New York Philosophy of Language Workshop

Our speaker on Monday, April 25th will be Kyle Blumberg, who is a Lecturer in Philosophy at the University of Melbourne. Kyle will give a talk called "Counterfactual Desire and Similarity"

Counterfactual attitudes such as wishing, imagining, supposing, etc. have recently received a great deal of attention from philosophers and linguists. One of the key insights which has emerged is that the appropriate analysis of these states involves a notion of worldly similarity. This paper is about the precise way in which wishing incorporates a notion of similarity. I argue that controlling for similarity, subjects cannot wish for what isn’t best. I show that existing analyses of wishing do not capture this constraint, and develop a positive proposal which does. The resulting theory has striking features, for example it brings counterexamples to various principles connecting wishing to subjective preference. It also raises interesting questions about the relationship between wishing and other desire states, and highlights possible connections between bouletic verbs and deontic modals.

The workshop will take place on Monday, April 25th from 5:15 until 7:15pm (Eastern Time) in room 202 of NYU's Philosophy Building (5 Washington Place).

Our speaker on Monday, April 13th will be Livia Polanyi, who is an Independent Scholar, formerly of Stanford University, Microsoft Research, Powerset, Fuji-Xerox Research, Rice University, BBN Labs, University of Amsterdam. Livia will give a talk called "The Topic of this Talk is Topic Shifting in Talk: Constraints on the Appropriateness of Next Utterances in Conversation"

Conversation is a form of linguistically encoded interaction in which speakers exchanging utterances with one another without acknowledged goals or tasks jointly construct a Discourse History. In framing an initial or next conversational utterance competent speakers adhere to social and pragmatic conventions to manage how talk flows appropriately from one thing, one topic, to another. After defining key terms and showing how a next utterance for the purpose of topic shifting need not be immediately prior in the flow of talk, I will introduce two distinct topic shifting processes, Step by Step (STEP) described by Sacks, 1973; Hobbs, 1990; Decker, 2022, and Disjunctive Topic Shifting (DTS) treated informally in Sacks,1973 and Jefferson, 1984. Both processes require utterances conform to constraints on Informativity, Coherence, and Relevance.

The Dynamic Feature Model of Topic in which topic is treated as an local phenomenon of individual clauses will then be introduced to explain how one moment we are talking about one thing and a few minutes later we might find ourselves talking about something all together different without anyone really noticing how we got from here to there (H. Sacks, 1973). In STEP, each clause encoding next and prior topic must be Informative to the extent that information novel to the immediate conversational interaction is expressed; while Coherence in STEP emerges from connecting lexical, syntactic and semantic properties of a next utterance to the properties of a previous utterance accessible according to the Right Edge Constraint of discourse syntax. introduced initially in Polanyi and Scha, 1976. Like Coherence, Relevance, is a text internal property in STEP and concerns how informational content in a next utterance relates in some way to some aspect of content in the utterance previous to it.

Unlike STEP, Disjunctive Topic Shift results when a clause without connection to any topic accessible on the Right Edge of the Discourse Parse Tree is uttered by a speaker. This is a fraught move in conversation because speakers are aware of being viewed as a less than competent conversational partner if the topic they introduce fails to adhere to the three constraints. While assessing the Informativity of the DTS utterance is essentially identical to assessing the Informativity of STEP next utterances, explicating how Coherence and Relevance are achieved requires rather different resources. The Be Coherent constraint is met in DS by simply being the next constituent in the interactive activity of Conversing. To explain how a next DTS utterance can meet the Relevant constraint, I will rely on the Egocentric Model of Relevance (EMR). The EMR situates EGO of all speakers at the center of a Relevance Space with the relative placement of each person’s Times, Places, Persons,Events, Activities, Interests etc. at various level of remove. Using the model, the relevance of information encoded in a next utterance to the conversational partners can be calculated. To the extent they are adhered to, respecting these three constraints determine the appropriateness of a next conversational utterance

The workshop will take place on Monday, April 13th from 5:15 until 7:15pm (Eastern Time) in room 202 of NYU's Philosophy Building (5 Washington Place).

