New papers
Recently Published
The grammar of knowledge. Above and beyond.
The empirical goal of this paper is to establish a generalization about the expression of knowledge. Unmarked declaratives are used in context of certainty: uncertainty has to be marked, but certainty need not be and if it is different strategies have different interpretive effects. The analytical goal is to argue for a grammatical analysis of this generalization. It is argued that the interpretation of certainty comes about (covertly) marking declaratives as being part of the speaker's epistemic state. This grammatical analysis is compared to an alternative according to which the interpretation of certainty is a pragmatic effect, based on Gricean reasoning. Empirical evidence is provided that favors the grammatical analysis. The generalization about the expression of certainty as well as its grammatical analysis has far-reaching consequences for cross-linguistic comparison, for philosophical theories of assertion, as well as for human AI interaction.
Wiltschko, Martina (2025). The grammar of knowledge. . Journal of the Linguistic Society of Germany, vol. 44, no. 2, Dec. 2025, https://doi.org/10.18148/zs/2025-2011.
Three Stages in The Acquisition of English Response Tokens: A Window into the Development of Common Ground.
(with Johannes M. Heim and Júlia Rovira Marí) Cahiers de praxématique, 83 2025.
This paper presents a longitudinal study of response tokens from one British English-speaking child aged 10 to 60 months. We demonstrate that response tokens can serve as a linguistic window into the development of common ground. Based on a qualitative and quantitative analysis of response tokens, including adjacency markers, repetitions, hesitation markers, repair requests, as well as verbal and non-verbal backchannels, a proposal is made for the development of common ground. It is shown that response tokens expand incrementally in frequency and lexical diversity, specifically three stages are identified. Initially, the child only marks responses without situating them in the child´s set of beliefs (their epistemic ground). From 28 months on, the child then signals agreement in addition to marking response. This is interpreted as signaling the emergence of a preliminary concept of common ground — one that does not distinguish between speaker and addressee’s beliefs. From 41 months on, the child begins to show nuanced forms of agreement and disagreement, which reflects a separation of speaker and addressee’s beliefs. Only at this stage, response marking will resemble adult common ground management. It is concluded that the different stages in the use of response tokens mirrors different stages in the development of common ground construction and management.
The Grammar of Knowledge: Above and Beyond.
Abstract
The empirical goal of this paper is to establish a generalization about the expression of knowledge. Unmarked declaratives are used in context of certainty: uncertainty has to be marked, but certainty need not be and if it is different strategies have different interpretive effects. The analytical goal is to argue for a grammatical analysis of this generalization. It is argued that the interpretation of certainty comes about (covertly) marking declaratives as being part of the speaker's epistemic state. This grammatical analysis is compared to an alternative according to which the interpretation of certainty is a pragmatic effect, based on Gricean reasoning. Empirical evidence is provided that favors the grammatical analysis. The generalization about the expression of certainty as well as its grammatical analysis has far-reaching consequences for cross-linguistic comparison, for philosophical theories of assertion, as well as for human AI interaction.
Journal of the Linguistic Society of Germany, vol. 44, no. 2, Dec. 2025,
Do chatbots dream of AI sheep? A semantic–pragmatic investigation of "naturalness" in human–AI interaction. AI & Society (2025). (with Iuliia Lysova, Lara Ahmed, Bethan Cunningham and Yonghua Huang) https://doi.org/10.1007/s00146-025-02595-1 Abstract This paper investigates what constitutes “natural” interaction between humans and AI chatbots, challenging the prevailing assumption that naturalness must equate to human-likeness. Drawing on key concepts of semantics and pragmatics, we analyze interactions with two state-of-the-art systems, namely, ChatGPT and Claude. Through qualitative analysis of prompt–response interactions, we assess how these systems manage meaning and social nuance. Our findings highlight significant limitations in AI’s pragmatic competence, especially in grounding communication and interpreting implicit cues. We argue that naturalness in human–AI interaction is not a fixed standard based on human norms, but a socially negotiated and context-dependent construct. Instead of designing AI to imitate human behavior perfectly, we advocate for redefining naturalness in terms of communicative functionality, user expectations, and adaptive interactional strategies. This reconceptualization foregrounds the importance of designing AI systems that foster trust, clarity, and usability within their specific social roles. Generative Interactions: No change no gain This is a paper which was commissioned for the Journal of Italian Linguistics. It is one of a series of comments on a paper by Cristiano Chesi entitled: Is it the end of (generative) linguistics as we know it?.
