Prof. Aniello De Santo Gave a Colloquium Talk Titled "What can Pseudo-relative clauses tell us about principles of parsing?" at Yale's Linguistics Department.
Prof. Aniello De Santo gave a colloquium talk titled "What can Pseudo-relative clauses tell us about principles of parsing?" at Yale's Linguistics Department.
Abstract:
(Joint work with Dr. So Young Lee, Miami University)
The cross-linguistic variability of attachment preferences for ambiguous relative clauses (RC) has been central to the study of the complex array of principles/mechanisms guiding ambiguity resolution in human sentence processing (Clifton Jr & Frazier, 1996; Cuetos & Mitchell, 1988; Gibson et al., 1996; Hemforth et al., 2000; Grillo & Costa, 2014; a.o).
Surveying past literature on RC attachment in languages like Italian and Spanish, Grillo & Costa (2014) observe an additional structural ambiguity available at the complementizer, beyond the classic low attachment vs. high attachment choice. That is, some languages allow for a structure --- Pseudo-Relative Clauses (PRs) --- denoting direct perception of events. While PRs and RCs are identical with respect to their surface strings, they are different at the semantic, syntactic and prosodic levels (Grillo, 2012; a.o.).
Grillo & Costa (2014) then posit that, when controlling for other factors, PRs are preferred over relative clauses (the PR-first hypothesis) during sentence processing, leading to only an apparent preference for “high attachment” in PR-licensing languages.
In this talk, I overview ongoing experimental and computational work investigating the effects of pseudo-relative clauses in Italian. First, I discuss data from self-paced reading examining how the availability of PRs within a language affects RC ambiguity resolution, and how it interacts with locality principles during online sentence processing (De Santo & Lee, 2022a; Lee & De Santo, u.r.). These SPR results add evidence to the PR-first hypothesis for Italian, while opening new questions about the interplay of structural ambiguity, locality, and language-specific properties. According to Grillo and Costa (2014), one possible explanation for a parser's preference for PRs is in terms of simplicity of their structure, and overall economy principles. Secondly then, I present work evaluating a naive implementation of this hypothesis by testing the predictions of a parser for Minimalist grammars for PR and RC structures in Italian (De Santo & Lee, 2022b).
I discuss the relevance of these results for PR-first explanations of the cross-linguistic variability of RC attachment biases, and highlight the role that interpretable computational models can play in evaluating the cognitive plausibility of economy considerations tied to fine-grained structural analyses.