Maintenance and retrieval in the processing of filler-gap dependencies: The view from a language with grammaticized resumption (Israel Science Foundation Grant)

This project targets two main questions: (1) What retrieval cues does the parser use in resolving filler-gap dependencies? are these the same cues used for retrieval in other syntactic dependencies? (2) Do island configurations, restricting filler-gap dependency formation, consist of structures in which filler maintenance or retrieval are impossible due to the structure's complexity, or do islands reflect grammatical constraints?

To date, research on memory mechanisms in filler-gap dependencies focused on a relatively small number of languages. Studying Hebrew, a language with grammaticized resumptive pronouns (RPs), can shed new light on these research questions. RPs provide the comprehender with the filler's agreement features, thus allowing us to test whether these features can serve as retrieval cues in filler-gap dependency resolution, on a par with other dependencies (e.g. subject-verb agreement). In addition, Hebrew RPs participate in the formation of grammatical island structures, allowing us to confirm whether maintenance and retrieval are possible in island configurations when these are grammatical.  


Feature maintenance and feature-driven integration in the processing of syntactic dependencies                                              (German Israel Foundation Young Scientists Grant)

Successful comprehension of filler-gap dependencies requires use of working memory resources to maintain the filler or some of its features throughout the dependency, and/or to retrieve it, or some of its features, at the gap site. Previous work has suggested that both maintenance and retrieval seem to be necessary, for different types of information associated with the filler (e.g. Wagers & Phillips, 2014). However, it is still not known precisely what information is maintained throughout the processing of a dependency, as well as what the implications of this maintenance are. 

This project targets three research questions: (1) What information is actively maintained during the processing of filler-gap dependencies? Specifically, is information with regard to the animacy of the filler maintained? (2) Is dependency resolution triggered by maintained information? i.e., is resolution attempted only with verbs matching the filler on this information? (3) What is the electrophysiological signature of feature maintenance and interference?


Remote sensing of negative mood tendencies and enhancing resilience in the general population: advanced computational models, novel behavioral tasks, eye-tracking, virtual reality and physiological measures                                                                   (Joy Ventures Grant, in collaboration with Dr. Tom Schonberg and Dr. Jonathan Berant)

In this collaborative research, we aim to adopt a computational approach to assess emotions and then enhance personal resilience from depression in the general population. We propose to develop a novel system combining eye-tracking, electrophysiological measures, natural language processing and machine learning algorithms to diagnose negative emotional states. Following the diagnosis of emotional states we will use a novel training paradigm to enhance mood.