abstract Chesi argues that generative linguists deserve Piantadosi's dismissal of the Chomskyan approach because of what he considers the generativists original sin: being content with incomplete pseudo-formalizations and data fragment explanations. This criticism is based on a narrow perspective on generativism. Here I outline my own (broader) perspective and the questions I ask, all inspired by core generative ideas. The discussion evolves around three areas: i) language variation; ii) interactional language; and iii) language acquisition. I argue that very large language models are not suitable to replace the theoretical assumptions I hold to explore these areas of research. Some of the assumptions I hold are not consistent with a narrow perspective on generativism a la Chesi, and one might conclude that this reflects his view that it is "the end of generativism as we know it". However, I invite a different conclusion: they simply reflect change. And just as change in language is a sign that the language is alive and spoken, so is change in a theoretical framework a sign that it is alive and used.
Wiltschko, Martina (2025). Generative Interactions: no change no gain. Journal of Italian Linguistics 37.1: 165-176 DOI:10.26346/1120-2726-247 Demonstratives locate referents in common space and ground: A comparative syntactic approach.
The paper investigates a generalization that´s well known for English: demonstratives, which allow speakers to locate referents in space, can also be used to locate referents in the common ground. Specifically, the proximate demonstrative this can be used to introduce novel discourse referents, while the distal demonstrative that can be used to refer to familiar discourse referents. Valentina and I asked ourselves if this pattern is found in other languages and so we studied some 20 languages. The result was pretty clear: if a language reuses spatial demonstratives for grounding, it follows the same pattern. We develop a syntactic analysis for this finding within the framework of the Interactional Spine Hypothesis. Colasanti, V., Wiltschko, M. Demonstratives locate referents in common space and ground: A comparative syntactic approach. Nat Lang Linguist Theory (2025). https://doi.org/10.1007/s11049-025-09675-3 Rethinking structural growth: Insights from the acquisition of interactional language. In this paper, we introduce a novel proposal for the acquisition of syntactic structure. Most studies of syntactic first language acquisition focus on the increasingly complex expression of propositional thought. Layer by layer, from the bottom up, child language matures into adult-like representations. We challenge this account of an upward growing syntax based on evidence from early interactional language, which has largely been ignored in the study of syntactic acquisition. Given that interactional units of language are associated with the topmost layers of syntactic structure, and given that they are acquired early, it follows that syntactic acquisition cannot be characterized by upward growth. Rather, we propose that syntactic acquisition proceeds in an inward fashion. We present a case study of the early uses of utterance-final huh (an interactional unit of language) in North American child English that supports this hypothesis. We thus introduce the Inward Growing Spine Hypothesis as a research agenda. Heim, Johannes, Martina Wiltschko (2025). Rethinking structural growth: Insights from the acquisition of interactional language. Glossa: a journal of general linguistics 10(1). doi: https://doi.org/10.16995/glossa.16396 The grammar of self-talk. What different modes of talking reveal about language Self-talk has played an important role in theorizing about the function of language in the psychological and philosophical literature. Linguistic investigations of self-talk, however, are scarce. In this paper, I argue that there are several modes of self-talk including (i) thinking out loud, which is characterized by the absence of an addressee and (ii) having a conversation with oneself, which is characterized by the presence of a grammatically represented addressee role. In the latter, the person engaged in self-talk may hold the role of the speaker or the addressee. Thus, the grammatical restrictions on self-talk serve as a hitherto underexplored window into the grammatical representation of speaker and addressee roles. Wiltschko, Martina. "The grammar of self-talk. What different modes of talking reveal about language" Theoretical Linguistics, 2025. https://doi.org/10.1515/tl-2024-2024 Emotions do not enter grammar because they are constructed (by grammar) I finally finished this paper on the relation between language and emotions. The core idea I pursue here is that functional categories are (universally) not dedicated to the expression of emotions because emotions are constructed. To make this claim, I had to immerse myself into the fascinating scholarship of the affective sciences and I learned (once again) that we should not trust our intuitive understanding of anything, including what emotions are. And in the