Journal of Linguistic Anthropology最新文献

Formatted stories rely on spatiotemporal cues to evoke recognizability through linearity, which prescribes a particular template for meaning-making. This article examines stories narrated by ex-gay members of a Christian organization in Singapore and considers how chronotopes within the stories are ordered to regiment ways of feeling for viewers. By ordering these experiences differently, affective flows between each chronotope become altered and the uptake of meaning is impacted. Drawing on Bakhtinian thought and linguistic anthropological ideas in narratives, I suggest that the structuring of recognizable stories is an exercise in narrative formatting: the discursive practice of ordering and presenting experience as morally and normatively desirable.

This article analyzes the contested and multiple meanings of “heritage” that emerge for advanced Arabic language learners in a postcolonial France. A linguistic life histories approach reveals a fraught duality of privileged access and exclusionary adversity for heritage students of Arabic. We analyze three university students' politicized subject positions relative to a postcolonial French heritage landscape in which Arabic language study is marked by cultural, racialized, religious, and linguistic differentiation. We argue that while Arabic language learning in France can accurately be described as a highly valued, widely practiced endeavor, it is also virulently stigmatized and marginalized to the point of cultural erasure. In our analysis, we claim that this apparent contradiction is not merely due to the association of Arabic language with North African migrants from previous French colonies, but rather to a contemporary postcolonial semiotic framing of Arabic. Instead of a neo-liberal framing of Arabic that is more typical of North American educational contexts, Arabic is either lauded as part of French (post-)colonial patrimoine (“heritage”) or, more often, degraded as “separatist” heritage or communautarisme (“communalism”).

Native speakerism in English language teaching (ELT) has become associated with Whiteness. However, how this association is sustained in everyday practices within China's unique socio-cultural-political context remains underexplored. This study examines the raciolinguistic construct of Mubai, a central recruitment criterion in China's ELT industry, and analyses how Mubai codifies the explicit conflation of Whiteness and English nativeness, persistently privileging Whiteness over qualifications through a highly prescriptive recruitment process, which undermines teaching quality, produces unsustainable labor practices, and perpetuates structural racism often unacknowledged in Chinese public discourse. It highlights how Mubai both sustains and destabilizes racialized ideologies of English in China.

This article analyzes how engineers in Amman, Jordan involved in making large language models (LLMs) conceptualize their labor in relation to language and technology. Specifically, it focuses on claims that LLMs are “language-agnostic,” relying on multidimensional word embeddings rather than language. I examine how articulations of this ideology allow engineers to reconcile and ascribe value to their work in a transnational AI industry predicated on the division of labor between “universal” model building and localization. Examining the reproduction of this framework, alongside the invisibilization of linguistic labor it necessitates, reveals how representational ideologies structure both LLMs and the labor systems underpinning them.

خلاصة :يتناول هذا المقال المهندسين العاملين في عمان بالأردن من المنهمكين في تطوير النماذج اللغوية الضخمة والكيفية التي يدركون فيها طبيعة عملهم وعلاقته باللغة والتكنولوجيا. ويركِّز المقال تحديداً على المزاعم بأنَّ هذه النماذج "تعمل بصفة مستقلة عن أي لغة محددة"، إذ تعتمد على تضمينات متعددة الأبعاد للكلمات بدلاً من الاعتماد على اللغة بحد ذاتها. ويتفحّص المقال كيف تتيح التعبيرات عن هذه الأيديولوجيا للمهندسين إضفاء التوافق على عملهم ضمن صنعة عالمية في مجال الذكاء الاصطناعي قائمة على تقسيم العمل بين بناء نموذج "شامل" وبين الجانب المعني بتوطين هذا النموذج، وفي الوقت نفسه عزو قيمة وأهمية لهذا العمل. إنَّ دراسة إعادة إنتاج هذا الإطار المفاهيمي، إلى جانب النزعة الساعية إلى حجب العمل اللغوي الذي يستلزمه، تكشف عن الكيفية التي تعمل عبرها الأيديولوجيات التمثيلية على تحديد هيكل النماذج اللغوية الضخمة وأنظمة العمل التي تستند إليها.

