Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams
Brain activity in Hindi & Nepali reflects language-adapted processing strategies
Poster C7 in Poster Session C, Friday, October 7, 10:15 am - 12:00 pm EDT, Millennium Hall
Dustin A. Chacón1,2, Subhekshya Shrestha1, Brian Dillon3, Diogo Almeida1, Rajesh Bhatt3, Alec Marantz1,4; 1New York University Abu Dhabi, 2University of Georgia, 3University of Massachusetts, 4New York University
[INTRODUCTION] A major goal of the cognitive neuroscience of language is to identify the neural substrates of syntax. However, “syntax” covers an array of processes and representations that vary between languages. Here, we leverage similarities and differences between two subject-object-verb languages with different agreement systems, Hindi and Nepali. In Hindi, verbs agree with the first NP that does not have a case suffix (‘bare’). In Nepali, verbs agree with the subject regardless of case. In two parallel MEG studies, we show that case features affect MEG activity in anterior perisylvan frontotemporal regions in both languages before the verb. In Nepali, only object NP case affects this activity. In Hindi, activity in right regions demonstrate an interaction of subject and object case, with greater negative activity in sentences with one bare NP. We suggest right frontotemporal regions facilitate selection of an agreement controller. Broadly, these results demonstrate that comprehenders make maximal use of available information as a function of the language-specific syntactic rules, which is directly reflected in brain activity. [METHODS] MEG activity of 22 Hindi and 18 Nepali speakers was continuously recorded. Participants read 46 (Hindi) and 50 (Nepali) three-word sentences per condition distributed in a within-subjects 2×2×2 design, manipulating SubjC (ERG/BARE), ObjC (DAT/BARE), and Verb Cloze (HIGH/LOW). Verb Cloze was included as an “N400” design as a validity check. Each phrase was presented using rapid serial visual presentation (RSVP) and involved a picture verification task. [RESULTS] Regressions by participant were conducted in object NP (0ms–1000ms) and verb (1000ms–2000ms) time windows, consisting of the critical manipulations (object NP: SubjC × ObjC; verb: Verb Cloze) and nuisance variables. One-sample t-tests were conducted over the resulting beta values, and spatiotemporal cluster tests were conducted over t-values in 3 time windows (object NP: 300-600ms; 600-900ms; verb: 300-500ms) in bilateral frontotemporal regions (object NP time window) and left temporal lobe (verb time window). We then fit a linear mixed effects model to the activity in each for both languages for each cluster, including factors of interest (object NP: SubjC × ObjC; verb: Verb Cloze), a between-subject factor Language and its interactions, and random effects. [OBJECT NP] A cluster (p < 0.0001) located in right anterior temporal, inferior frontal, and ventromedial prefrontal cortex ~500ms post-NP onset showed greater negative activity in conditions with one bare NP in Hindi, reflected as a three-way interaction between SubjC×ObjC×Language (p = 0.01). A spatially and temporally overlapping cluster (p < 0.001) showed a main effect of object case, with greater positive activity for bare object NPs in both languages (p = 0.02). [VERB] A cluster (p = 0.01) located in left anterior inferior temporal lobe ~400ms showed a main effect of Verb Cloze (p < 0.01). [CONCLUSION] Superficially-similar structures may provide different information to the comprehender depending on the grammatical rules of their language. Brain activity reflects how comprehenders make maximal use of this information.
Topic Areas: Syntax, Control, Selection, and Executive Processes