Hadas Kotek » Research

My research concentrates on different aspects of the syntax-semantics interface, using both traditional and experimental tools. This work can be divided into several different (but related) research projects.

Gender bias and stereotypes

This work explores the representation of gender in linguistics through the lens of examples sentences in textbooks and journal papers. We find pervasive gender bias both in the distribution of arguments and in the lexical choices in examples referring to male and female arguments. This work replicates and extends work conducted in the mid-90s on then-contemporary textbooks.

In earlier co-authored work, we replicated and extended an influential norming study of occupation-denoting nouns, showing that biases and stereotypes associated with men and women have hardly changed between 2003 and 2016.

2021: Cepeda, Paola, Hadas Kotek, Katharina Pabst, and Kristen Syrett. Gender bias in linguistics textbooks: Has anything changed since Macaulay & Brice (1997)?. Language 97(4): 678–702.
2021: Kotek, Hadas, Rikker Dockum, Sarah Babinski, and Christopher Geissler. Gender bias and stereotypes in linguistic example sentences. Language 97(4): 653–677.
2021: Kotek, Hadas, Sarah Babinski, Rikker Dockum, and Christopher Geissler. 2021. Gender stereotypes and inclusion in language teaching. Babylonia 1: 66–70.
2020: Kotek, Hadas, Sarah Babinski, Rikker Dockum, and Christopher Geissler. Gender representation in linguistic example sentences. Proceedings of the Linguistic Society of America (LSA), Volume 5, Number 1 (2020), 514–528. ISSN (online): 2473-8689.

Related (non-peer-reviewed) work on this issue can be found in the REIL guidebook, which provides resources for equity and inclusivity for linguistic conference organizers.

2017: Grant, Margaret, Hadas Kotek, Jayun Bae and Jeffrey Lamontagne. Stereotypical Gender Effects in 2016. CUNY Conference on Human Sentence Processing 30, MIT, April 2017. Full dataset here (csv)

NLP

This is joint work with members of the Siri and Language Technologies Data Science team and other collaborators.

2023: David Q. Sun, Artem Abzaliev, Hadas Kotek, Zidi Xiu, Christopher Klein, Jason D. Williams. 2023. DELPHI: Data for Evaluating LLMs’ Performance in Handling Controversial Issues. The 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP).
2023: Kotek, Hadas, Rikker Dockum, and David Q. Sun. Gender bias and stereotypes in Large Language Models. ACM Collective Intelligence conference.
2023: Xiu, Zidi, Kai-Chen Cheng, David Q. Sun, Jiannan Lu, Hadas Kotek, Yuhan Zhang, Paul McCarthy, Christopher Klein, Stephen Pulman, Jason D. Williams. Feedback Effect in User Interaction with Intelligent Assistants: Delayed Engagement, Adaption and Drop-out. The 27th meeting of the Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD).
2021: Patel, Alkesh, Joel Moniz, Roman Nguyen, Hadas Kotek, Nick Tzou, Vincent Renkens. 2021. MMIU: Dataset for Intent Understanding in Multimodal Assistant”. West Coast NLP (WeCNLP).
2021: Patel, Alkesh, Akanksha Bindal, Hadas Kotak, Christopher Klein, and Jason D. Williams. Generating Natural Questions from Images for Multimodal Assistant. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).
2020: Sun, David Q., Hadas Kotek, Christopher Klein, Mayank Gupta, William Li and Jason D. Williams. Improving Human-Labeled Data through Dynamic Automatic Conflict Resolution. The 28th International Conference on Computational Linguistics (COLING).

Arguments for top-down derivations in syntax

In joint work with Bob Frank, we collect arguments in favor of a top-down approach to linguistic derivations. Arguments come in the form of phenomena where properties of higher parts of the structure condition elements or operations in a lower domain. This would follow from a top-down approach to structure-building but require additional assumptions in bottom-up approaches.

2022: Frank, Rober, and Hadas Kotek. 2022. Top-down derivations: Flipping syntax on its head. Proceedings of the Linguistic Society of America (LSA), Volume 7, Number 1 (2022), 5264. https://doi.org/10.3765/plsa.v7i1.5264.

Scope restrictions in sluicing and ellipsis licensing

In joint work with Matt Barros, we defend a semantic identity account of ellipsis licensing. We notice a surprising asymmetry in sluicing constructions with quantified antecedents, where surface-scope antecedents can license a multiple sluice, but inverse-scope antecedents cannot. We argue that this finding is explained by semantic accounts of ellipsis licensing, where ellipsis is licensed when the sluice corresponds to an (implicit) Question under Discussion. We show that QuDs cannot be computed based on the truth-conditional content of the antecedents alone; instead, they must be computed only after (scalar) implicatures have been calculated and added to the common ground, along with the context of utterance. We additionally examine the commitments required of a syntactic LF-identity account of ellipsis licensing in order to account for multiple sluicing with quantified antecedents, and argue that accounts along these lines would run into serious trouble, making them practically untenable.

