Poster D64, Monday, March 27, 8:00 – 10:00 am, Pacific Concourse
Decoding phonology and lexicality from MEG data
Keith Doelling1, Bijan Pesaran1, David Poeppel2; 1New York University, New York, NY, 2Max Planck Institute for Empirical Aesthetics, Frankfurt, Germany
There exists a large body of research on effects of lexical status on speech recognition, including in the neuroimaging literature. Often, these studies have used pseudowords (i.e. items that conform to the phonological rules of a language but have no meaning) as a control to isolate semantic processing from acoustic and phonological analysis. Is this comparison accomplishing what it promises, i.e. does the brain process pseudowords just as it would words except for semantic analysis? Or is more processing required to deal with items that are not stored in our lexicon and could potentially be incorporated as a new word? Here we apply multivariate decoding techniques to MEG data in order to investigate the time course of semantic and phonological pathways when listening to words and pseudowords. We were able successfully to decode lexicality directly after item offset, suggesting a divergence of semantic pathways and functions for words and pseudowords at this timepoint. We then decoded a phonological feature (i.e. voicing of the first phoneme) as a marker of phonological processing. While classification accuracy is significantly above chance for both words and nonwords during the word, only the pseudowords show a resurgence of phonologically related activity after word offset. This re-activation coincides temporally with the onset of the lexicality effect. These results suggest that at least part of the divergence between word and pseudoword processing lies not only in the addition of semantic pathways for words but also in additional phonological (and potentially other) forms of processing of pseudowords.
Topic Area: LANGUAGE: Lexicon