報告題目: Parsing and understanding wh-in-situ dependencies

主講人：Ming Xiang (向明, Associate Professor, University of Chicago)

報告地點：beat365文南樓116會議室

報告時間：2018年6月13日，14:00-16:00

關鍵詞：wh-in-situ, covert dependencies, processing dependency, interpretation bias, Bayesian reasoning

Title: Parsing and understanding wh-in-situ dependencies

Abstract

Much previous processing research has examined wh-fronting languages like English to ask how the indeterminacy of the gap position can be resolved in the incremental processing of filler-gap dependencies. In this talk, we look at the inverse problem that has received much less attention—how does the parser identify the correct scope position for a wh-in-situ phrase in a wh-in-situ language like Mandarin Chinese? First, we show some of the first experimental evidence, from eye tracking reading data, for the long-held assumption in the syntactic literature that a covert dependency is constructed between the wh-in-situ phrase and its scope position, which in Mandarin has no morpho-phonological reflex on the surface string. Second, we illustrate a clear locality bias in the online processing of wh-in-situ dependencies, suggesting that the parser serially searches through each intermediate clause edge CP position until getting to the right scope position, instead of simultaneously accessing all CP positions. Finally, we investigate the relationship between processing complexity and actual comprehension by zooming in on those cases for which processing complexity is greater for the distant vs. local scope position. Surprisingly, results from the truth value judgment task show a clear preference for the distant scope interpretation even though associating the wh-in-situ phrase with a distant scope position is more costly. We will approach the scope interpretation bias from a Bayesian perspective, and show that the interpretation preference is partially, but not completely, predicted by the Bayesian reasoning. In particular, the scope preference is predicted by the mirroring production bias, but it is not affected by the prior plausibility of each scope interpretation. 

Bio Sketch

Dr. Xiang is an Associate Professor of Linguistics and the director of the Language Processing Lab at the University of Chicago. Her research is geared towards better understanding the processing and neural mechanisms that support the rapid, real-time construction of sophisticated linguistic representations. She primarily works on sentence processing, including syntax, semantics and discourse comprehension. She has also worked on the syntax and semantics of Mandarin comparatives and the famous dou.

(向教授的個人網站https://lucian.uchicago.edu/blogs/mingxiang/)