2019: Barros, Matthew and Hadas Kotek. Ellipsis licensing and redundancy reduction: A focus-based approach. Glossa 4(1): 100. 1-36.
2018: Kotek, Hadas and Matthew Barros. Multiple sluicing, scope, and superiority: Consequences for ellipsis identity. Linguistic Inquiry 49(4): 781–812.

Operations in Grammar

This work investigates the nature of (overt and covert) movement in natural language, how focus is calculated, and how the two interact. First, it is proposed that Association with Focus must involve covert focus movement with pied-piping. I additionally argue that the derivation of English wh-questions must involve both covert movement and focus-alternatives composition, and that the pattern and distribution of intervention effects in wh-questions suggests that syntactic derivations are better understood as constructed top-down, left-to-right, rather than bottom-up. Finally, I argue that intervention effects are the logical consequence of the inability of movement and focus to operate over the same structures, leading to the (correct) expectation that they are in fact quite a prevalent phenomenon which can be found in a variety of linguistic structures.

manuscript: Erlewine, Michael Y. and Hadas Kotek. Intervention tracks scope-rigidity in Japanese.
2018: Erlewine, Michael Yoshitaka and Hadas Kotek. Focus association by movement: Evidence from binding and parasitic gaps. Proceedings of Sinn und Bedeutung (SuB) 21.
2018: Erlewine, Michael Yoshitaka and Hadas Kotek. Covert focus movement with pied-piping: Evidence from Tanglewood. Linguistic Inquiry 49(3): 441–463.
2017: Erlewine, Michael Yoshitaka and Hadas Kotek. 2017. Movement and alternatives don’t mix: Evidence from Japanese. Proceedings of the Amsterdam Colloquium (AC) 2017.
2017: Erlewine, Michael Yoshitaka and Hadas Kotek. Intervention tracks scope-rigidity in Japanese. Proceedings of Logic and Engineering of Natural Language Semantics (LENLS) 14.
2017: Kotek, Hadas. Intervention effects arise from scope-taking across alternatives. Proceedings North East Linguistic Society (NELS) 47, eds. Andrew Lamont and Katerina Tetzloff. GSLA: Amherst, MA. Volume 2, 153–166.
2017: Kotek, Hadas. Dissociating intervention effects from superiority in English wh-questions. The Linguistic Review 34(2): 397–417.
2017: Kotek, Hadas. Questioning Superiority. A Pesky Set: Papers for David Pesetsky. MIT Working Papers in Linguistics 80, eds. Halpert, Claire, Hadas Kotek, and Coppe van Urk, 457–466. Cambridge, MA: MITWPL.
2016: Hadas Kotek. Covert partial wh-movement and the nature of derivations. In Glossa: a journal of general linguistics 1(1), 25: 1-19.
2016: Erlewine, Michael Yoshitaka and Hadas Kotek. Tanglewood untangled. Proceedings of Semantics and Linguistic Theory (SALT) 26, eds. Mary Moroney, Carol-Rose Little, Jacob Collard, and Dan Burgdorf, 224–243.
2015: Kotek, Hadas. Intervention everywhere! (handout, slides). Presentation at GLOW 38, Paris, April 2015.

The syntax/semantics of wh-questions

This work studies the syntax and semantics of wh-questions, concentrating on English and German multiple questions. I argue for covert scrambling of in-situ wh-phrases in English multiple questions, parallel to the overt scrambling of wh-in-situ in German. Evidence is provided from experimental data as well as offline judgments. My dissertation develops a new semantic framework for questions building on the syntax that I motivate. I am currently writing an LI Monograph based on this work.

2019: Kotek, Hadas. Composing Questions. Linguistic Inquiry Monographs series. Cambridge, MA: MIT Press.
2016: Kotek, Hadas. On the semantics of wh-questions. Proceedings of Sinn und Bedeutung (SuB) 20, eds. Nadine Bade, Polina Berezovskaya and Anthea Scholler.
2015: Kotek, Hadas. Against a feature driven view of wh-movement (slides). Presentation at the Deutsche Gesellschaft für Sprachwissenschaft (DGfS) 37, workshop: What drives syntactic computation? Alternatives to formal features, Leipzig, March 2015.
2014: Kotek, Hadas. Composing Questions. Doctoral dissertation, Massachusetts Institute of Technology.
2013: Kotek, Hadas and Martin Hackl. An experimental investigation of interrogative syntax/semantics. Proceedings of the 2013 Amsterdam Colloquium, eds. Maria Aloni, Michael Franke and Floris Roelofsen.