course of this research, I found myself surprised by the idea that emerged: the same system that is responsible for constructing language (call it grammar) may just be responsible for constructing emotions. The paper was recently published (completely open access) as a discussion paper in the online journal Language under Discussion. So maybe you feel inclined to write a comment or reply! Wiltschko, M. (2024). Emotions do not enter grammar because they are constructed (by grammar). Language Under Discussion, 7(1), 1–62. https://doi.org/10.31885/lud.7.1.261 Is grammar affected in Schizophrenia. A short reply to a recent paper by Zhang et al. that argues that grammar is affected in Schizophrenia. While this may well be, the criteria they use do not support this conclusion. Wiltschko Martina (2024). Is grammar affected in Schizophrenia? Psychiatry Research. Jul 2;339:116061. doi: 10.1016/j.psychres.2024.116061. Epub ahead of print. PMID: 38968919. The meaning of what A short paper on a use of what that does not question a propositional argument, as in a scenario when someone is staring at you and you utter What? I show that this use of what cannot be analyzed as an elliptical question. And I argue that it serves as an interactional question in that it questions an initiation move. I wrote this paper for a Festscrhift for Hotze Rullmann. The meaning of what. In M. Ryan Bochnak, Eva Csipak, Lisa Matthewson, Marcin Morzycki and Daniel K.E. Reisinger, eds. The title of this volume is shorter than its contributions are allowed to be: Papers in honour of Hotze Rullmann (UBC Occasional Papers in Linguistics 9). Vancouver: University of British Columbia. A new argument for linguistic determinants of human thought. The fact that grammatical categories (i.e., PERSON or TENSE), are necessary to construct sentences and define the content of the thought expressed suggests that thought cannot be independent of language. This is because grammatical categories are strictly grammatical, with not non-grammatical analogue. Hinzen, W., Martin, T. & Wiltschko, M. A new argument for linguistic determinants of human thought. Linguist and Philos (2024). https://doi.org/10.1007/s10988-024-09414-8 The Syntax of Talking Heads In this paper, I explore two discourse markers of Upper Austrian German: ma indicates surprise, while geh indicates a discrepancy between speaker and addresee. I argue that they are interactional pro-forms (ProGroundP) which mark a reaction to the speaker's or the addresee's current epistemic state. This extends the typology of proforms I have been working on for decades now. Wiltschko, M. 2024. The syntax of talking heads, Journal of Pragmatics, Volume 232: 182-198. Published online. https://doi.org/10.1016/j.pragma.2024.08.011. Pronouns beyond phi-features: The speaker–addressee relation in Japanese pronouns and its implication for formal pronouns In this paper, we demonstrate that Japanese pronouns bear the interactional roles of speaker and addressee but lack phi-features and that the separation of interactional roles from phi-features is structurally conditioned. Phi-based pronouns are introduced DP-internally while Japanese pronouns instantiate syntactic categories above the DP that realize interactional content including speaker and addressee roles. Ritter E, Wiltschko M. 2025. Pronouns beyond phi-features: the speaker–addressee relation in Japanese pronouns and its implications for formal pronouns. Journal of Linguistics. 61(3):653-688. doi:10.1017/S0022226724000306
Do chatbots dream of AI sheep? A semantic–pragmatic investigation of "naturalness" in human–AI interaction. AI & Society (2025). (with Iuliia Lysova, Lara Ahmed, Bethan Cunningham and Yonghua Huang) https://doi.org/10.1007/s00146-025-02595-1 Abstract This paper investigates what constitutes “natural” interaction between humans and AI chatbots, challenging the prevailing assumption that naturalness must equate to human-likeness. Drawing on key concepts of semantics and pragmatics, we analyze interactions with two state-of-the-art systems, namely, ChatGPT and Claude. Through qualitative analysis of prompt–response interactions, we assess how these systems manage meaning and social nuance. Our findings highlight significant limitations in AI’s pragmatic competence, especially in grounding communication and interpreting implicit cues. We argue that naturalness in human–AI interaction is not a fixed standard based on human norms, but a socially negotiated and context-dependent construct. Instead of designing AI to imitate human behavior perfectly, we advocate for redefining naturalness in terms of communicative functionality, user expectations, and adaptive interactional strategies. This reconceptualization foregrounds the importance of designing AI systems that foster trust, clarity, and usability within their specific social roles. Generative Interactions: No change no gain This is a paper which was commissioned for the Journal of Italian Linguistics. It is one of a series of comments on a paper by Cristiano Chesi entitled: Is it the end of (generative) linguistics as we know it?.
