Many languages of the world show grammatical agreement of some sort, where the grammatical features of a linguistic constituent are visible on another associated constituent. Recent psycholinguistic research has uncovered a number of semantic and syntactic factors that influence how we compute agreement during language production, though much less is known about how these factors interact during language comprehension. Understanding the processing of agreement is also of critical importance in the field of second language acquisition (SLA), as it is frequently observed that second language (L2) learners show considerable trouble with grammatical agreement, often despite many years of formal language instruction and immersion in an L2 environment. Many SLA scholars have proposed that L2 learners' problems with agreement are the result of processing problems, though little research has directly investigated this.
This dissertation research project investigates how subject-verb agreement is computed during real-time sentence comprehension by native speakers of English and by advanced second language learners of English from different linguistic backgrounds, specifically focusing on how a sentence's syntactic complexity influences agreement processing. The experiment uses grammatical agreement phenomena to additionally investigate recent hypotheses that nonnative grammatical contrasts are unacquirable by L2 learners and that L2 learners' language processing systems are radically different from native language systems.
These questions will be addressed by recording event-related brain potentials (ERPs) while participants read sentences in English. ERPs have been shown to be highly sensitive to many facets of native language agreement processing, though no studies have investigated how syntactic complexity affects agreement computation. Additionally, ERPs have only recently been used to study L2 processing. There is thus a paucity of knowledge about the neural substrates underlying L2 agreement processing or how agreement, syntax, and L1 experience interact during L2 processing. The inclusion of modern psycholinguistic and neurophysiological techniques in studying L2 acquisition and processing will therefore complement the existing rich theoretical linguistic literature on L2 grammatical representation. The outcomes of this research will be of interest to cognitive neuroscientists interested in language and to theoretical linguists and psycholinguists studying agreement, its relationship to syntactic structure, and how agreement interacts with syntactic parsing systems. The results will additionally be of interest to theoretical and applied SLA researchers, having potentially significant implications for language acquisition theory and language pedagogy.
The primary goals of the research conducted under this grant were to understand the neural correlates of language processing and comprehension in native English speakers and bilinguals who have acquired English as a second language. In particular, we investigated how individuals process long-distance grammatical dependencies using event-related brain potentials. Our particular questions revolved around how the grammatical complexity of a sentence impacts comprehension, what the neural correlates of syntactic complexity are, whether second language learners show similar profiles of complexity modulations, and whether the neural resources underlying second language syntactic comprehension maybe similar to or are fundamentally different from those of native language comprehension. To answer these questions, we recorded the brainwaves of native English speakers and second language learners of English while they read English sentences. The sentences were either grammatically correct, or violated rules of subject-verb agreement. In order to identify what types of information individuals use when establishing long-distance agreement dependencies (e.g., syntactic, morphological), we also manipulated the syntactic complexity of the subject noun phrases of the sentences as well as the number and type of morphological cues to the agreement dependency. Of particular interest in the group of second language learners was whether they used the same morphological and syntactic cues to agreement dependencies as native English speakers, as some theories of second language learning posit that second language processing uses fundamentally different processing strategies and neurocognitive resources from native language processing. Results from native English speakers showed that processing of subject-verb agreement was subject to significant interference from nouns which intervened between the head noun of the subject and the verb when the morphological number marking on the intervening noun conflicted with that of the head noun. There was an additional effect of syntactic complexity, such that brain responses to agreement violations were larger following more complex subject noun phrases. However, contrary to findings from the speech production literature, the degree of interference from intervening nouns was not modulated by the complexity of the subject noun phrase. Overall this indicates that we rapidly integrate multiple streams of information when establishing agreement dependencies. We argue that language comprehension relies both on bottom-up feature checking mechanisms as well as rapid top-down syntactic predictions, and moreover, that the mechanisms underlying grammatical agreement processing in comprehension are partly distinct from those used in speech production. Results from the second language learner group largely paralleled native speakers, as there were effects of both interference and syntactic complexity. This suggests that second language processing may not be as fundamentally different as some have suggested, as bilinguals processing their second language rapidly integrated multiple streams of information and showed similar modulations of brain responses to agreement violations as those seen in native speakers. Perhaps the most striking findings from this research were the discovery of marked individual differences in how individuals processed agreement, both among the native English speakers and second language learners. Previous work using event-related potentials to study agreement had shown that agreement violations consistently elicit large positive-going brain responses known as the P600 effect. However, our data showed that individuals’ brain responses varied along a continuum of responses between the P600 effect and the N400 effect, which is more typically elicited by semantic, pragmatic, or conceptual violations. While this variation existed in all groups studied, it was most pronounced in the bilingual group. In order to investigate this variation among the bilinguals, we calculated two new measures which differentially quantify the size and type of brain response to agreement violations. The Response Magnitude Index is a measure of the overall neural sensitivity to agreement violations, while the Response Dominance Index is a measure of the relative dominance of the N400 or P600 effect in an individual’s brain response. Multivariate analyses showed that the size of a learner’s brain response was most strongly associated with English proficiency, while the type of brain response a learner showed was most strongly associated with the individual’s motivation to learn English and age of arrival in an English-speaking environment. Thus, in addition to investigating theoretical questions related to language processing and language learning, this research yielded methodological innovations related to the quantification of individual differences in brain responses.
