About

Gregory Scontras is an Associate Professor and Undergraduate Director in the Department of Language Science at the University of California, Irvine, where he also directs the meaning lab. His research centers on the theories of natural language, specifically focusing on speakers’ mental representations of words and how these representations are employed in computing sentence meaning. His work aims to deepen the understanding of language and the psychological systems that influence it, with recent projects exploring measurement in linguistic phenomena, the role of context in ambiguity resolution, and the profiles of heritage speakers, who are unbalanced bilinguals. Originally from Maine, Scontras earned his BA in Linguistics and Philosophy from Boston University, followed by a PhD in Linguistics from Harvard University, where he wrote his thesis on the semantics of measurement in 2014. He then joined Noah Goodman’s Computation and Cognition Lab at Stanford University in the Department of Psychology. His academic and research pursuits are reflected in his extensive publication record, which includes contributions to understanding adjective ordering, heritage language processing, social inference, and the semantics of number morphology, among other topics.

Research topics

• Computer Science
• Philosophy
• Linguistics
• Psychology
• Mathematics
• Statistics
• History

Selected publications

• How Novices Interpret Generalizations and How Experts Use Them

  Cognitive Science · 2026-03-01

  article

  Generalizations, such as "ducks are birds" and "ducks carry avian flu," are a common way of conveying information about the world, yet their implied prevalence-how broadly they should be applied-can vary widely. To interpret how broadly a generalization should be applied, listeners rely on prior knowledge. Listeners who have considerable experience ("experts") with the subject being discussed may thus interpret a generalization differently than those without such experience ("novices"). In the present study, we investigated the ways in which experts and novices differ in how they interpret generalizations, using the esport League of Legends as a cultural microcosm. In the process, we investigated the extent to which expert listeners discount generalizations with which they disagree. We found that novices tended to interpret generalizations more broadly than experts, with only experts adjusting their interpretations based on the context. We also investigated whether expert speakers, when addressing novices, avoid generalizations that novices are likely to interpret differently. In line with research investigating the curse of knowledge and the challenge of designing utterances for specific audiences, we found that speakers did not adjust their use of generalizations when explicitly told that their audience was inexperienced. Taken together, these results point to novice listeners interpreting generalizations as applying more broadly than expert speakers intend. Future research can help clarify the practical impact of such a mismatch by examining how generalizations are used in relation to speakers' and listeners' goals.

  PublisherDOI

• Stress Patterns in Intra-word Code-switching

  University of Massachusetts (UMass) Amherst · 2026-03-14

  articleOpen accessSenior author

  Code-switching has primarily been studied at the sentence level. More recent work, however, shows that intra-word code-switching is cross-linguistically robust and widespread, yet its phonological properties remain underexplored. As a result, a central open question is whether intra-word code-switching maintains a single phonological system or allows phonological processes from both languages to surface within a single word. To address this gap, we investigate intra-word code-switching between Kazakh and Russian, a societally-widespread but understudied language pair, focusing on stress patterns. Using experimentally-derived acoustic data, we examine whether Russian stress patterns are preserved when Russian stems are inflected with Kazakh suffixes within a Kazakh morphosyntactic frame. Our results show that Russian stress persists in inflected forms, most robustly through duration, while also exhibiting Kazakh-style final lengthening. Vowel quality analyses further reveal a Russian-style effect of stress on vowel reduction also in code-switched tokens. Overall, these findings point to hybrid stress patterns in intra-word code-switching, reflecting interaction rather than categorical dominance of a single phonological system. We analyze these patterns within a Stratal Optimality Theory framework.

