Presentation

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

Neural mechanisms of learning and consolidation of novel morphologically derived words: Evidence from Hebrew speakers

Poster D41 in Poster Session D with Social Hour, Friday, October 7, 5:30 - 7:15 pm EDT, Millennium Hall

Upasana Nathaniel¹, Stav Eidelsztein¹, Kate Girsh Geskin¹, Brianna L. Yamasaki², Bracha Nir¹, Vedran Dronjic³, James R. Booth⁴, Tali Bitan^1,5; ¹University of Haifa, Israel, ²Emory University, Atlanta, GA, ³Northern Arizona University, Flagstaff, AZ, ⁴Vanderbilt University, Nashville, TN, ⁵University of Toronto, Toronto, Canada

We examined the neural mechanisms associated with learning of morphologically complex words within the Complementary Learning Systems (CLS) framework (McClelland et al., 1995), and its extension to word learning (Davis & Gaskell, 2009). According to CLS, new words are initially encoded within the hippocampus as episodic memories and over time consolidated into long-term lexical representations within neocortical structures. During consolidation repeated aspects of discrete events are extracted to form regularities and induce generalization (Lewis & Durrant, 2011). Extraction of repeated units and generalizing them is fundamental to learning morphological regularities. We therefore predict that morphologically complex words, which can be decomposed and generalized, will be learned and consolidated faster than morphologically simple words, and will show more reliance on neocortical regions, and independence from hippocampal areas, in the earlier stages of acquisition. We included two types of complex words: Linear which have a more salient structure, and Non-Linear, which is a prominent derivational structure in participants’ first language. 29 adult native-Hebrew speakers were trained across 4 sessions on 72 novel nouns from three morphological conditions: Complex Non-Linear (CNL, ‘gulbaki’=fisherman, root: g-l-b-k=fish, pattern: XuXXaXi=person), Complex Linear (CL, ‘zomgine’=birdwatcher, root: zomg=bird, suffix: -ine=person), and a mono-morphemic Simple condition (‘bunkut’=shoemaker) which served as a baseline. In the first and last sessions, training was followed by testing in an MRI scanner. In the last session, participants were tested on their ability to generalize learning. Behavioural results showed better overall learning and generalization of the morphologically complex conditions compared to the simple condition, particularly CL. fMRI results across scans revealed greater activity in left inferior frontal gyrus (LIFG) pars orbitalis for CNL > Simple. Only for CNL, overall performance across scans correlated with activity in LIFG pars opercularis, suggesting reliance on morphological decomposition. Similarly, only for CNL, improvement in performance during session 1 correlated with activity in LIFG pars triangularis in the first scan. The CNL condition also showed greater recruitment of hippocampus in comparison to the Simple condition. Furthermore, performance across sessions in the CL condition correlated with activity in hippocampus, suggesting reliance on episodic retrieval. Finally, the Simple condition showed greater activity compared to CL in temporal regions, associated with whole-word processing. Overall, behavioural findings show better learning and generalization of the morphologically complex conditions, indicating decomposition of morphologically complex words in the early stages of learning. The linear condition being more salient and easily learned (based on behavioural results) showed less cortical involvement in comparison to the non-linear condition, while processing of the non-linear structure involved higher load on mechanisms of decomposition in frontal regions. Finally, while our fMRI results did not reveal an effect of session, the learning of both morphologically complex conditions across sessions involved greater engagement of hippocampus than the Simple condition, suggesting that in contrast to our hypothesis, extraction of morphological regularities and neocortical activation were not associated with independence of hippocampal processing.

Topic Areas: Morphology, Multilingualism