Language models (LMs) call for a theoretical rethinking grounded within linguistics itself. Rather than signalling the “end of linguistics” or merely encouraging interdisciplinarity, LMs function as an empirical testing ground for formal linguistic concepts. They prompt a renewed examination of form–meaning mapping, theoretical autonomy, and the conditions under which computational systems can genuinely inform linguistic explanation.

Norm, the formal theoretical linguist, and Claudette, the computational language scientist, have a lovely time discussing whether modern language models can inform important questions in the language sciences. Just as they are about to part ways until they meet again, 25 of their closest friends show up – from linguistics, neuroscience, cognitive science, psychology, philosophy, and computer science. We use this discussion to highlight what we see as some common underlying issues: the String Statistics Strawman (the mistaken idea that LMs can’t be linguistically competent or interesting because they, like their Markov model predecessors, are statistical models that learn from strings) and the As Good As It Gets Assumption (the idea that LM research as it stands in 2026 is the limit of what it can tell us about linguistics). We clarify the role of LM-based work for scientific insights into human language and advocate for a more expansive research program for the language sciences in the AI age, one that takes on the commentators’ concerns in order to produce a better and more robust science of both human language and of LMs.

Language models excel at finding patterns in linguistic data, and can therefore prove insightful for statistical approaches to linguistics in that they provide further evidence for the strong reliance of natural languages on recurrent, fixed patterns. Nevertheless, regarding actual usage-based language processing, their implications are severely limited as they lack a crucial aspect of language use: interaction.

Futrell and Mahowald claim language models (LMs) “serve as model systems,” but an assessment at each of Marr’s three levels suggests the claim is clearly not true at the implementation level, poorly motivated at the algorithmic-representational level, and problematic at the computational theory level. LMs are good candidates as tools; calling them cognitive models overstates the case and unnecessarily feeds large language model hype.

This article addresses a central challenge in political science: how to choose between competing conceptions and structures of concepts. Existing approaches to concept validity offer useful criteria—such as resonance, consistency, differentiation, causal utility, and operationalization—but tend to omit criteria for evaluating the normative considerations that often underpin conceptual choices. As a result, conceptualization may face a fundamental indeterminacy when multiple conceptions appear equally well grounded. To address this lacuna, I introduce a “concern” criterion, which evaluates concepts according to the extent to which they capture what is most worrisome in the political world. Building on the semantic–pragmatic approach to conceptualization, I argue that normative considerations can be disciplined, rather than avoided. The argument is illustrated through the case of political polarization, where a shift from issue-based to affect-based conceptions reflects changing concerns about political division. I also examine the implications of the concern criterion for structuring multidimensional concepts and address objections concerning objectivity, stability, and legitimacy. A concern criterion, although not sufficient on its own, provides a valuable complement to existing criteria and helps ground conceptual choices in a normatively informed yet methodologically disciplined manner.

Large Language Models (LLMs) can serve as tools for understanding how probabilistic constraints interact during language acquisition. To motivate such use cases of LLMs, we discuss several examples from allied fields, including neurobiology and animal behavior, of how soft constraints shape learning and development in cognitive systems. We end by outlining four challenges that LLM cognitive modeling should address in the coming decade.

Under the lens of Marr’s (levels of analysis, we critique and extend the authors’ two points about language models (LMs) and language processing: first, predicting upcoming linguistic information based on context is key to language processing, and second, that many advances in psycholinguistics would be impossible without LLMs. We also outline directions combining LLMs’ strengths with psycholinguistic models.

Evidence-based nutrition guidance for female athletes remains limited relative to that available for males; in part, this has contributed to widespread reliance on social media for dietary information. Whilst social media can enhance health communication, it also facilitates the rapid dissemination of unverified, commercially driven nutrition claims. This narrative review critically synthesises the current scientific literature underpinning four prevalent claims targeting nutrition close to exercise for active females; (1) fasted training is harmful for all females, (2) menstrual cycle–related hormonal fluctuations require sex-specific hydration strategies, (3) carbohydrate differences by sex and menstrual cycle phase, and (4) precise protein timing is essential for optimal adaptation in females. Despite social media and ‘influencer’ claims of no evidence in humans for many of the claims, there is some, albeit limited, evidence. This review evaluates the available research and the evidence supporting these claims to provide practical advice for active females. Collectively, this review demonstrates that many widely circulated nutrition claims directed at active females lack robust scientific support. The findings emphasise the importance of individual context, including training load, energy availability, environmental conditions and total dietary intake, over rigid, sex-specific nutrition rules. Improved translation of female-specific sports nutrition research into accurate, accessible public messaging is urgently needed to counter persistent misinformation in digital media.

This paper examines virtual reality gaming as a form of embodied interaction at the intersection of digital mediation, improvisation, and agency. In VR environments, players act through avatars, and their actions are shaped in real time by shifting relations among embodiment, disembodiment, subjectivity, and intersubjectivity. The analysis brings together Gilles Deleuze and Félix Guattari’s concept of the Body without Organs, Charles Goodwin and Marjorie Harness Goodwin’s work on cooperation and multimodal interaction, and Alessandro Duranti’s account of improvisation. Focusing on Population: One and Richie’s Plank Experience, I argue that improvisation emerges through the unstable relation between the biophysical body and the digital body. Glitches, misalignments, and other breakdowns create moments in which participants must adjust ongoing action spontaneously, thereby destabilizing established physical and linguistic categories. These moments reveal a continuing process of deterritorialization and reterritorialization through which bodies, joint action, and agency are continuously reconfigured.

