Poster D59, Friday, August 17, 4:45 – 6:30 pm, Room 2000AB

Working Memory Filtering and Individual Differences in Second Language Aptitude

Chantel Prat1, Malayka Mottarella1, Brianna Yamasaki1;1University of Washington

Individual differences in second language aptitude (L2A) have been linked to working memory capacity (Miyake & Friedman, 1998), although the neurocognitive mechanisms proposed to explain the relation vary. One complication is that the construct of working memory has evolved dynamically, as have the tasks for measuring capacity. Additionally, little work has investigated the shared neurocomputations underpinning working memory capacity and L2A. In the current study, we adopt a selective attention view of working memory, and explore the hypothesis that L2A relies not only on the ability to maintain information in working memory, but also on the ability to flexibly filter or block task-irrelevant information. We propose that such an ability should be reflected by differences in fronto-striatal circuit functioning. To test this hypothesis, 34 adults completed a verbal working memory fMRI task modeled after McNab & Klingberg, 2008, in which words to-be-remembered appeared on a 3x3 grid in locations to-be-remembered. Memory load (3 vs. 5 words) and filtering demands (Filter = remember only words from a particular category vs. No-Filter = remember all words) varied orthogonally. Participants subsequently completed a 4-week French language training program in the laboratory using a virtual immersive software (Prat et al., 2016). Differences in French vocabulary and grammatical proficiency were assessed independently following training. Top-down filtering mechanisms were measured by contrasting activation during “Filter” and “No-Filter” instructions. Activation for the “Filter- 5” condition, in which 3 of 5 presented words needed to be remembered, was compared to the “No-Filter” conditions, which required remembering 3 or 5 words, to assess the effectiveness of filtering. Specifically, poor filtering would result in a Filter-5 condition that more closely resembled a No-Filter 5 condition (with all 5 words remembered); whereas good filtering would result in a Filter-5 condition that more closely resembled a No-Filter 3 condition (with only the 3 task-relevant words remembered). The three fMRI contrasts were then correlated with individual French proficiency following training. Results showed that better working memory filtering, indexed by activity in the left dorsolateral prefrontal cortex (DLPFC), left parietal lobe, and anterior cingulate cortex (ACC), predicted better subsequent vocabulary learning. Better grammatical learning was also correlated with filtering efficiency in the ACC. With respect to top-down preparation during filter instructions, smaller activation changes in the left hippocampus and parahippocampus predicted better vocabulary learning, whereas, greater involvements of bilateral parietal and right fronto-temporal regions predicted better grammatical learning. These results suggest that more effective top-down filtering, involving a network of memory and cognitive control regions, is associated with better L2 learning.

Topic Area: Multilingualism

Back