abstract Chesi argues that generative linguists deserve Piantadosi's dismissal of the Chomskyan approach because of what he considers the generativists original sin: being content with incomplete pseudo-formalizations and data fragment explanations. This criticism is based on a narrow perspective on generativism. Here I outline my own (broader) perspective and the questions I ask, all inspired by core generative ideas. The discussion evolves around three areas: i) language variation; ii) interactional language; and iii) language acquisition. I argue that very large language models are not suitable to replace the theoretical assumptions I hold to explore these areas of research. Some of the assumptions I hold are not consistent with a narrow perspective on generativism a la Chesi, and one might conclude that this reflects his view that it is "the end of generativism as we know it". However, I invite a different conclusion: they simply reflect change. And just as change in language is a sign that the language is alive and spoken, so is change in a theoretical framework a sign that it is alive and used.
Wiltschko, Martina (2025). Generative Interactions: no change no gain. Journal of Italian Linguistics 37.1: 165-176 DOI:10.26346/1120-2726-247 Demonstratives locate referents in common space and ground: A comparative syntactic approach.
The paper investigates a generalization that´s well known for English: demonstratives, which allow speakers to locate referents in space, can also be used to locate referents in the common ground. Specifically, the proximate demonstrative this can be used to introduce novel discourse referents, while the distal demonstrative that can be used to refer to familiar discourse referents. Valentina and I asked ourselves if this pattern is found in other languages and so we studied some 20 languages. The result was pretty clear: if a language reuses spatial demonstratives for grounding, it follows the same pattern. We develop a syntactic analysis for this finding within the framework of the Interactional Spine Hypothesis. Colasanti, V., Wiltschko, M. Demonstratives locate referents in common space and ground: A comparative syntactic approach. Nat Lang Linguist Theory (2025). https://doi.org/10.1007/s11049-025-09675-3 Rethinking structural growth: Insights from the acquisition of interactional language. In this paper, we introduce a novel proposal for the acquisition of syntactic structure. Most studies of syntactic first language acquisition focus on the increasingly complex expression of propositional thought. Layer by layer, from the bottom up, child language matures into adult-like representations. We challenge this account of an upward growing syntax based on evidence from early interactional language, which has largely been ignored in the study of syntactic acquisition. Given that interactional units of language are associated with the topmost layers of syntactic structure, and given that they are acquired early, it follows that syntactic acquisition cannot be characterized by upward growth. Rather, we propose that syntactic acquisition proceeds in an inward fashion. We present a case study of the early uses of utterance-final huh (an interactional unit of language) in North American child English that supports this hypothesis. We thus introduce the Inward Growing Spine Hypothesis as a research agenda. Heim, Johannes, Martina Wiltschko (2025). Rethinking structural growth: Insights from the acquisition of interactional language. Glossa: a journal of general linguistics 10(1). doi: https://doi.org/10.16995/glossa.16396 The grammar of self-talk. What different modes of talking reveal about language Self-talk has played an important role in theorizing about the function of language in the psychological and philosophical literature. Linguistic investigations of self-talk, however, are scarce. In this paper, I argue that there are several modes of self-talk including (i) thinking out loud, which is characterized by the absence of an addressee and (ii) having a conversation with oneself, which is characterized by the presence of a grammatically represented addressee role. In the latter, the person engaged in self-talk may hold the role of the speaker or the addressee. Thus, the grammatical restrictions on self-talk serve as a hitherto