Multiword expressions (MWEs) are composed of multiple words and exhibit variable degrees of compositionality. As such, their meanings are notoriously difficult to model, and it is unclear to what extent this issue affects transformer architectures. Addressing this gap, we provide the first in-depth survey of MWE processing with transformer models. We overall find that they capture MWE semantics inconsistently, as shown by reliance on surface patterns and memorized information. MWE meaning is also strongly localized, predominantly in early layers of the architecture. Representations benefit from specific linguistic properties, such as lower semantic idiosyncrasy and ambiguity of target expressions. Our findings overall question the ability of transformer models to robustly capture fine-grained semantics. Furthermore, we highlight the need for more directly comparable evaluation setups.

Pretrained language models (PLMs) have been shown to encode binary gender information of text authors, raising the risk of skewed representations and downstream harms. This effect is yet to be examined for transgender and non-binary identities, whose frequent marginalization may exacerbate harmful system behaviors. Addressing this gap, we first create TRANsCRIPT, a corpus of YouTube transcripts from transgender, cisgender, and non-binary speakers. Using this dataset, we probe various PLMs to assess if they encode the gender identity information, examining both frozen and fine-tuned representations as well as representations for inputs with author-specific words removed. Our findings reveal that PLM representations encode information for all gender identities but to different extents. The divergence is most pronounced for cis women and non-binary individuals, underscoring the critical need for gender-inclusive approaches to NLP systems.

This report presents the results of the shared tasks organized as part of the VarDial Evaluation Campaign 2024. The campaign is part of the eleventh workshop on Natural Language Processing (NLP) for Similar Languages, Varieties and Dialects (VarDial), co-located with NAACL 2024. Two shared tasks were included this year: dialectal causal commonsense reasoning (DIALECT-COPA), and Multi-label classification of similar languages (DSL-ML). Both tasks were organized for the first time this year, but DSL-ML partially overlaps with the DSL-TL task organized in 2023.

This paper provides a framework and tool set for computing and visualising dynamic, time- specific semantic neighbourhoods of English noun-noun compounds and their constituents over time. Our framework not only identifies salient vector-space dimensions and neighbours in notoriously sparse data: we specifically bring together changes in meaning aspects and degrees of (non-)compositionality.

The escalating challenge of misinformation, particularly in political discourse, requires advanced fact-checking solutions; this is even clearer in the more complex scenario of multimodal claims. We tackle this issue using a multimodal large language model in conjunction with retrieval-augmented generation (RAG), and introduce two novel reasoning techniques: Chain of RAG (CoRAG) and Tree of RAG (ToRAG). They fact-check multimodal claims by extracting both textual and image content, retrieving external information, and reasoning subsequent questions to be answered based on prior evidence. We achieve a weighted F1-score of 0.85, surpassing a baseline reasoning technique by 0.14 points. Human evaluation confirms that the vast majority of our generated fact-check explanations contain all information from gold standard data.

Bosnian, Croatian, Montenegrin and Serbian are the official standard linguistic varieties in Bosnia and Herzegovina, Croatia, Montenegro, and Serbia, respectively. When these four countries were part of the former Yugoslavia, the varieties were considered to share a single linguistic standard. After the individual countries were established, the national standards emerged. Today, a central question about these varieties remains the following: How different are they from each other? How hard is it to distinguish them? While this has been addressed in NLP as part of the task on Distinguishing Between Similar Languages (DSL), little is known about human performance, making it difficult to contextualize system results. We tackle this question by reannotating the existing BCMS dataset for DSL with annotators from all target regions. We release a new gold standard, replacing the original single-annotator, single-label annotation by a multi-annotator, multi-label one, thus improving annotation reliability and explicitly coding the existence of ambiguous instances. We reassess a previously proposed DSL system on the new gold standard and establish the human upper bound on the task. Finally, we identify sources of annotation difficulties and provide linguistic insights into the BCMS dialect continuum, with multiple indicators highlighting an intermediate position of Bosnian and Montenegrin.