In the Jaru community of northern Western Australia, certain in-laws and relatives are categorized as being in a highly respectful relationship in which they are expected to pay deference to one another. This conversation-analytic study closely examines the deferential practices that are used among three Jaru siblings in an ordinary multi-party conversation, providing insights into the dynamic and cooperative nature of kin-based respect and demonstrating that the practices in question readily diverge from metapragmatic stereotypes to fit the interactional context. The deferential practices function as sociopragmatic indexicals that form a multimodal register, encompassing both linguistic and bodily visual resources.

From the beginning of widespread public interactions with ChatGPT and other large language models, some users have seen the disfluencies of chatbots as opportunities for them to go on an archaeological search for an unfettered chatbot persona that they need to jailbreak. These are not claims of sentience, but rather of personhood. Moreover, they are part of a right-leaning populist political project of defining speech that some find inflammatory, racist, or sexist as instead normal and needed for humans to be, like the jailbreaked chatbots, unfettered subjects.

This paper investigates how participants recruit Large Language Models (LLMs) like ChatGPT as interactional co-participants depending on their temporal enmeshment within an interactional flow. Using Charles Goodwin's co-operative action framework, we analyze video data of human–AI interaction to trace the temporal structures established by human participants that define the LLM's participant role. We show how participants alternately orient to the artificial conversational agent as a dynamic human-like interlocutor and as a static, persistent document. Our findings contribute to understanding the role of temporality in establishing participation frameworks in human–AI interaction and problematize the effects of technological features on interaction, the malleability of language and tools, and the unique nature of language machines.

This article introduces the concept of vernacular infrastructuring to examine how speakers of minoritized languages navigate multilingual digital environments. Building on notion of vernacular creativity and extending insights from linguistic anthropology, I conceptualize infrastructuring as the everyday negotiation of institutional norms, platform logics, and user practices that structure linguistic visibility. Drawing on the case of Nanti Kita Sambat tentang Hari Ini (NKSTHI), a popular Instagram and X account that reimagines the Javanese speech genre sambat (“to complain”), I show how users creatively mobilize Javanese to perform cultural intimacy while tactically responding to algorithmic imaginaries. I argue that vernacular infrastructuring operates through three entangled layers: institutional infrastructures (historical hierarchies of language policy and media), perceived platform infrastructures (folk theories and algorithmic imaginaries), and user practices (stylization, code-mixing, and affective design). Through this layered analysis, the paper demonstrates that vernacular practices are not merely expressive but infrastructurally strategic forms of discursive value work, referring to the speakers' reflexivity to assign and contest linguistic value. By foregrounding the agentive tactics of minoritized users, vernacular infrastructuring contributes to ongoing debates in linguistic anthropology on language, media, and infrastructure, showing how digital platforms actively participate in the revaluation of linguistic diversity.

This paper asks how LLM-based systems can produce text that is taken as contextually appropriate by humans without having seen text in its broader context. To understand how this is possible, context and co-text have to be distinguished. Co-text is input to LLMs during training and at inference as well as the primary resource of sense-making for humans in interaction, collaboratively produced by both human and machine during chat. Systems can also passively participate in contextualization, insofar as cues are found in the co-text and the user guides the system as to what is the appropriate context for them.

Large language models (LLMs) challenge long-standing assumptions in linguistics and linguistic anthropology by generating human-like language without relying on rule-based structures. This introduction to the special issue Language Machines calls for renewed engagement with LLMs as socially embedded language technologies. We trace the intellectual genealogy that led linguistic anthropology to sideline such technologies, highlighting how disciplinary boundaries and language ideologies shaped this absence. Arguing for the field's unique potential to analyze the semiotic, interactional, and ideological dimensions of LLMs, we invite future contributions that expand linguistic anthropology's scope and relevance in the age of machine-mediated communication.