Our speaker on Monday, April 6th will be Evan Westra, who is Associate Professor of Philosophy at Purdue. Evan will give a talk called "The dynamics of articulated norms":

Of the myriad social norms that structure our everyday lives, some are implicit and unarticulated, while others are explicit and articulated. In this tall, I analyze the notion of norm articulation and explore its consequences for the social dynamics of norms – how they are acquired, how they are enforced, how they spread throughout a population. I will also discuss the implications of articulation for "norm engineering," the practical project of attempting to instigate positive norm change. While articulation can help us attend to norms that we might otherwise take for granted, the act of articulation can also trigger further normative dynamics that inevitably complicate even the simplest attempts at norm engineering.

The workshop will take place on Monday, April 6th from 5:15 until 7:15pm in room 202 of NYU's Philosophy Building (5 Washington Place).

Our speaker on Monday, March 2nd will be Endre Begby, who is Professor of Philosophy at Simon Fraser University. Endre will give a talk called "Knowledge of Language: Perspectives from Sign and Gesture Studies"

Recent and emerging discoveries in Sign and Gesture studies put serious pressure on the widespread idea that we can model a subject’s knowledge of a language in terms of their knowledge of a well-defined set of rules governing, among other things, the grammar, semantics, and phonology of that language. First, I point out that all extant Sign languages are historically recent creations, which have emerged spontaneously (though gradually) from patterns of presumptively non-linguistic gestural communication among Deaf people. I argue that there is no unique point in their history at which these gestural systems would have transitioned to become “languages proper” (even as they are, indisputably, languages by now). Second, I point out that speech, no less than Sign, continues to functionally incorporate non-categorical “gestural” complements in both semantic and grammatical roles, pointing toward a conception of language – in any form – as a “multimodal” phenomenon. 

The workshop will take place on Monday, March 2nd from 5:15 until 7:15pm (Eastern Time) in room 202 of NYU's Philosophy Building (5 Washington Place).

NOTE: THIS WORKSHOP HAS BEEN CANCELLED DUE TO SNOW STORM. WE WILL TRY TO RESCHEDULE.

Our speaker on Monday, February 23rd will be Livia Polanyi, who is an Independent Scholar, formerly of Stanford University, Microsoft Research, Powerset, Fuji-Xerox Research, Rice University, BBN Labs, University of Amsterdam. Livia will give a talk called "The Topic of this Talk is Topic Shifting in Talk: Constraints on the Appropriateness of Next Utterances in Conversation"

Conversation is a form of linguistically encoded interaction in which speakers exchanging utterances with one another without acknowledged goals or tasks jointly construct a Discourse History. In framing an initial or next conversational utterance competent speakers adhere to social and pragmatic conventions to manage how talk flows appropriately from one thing, one topic, to another. After defining key terms and showing how a next utterance for the purpose of topic shifting need not be immediately prior in the flow of talk, I will introduce two distinct topic shifting processes, Step by Step (STEP) described by Sacks, 1973; Hobbs, 1990; Decker, 2022, and Disjunctive Topic Shifting (DTS) treated informally in Sacks,1973 and Jefferson, 1984. Both processes require utterances conform to constraints on Informativity, Coherence, and Relevance.

The Dynamic Feature Model of Topic in which topic is treated as an local phenomenon of individual clauses will then be introduced to explain how one moment we are talking about one thing and a few minutes later we might find ourselves talking about something all together different without anyone really noticing how we got from here to there (H. Sacks, 1973). In STEP, each clause encoding next and prior topic must be Informative to the extent that information novel to the immediate conversational interaction is expressed; while Coherence in STEP emerges from connecting lexical, syntactic and semantic properties of a next utterance to the properties of a previous utterance accessible according to the Right Edge Constraint of discourse syntax. introduced initially in Polanyi and Scha, 1976. Like Coherence, Relevance, is a text internal property in STEP and concerns how informational content in a next utterance relates in some way to some aspect of content in the utterance previous to it.

