Research

The plan is to look at some data collected as part of Project Prakash, study the performance of test and control subjects on some simple visual tasks, and examine how computational models might give us insight into variations in attentional mechanisms in these populations.

Abstract: Light verb constructions (e.g. give a sigh, take a walk) are a linguistic puzzle, as they consist of two predicating elements in a monoclausal structure. In the theoretical linguistics literature, there has been much interest in the linguistic analysis of such constructions across a range of grammatical frameworks. One such proposal is event co-composition, where the argument structures of noun and light verb merge, resulting in a composite argument structure, which has been shown to incur processing cost in English and German. In contrast to these languages, a larger proportion of the predicates in Hindi are light verb constructions. Hence, we may ask whether a Hindi speaker’s experience with light verb constructions allow them to go through the same co-composition operation faster than a speaker of English. Our results show that Hindi speakers are adept at the process of using light verb constructions to verbalize predicates, much more so than speakers of Germanic languages. We argue that these data provide evidence for a case of specific linguistic experiences shaping cognition: cost disappears with practice.

Abstract: Formal constraints on crossing dependencies have played a large role in research on the formal complexity of natural language grammars and parsing. Here we ask whether the apparent evidence for constraints on crossing dependencies in treebanks might arise because of independent constraints on trees, such as low arity and dependency length minimization. We address this question using two sets of experiments. In Experiment 1, we compare the distribution of formal properties of crossing dependencies, such as gap degree, between real trees and baseline trees matched for rate of crossing dependencies and various other properties. In Experiment 2, we model whether two dependencies cross, given certain psycholinguistic properties of the dependencies. We find surprisingly weak evidence for constraints originating from the mild context-sensitivity literature (gap degree and well-nestedness) beyond what can be explained by constraints on rate of crossing dependencies, topological properties of the trees, and dependency length. However, measures that have emerged from the parsing literature (e.g., edge degree, end-point crossings, and heads’ depth difference) differ strongly between real and random trees. Modeling results show that cognitive metrics relating to information locality and working-memory limitations affect whether two dependencies cross or not, but they do not fully explain the distribution of crossing dependencies in natural languages. Together these results suggest that crossing constraints are better characterized by processing pressures than by mildly context-sensitive constraints. Assessing corpus evidence for formal restrictions on crossing dependencies. Himanshu Yadav, Samar Husain and Richard Futrell. Computational Linguistics. 2022. https://doi.org/10.1162/coli_a_00437