Endowing models with consistent multilingual performance can be achieved by _mixing_ pre-training data, or post-training approaches such as language-specific model _merging_. In this work, we test whether merging can be applied to monolingually pre-trained models. We conduct a controlled study on the efficacy of mixed, merged, and monolingual pre-training setups. We find that while monolingual pre-training results in strong in-language performance, merging any combination of monolingual models leads to performance collapse due to interference. Our analysis suggests representational similarity is a prerequisite for model merging. We therefore conclude that the flexibility of merging in fine-tuning does not extend trivially to language-specific pre-training.

This paper describes the system submitted by our team to the Multilingual Euphemism Detection Shared Task for the Fourth Workshop on Figurative Language Processing (FigLang 2024). We propose a novel model for multilingual euphemism detection, combining contextual and behavior-related features. The system classifies texts that potentially contain euphemistic terms with an ensemble classifier based on outputs from behavior-related fine-tuned models. Our results show that, for this kind of task, our model outperforms baselines and state-of-the-art euphemism detection methods. As for the leader-board, our classification model achieved a macro averaged F1 score of [anonymized], reaching the [anonymized] place.