Unlike STEP, Disjunctive Topic Shift results when a clause without connection to any topic accessible on the Right Edge of the Discourse Parse Tree is uttered by a speaker. This is a fraught move in conversation because speakers are aware of being viewed as a less than competent conversational partner if the topic they introduce fails to adhere to the three constraints. While assessing the Informativity of the DTS utterance is essentially identical to assessing the Informativity of STEP next utterances, explicating how Coherence and Relevance are achieved requires rather different resources. The Be Coherent constraint is met in DS by simply being the next constituent in the interactive activity of Conversing. To explain how a next DTS utterance can meet the Relevant constraint, I will rely on the Egocentric Model of Relevance (EMR). The EMR situates EGO of all speakers at the center of a Relevance Space with the relative placement of each person’s Times, Places, Persons,Events, Activities, Interests etc. at various level of remove. Using the model, the relevance of information encoded in a next utterance to the conversational partners can be calculated. To the extent they are adhered to, respecting these three constraints determine the appropriateness of a next conversational utterance

The workshop will take place on Monday, February 23rd from 5:15 until 7:15pm (Eastern Time) in room 202 of NYU's Philosophy Building (5 Washington Place).

Our speaker on Monday, February 9th will be Wade Munroe, who is Assistant Professor at the University of Nebraska-Lincoln. Wade will give a talk called "Unintentionally Speaking: Inner Speech and Intentionalist Accounts of Language"

It is widely assumed that language use is intentional; we generate speech acts and linguistic entities by producing utterances with the appropriate intentions. In this talk, I offer a novel challenge to intentionalist views on the grounds that they fail to account for inner speech, that is, our use of language without overt articulation. I argue that speech acts and linguistic entities generated via inner speech function analogously to their overt counterparts; however, inner speech is produced non-intentionally. As I show, inner speech allows for the monitoring and control of speech planning processes, enabling intentional control over our overt uses of language. However, there are no second-order monitoring and control processes that govern inner speech. The talk thus offers a model of speech planning that explains why overt speech can be intentional, why inner speech cannot, and why inner speech must figure centrally in any account of language control.

The workshop will take place on Monday, February 9th from 5:15 until 7:15pm (Eastern Time) in room 202 of NYU's Philosophy Building (5 Washington Place).

During Spring, 2026, we'll meet on Monday evenings, 5:15–7:15, in room 202 of NYU's philosophy building (5 Washington Place). Here is our lineup:

February 9th Wade Munroe (Nebraska)

March 3rd Endre Begby (Simon Fraser)

April 6th Evan Westra (Purdue)

April 13th Livia Polanyi

April 20th Eliot Watkins (Bielefeld) and Daniel Fogal (NYU)

April 27th Kyle Blumberg (Melbourne)

May 4th Alejandro Pérez-Carballo (UMass)

We have a lighter-than-usual lineup this term, in part to leave space for the unusually busy slate of other philosophy of language events that will be happening at NYU. Some highlights that you may wish to attend:

1. Matt Mandelkern and Cian Dorr will be teaching the Mind and Language Seminar, which meets on Tuesday afternoons, and which will host a visiting philosopher of language most weeks (more info here).

2. François Recanati will give the NYU/College de France lecture on Thursday, March 12th, 4–6pm (more info will be posted here).

Our speaker on Monday, December 8th will be Dean McHugh, who is a postdoctoral fellow at the University of Edinburgh. Dean will give a talk called "Modal Reasoning as a Species of Conditional Reasoning"

Alice is at the library. She is not allowed to make noise. If she made noise, she’d be in trouble. At the same time, she is allowed to leave the library and make noise. If she left the library and made noise, she wouldn’t be in trouble.

