Previous studies on a variety of languages have demonstrated that manual gesture is temporally aligned with prosodic prominence. However, the majority of these studies have been conducted on languages with word-level stress. In this paper, we investigate the alignment of manual beat gestures to speech in local varieties of Standard Indonesian, a language whose word prosodic system has been the subject of conflicting claims. We focus on the varieties of Indonesian spoken in the eastern part of the archipelago and Java. Our findings reveal that there is a strong tendency to align gesture to penultimate syllables in the eastern variety and a tendency to align gesture to final syllables in the Javanese variety. Additionally, while the eastern patterns appear to be word based, the Javanese pattern shows evidence of being phrase based. Surprisingly, the penultimate syllable emerges as a gestural anchor in the eastern variety even for two of the three speakers who showed little to no regular prosodic prominence on this syllable. This suggests that gestural alignment may serve to uncover prosodic anchors even when they are not employed by the phonology proper.

This contribution presents a perceptual dialectology study conducted with 123 Albanian-speaking participants, who rated the correctness and pleasantness of speech around Albania. We investigate how ratings were modulated by three factors: a well-established dialectal division within Albania, relative urbanization across the country, and the participants’ dialect backgrounds. These three factors were found to interact in the correctness and pleasantness ratings given by the participants, which is generally consistent with previous perceptual dialectology studies conducted in other linguistic settings but also highlights some nuances and complexities in this relationship. While heavily urbanized centers in central Albania were rated as highly correct and pleasant independently from prior dialect descriptions or dialect background of the participants, in one dialect area, less urbanized counties were rated more pleasant. We argue that these insights from non-linguists could serve as starting point for future scientific inquiry.

The present study examines how L1-English learners acquire Korean subject honorification – a system that is socio-pragmatic in interpretation but syntactically constrained. Using a multi-method design (corpus analysis, politeness ratings, and self-paced reading), we find that learners show increasing sensitivity to politeness norms yet limited awareness of morphosyntactic constraints. In corpus analysis, learners used subject honorification almost exclusively alongside addressee honorification, indicating limited functional differentiation. In politeness ratings, learners consistently associated the subject honorific suffix with greater politeness, regardless of subject type, diverging from native speakers’ judgments. In self-paced reading, learners were sensitive to semantic anomalies (e.g., inanimate subjects) but not to morphosyntactic violations. Together, these findings suggest that learners interpret the subject honorific suffix as a general politeness marker, likely due to its low cue validity and frequent co-occurrence with pragmatically salient features. Our results highlight how cue reliability and competition shape L2 acquisition pathways under conditions of noisy linguistic representations.

In this article, I analyse the word-prosodic system of Drubea and Numèè, two of the rare tonal Oceanic languages. Building on Rivierre’s (1973) seminal work, I show that the word-prosodic system of these two languages can be analysed as involving only register features: an underlying downstep and a postlexical epenthetic upstep. Drubea and Numèè are thus tonal languages without tones stricto sensu. This new type of word-prosodic system has both theoretical and typological implications: (i) register features, defined as in Snider’s (1999) Register Tier Theory, need not be subordinate to or associated with tones, and may exist in the absence of tone, including in underlying representation; (ii) tonal systems come in two types: tone-based systems in which the tonal contrasts are defined paradigmatically, as in most tone languages, and register-based systems where tonal contrasts are defined syntagmatically, as in Drubea and Numèè.

Autoregressive language models generate text by predicting the next word from the preceding context. The regularities internalized from specific training data make this mechanism a useful proxy for historically situated readerly expectations, reflecting what earlier linguistic communities would find probable or meaningful. In this article, I pre-train a GPT model (223M parameters) on a broad corpus of Chinese texts (FineWeb Edu Chinese V2.1) and fine-tune it on the collected writings of Mao Zedong (1893–1976) to simulate the evolving linguistic landscape of post-1949 China. Identifying token sequences with the sharpest drops in perplexity – a measure of the model’s surprise – reveals the core phraseology of “Maospeak,” the militant language style that developed from Mao’s writings and pronouncements. A comparative analysis of modern Chinese fiction demonstrates how literature becomes unfamiliar to the fine-tuned model, generating perplexity spikes of increasing magnitude. The findings suggest a mechanism of attentional control: whereas propaganda backgrounds meaning through repetition (cognitive overfitting), literature foregrounds it through deviation (non-anomalous surprise). By visualizing token sequences as perplexity landscapes with peaks and valleys, the article reconceives style as a probabilistic phenomenon and showcases the potential of “cognitive stylometry” for literary theory and close reading .

Though the US Supreme Court is famous for ideological disagreements among its Justices, agreement may in fact be the norm: most appeals are not politically salient, unanimous rulings are common, and even divided rulings require at least five Justices to agree. Because nearly all speaking turns of Justices in oral arguments are in the form of questions to an attorney, any linguistic evidence of agreement would have to be in the ways that these questions are asked. In this study, I review an oral argument for evidence of agreement, with a focus on supportive alignment, that is, when one party ratifies or approves of another’s conversation turn. I analyze two questions from Justices that were later repeated and endorsed by other Justices, and I argue that these reuses are a form of supportive alignment driven by the unique interactional constraints of the setting. (Institutional discourse, legal discourse, US Supreme Court, multiparty interaction, alignment)