Processing of verbal versus adjectival agreement: Implications for syntax and psycholinguistics

Abstract

Linguistic theories distinguish between external and internal agreement (e.g., noun-verb agreement vs. noun-modifier agreement, the latter also known as concord) and model them using different mechanisms. While this distinction has garnered considerable attention in syntactic theory, it remains largely unexplored in experimental work. In an effort to address this gap, we conducted two studies of external/internal agreement in Russian using self-paced reading and eye-tracking while reading. We measured the response to violations generated when native speakers encounter a noun that mismatches the features on an earlier element inflected for agreement (verb, modifying adjective, and predicative adjective). Both experimental studies found strong effects of ungrammaticality: participants were sensitive to agreement mismatches between the agreeing element and the trigger. However, there was no interaction observed between the effect of grammaticality and the type of agreeing element, suggesting that, while participants are sensitive to mismatches, the processing of the mismatches does not differ between external and internal agreement. Despite the cross-methodological replication of the null interaction effect, interpreting this result is necessarily tentative. We discuss possible implications, should the result be further replicated by future high-powered studies. We suggest that this outcome may indicate that differences in real-time processing of internal vs. external agreement may not be observable in time-course measures, or that the lack of such differences constitutes support for analyses of agreement as a two-step process, with one step in syntax, and the other, post-syntactic. We invite future work to test these hypotheses further.