  PublisherDOI

• Predicting within-utterance projection variation

  Meaning a journal of linguistics and philosophy · 2026-05-05

  articleOpen access1st authorCorresponding

  Experimental research over the past decade suggests that projection inferences are more ubiquitous than previously assumed and that there are fine-grained differences in the strength of projection inferences (e.g., de Marneffe, Simons & Tonhauser 2019, Degen & Tonhauser 2022, Tonhauser, Beaver & Degen 2018, White & Rawlins 2018). This paper contributes to the question of how to predict the observed projection variation by empirically investigating and theoretically analyzing a case of within-utterance projection variation. Specifically, we consider projection variation between two contents of declarative know-utterances (like Cole knows that Charley speaks Spanish), namely between the content of the clausal complement (CC; here, that Charley speaks Spanish) and the attitude holder’s belief in this content (BEL; here, that Cole believes that Charley speaks Spanish). We present the results of an experiment that suggest that CC and BEL differ in how likely they are to be inferred from utterances of negative variants (like Cole doesn’t know that Charley speaks Spanish): Specifically, the results suggest that (a) CC is projective regardless of whether the Question Under Discussion (QUD) condition of the experiment is about CC or about BEL, (b) BEL is projective when the QUD is about CC, and (c) CC is more projective than BEL regardless of the QUD. Observing that the reasoning-based analysis in Scontras & Tonhauser 2025 predicts the first two results but not the last one, we provide a modification that predicts all three results. The paper also discusses the predictions of other contemporary projection analyses and the viability of reasoning-based analyses for predicting projection variation.

  PublisherOA PDFDOI

• English Restrictive Relative Clauses Are Subject to Crossover Violations

  Linguistic Inquiry · 2025-09-18 · 1 citations

  article

  Abstract The literature on crossover effects has yet to arrive at a consensus on the status of weak crossover (WCO) in restrictive relative clauses (RRCs). We present the first experimental investigation of this construction, finding clear evidence for the presence of WCO effects in English RRCs, contra claims in the theoretical literature (e.g., Chomsky 1982, Reinhart 1997, Rouveret 2002). In our large-scale acceptability judgment task, participants consistently rated WCO-inducing constructions lower than their non-WCO counterparts. Our results also show similarly strong crossover effects for both referential and quantificational antecedents, a finding that aligns more closely with a head-external analysis of relative clauses rather than a head-raising analysis. While our results do not adjudicate among the many analyses of crossover effects in the literature, they confirm the presence of WCO in English RRCs, thereby refuting the empirical claim that English RRCs do not exhibit this effect whatsoever. Accordingly, our findings can be used to argue against analyses of WCO that predict its absence in RRCs. This work also sets the stage for cross-linguistic investigations of WCO, extending our paradigm to other, lesser-studied languages.

  PublisherDOI

• Italian Adjective Order Is Not Predicted by Syntactic Hierarchies and Movement

  Linguistic Inquiry · 2025-07-30

  article1st authorCorresponding

  Abstract Speakers have robust preferences for the relative order of adjectives when multiple adjectives modify a noun (e.g., big blue box vs. blue big box; see Scontras 2023 for an overview). Here we investigate adjective ordering in Italian, a language with postnominal adjectives that also allows for noun-medial ANA configurations. We present two experiments on Italian AAN, ANA, and NAA orders. We find strong ordering preferences in the ANA template but weaker preferences for NAA; our results are especially problematic for the idea that NAA can be derived cross-linguistically on the basis of movement from an underlying AAN structure.

  PublisherDOI

• Weak crossover in Spanish relative clauses

  Probus · 2025-09-12

  articleOpen access

  Abstract We investigate the status of weak crossover violations (WCO) in Spanish restrictive relative clauses. Claims in the literature lead to the expectation that Spanish does not show sensitivity to WCO in this context; however, ours is the first systematic investigation of the phenomenon in the language. We find that Spanish restrictive relative clauses do exhibit WCO violations: when the relative clause is headed by que , participants rate crossover configurations lower than configurations without crossover; when the relative clause is introduced by a quien , the status of WCO violations is less clear. We consider two explanations for our results in a quien clauses: potential structural differences between que and a quien , and the possibility for an appositive parse for some of our a quien configurations. Our results set the stage for further cross-linguistic investigations of crossover phenomena.

