Slide Slam

< Slide Slam Sessions

Slide Slam L10

The Role of RTPJ and LATL in Processing Agreement: Evidence from Hindi

Slide Slam Session L, Thursday, October 7, 2021, 6:00 - 8:00 am PDT Log In to set Timezone

Dustin Chacón1, Subhekshya Shrestha1, Brian Dillon2, Rajesh Bhatt2, Diogo Almeida1, Alec Marantz1,3; 1New York University Abu Dhabi, 2University of Massachussets, 3New York University

Verb agreement is important because it facilitates identifying the arguments of a verb. Theories differ on whether agreement is interpreted by actively encoding information pre-verbally or by retrieval processes post-verbally. Identifying the neural generators of agreement has been challenging, as they are confounded with other syntactic relations, e.g., thematic roles. Here, we exploit Hindi split-ergative agreement, in which verbs may agree in gender with the subject, object, or neither depending on case marking. By examining the neural response to bare objects that control agreement vs. those that do not in magnetoencephalography (MEG), we show that RTPJ and LATL are both implicated in preverbal agreement processes. [MATERIALS] We prepared 46 sets of 8 items, manipulating SubjCase {Erg/Bare}, ObjCase {Acc/Bare}, and Verb Cloze {High/Low}. Sentences consisted of a subject, object, and verbal complex. NP genders differed and were counterbalanced. [PROCEDURE] MEG signals were recorded for 12 (/24 planned) native Hindi speakers while reading Standard Hindi sentences. Each phrase was displayed 900ms on 100ms off. After 25% of trials, participants completed a picture verification task. [RESULTS] Five lateral ROIs per hemisphere plus bilateral vmPFC were selected. Test statistics were computed over average activation levels (dSPM) in each time point per ROI, and in each time and source in the whole brain. We then conducted cluster permutation tests to identify significant clusters. Test statistics were F-values from ANOVAs (dSPM ~ SubjCase × ObjCase in object NP time window, 0-1000ms; … × Cloze in verb time window, 1000ms–2000ms), or t-values resulting from one-tailed t-tests over beta coefficients resulting from regressions at each time point and source per subject (dSPM ~ SubjCase × ObjCase + NP Gender + Verb Gender + List Position + log(Lexical Frequency) + Verb Entropy, 0–1000ms; … + Cloze, 1000–2000ms). ROI analyses revealed significant clusters in right temporo-parietal junction (RTPJ) in object time window: ObjCase × SubjCase interaction in ANOVAs (p = 0.01; 360–440ms) and ObjCase effect in regressions (p = 0.04, 360–430ms). Cloze results were identified in ventro-medial prefrontal cortex (vmPFC) (p = 0.01, 1480–1580ms; p = 0.01, 1630–1690ms) in ANOVAs, and left posterior temporal lobe and occipital lobe (LPTL+lOCC) in regressions (p = 0.04, 1400–1450ms; p = 0.05, 1330–1380ms. There was an effect of Verb Gender in the pre-verbal object time window in left anterior temporal lobe (LATL) (p = 0.03, 580ms–630ms), also in whole-brain regressions (p = 0.05; 510–920ms). No effects were found in other ROIs. [DISCUSSION] These preliminary findings suggest that RTPJ is sensitive to case around 400ms after onset of object, and LATL is sensitive to the predicted gender feature 200ms afterwards. RTPJ is implicated in the ventral attentional network and in shifting attention to relevant stimuli. We suggest that processing the lexical features of the object activates its morphosyntactic features. If the structure requires object agreement, then RTPJ is engaged to shift attention from the subject’s features to those of the object. These features are represented in LATL, a "conceptual hub" implicated in linguistic composition.

< Slide Slam Sessions

SNL Account Login

Forgot Password?
Create an Account

News