Title: Information exchange and Opinion discussion as different communicative contexts: a corpus-based conversational analysis.

Abstract: In this presentation, I will introduce an annotation scheme for online conversations grounded in three key pragmatic concepts: Common Ground, Speech Acts, and Discourse Relations. I will discuss the results of the annotation process, which reveal two primary conversational contexts and allow us to characterize each with distinctive patterns in the interaction between speech acts and discourse relations. Finally, I will address some challenges encountered in the annotation process and outline strategies for scaling it using large language models (LLMs). 

Title: Developmental change in children’s mental representation of kinship terms: Preliminary data from Mandarin Chinese

Abstract: Kinship system is a fundamental aspect of human societies, and its transmission across generations is crucial. However, children typically take an extended period to acquire kinship terms. While the cognitive constraints on this process are well established, emerging research highlights significant cross-cultural variation in the acquisition trajectory, emphasizing the critical role of sociocultural factors. This project investigates the acquisition of kinship terms among Mandarin-speaking children in China, a linguistically and culturally rich context that offers a unique opportunity to examine how early word learning is shaped by linguistic components, children’s daily experiences, and culturally specific cognitive tendencies. I will present the preliminary findings of my field work last summer and look forwards to brainstorming the next steps.

Title: Data-driven Cross-linguistic Variation in Languages and Machines

Abstract: Progress in computational linguistics and natural language processing (NLP), albeit impressive, has predominantly focused on (written) English and languages with large speaker populations. In this talk, I outline a research program that advances the scope of computational language science by promoting linguistic diversity. I first present work on cross-linguistic variation in syntactic ordering preferences as well as morphosyntactic structural transfer in second language learning from a typological perspective. I then turn to variation in computational models and datasets, highlighting challenges in common evaluation practices in multilingual NLP. Building on this, I discuss strategies for designing datasets for under-resourced languages, and conclude with reflections on the ethical considerations of working with indigenous speech communities.

Title: Probing the noun-verb distinction in Zinacantec Family Homesign

Abstract: A distinction between nouns and verbs is thought to be fundamental to human language. Indeed, recent studies report that the N/V distinction is apparent even in homesign and emerging sign languages created de novo by deaf individuals without access to an existing language. Building on these studies, I explore strategies for marking the distinction in Zinacantec Family Homesign (‘Z sign’), a sign language developed over the last four decades by deaf and hearing members of an extended family in an indigenous community of southern Mexico. I first examine production data, elicited using an extended version of a task from a prior study (Abner et al. 2019). I focus on several features attested in other languages (movement size, reduplication, and utterance position); additionally, I examine a feature that has not been examined before (eye gaze). Second, I will seek feedback on the development of a novel comprehension task that incorporates the production data, intended to test whether the Z signers can distinguish de-contextualized noun signs from verb signs.

Title: Hyperlinks on the social media platform X: function-anchoring and knowledge-state synchronization

Abstract: 

In this presentation, I will discuss the role of hyperlinks in online conversations. Traditionally defined as mere tools for ‘relaying information’, I will offer a more complex view of what a hyperlink can do in an online conversation context: (i) anchoring parameters of an associated textual situation (ii) operating synchronization of the knowledge state of a speaker with the one of her future (often unknown) addressee.

First, drawing on an empirical corpus study, I present a descriptive analysis of the hyperlink functions. Based on this survey, I propose an initial – and non-exhaustive – typology of hyperlink functions observed in this online context, which includes a.o. source anchoring (direct, indirect), illustration (focal, generic), referent specification (discourse, object entities), justification of speech act production, rhetorical, prospeech.

Second, I will provide a preliminary formalization of hyperlinks within the situation theory framework. I suggest that a hyperlink is a function that anchors the hyperlink destination to the textual situation. In this formalization, we will look closer at the relationships between demonstratives and the hyperlink’s destination situation.

The discourse-level implications of hyperlink use are as follows. When a reader clicks on a hyperlink, their informational state may be updated, hereby synchronized with the speaker’s knowledge state. This update, however, depends on the novelty of the information conveyed within the hyperlink’s destination. I argue that the notion of novelty is closely tied to a notion of accessibility: the more accessible a piece of information is (e.g., public, widely known, already widely shared online), the more likely it is to be shared knowledge or already known by the addressee; conversely, the less accessible it is (e.g., private information of the speaker or context-specific), the less likely it is to be commonly shared between the two informational states.

As this project is still ongoing, I am happy to have some feedback or criticisms on these preliminary findings.

Title: Is dos possessive? A study of variation in Yiddish possessive determiner use

Abstract: In certain contexts, Yiddish permits the definite article ‘the’ to be used possessively, especially with body parts and kinship terms. Textbooks and grammars have claimed that using a possessive determiner (e.g. mayn ‘my’) is marked where possessive ‘the’ is possible. Examples from naturalistic speech, however, demonstrate that possessive determiners are regularly used outside of the contexts these sources would predict. To investigate the (non-)linguistic factors underlying the distribution of these determiners, I annotated 2,200 noun phrases from 30 speakers in the Corpus of Spoken Yiddish in Europe. This talk will highlight methodological challenges in analyzing Yiddish at scale, situate my findings in the typology of external possession, and discuss prospects for further research in this area and the development of computational tools for Yiddish corpus linguistics.