Jon Brennan recently presented two papers at the Annual Meeting of the Society for the Neurobiology of Language that was held during August in Amsterdam. These two papers (both presented with colleagues from UM and other institutions) are typical of the breadth of Jon's research.  The first paper uses MEG technology to explore the processing of lexical access during listening. The second paper shows how simple language processing tasks, as measured with MEG, could be used as a tool for diagnosing autism. This paper also shows that basic linguistic research can have real life clinical applications - a message that the field of linguistics would do well to communicate better. The full titles and abstracts for both presentations are given below. 

 

Brennan, J., Lignos, C., Cantor, M., Embick, D., & Roberts, T. P. L. Oscillatory dynamics to time-stretched speech during lexical decision. Poster presented at the 2014 Annual Meeting of the Society for the Neurobiology of Language, Amsterdam, August 2014.

Neurophysiological models of spoken word recognition implicate peri-auditory activation within the first few hundred milliseconds after stimulus onset (Friederici 2012), however the mechanism by which these regions incrementally map spoken stimuli on to candidate lexical entries remains poorly understood. Recent attention to neural oscillations suggest that alpha-centered power reduction (event-related de-synchronization; ERD) beginning between 250-300ms after stimulus onset reflects early stages of lexical processing (e.g. Wang et al. 2012; Brennan et al., 2014); left unspecified is whether this activation is sensitive to the onset of lexical activation (reduced power = earlier activation) or the speed of activation (reduced power = more rapid lexical convergence). We predict that stimuli with more rapidly presented cues will lead to early onset of lexical activation. We quantify the mapping of speech cues to lexical identity using cohort entropy, an information theoretic measurement of uncertainty over possible lexical items given partial speech input. Recent work has shown that cohort entropy correlates with evoked activity in peri-auditory cortex with a 200ms lag (Ettinger et al. 2014). Using magnetoencephalography (MEG), we test for oscillatory dynamics sensitive to lexical activation onset; using time-stretched speech we manipulate the speed that sub-lexical features incrementally unfold while keeping lexical identity constant. We find that alpha power and gamma coherence are sensitive to time-stretched speech; cohort entropy and power in the alpha, beta, and gamma range correlate at latencies suggesting that these components may reflect later stages of lexical processing.

 

Brennan, J., Wagley, N., Ugolini, M., Richard, A., Kovelman, I., Bowyer, S., & Lajiness-O’Neill, R. Neural coherence during natural story listening as a biomarker for Autism. Poster presented at the 2014 Annual Meeting of the Society for the Neurobiology of Language, Amsterdam, August 2014.

A major thread in current neuroimaging research on Autism Spectrum Disorders (ASD) focuses on biomarkers sensitive to diagnostic status and/or treatment efficacy that can be passively measured in pediatric populations. Recent work on resting state brain activity has suggested that ASD may be associated with impaired functional connectivity between brain networks (e.g. Dinstein et al., 2011; Coben et al., 2008; Lajiness-O’Neill et al., 2014), raising the possibility that a short segment of electrophysiological activity, easily collected with minimal task demands, may provide clinically useful information. In parallel, neural responses to simple auditory stimuli and speech sounds also show atypical patterns that correlate with clinical language scores (Roberts et al., 2010; Edgar et al., 2013). This correlation raises the possibility that connectivity patterns during language processing may be more divergent between children with ASD and those with neurotypical (NT) development compared to resting state connectivity, where variance in brain states between individuals is expected to be high. Listening to a short story is a simple, familiar and, for many children, enjoyable activity that requires minimal extraneous task demands. Accordingly, we tested whether neural coherence, measured using magnetoencephalography (MEG), during passive story listening could distinguish children with and without ASD. We find that data derived from passive story-listening improves diagnostic accuracy compared to data derived from resting state brain activity.