Standard theories of conditionals account for the conditional data above, allowing “if A, C” and “if (A and B), not C” to be compatible (both true, for variably strict theories, or both true modulo contextual shifts, for strict theories). But standard theories of modals, without further refinement, predict the two modal claims to be contradictory. They validate Inheritance: if A & B is permitted then A is too, so if A is not permitted, neither is A & B: if Alice is not allowed to make noise, she is not allowed to leave the library and make noise either.

We have two phenomena: the failure of antecedent strengthening for conditionals, and the failure of inheritance for modals. They seem to have the same character, involving a shift in the set of possibilities under consideration. It is, then, quite remarkable that they should standardly receive such different treatments.

This talk proposes an alternative analysis of modality, going back to Alan Ross Anderson in the 1950’s, with conditionals at its heart. It provides a uniform account of the modal and conditional data above, on which they are two instances of the general same mechanism of conditional reasoning, one that shifts the set of relevant possibilities in response to a sentence. I further show that this account also provides a novel solution to independence inferences, observed by Paula Menéndez Benito and Richard Booth, and David Lewis’s problem of permissive update.

The workshop will take place on Monday, December 8th from 6:15 until 8:15pm (Eastern Time) in room 202 of NYU's Philosophy Building (5 Washington Place).

Our speaker on Monday, December 1st will be Kate Stanton, who is Assistant Professor of Philosophy at the University of Pittsburgh. Kate will give a talk called "The Semantics and Pragmatics of Air Quotation"

Much recent work on the semantics and pragmatics of gesture has focused on gestures that are both iconic and whose contribution is pragmatically derived. This paper explores the interpretation of air quotation, a gesture that I will argue falls squarely into neither category. I claim that the association of air quotation with orthographic quotation has conventionalized and that this facilitates its use as a quotational shield: a pragmatic marker that allows the speaker to distance themselves from the enquoted content. I then offer a case study to exemplify and refine the analysis, arguing that air quotes applied to nominals are `reverse clones' that can initiate truth-conditional narrowing of the expression they enclose.

The workshop will take place on Monday, December 1st from 6:15 until 8:15pm (Eastern Time) in room 202 of NYU's Philosophy Building (5 Washington Place).

Our speaker on Monday, November 24th will be James Gu, who is a PhD Student at UCLA. James will give a talk called "No thing of content : CP predicativism & the enigma of intensional nouns"

According to the CP-as-predicate hypothesis, finite CPs (embedded tensed sentences) do not function as saturating arguments but rather as predicates which compose with other lexical items (e.g., nouns) by way of intersective Predicate Modification (for concrete implementations of this increasingly-popular view, see Moltmann 1989, Kratzer 2013, Moulton 2015, Elliott 2020, Bondarenko 2022, and Srinivas & Legendre 2024). In this paper I articulate a new puzzle for the CP-as-predicate hypothesis. Very roughly, the proposed puzzle is as follows: once we adopt the hypothesis in its usual formulation, then there can be no coherent semantic account of why certain nouns are CP-taking (i.e., can appear w/ CPs) while others are CP-allergic (i.e., cannot appear w/ CPs). I then proceed to sketch out a possible fix on behalf of the CP-as-predicate hypothesis. The proposal, in brief, is that we should combine the CP-as-predicate hypothesis with the Generative Lexicon (GL) framework (see Pustejovsky 1995 & Bouillon 1999) and model CPs as ‘subsective’ (or ‘selective’) modifiers which apply to the Telic quale of CP-taking nouns. Along the way, I will also discuss (i) how the resulting GL-enriched view deals with Safir’s (1985) proof-class nominals and (ii) whether or not there are any genuine reasons for preferring a GL-enriched version of CP-predicativism over a suitably GL-enriched version of CP-propositionalism.

The workshop will take place on Monday, November 24th from 6:15 until 8:15pm (Eastern Time) in room 202 of NYU's Philosophy Building (5 Washington Place).

