BLIT (Brain science of Language, Inference, and Thought research unit) Colloquium Series 2016
Jointly organized by BLIT and CELESE (Center for English Language Education for Science and Engineering) in Waseda University
Date: January 8th, 2015 (Friday)
Venue: Meeting Room 12, 5th floor, Building 51, School of Science and Engineering, Waseda University (Nishi Waseda Campus) http://www.sci.waseda.ac.jp/eng/access/
Title: Mismatch negativity and abstractness of sound representations
Speaker: Arild Hestvik (Professor, Department of Linguistics and Cognitive Science, University of Delaware)
Abstract: Underspecification theory states that speech sound representations in long-term memory are “economical” in the sense that they contain only the non-redundant subset of the features needed for pronunciation (Halle, 1959; Hall, 2007). Recent event-related potentials experiments (Eulitz & Lahiri, 2004; Lahiri & Reetz, 2010) provide evidence for this claim for German vowels. We examined the predictions arising from Lahiri et al’s theory for English, with a series of MMN studies of the /d/-/t/ contrast. In laryngeal underspecification theory, voiceless stops are linked to a voicelessness feature in lexical representations, while voiced stops are underspecified and have no voicing specification (Iverson & Salmons, 1995). Using the multiple-standard MMN paradigm (Phillips et al., 2000), where the MMN response of phonetic oddball tokens are compared to underlying phonological representations of standard token memory representations, we found that the MMN to phonetic oddball [d] was significantly greater than the MMN to phonetic oddball [t]. This observation matches the predictions of laryngeal underspecification theory.
We also tested the prediction arising from this literature that non-varying standards should remove the asymmetry in MMNs, because it would entail comparing purely phonetic (and therefore fully specified) representations ―these predictions were also confirmed. Finally, we will discuss the ultimate test of this theory, namely that a language with same contrast but with the opposite phonological underspecification system should give rise to the opposite MMN-asymmetry. This would provide compelling evidence that the observed MMN-asymmetries arise from abstract representations and not phonetic/perceptual contrasts. Japanese would be such a language, and we offer some preliminary data and a suggestion for an experiment.
This colloquium is presented by BLIT and supported by CELESE (Center for English Language Education for Science and Engineering) at Waseda University
For further information, contact Hiromu Sakai.