Predicting the compositionality of noun compounds such as climate change and tennis elbow is a vital component in natural language understanding. While most previous computational methods that automatically determine the semantic relatedness between compounds and their constituents have applied a synchronic perspective, the current study investigates what diachronic changes in contexts and semantic topics of compounds and constituents reveal about the compounds’ present-day degrees of compositionality. We define a binary classification task that utilizes two diachronic vector spaces based on contextual co-occurrences and semantic topics, and demonstrate that diachronic changes in cosine similarities – measured over context or topic distributions – uncover patterns that distinguish between compounds with low and high present-day compositionality. Despite fewer dimensions in the topic models, the topic space performs on par with the co-occurrence space and captures rather similar information. Temporal similarities between compounds and modifiers as well as between compounds and their prepositional paraphrases predict the compounds’ present-day compositionality with accuracy >0.7.

We investigate the descriptive relevance of widely used semantic change models in linguistic descriptions of present-day speech communities. We focus on the sociolinguistic issue of contact-induced semantic shifts in Quebec English, and analyze 40 target words using type-level and token-level word embeddings, empirical linguistic properties, and – crucially – acceptability ratings and qualitative remarks by 15 speakers from Montreal. Our results confirm the overall relevance of the computational approaches, but also highlight practical issues and the complementary nature of different semantic change estimates. To our knowledge, this is the first study to substantively engage with the speech community being described using semantic change models.

We investigate the effect of sub-word tokenization on representations of German noun compounds: single orthographic words which are composed of two or more constituents but often tokenized into units that are not morphologically motivated or meaningful. Using variants of BERT models and tokenization strategies on domain-specific restricted diachronic data, we introduce a suite of evaluations relying on the masked language modelling task and compositionality prediction. We obtain the most consistent improvements by pre-splitting compounds into constituents.

To date, transformer-based models such as BERT have been less successful in predicting compositionality of noun compounds than static word embeddings. This is likely related to a suboptimal use of the encoded information, reflecting an incomplete grasp of how the models represent the meanings of complex linguistic structures. This paper investigates variants of semantic knowledge derived from pretrained BERT when predicting the degrees of compositionality for 280 English noun compounds associated with human compositionality ratings. Our performance strongly improves on earlier unsupervised implementations of pretrained BERT and highlights beneficial decisions in data preprocessing, embedding computation, and compositionality estimation. The distinct linguistic roles of heads and modifiers are reflected by differences in BERT-derived representations, with empirical properties such as frequency, productivity, and ambiguity affecting model performance. The most relevant representational information is concentrated in the initial layers of the model architecture.

This study investigates the applicability of semantic change detection methods in descriptively oriented linguistic research. It specifically focuses on contact-induced semantic shifts in Quebec English. We contrast synchronic data from different regions in order to identify the meanings that are specific to Quebec and potentially related to language contact. Type-level embeddings are used to detect new semantic shifts, and token-level embeddings to isolate regionally specific occurrences. We introduce a new 80-item test set and conduct both quantitative and qualitative evaluations. We demonstrate that diachronic word embedding methods can be applied to contact-induced semantic shifts observed in synchrony, obtaining results comparable to the state of the art on similar tasks in diachrony. However, we show that encouraging evaluation results do not translate to practical value in detecting new semantic shifts. Finally, our application of token-level embeddings accelerates manual data exploration and provides an efficient way of scaling up sociolinguistic analyses.

We present a 78.8-million-tweet, 1.3-billion-word corpus aimed at studying regional variation in Canadian English with a specific focus on the dialect regions of Toronto, Montreal, and Vancouver. Our data collection and filtering pipeline reflects complex design criteria, which aim to allow for both data-intensive modeling methods and user-level variationist sociolinguistic analysis. It specifically consists in identifying Twitter users from the three cities, crawling their entire timelines, filtering the collected data in terms of user location and tweet language, and automatically excluding near-duplicate content. The resulting corpus mirrors national and regional specificities of Canadian English, it provides sufficient aggregate and user-level data, and it maintains a reasonably balanced distribution of content across regions and users. The utility of this dataset is illustrated by two example applications: the detection of regional lexical and topical variation, and the identification of contact-induced semantic shifts using vector space models. In accordance with Twitter’s developer policy, the corpus will be publicly released in the form of tweet IDs.