underexplored window into the grammatical representation of speaker and addressee roles. Wiltschko, Martina. "The grammar of self-talk. What different modes of talking reveal about language" Theoretical Linguistics, 2025. https://doi.org/10.1515/tl-2024-2024 Emotions do not enter grammar because they are constructed (by grammar) I finally finished this paper on the relation between language and emotions. The core idea I pursue here is that functional categories are (universally) not dedicated to the expression of emotions because emotions are constructed. To make this claim, I had to immerse myself into the fascinating scholarship of the affective sciences and I learned (once again) that we should not trust our intuitive understanding of anything, including what emotions are. And in the course of this research, I found myself surprised by the idea that emerged: the same system that is responsible for constructing language (call it grammar) may just be responsible for constructing emotions. The paper was recently published (completely open access) as a discussion paper in the online journal Language under Discussion. So maybe you feel inclined to write a comment or reply! Wiltschko, M. (2024). Emotions do not enter grammar because they are constructed (by grammar). Language Under Discussion, 7(1), 1–62. https://doi.org/10.31885/lud.7.1.261 Is grammar affected in Schizophrenia. A short reply to a recent paper by Zhang et al. that argues that grammar is affected in Schizophrenia. While this may well be, the criteria they use do not support this conclusion. Wiltschko Martina (2024). Is grammar affected in Schizophrenia? Psychiatry Research. Jul 2;339:116061. doi: 10.1016/j.psychres.2024.116061. Epub ahead of print. PMID: 38968919. The meaning of what A short paper on a use of what that does not question a propositional argument, as in a scenario when someone is staring at you and you utter What? I show that this use of what cannot be analyzed as an elliptical question. And I argue that it serves as an interactional question in that it questions an initiation move. I wrote this paper for a Festscrhift for Hotze Rullmann. The meaning of what. In M. Ryan Bochnak, Eva Csipak, Lisa Matthewson, Marcin Morzycki and Daniel K.E. Reisinger, eds. The title of this volume is shorter than its contributions are allowed to be: Papers in honour of Hotze Rullmann (UBC Occasional Papers in Linguistics 9). Vancouver: University of British Columbia. A new argument for linguistic determinants of human thought. The fact that grammatical categories (i.e., PERSON or TENSE), are necessary to construct sentences and define the content of the thought expressed suggests that thought cannot be independent of language. This is because grammatical categories are strictly grammatical, with not non-grammatical analogue. Hinzen, W., Martin, T. & Wiltschko, M. A new argument for linguistic determinants of human thought. Linguist and Philos (2024). https://doi.org/10.1007/s10988-024-09414-8 The Syntax of Talking Heads In this paper, I explore two discourse markers of Upper Austrian German: ma indicates surprise, while geh indicates a discrepancy between speaker and addresee. I argue that they are interactional pro-forms (ProGroundP) which mark a reaction to the speaker's or the addresee's current epistemic state. This extends the typology of proforms I have been working on for decades now. Wiltschko, M. 2024. The syntax of talking heads, Journal of Pragmatics, Volume 232: 182-198. Published online. https://doi.org/10.1016/j.pragma.2024.08.011. Pronouns beyond phi-features: The speaker–addressee relation in Japanese pronouns and its implication for formal pronouns In this paper, we demonstrate that Japanese pronouns bear the interactional roles of speaker and addressee but lack phi-features and that the separation of interactional roles from phi-features is structurally conditioned. Phi-based pronouns are introduced DP-internally while Japanese pronouns instantiate syntactic categories above the DP that realize interactional content including speaker and addressee roles. Ritter E, Wiltschko M. 2025. Pronouns beyond phi-features: the speaker–addressee relation in Japanese pronouns and its implications for formal pronouns. Journal of Linguistics. 61(3):653-688. doi:10.1017/S0022226724000306
.