  PublisherDOI

• Explaining gender

  Studies in bilingualism · 2024-03-23

  book-chapterSenior author

  Abstract This work examines the knowledge of grammatical gender by bilingual speakers for whom Spanish is the weaker language in the dyad, that is, heritage speakers of Spanish. In modeling heritage speakers’ knowledge and use of grammatical gender, we distinguish between gender assignment (how nouns get associated with specific genders) and gender agreement (the process by which features of a noun get shared by other sentential constituents). We show that divergent instances of gender assignment constitute the use of a default strategy; with masculine as the default class in Spanish, the absence of gender assignment surfaces as masculine morphology. We also review evidence suggesting that heritage speakers differ from the baseline in relying on surface gender cues more categorically. To explain the intricacies of divergent agreement behavior, we argue for genuine representational difference between how baseline and heritage speakers represent gender (and number) features. Taken together, these observations indicate that heritage speakers are sensitive to cues for gender assignment despite the decreased input they receive compared to baseline speakers; their assignment strategies then translate into the observed patterns of agreement, which as a syntactic phenomenon is alive and well in the heritage grammar. We use our results to highlight the utility of non-lexicalist approaches to the representation of gender.

  PublisherDOI

• On the compatibility of models with experiments

  Linguistic Approaches to Bilingualism · 2024-02-01 · 1 citations

  article1st authorCorresponding

  Comment This is a commentary article in response to the following content: The importance of features and exponents

  PublisherDOI

• On the Role of Loopholes in Polite Communication: Linking Subjectivity and Pragmatic Inference

  Open Mind · 2024-01-01 · 6 citations

  articleOpen accessSenior author

  ) agree that speakers exploit indirectness for pragmatic purposes but differ on the underlying sources they attribute to these uses. Here, we synthesize existing proposals via adjective subjectivity, which operationalizes the notion of loopholes for plausible deniability. We present experimental evidence that the degree of subjectivity of an adjective predicts the degree to which participants strengthen the negated adjective's meaning, but only if the adjective under consideration has an evaluatively-positive meaning. This finding indicates that speakers may intentionally use negation to leave themselves the option to retract the implicated face-threatening meaning if openly challenged.

  PublisherOA PDFDOI

• Corpus evidence for the role of world knowledge in ambiguity reduction: Using high positive expectations to inform quantifier scope

  Experiments in Linguistic Meaning · 2023-01-27

  articleOpen accessSenior author

  Every-negation utterances (e.g., Every vote doesn’t count) are ambiguous between a surface scope interpretation (e.g., No vote counts) and an inverse scope interpretation (e.g., Not all votes count). Investigations into the interpretation of these utterances have found variation: child and adult interpretations diverge (e.g., Musolino 1999) and adult interpretations of specific constructions show considerable disagreement (Carden 1973, Heringer 1970, Attali et al. 2021). Can we concretely identify factors to explain some of this variation and predict tendencies in individual interpretations? Here we show that a type of expectation about the world (which we call a high positive expectation), which can surface in the linguistic contexts of every-negation utterances, predicts experimental preferences for the inverse scope interpretation of different every-negation utterances. These findings suggest that (1) world knowledge, as set up in a linguistic context, helps to effectively reduce the ambiguity of potentiallyambiguous utterances for listeners, and (2) given that high positive expectations are a kind of affirmative context, negation use is felicitous in affirmative contexts (e.g., Wason 1961).

  PublisherOA PDFDOI

Frequent coauthors

• Maria Polinsky

  12 shared

• Zuzanna Fuchs

  8 shared

• Lisa Pearl

  University of California, Irvine

  7 shared

• Noah D. Goodman

  5 shared

• Kenneth Mai

  Harvard University Press

  4 shared

• Zeinab Kachakeche

  4 shared

• Richard Futrell

  4 shared

• Michael Franke

  3 shared

Labs

• The Meaning LabPI

Education

• B.A., Linguistics and Philosophy

  Boston University

• Ph.D., Linguistics

  Harvard University

  2014