Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Neural indices of asymmetries in processing stop consonant place distinctions: replication and extension
Poster A20 in Poster Session A, Friday, September 12, 11:00 am - 12:30 pm, Field House
Valerie Shafer1, Thomas Jacobsen2, Katarina Antolovic1, Elyse Sussman3; 1The Graduate Center, City University of New York, 2Helmut Schmidt University, 3Albert Einstein College of Medicine
Several studies have suggested that the nature of linguistic-phonological representations accounts for asymmetries in the neural mismatch negativity (MMN) measure to speech sounds which contrast by one phonological feature (e.g., “anterior” versus “coronal” place of articulation). Specifically, the claim is that some features are not specified in neural representations (e.g., “coronal”), and thus, result in a less robust MMN when serving as the central sound representation in an MMN oddball protocol. The current study employed event-related potentials (ERPs) to /b/ and /d/-onset stimuli to attempt to replicate the finding that the MMN is larger when the underspecified coronal (/d/) is the deviant, using three stimulus pairs: “ba” vs. “da” (Experiment 1), “baz” vs. “daz” (Experiment 2) and “bad” vs. “dad” (Experiment 3). Method: Twenty participants listened to the stimuli while the electroencephalogram (EEG) was recorded from 32 scalp sites, time-locked to the stimuli, flipping which stimulus served as the standard and deviant in each experiment. Data were cleaned and the normalized difference between each stimulus as the deviant and standard were subjected to a principal components analysis (PCA) followed by k-Means cluster analysis of the site factor scores. Two clusters reflected the MMN negativity over left versus right frontal region and two clusters reflected the MMN inversion over left and right inferior posterior sites. The data at the sites in a cluster were averaged and the four models (Right Fronto-Central (FC) Model: Fz, Cz, F4, FC1, FC2, C4, CP1, CP2; Left Frontal Model: F7, F3, FC5, T7, C3, CP5; Right Inferior model: RM, P8, CP6; Left Inferior Model: LM, P7) were subjected to statistical analyses. Results: In Experiment 1, a significantly larger MMN was observed to /dæ/ deviant compared to /bæ/ deviant, peaking at 200 ms. In Experiment 2, however, /dæz/ as the deviant did not show a larger negativity than /bæz/ as the deviant. In fact, frontal negativity was only robust between 350 and 400 ms. In Experiment 3, /dæd/ as the deviant elicited a larger MMN than /bæd/ as the deviant, peaking around 200 ms. The stimulus /dæd/ as a deviant also elicited an enhanced N1 pattern peaking at 88 ms at frontal sites and inverting at the mastoids. The other stimulus conditions also show an increase in N1 for the FC model, but the response did not invert at the inferior sites. Not observing an MMN at 200 ms for the /dæz/ deviant was unexpected because the /b/ vs /d/ difference was identical to the onsets used for /bæd/ and /dæd/. Discussion: These findings suggest a more complex explanation of asymmetries of the MMN, where acoustic-phonetic properties (e.g., formant frequencies of [z]) and/or lexical properties of the stimuli modulate neural discrimination. Future studies will need to systematically explore how phonetic features and lexical information modulate neural discrimination before concluding that asymmetries in MMN to various phonological feature contrasts support a model of phonological underspecification.
Topic Areas: Language Development/Acquisition,