One of the goals of fairness research in NLP is to measure and mitigate stereotypical biases that are propagated by NLP systems. However, such work tends to focus on single axes of bias (most often gender) and the English language. Addressing these limitations, we contribute the first study of multilingual intersecting country and gender biases, with a focus on occupation recommendations generated by large language models. We construct a benchmark of prompts in English, Spanish and German, where we systematically vary country and gender, using 25 countries and four pronoun sets. Then, we evaluate a suite of 5 Llama-based models on this benchmark, finding that LLMs encode significant gender and country biases. Notably, we find that even when models show parity for gender or country individually, intersectional occupational biases based on both country and gender persist. We also show that the prompting language significantly affects bias, and instruction-tuned models consistently demonstrate the lowest and most stable levels of bias. Our findings highlight the need for fairness researchers to use intersectional and multilingual lenses in their work.

We present the LCT-LAP proposal for the shared task on Translation into Low-Resource Languages of Spain at WMT24 within the constrained submission category. Our work harnesses encoder-decoder models pretrained on higher-resource Iberian languages to facilitate MT model training for Asturian, Aranese and Aragonese. Furthermore, we explore the robustness of these models when fine-tuned on datasets with varying levels of alignment noise. We fine-tuned a Spanish-Galician model using Asturian data filtered by BLEU score thresholds of 5, 15, 30 and 60, identifying BLEU 15 as the most effective. This threshold was then applied to the Aranese and Aragonese datasets. Our findings indicate that filtering the corpora reduces computational costs and improves performance compared to using nearly raw data or data filtered with language identification. However, it still falls short of the performance achieved by the rule-based system Apertium in Aranese and Aragonese.