Our speaker on Monday, November 17th will be Noga Zaslavsky, who is Assistant Professor of Psychology at NYU. Noga will give a talk called "Cultural evolution of efficiently compressed semantic systems in humans and LLMs"

It has been shown that human languages efficiently compress meanings into words, but how did our semantic systems evolve to be that way? Are modern AI systems capable of evolving efficient semantic systems as we do? In this talk, I address these open questions by integrating cognitive, developmental, cultural, and computational perspectives. First, across two human studies, I show that both children and adult learners operate under pressure to maintain efficiently compressed semantic systems. This inductive learning bias, when amplified via cultural transmission, drives the evolution of near-optimally efficient semantic systems. Second, I present two parallel studies with LLMs, designed to test whether LLMs exhibit a similar human-like inductive bias toward efficient compression. These studies reveal that while LLMs vary widely in their semantic alignment with humans, the best performing models have acquired a human-like tendency toward efficient compression: when simulating cultural evolution, they iteratively restructure initially random semantic systems towards greater efficiency, similar to the pattern observed in humans. Taken together, these results illuminate the cultural and cognitive mechanisms that drive the evolution of semantic systems in humans and modern AI systems. More broadly, this work suggests that efficient compression may be a fundamental ingredient of intelligent behavior.

The workshop will take place on Monday, November 17th from 6:15 until 8:15pm (Eastern Time) in room 202 of NYU's Philosophy Building (5 Washington Place).

Our speaker on Monday, Monday, November 10th will be Frank Pupa, who is Professor of Philosophy at Nassau Community College. Frank will give a talk called "Can roots break chains?"

The causal theory of names implies that a language user can only acquire an established proper name from someone who is already competent in its use. This position appears to place the causal theory of names in conflict with root-based approaches to morphology, as the decomposition of words into roots and categorizers seems to provide language users with the resources to break causal chains of reference (Dupre 2023). In this talk, I will explore the tension between the causal theory of names and root-based approaches to morphology. I hope to show that the causal theory of names is compatible with such approaches. Their compatibility, I will argue, turns on (a) adopting a particular understanding of roots, (b) recognizing the interconnectedness of morphosyntactic structures, semantic conventions, and long-term memory, and (c) acknowledging the special standing of proper names in language and within linguistic communities.

The workshop will take place on Monday, November 10th from 6:15 until 8:15pm (Eastern Time) in room 202 of NYU's Philosophy Building (5 Washington Place).

Our speaker on Monday, November 3rd will be Tanya Bondarenko, who is Assistant Professor of Linguistics at Harvard. Tanya will give a talk called "Mereology of thoughts, and why you can't (statively) think a question"

Cross-linguistically, the ability of verbs meaning 'think' to embed questions is contingent on their lexical aspect (Dayal 2016; White and Rawlins 2018; Özyıldız 2021; Özyıldız 2024; White 2021, Bondarenko et. al. 2024, a.o.): stative ‘think’ cannot compose with questions, (2a), whereas dynamic uses of ‘think’, e.g. when the verb describes an activity, (2b), can.

(1) Özyıldız’s (2021) Generalization about ‘think’ When ‘think’ composes with a question, the resulting eventuality description must be dynamic. (Özyıldız 2021, p. 34)

(2) a. *I thought whether to invite Bill to the party. b. I'm thinking whether to invite Bill to the party. (Dayal 2016)

In this talk I attempt to account for why stative ‘think’ is incompatible with questions by using the idea that contentful objects such as thoughts form structures based on part-whole relations, and are systematically mapped to the part-whole structures of their contents (Bondarenko & Elliott 2024). I adopt the idea that stative ‘think’ is derived via a genericity operator (Özyıldız 2024), and propose that embedding questions under stative ‘think’ is impossible because it leads to a deviant meaning - one that requires attitude holders to only ever hold mental states that are inquisitive. This effect arises because as soon as one admits a thought with an inquisitive content into a thought structure, all thoughts that it is part of become inquisitive. The same is not true of declarative content: a thought with inquisitive content could have a thought with declarative content as its subpart.

