Presentation

Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams

Dissociations in statistical word learning in aphasia

Poster E49 in Poster Session E, Saturday, October 8, 3:15 - 5:00 pm EDT, Millennium Hall

Claudia Penaloza¹, Nadine Martin², Matti Laine³, Antoni Rodríguez Fornells¹; ¹University of Barcelona, Barcelona, Spain, ²Temple University, Philadelphia, PA, USA, ³Abo Akademi University, Turku, Finland

Introduction: Statistical learning (SL) is a cognitive mechanism that supports the ability to parse unknown words from running speech and to acquire novel word-referent mappings in referentially ambiguous contexts by computing statistical patterns in the learning context. Here we assessed the functionality of these mechanisms in aphasia with SL tasks tapping phonological and lexical-semantic word acquisition by using a Bayesian method for the examination of dissociations in single-case studies. Methods: Participants were 24 healthy older adults (10 male, age: 60±11.63 years) and 3 participants with chronic post-stroke aphasia (PWA): P1 (male, 78 years, global aphasia), P2 (male, 42 years, mixed non-fluent aphasia), and P3 (male, 73 years, fluent aphasia). The PWA underwent language assessments including the BDAE III (Goodglass, Kaplan, & Barresi, 2005) to determine the presence of aphasia and aphasia severity and subtests of the TALSA battery (Martin et al., 2018) to evaluate verbal short-term memory. All participants completed a speech segmentation (SS) task (Peñaloza et al. 2015) tapping phonological learning, and a cross-situational learning (CSL) task (Peñaloza et al., 2017) examining lexical-semantic learning. In the SS task, participants were exposed to a spoken artificial language composed of four trisyllabic pseudowords and needed to learn the words by identifying word boundaries (i.e., computing transitional probabilities between adjacent syllables, higher between syllables within words and lower between syllables spanning word boundaries). They also completed a test that required discriminating words from nonwords. In the CSL task, participants needed to learn 9 pseudoword-novel referent pairs across 4 training blocks. In each trial, two objects of the training set were presented together with two spoken pseudowords, and the participants needed to figure out the correct associations between words and objects. The referential ambiguity of each learning trial (i.e., 4 possible word-referent associations) could be resolved by tracking the co-occurrence between words and objects across learning trials. Each training block was followed by a recognition test that required identifying the correct object associated to each trained word among four objects of the training set. Dissociations in word learning were assessed using the Bayesian Standardized Difference Test (Crawford & Garthwhite, 2007) by comparing the difference between each PWA’s performance on task X and task Y relative to the differences of the control group performance on these tasks. Results: All PWA showed a putative classical dissociation between phonological and semantic word learning (all p values ≤.035), presenting deficits in CSL but not in SS relative to the control group average performance (task X: SS=.66±.14; task Y: CSL=.81±.20). Conclusion: All 3 PWA met criteria for a putative classical dissociation in novel word learning suggesting that relatively more automatic, phonologically based statistical word learning appears to be better preserved in PWA than the ability to discover word-referent associations in referentially complex contexts.

Topic Areas: Disorders: Acquired, Meaning: Lexical Semantics