Non-interrogative uses of wh-words

Aside from their canonical interrogative use, wh-words in many languages are used to express a variety of quantificational expressions, including free relatives, NPIs, wh-quantification, and free choice items. A key question is what syntactic and semantic strategies are employed to quantify over the alternatives introduced by wh-words. To answer this question, this project studies a number of non-interrogative wh-constructions through original elicitation on Chuj (Mayan; Guatemala) and Dharamsala Tibetan (India).

2019: Kotek, Hadas and Michael Yoshitaka Erlewine. Wh-indeterminates in Chuj (Mayan). Canadian Journal of Linguistics 64(1): 62–101.
2018: Kotek, Hadas and Michael Yoshitaka Erlewine. Non-interrogative wh-constructions in Chuj (Mayan). Proceedings of the Workshop on the Structure and Constituency of the Languages of the Americas (WSCLA) 21.
2016: Kotek, Hadas and Michael Yoshitaka Erlewine. Unifying definite and indefinite free relatives: Evidence from Mayan. Proceedings North East Linguistic Society (NELS) 46, eds. Christopher Hammerly and Brandon Prickett, GSLA: Amherst, MA. Volume 2, 241–254.
2016: Erlewine, Michael Yoshitaka and Hadas Kotek. Even-NPIs in Dharamsala Tibetan. Linguistic Analysis: Special Issue on South Asian morphosyntax 40(3-4): 129–166.
submitted: Kotek, Hadas and Michael Yoshitaka Erlewine. Relative pronoun pied-piping in English non-restrictive relatives. Manuscript under review.

Intervention effects

This work investigates the nature of intervention effects in A-bar constructions. It is shown that intervention can be used as a diagnostic of underlying syntactic structures, in particular distinguishing between regions where focus-alternatives are computed and areas where covert movement has occurred. This diagnostic informs our understanding of a variety of A-bar and pied-piping phenomena in wh-questions, Association with Focus constructions, and relative clauses.

2016: Kotek, Hadas and Michael Yoshitaka Erlewine. Covert pied-piping in English multiple wh-questions. Linguistic Inquiry 47(4): 669–693.
2016: Kotek, Hadas and Michael Yoshitaka Erlewine. Intervention effects in relative pronoun pied-piping: experimental evidence. Proceedings of Sinn und Bedeutung (SuB) 20, eds. Nadine Bade, Polina Berezovskaya and Anthea Scholler.
2016: Erlewine, Michael Yoshitaka and Hadas Kotek. The structure and interpretation of non-restrictive relatives: Evidence from relative pronoun pied-piping. Proceedings of Chicago Linguistic Society (CLS) 51. (slides), Chicago, IL.: Chicago Linguistic Society, pp. 149-163.
2014: Kotek, Hadas. Intervention out of islands. Proceedings North East Linguistic Society (NELS) 44, eds. Jyoti Iyer and Leland Kusmer. GSLA: Amherst, MA. Volume 1, 234–246.
2014: Erlewine, Michael Yoshitaka and Hadas Kotek. Intervention in focus pied-piping. Proceedings North East Linguistic Society (NELS) 43, eds. Hsin-Lun Huang, Ethan Poole and Amanda Rysling. GLSA: Amherst, MA. Volume 1, 117-130. (Handout)
2013: Kotek, Hadas. Intervention, covert movement, and focus computation in multiple wh-questions. LSA Annual Meeting extended abstracts.

Tools for online linguistic surveys

Together with Michael Yoshitaka Erlewine, we develop a set of free, open-source tools that aid in the construction of linguistic surveys for Amazon’s Mechanical Turk, along with a supplied server-side component which allows researchers to host these randomized experiments on their own servers. The tools allow for the creation of a wide range of linguistic tasks, including linguistic grammaticality surveys, sentence completion tasks, and picture-matching tasks. These tools further help streamline the design of such experiments and assist in the extraction and analysis of the resulting data.

Tools available at: turktools.net.

2016: Erlewine, Michael Yoshitaka and Hadas Kotek. A streamlined approach to online linguistic surveys. Natural Language & Linguistic Theory 34(2), 481-495. (includes an online Appendix). Tools available at: turktools.net.

The syntax and semantics of most

This work focuses on the syntax, semantics, and processing behavior of most in subject position. We argue that most is ambiguous between a proportional reading, which can be paraphrased as ‘more than half,’ and a superlative reading, which is usually associated with the determiner the most. Evidence comes from an array of experiments and from judgment data.

