2021年3月10号15:00-16:30—Shravan Vasishth—Case and Agreement Attraction in Armenian: Experimental and Computational Investigations

 Case and Agreement Attraction in Armenian: Experimental and Computational Investigations Shravan Vasishth 15:00 – 16:30, Wed, 10 March 2021           (Beijing, Hong Kong time)Venue: https://cuhk.zoom.us/j/779556638            https://cuhk.zoom.cn/j/779556638

About the SpeakerShravan Vasishth received a BA (Honours) in Japanese from Jawaharlal Nehru University (JNU), New Delhi (1989), an MA in Linguistics (1992) also from JNU, an MS in Computer and Information Science and a PhD in Linguistics from Ohio State University (2002), and an MSc in Statistics from Sheffield, UK (2015).  After a two-year postdoc at Saarland University’s Computational Linguistics department, he joined the faculty at the Department of Linguistics at the University of Potsdam, Germany, in 2004, where he has been full professor since 2008. He has worked as a patent translator in a law firm in Osaka, translating patents from Japanese to English. Fun fact: Shravan owns a real Japanese sword which he knows how to use; he has a second-degree black belt in the Japanese martial art Iaido (居合道, swordsmanship). When he was living in Japan, he won a state (Hyogo prefecture) and a national Iaido championship in Japan. He also practises Tai Chi and Qi Gong in Berlin, with Klaus-Dieter Zarn.
Vasishth’s research group develops computational models of human sentence comprehension. He is co-developer of the leading model of retrieval processes in sentence processing (Lewis and Vasishth, Cognitive Science, 2005; Vasishth et al., TiCS, 2019). A major research interest is in developing computational models of impairment in aphasia (Patil et al., 2016, Maetzig et al. 2018, Lisson et al., 2020). He is also interested in statistical theory and practice, particularly the applications of Bayesian methods to data analysis and computational modeling, and in open science, transparency in research, and good scientific practice. He has co-authored many tutorial articles on applying modern statistical methods in psycholinguistics (e.g., Schad, Betancourt, Vasishth, Psychological Methods, 2020), and methodological articles illustrating the consequences of basing inferences on underpowered studies in psycholinguistics (Vasishth et al., JML 2018). Shravan runs an annual week-long summer school in statistical methods for linguistics and psychology that covers both frequentist and Bayesian methods and attracts approximately 350 applications every year (https://vasishth.github.io/smlp/). 
Case and Agreement Attraction in Armenian: Experimental and Computational Investigations
Shravan Vasishth Department of Linguistics, University of Potsdam, GermanyEmail: vasishth@uni-potsdam.de        URL: vasishth.github.io
In reading studies, agreement attraction is often taken to refer to the fact that the auxiliary verb are in (1) is read faster than in (2):(1)The key to the cabinets are on the table.(2)The key to the cabinet are on the table.What’s surprising about this finding is that both sentences are equally ungrammatical. Why would (1) be easier to process than (2)? There are various theoretical explanations for this effect, ranging from a presumed misretrieval of the non-subject noun cabinets (the retrieval account), to feature overwriting of the subject noun’s number feature (hereafter, encoding accounts). In this talk, I discuss two issues related to agreement attraction. First, can distinctive case marking on the nouns attenuate interference effects? Previous studies have suggested that, in production, distinctive case marking on noun phrases reduces agreement attraction; theory predicts that this should happen in sentence comprehension as well. To answer this question, we conducted three attraction experiments in Armenian, a language with a rich and productive case system. The experiments showed clear attraction effects, and they also revealed an overall role of case marking such that participants showed faster response and reading times when the nouns in the sentence had different cases. However, we found little indication that distinctive case marking modulated attraction effects. We present a theoretical proposal of how case and number information may be used differentially during agreement licensing in comprehension. The second issue we consider is: which theoretical account explains agreement attraction data better, encoding or retrieval? We carried out a self-paced reading study in which we elicited information about which noun might have been retrieved; then we computationally implement several competing models of agreement attraction, and show through a quantitative evaluation of the models that encoding accounts  provide a superior explanation for the data than retrieval theories, at least for the Armenian data considered here. 
Background reading:Serine Avetisyan, Sol Lago, and Shravan Vasishth. Does case marking affect agreement attraction in comprehension? Journal of Memory and Language, 112, 2020. doi: 10.1016/j.jml.2020.104087Dario Paape, Serine Avetisyan, Sol Lago, and Shravan Vasishth. Modeling misretrieval and feature substitution in agreement attraction: A computational evaluation. 2020. Preprint: https://psyarxiv.com/957e3/
Virtual Psycholinguistics Forum: (https://cuhklpl.github.io/forum.html)