Cue based semantic prediction in Hindi: an ERP study

Abstract

The human brain is a predictive machine (Clark, 2013), and it is well established in psycholinguistics that it routinely and inherently predicts upcoming words during realtime sentence comprehension. some central questions in psycholinguistics have been how predictive processing takes place and which linguistic information triggers what kind of predictive processing. Previous ERP studies have explored predictive processing by exploiting the sentential contextual, discourse, pragmatic or world knowledge information during the anticipation of upcoming information. Very few studies have studied the role of syntactic information like sub-categorisation, transitivity, and morpho-syntactic information during the anticipation of upcoming nominal or verbal arguments (Chow & Phillips, 2013; Friederici & Frisch, 2000; Wicha, Moreno, & Kutas, 2004; Yano, 2018). This thesis aimed to explore the neural underpinnings of the role of syntactic (caseclitics) and structural information (thematic structure) in the semantic prediction of the upcoming words in a Hindi sentence. Hindi is a split-ergative language with a rich morphological case-marking system. According to Kar (2006); Sahay (1986), these case-markers of Hindi present selectional restrictions to upcoming nominal or verbal arguments. In this thesis, we explored the role of kaarak-chihn ’case-clitics’ of Hindi (especially the nominative ’ϕ’, ergative ’-ne’, accusative ’-ko’ , & locative-in ’-meM’) of Hindi in anticipation of upcoming words. All the thesis experiments have examined the interaction of these case-clitics’ predictive nature with the upcoming word’s plausibility. The first study examined the effect of the predictive nature of nominative and ergative clitics on the semantic prediction of the sentence-final verb. The results showed that the additional information of transitivity and telicity carried by ergative clitic -ne causes its processing differently than the nominative clitic. The second study investigated the interaction of plausibility information with the thematic structure of arguments and the selectional restriction of the nominative ϕ, accusative -ko and ergative -ne. The result exhibited the pattern for the thematic reversal anomaly similar to the nominative-accusative languages, and the accusative-marked NP2 seemed to increase the complexity of semantic prediction due to its specificity nature. The third study examined the interaction of animacy with the nominal restriction of locative clitic -meM for its complementary argument. The results suggested that the locative clitic presented a split pattern for predicting upcoming nominal words and demonstrated a different neural behaviour than the other clitics like nominative, accusative and ergative. In sum, the results of this thesis work demonstrate that the different case-clitics of Hindi are processed differently by the human brain, and they are highly sensitive to the linguistic information carried by the nominals with which they occur in the sentence. Further, many other linguistic factors like animacy, thematic roles, abstractness etc., seem to influence the predictive processing during meaning computation.