Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Dissociating the Neural Substrates Underlying Phonological and Articulatory-Motor Processing During Language Production
Poster Session F, Friday, October 2, 2:45 - 4:45 pm, Wangari Maathai
Yusheng Wang1, Elizabeth J. Anderson2, Kyle Shattuck2, Catherine Kelly2, Peter E. Turkeltaub2,3,4, Stephanie K. Riès5,6; 1SDSU/UCSD joint doctoral program in Language and Communicative Disorders, 2Center for Brain Plasticity and Recovery, 3Center for Aphasia Research and Rehabilitation, 4Departments of Neurology and Rehabilitation Medicine, 5School of Speech, Language, and Hearing Sciences, 6Center for Clinical and Cognitive Neuroscience
Phonological and articulatory processing have been difficult to disentangle during language production. Pseudoword repetition is a classically used paradigm that involves both phonological processing and speech-motor production. We used word-level complexity scores to test whether partially overlapping production demands could be distinguished behaviorally and neurobiologically. The Kent-based complexity score was treated as an articulatorily motivated index of cumulative segmental production load because it sums the motor complexity of individual segments in a word. The Word Complexity Measure (WCM) was treated as an index of developmental phonological-structural complexity because it scores word patterns, syllable structures, and sound classes. Because part of the WCM variance can be explained by articulatory complexity captured by Kent scores, we used residualized WCM (WCMres) computed at the item level by removing variance in WCM shared with Kent, providing a measure of the Kent-independent WCM component hypothesized to reflect relatively more phonological-structural than articulatory-motor demand. Participants were 73 individuals with left-hemisphere stroke who completed a pseudoword repetition task and underwent diffusion MRI scanning. Real words spanning WCM-defined complexity levels were used to generate matched pseudowords by segment substitution while preserving WCM structure. All pseudowords were rescored using Kent scores. For each complexity score (Kent and WCMres), mixed-effects models tested effects on accuracy, response time (RT), and duration while adjusting for syllable count. Subject-specific complexity slopes were extracted from each model as indices of sensitivity to increasing item articulatory versus phonological complexity and entered into parallel diffusion tensor imaging (DTI) regression analyses. Tracts of interest were the left arcuate fasciculus (AF_L) and the left superior longitudinal fasciculus III (SLF3_L) for phonological processing and SLF3_L and the left frontal aslant tract (FAT_L) for speech-motor processing. White matter models controlled for age and lesion volume. Behavioral results showed that pseudowords with higher Kent scores, reflecting greater cumulative articulatory-motor load, were repeated less accurately (βraw = -0.55, p = .004), initiated more slowly (βraw = 0.05, p < .001), and produced with longer durations (βraw = 0.06, p < .001). Pseudowords with higher WCMres scores, reflecting phonological complexity, were also produced with longer duration (β = 0.03, p = .003), suggesting an additional cost of phonological complexity beyond articulatory-motor load. For syllable-adjusted accuracy slopes, positive FA-slope associations indicated that participants with greater tract integrity showed smaller accuracy declines as item complexity increased. Kent slopes were associated only with AF_L FA (β = 1.11, pFDR = .002). WCMres slopes were associated with both AF_L FA (β = 0.92, pFDR = .002) and SLF3_L FA (β = 0.65, pFDR = .049). FAT_L was not significantly associated with slopes from either complexity score. Together, Kent and WCMres distinguished different emphases within pseudoword repetition demands: Kent captured cumulative segmental/articulatory load, whereas WCMres captured a Kent-independent phonological-structural contribution. Their shared AF_L association, broader AF_L/SLF3_L involvement for WCMres, and lack of FAT_L effects suggest partially overlapping contributions within the dorsal repetition network rather than fully separable articulatory-motor and phonological systems.
Topic Areas: Phonology, Language Production