2015: Kotek, Hadas, Yasutada Sudo, and Martin Hackl. Experimental investigations of ambiguity: The case of most. Natural Language Semantics 23(2): 119-156.
2012: Kotek, Hadas, Yasutada Sudo, and Martin Hackl. Many readings of Most (handout). Presentation at Chicago Linguistic Society (CLS) 48.
2011: Kotek, Hadas, Yasutada Sudo, Edwin Howard, and Martin Hackl. Most Meanings are Superlative. Syntax and Semantics 37: Experiments at the Interfaces, ed. Jeff Runner, 101-145. Bingley, UK: Emerald Publications.
2011: Kotek, Hadas, Yasutada Sudo, Edwin Howard, and Martin Hackl. Three Readings of Most. Proceedings of Semantics and Linguistic Theory (SALT) 21, eds. Ashton, Neil, Anca Chereches and David Lutz, 353-372.

Understanding Hebrew questions

This work concentrates on the syntax and semantics of Hebrew questions. I argue that Hebrew syntax distinguishes two kinds of wh-phrases: those headed by a wh-element and those headed by another element. I furthermore investigate the possible meaning multiple questions in Hebrew.

2014: Kotek, Hadas. Wh-Fronting in a Two-Probe System. Natural Language & Linguistic Theory 32(4): 1105-1143.
2012: Kotek, Hadas. Readings of Hebrew multiple questions. Proceedings West Coast Conference on Formal Linguistics (WCCFL) 30, eds. Nathan Arnett and Ryan Bennett, 216-225. Somerville, MA: Cascadilla Proceedings Project.

The acquisition of covert movement

This work concerns children’s acquisition of covert movement. We concentrate on the phenomenon of Antecedent Contained Deletion (ACD) in English. We argue that children are able to perform both short and long-distance covert movement to correctly interpret ACD. We furthermore argue that around the age of 5;6, children acquire a ‘Scope Matching Preference,’ such that the size of movement in the sentence should correspond to the size of ellipsis in that sentence, with mismatches being perceived as sub-optimal.

2013: Sugawara, Ayaka, Hadas Kotek, Martin Hackl and Ken Wexler. Long vs. Short QR: Evidence from the Acquisition of ACD. Proceedings of Boston University Conference on Language Development (BUCLD) 37. Somerville, MA: Cascadilla Press.

Morphological blocking effects in causative constructions

A project studying the structure of causative constructions in English and Japanese and arguing for a decompositional approach, as in Distributed Morphology, and against a lexical approach where causatives are formed and stored in the lexicon. Evidence comes from English judgment data and from English and Japanese experimental evidence from MEG and ERPs.

submitted: Kotek, Hadas and Michael Yoshitaka Erlewine. Blocking effects in English causatives.
ongoing: Morphological blocking effects in English and Japanese using MEG and ERPs (with Michael Yoshitaka Erlewine, Ayaka Sugawara, Shigeru Miyagawa, and Tohoku University Koizumi Lab)

Automaticity in L2 language acquisition

A project studying the automaticity of speech perception in second language acquisition of Japanese by native speakers of English, using ERPs and MEG. We show that as little as one semester of classroom experience with Japanese enhances discrimination of a phonemic contrast not found in English, but that L2 speech processing is still less automatic than for L1 listeners. The project furthermore examines whether changes in performance are more strongly related to the number of contact classroom hours, or the intensity of hours, and shows that duration of study plays an important role. (Lead investigator: Miwako Hisagi.)

2016: Hisagi, Miwako, Valerie L. Shafer, Shigeru Miyagawa, Hadas Kotek, Ayaka Sugawara, and Dimitrios Pantazis. Second-language learning effects on automaticity of speech processing of Japanese phonetic contrasts: An MEG study. Brain Research 1652: 111–118.
2014: Hisagi, Miwako, Valerie L. Shafer, Shigeru Miyagawa, Hadas Kotek, Ayaka Sugawara, and Dimitrios Pantazis. Perception of Japanese vowel contrasts by L1 and L2 learners of Japanese: An EEG study. Proceedings of Formal Approaches to Japanese Linguistics (FAJL) 7, MIT Working Papers in Linguistics #73, pp. 45-55.
2014: Hisagi, Miwako, Valerie L. Shafer, Shigeru Miyagawa, Hadas Kotek, Ayaka Sugawara, and Dimitrios Pantazis. Changes in Automaticity of Speech Processing of Japanese Phonetic Contrasts in Second-Language Learning: An MEG study. Proceedings of the Japanese Society for Language Sciences 2014.