Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Functional correlates of word learning and inflection generalisation
Poster B104 in Poster Session B, Tuesday, October 24, 3:30 - 5:15 pm CEST, Espace Vieux-Port
This poster is part of the Sandbox Series.
Kshipra Gurunandan1,2, Lydia Vinals1, Matthew H. Davis1; 1University of Cambridge, 2BCBL Basque Center on Cognition, Brain and Language
Naturalistic word learning in languages such as English often involves acquisition of quasi-regular systems consisting largely of systematic, regular forms (for example, in the case of past-tenses: work–worked, live–lived) along with a smaller number of high frequency irregulars (e.g. bring–brought, keep–kept). In the current study, we will examine the neural correlates of morphological learning and generalisation of quasi-regular inflectional morphology in the framework of a Complementary Learning Systems (CLS) account of lexical acquisition (Davis and Gaskell, 2009). The CLS account posits a two-stage learning process in which rapid initial familiarisation is followed by slow lexical consolidation, underpinned by complementary systems in the hippocampus and the neocortex. Differences in the encoding schemes of these two systems may have particular implications for learning of regular and irregular forms: hippocampal sparse encoding and pattern separation may be well-suited for mapping of exceptional irregular forms, while a distributed neocortical representational scheme may better capture systematically regular forms. These differences predict distinct patterns of generalisation for regular and irregular forms before and after lexical consolidation. In a dataset from Vinals-Castonguay (2018), twenty-two adult participants aged 18–34 (mean age = 23, SD = 4, 14 female) were trained and tested in an artificial language learning paradigm in 3 sessions over a 9-day period. On day 1, participants learned novel names for professions (e.g. gleet, shiln), followed by training with separate sets of gender-marked plural affixes (e.g. -aff, -opp; gleetaff, shilnopp) on days 8 and 9. The plurals varied in their phonological consistency (diverse, ambiguous or consistent), and their type and token frequency (high or low). Training sessions involved word-repetition and picture naming with feedback, and testing sessions consisted of picture naming and 2-alternative forced choice (AFC) tasks. On day 9, testing involved generalisation to previously unseen words (plural elicitation and 3AFC tasks), followed by fMRI scanning during a 4AFC task to compare neural correlates of the plurals that were learned 24h prior versus just before scanning. In our re-analyses, we will test behavioural changes in generalisation of regular and irregular forms before and after lexical consolidation, predicting differences indicative of overnight changes in their underlying representations. We will perform three types of fMRI analyses: ROI analyses to examine hippocampal and neocortical contributions to encoding of regular and irregular forms, Representational Similarity Analysis (RSA) to examine effects of initial learning and consolidation-related changes in neural representations of the newly-learned words, and Psychophysiological Interactions (PPI) analysis to examine changes in hippocampal-neocortical connectivity and neocortical-neocortical connectivity. Based on the CLS model, we predict greater hippocampal involvement in encoding of irregular forms versus greater neocortical involvement for regular forms; contrasting patterns of changes in representational similarity for regular and irregular forms before and after consolidation; and post-consolidation weakening of hippocampal-neocortical connectivity and strengthening of neocortical-neocortical connectivity. These results will shed new light on the long-term memory processes underlying learning and consolidation of novel linguistic forms with varying structural and statistical properties.
Topic Areas: Language Development/Acquisition, Morphology