This special issue pursues a social semiotic study of improvisation. The approach considers the phenomenon both as a constitutive dimension of action and as a socially recognizable achievement. Contributions share a common focus on interaction, which is analyzed across multiple modalities including virtual reality, heavy machinery, paint and canvas, rock, theater, war, and the ethical relation between self and other.

Large language models are interesting, but linguistics and cognitive science should be cautious about centering any new technology as a magic bullet. Doing so reinforces the historically “narrow focus” of linguistics identified in the target article, rather than expanding our understanding of human language. We argue that centering values instead of technology is the best way to future-proof scholarly work and to keep finding out new things about language.

This study examines how love is metaphorically represented by non‑native English speakers, focusing on German learners of English as a second language. Drawing on interviews conducted in 2024, the research investigates the metaphorical linguistic features and conceptualizations used by 96 bachelor’s degree students when expressing love in English. Participants were asked to describe situations in which they would express love and to provide examples of how they would formulate such expressions. As the interviews took place in English used as a lingua franca, the dataset reflects communicative practices characteristic of ELF interactions, including creative or non‑standard metaphor use. Conceptual Metaphor Theory (CMT) serves as the analytical framework for identifying and interpreting the metaphorical patterns in the data. By relying on naturally occurring language rather than introspective intuition, the study addresses critiques of CMT, including its dependence on researcher interpretation and limited empirical grounding. The findings show how socio‑cultural background, L1 conceptual structures and L2 proficiency shape metaphorical expressions of love, contributing to a deeper understanding of emotional metaphor use in second language contexts.

In this response, we outline the central idea of a real pattern as applied in the philosophy of science as well as its relationship to the modal structure of models. We suggest that there may be additional concerns with linguistically real patterns in terms of convergence between natural language processing and formal linguistic theory.

This roundtable discussion convenes contributors to this special issue for reflections on the diversity of their research questions, approaches, and findings, as well as avenues for further inquiry. The discussion touches on several shared concerns, including the role of lexicalization in improvised forms becoming intelligible, enregistered, and reusable; how improvisational action might implicate action below the threshold of awareness; and more generally the relation between sign-processes and embodied action. In doing so, this roundtable discussion considers the importance of social semiotic analysis for understanding improvisation.

Futrell and Mahowald’s assumptions about parallels between language models and human language are primarily based on LMs’ performance in English. In our commentary, we discuss how this impacts some of their key takeaways and highlight LM shortcomings in languages other than English (specifically, Icelandic), which are only alluded to in the target article.

China’s persistently low fertility is associated with fertility inequality, reflected in a U-shaped relationship between household human capital and fertility. We develop an overlapping-generations model showing that this pattern depends on the substitutability of educational inputs. When educational inputs are complementary, fertility is U-shaped in household human capital, with middle-human-capital households having the fewest children; when inputs are substitutable, the relationship is inverted U-shaped. Using China Family Panel Studies data, we find a robust U-shaped relationship between household human capital and fertility, significant complementarity among educational time, monetary investment, and household human capital in children’s human-capital formation, and similar patterns across eastern, central, and western China. Complementarity requires households to increase time and monetary inputs jointly, intensifying the quantity–quality trade-off, particularly for middle-human-capital households. Policies that enhance substitutability among educational inputs may therefore mitigate fertility inequality and raise aggregate fertility.

Acknowledging that large language models have learned to use language can open doors to breakthrough language science. Achieving these breakthroughs may require abandoning some long-held ideas about how language knowledge is evaluated and reckoning with the difficult fact that we have entered a post-Turing test era.

Food habits vary across ethnic groups and geographical regions. However, validated dietary assessment tools accounting for such diversity remain limited. A semi-quantitative food frequency questionnaire (FFQ) was developed and validated to assess the habitual food intake of adolescents and adults across Malaysia. The 147-item FFQ was constructed using commonly consumed foods from five main ethnicities (Malay, Chinese, Indian, and Sabah and Sarawak indigenous groups) identified from national surveys. A cross-sectional validation study was conducted among purposively sampled healthy individuals aged 10–59 years from 16 administrative regions. Trained community nutritionists administered the FFQ to assess monthly intake, alongside a three-day dietary record and recall (3DRR) covering two weekdays and one weekend. Spearman’s correlation, Bland–Altman plots, and quartile cross-classification evaluated the agreement between the FFQ and 3DRR for energy, macronutrients, and selected micronutrients (Vitamin C, thiamine, calcium, and iron). Respondents (n = 361; 50.3% adults, 49.7% adolescents) were 50.4% female and represented five main ethnicities (range: 15.8–25.2%), with 60.4% from Peninsular Malaysia. Energy intake estimated by the FFQ (median: 2285 kcal) was significantly higher than by the 3DRR (median: 1785 kcal; Wilcoxon p < 0.001). Spearman’s correlation coefficients observed for energy (crude r = 0.31), and selected nutrients (energy-adjusted r range: 0.19–0.38), along with <10% of extreme quartile misclassification indicated acceptable ranking ability and agreement for most nutrients. Bland–Altman plots indicated no proportional bias for energy and macronutrients. In conclusion, the FFQ is a valid tool for assessing dietary intake within the multi-ethnic Malaysian population nationwide.