-level lexical-semantic word properties (Model 2a), individual phonological skills and stimulus-level phonological word properties (Model 2b), or individual phonological skills and stimulus-level phonotactic word properties (Model 2c). RESULTS: For Question 1, both the stimulus-level phonological (p<.05) and lexical-semantic (p<.05) component scores were significant predictors of naming accuracy (less complex words showed higher accuracy) (Model 1). For Question 2, there was a significant interaction between individual semantic processing skills and the stimulus-level lexical-semantic component (p<.05) (Model 2a); namely, in individuals with more severe semantic processing impairment, lexical-semantic word properties had little influence on response accuracy. In contrast, in individuals with more mild semantic processing impairment, words with less complex lexical-semantic properties showed greater accuracy. The interaction was not significant between individual phonological processing skills and the stimulus-level phonological (Model 2b) or phonotactic (Model 2c) components. CONCLUSION: In brief, lexical-semantic and phonological properties of nouns influence naming accuracy in individuals with aphasia. For lexical-semantic word properties, this interacts with individual semantic processing skills, with individuals with stronger semantic skills showing greater response accuracy for words with less complex lexical-semantic properties. This work confirms the influence of multiple psycholinguistic factors on word production in individuals with aphasia and demonstrates an interplay between stimulus-level lexical-semantic word properties and individual semantic impairment.