Large language models (LLMs) achieve impressive results in terms of fluency in text generation, yet the nature of their linguistic knowledge – in particular the human-likeness of their internal lexicon – remains uncertain. This study compares human and LLM-generated word associations to evaluate how accurately models capture human lexical patterns. Using English cue-response pairs from the SWOW-EN dataset and newly generated associations from three LLMs (Mistral-7B, Llama-3.1-8B, and Qwen-2.5-32B) across multiple temperature settings, we examine (i) the influence of lexical factors such as word frequency and concreteness on cue-response pairs, and (ii) the variability and typicality of LLM responses relative to humans. Results show that all models mirror human trends for frequency and concreteness but differ in response variability and typicality. Larger models such as Qwen tend to emulate a single “prototypical” human participant, generating highly typical but minimally variable responses, while smaller models such as Mistral and Llama produce more variable yet less typical responses. Temperature settings further influence this trade-off, with higher values increasing variability but decreasing typicality. These findings highlight both the similarities and differences between human and LLM lexicons, emphasizing the need to account for model size and temperature when probing LLM lexical representations.

Word associations have a longstanding tradition of being instrumental for investigating the organization of the mental lexicon. Despite their wide application in psychology and psycholinguistics, analyzing word associations remains challenging due to their inherent heterogeneity and variability, shaped by linguistic and extralinguistic factors. Existing word-association taxonomies often suffer limitations due to a lack of comprehensive frameworks that capture their complexity.To address these limitations, we introduce a linguistically motivated taxonomy consisting of co-existing meaning-related and form-related relations, while accounting for the directionality of word associations.We applied the taxonomy to a dataset of 1,300 word associations (FAMWA) and assessed it using various LLMs, analyzing their ability to classify word associations.The results show an improved inter-annotator agreement for our taxonomies compared to previous studies (𝜅 = .60 for meaning and 𝜅 = .58 for form). However, models such as GPT-4o perform only modestly in relation labeling (with accuracies of 46.2% for meaning and 78.3% for form), which calls into question their ability to fully grasp the underlying principles of human word associations.

In this paper, we present and test an annotation scheme designed to analyse the semantic properties of derived nouns in context. Aiming at a general semantic comparison of morphological processes, we use a descriptive model that seeks to capture semantic regularities among lexemes and affixes, rather than match occurrences to word sense inventories. We annotate two distinct features of target words: the ontological type of the entity they denote and their semantic relationship with the word they derive from. As illustrated through an annotation experiment on French corpus data, this procedure allows us to highlight semantic differences and similarities between affixes by investigating the number and frequency of their semantic functions, as well as the relation between affix polyfunctionality and lexical ambiguity.

French, as many languages, lacks semantically annotated corpus data. Our aim is to provide the linguistic and NLP research communities with a gold standard sense-annotated corpus of French, using WordNet Unique Beginners as semantic tags, thus allowing for interoperability. In this paper, we report on the first phase of the project, which focused on the annotation of common nouns. The resulting dataset consists of more than 12,000 French noun occurrences which were annotated in double blind and adjudicated according to a carefully redefined set of supersenses. The resource is released online under a Creative Commons Licence.

The Asfalda project aims to develop a French corpus with frame-based semantic annotations and automatic tools for shallow semantic analysis. We present the first part of the project: focusing on a set of notional domains, we delimited a subset of English frames, adapted them to French data when necessary, and developed the corresponding French lexicon. We believe that working domain by domain helped us to enforce the coherence of the resulting resource, and also has the advantage that, though the number of frames is limited (around a hundred), we obtain full coverage within a given domain.