NLP research frequently grapples with multiple sources of variability—spanning runs, datasets, annotators, and more—yet conventional analysis methods often neglect these hierarchical structures, threatening the reproducibility of findings. To address this gap, we contribute a case study illustrating how linear mixed-effects models (LMMs) can rigorously capture systematic language-dependent differences (i.e., population-level effects) in a population of monolingual and multilingual language models. In the context of a bilingual hate speech detection task, we demonstrate that LMMs can uncover significant population-level effects—even under low-resource (small-N) experimental designs—while mitigating confounds and random noise. By setting out a transparent blueprint for repeated-measures experimentation, we encourage the NLP community to embrace variability as a feature, rather than a nuisance, in order to advance more robust, reproducible, and ultimately trustworthy results.

In this article we present UNED-ACCESS 2024, a bilingual dataset that consists of 1003 multiple-choice questions of university entrance level exams in Spanish and English. Questions are originally formulated in Spanish and manually translated into English, and have not ever been publicly released, ensuring minimal contamination when evaluating Large Language Models with this dataset. A selection of current open-source and proprietary models are evaluated in a uniform zero-shot experimental setting both on the UNED-ACCESS 2024 dataset and on an equivalent subset of MMLU questions. Results show that (i) Smaller models not only perform worse than the largest models, but also degrade faster in Spanish than in English. The performance gap between both languages is negligible for the best models, but grows up to 37% for smaller models; (ii) Model ranking on UNED-ACCESS 2024 is almost identical (0.98 Pearson correlation) to the one obtained with MMLU (a similar, but publicly available benchmark), suggesting that contamination affects similarly to all models, and (iii) As in publicly available datasets, reasoning questions in UNED-ACCESS are more challenging for models of all sizes.

Recent studies comparing AI-generated and human-authored literary texts have produced conflicting results: some suggest AI already surpasses human quality, while others argue it still falls short. We start from the hypothesis that such divergences can be largely explained by genuine differences in how readers interpret and value literature, rather than by an intrinsic quality of the texts evaluated. Using five public datasets (1,471 stories, 101 annotators including critics, students, and lay readers), we (i) extract 17 reference-less textual features (e.g., coherence, emotional variance, average sentence length...); (ii) model individual reader preferences, deriving feature importance vectors that reflect their textual priorities; and (iii) analyze these vectors in a shared “preference space”. Reader vectors cluster into two profiles: _surface-focused readers_ (mainly non-experts), who prioritize readability and textual richness; and _holistic readers_ (mainly experts), who value thematic development, rhetorical variety, and sentiment dynamics. Our results quantitatively explain how measurements of literary quality are a function of how text features align with each reader’s preferences. These findings advocate for reader-sensitive evaluation frameworks in the field of creative text generation.

In the context of text representation, Compositional Distributional Semantics models aim to fuse the Distributional Hypothesis and the Principle of Compositionality. Text embedding is based on co-ocurrence distributions and the representations are in turn combined by compositional functions taking into account the text structure. However, the theoretical basis of compositional functions is still an open issue. In this article we define and study the notion of Information Theory–based Compositional Distributional Semantics (ICDS): (i) We first establish formal properties for embedding, composition, and similarity functions based on Shannon’s Information Theory; (ii) we analyze the existing approaches under this prism, checking whether or not they comply with the established desirable properties; (iii) we propose two parameterizable composition and similarity functions that generalize traditional approaches while fulfilling the formal properties; and finally (iv) we perform an empirical study on several textual similarity datasets that include sentences with a high and low lexical overlap, and on the similarity between words and their description. Our theoretical analysis and empirical results show that fulfilling formal properties affects positively the accuracy of text representation models in terms of correspondence (isometry) between the embedding and meaning spaces.

In this paper we conduct a set of experiments aimed to improve our understanding of the lack of semantic isometry in BERT, i.e. the lack of correspondence between the embedding and meaning spaces of its contextualized word representations. Our empirical results show that, contrary to popular belief, the anisotropy is not the root cause of the poor performance of these contextual models’ embeddings in semantic tasks. What does affect both the anisotropy and semantic isometry is a set of known biases: frequency, subword, punctuation, and case. For each one of them, we measure its magnitude and the effect of its removal, showing that these biases contribute but do not completely explain the phenomenon of anisotropy and lack of semantic isometry of these contextual language models.

In this paper we introduce TweetNorm_es, an annotated corpus of tweets in Spanish language, which we make publicly available under the terms of the CC-BY license. This corpus is intended for development and testing of microtext normalization systems. It was created for Tweet-Norm, a tweet normalization workshop and shared task, and is the result of a joint annotation effort from different research groups. In this paper we describe the methodology defined to build the corpus as well as the guidelines followed in the annotation process. We also present a brief overview of the Tweet-Norm shared task, as the first evaluation environment where the corpus was used.