The workshop will take place on Monday, November 3rd from 6:15 until 8:15pm (Eastern Time) in room 202 of NYU's Philosophy Building (5 Washington Place).

Our speaker on Monday, October 27th will be Chris Kennedy, who is William H. Colvin Professor of Linguistics at the University of Chicago. Chris will give a talk called "Pragmatic Indecision (or "When meaning is like voting")"

Vague predicates are obligatorily tolerant. For example, the positive form gradable adjective ‘long’ cannot be used to draw a sharp distinction, even when the facts of the context of utterance and the semantic properties of the sentence in which the adjective occurs otherwise conspire to make such uses possible. In this talk, I will do three things. First, I will argue that no existing account of vagueness — neither epistemic nor expressivist nor contextual nor semantic nor cognitive nor distributional — provides a satisfactory explanation for why tolerance persists in such cases. Second, I will show that state-of-the-art attempts to eliminate the challenge presented by these kinds of examples by appeal to a decompositional semantics for positive form gradable predicates fail, because they do not generalize to other kinds of examples that manifest exactly the same behavior but cannot plausibly be analyzed in the same way. And finally, I will outline an alternative, pragmatic account of tolerance, in which application of a vague predicate constitutes a “vote” for a particular resolution of linguistic indeterminacy. On this view, tolerance emerges from a normative constraint treating such votes as subject to principles of social choice.

The workshop will take place on Monday, October 27th from 6:15 until 8:15pm (Eastern Time) in room 202 of NYU's Philosophy Building (5 Washington Place).

Our speaker on Monday, October 20th will be David Liebesman, who is Professor of Philosophy and Linguistics at the University of Calgary. David will present joint work with Ofra Magidor called "Polysemy and Observations in Linguistics"

In linguistics it is standard to distinguish between two types of ambiguity: homonymy, where (roughly) a term has two completely unrelated senses (e.g. ‘ball’ as used for a spherical play object, and ‘ball’ as used for a party); and polysemy, where (roughly) a word has two closely related senses (e.g. ‘lunch’ as used for used for a an event taking place midday and for a portion of food eaten at midday). The literature on polysemy reports a wide-range of alleged observations regarding the concept based on both straightforward intuitions of theorists, as well as experimental data on wide-scale speaker judgements and psychological processing. In this talk, we argue that many of these observations implicitly presuppose some theoretical assumptions about the phenomenon of copredication (sentences such as ‘Lunch was delicious, but took hours’). Indeed, they implicitly rely on false theoretical assumptions. After explaining what copredication is and our own preferred theoretical account of the phenomenon, we return to show how all this bears on the range of linguistic observations about polysemy.

The workshop will take place on Monday, October 20th from 6:15 until 8:15pm (Eastern Time) in room 202 of NYU's Philosophy Building (5 Washington Place).

Note: Next week's workshop will take place on a Tuesday. Our speakers on Tuesday, October 14th will be Tue Trinh, who is Associate Professor of Linguistics at the University of Nova Gorica, and Salvatore Pistoia-Reda, who is a Tenured Senior Researcher at the University of Siena. They will give a talk called "Tractarian non-sense and modern linguistic theory"

We argue that theory of language proposed by Wittgenstein in the Tractatus makes two seemingly counterintuitive predictions: (A) the non-sensicality of a combination of words (e.g. "a man and is") cannot result from it being ill-formed but can only result from one or more of its constituents lacking in meaning; and (B) trivialities (e.g. "it's raining and not raining") are non-sensical. We provide textual evidence that Wittgenstein endorses (A) but tries to circumvent (B) by introducing a problematic distinction between the non-sensical (unsinnig) and the senseless (sinnlos). Drawing on recent advances in linguistics, we explore a perspective under which both (A) and (B) square with natural language intuitions and the distinction between non-sensicality and senselessness can be dispensed with.

The workshop will take place on Tuesday, October 14th from 6:15 until 8:15pm (Eastern Time) in room 202 of NYU's Philosophy Building (5 Washington Place).