We investigate the effect of terminology injection for terminology-constrained translation in a low-resource language variety, with a particular focus on off-the-shelf instruction-tuned Large Language Models (LLMs). We compare a total of 9 models: 4 instruction-tuned LLMs from the Tower and EuroLLM suites, which have been specifically trained for translation-related tasks; 2 generic open-weight LLMs (LLaMA-8B and Mistral-7B); 3 Neural Machine Translation (NMT) systems (an adapted version of MarianMT and ModernMT with and without the glossary function). To this end, we release LegISTyr, a manually curated test set of 2,000 Italian sentences from the legal domain, paired with source Italian terms and target terms in the South Tyrolean standard variety of German. We select only real-world sources and design constraints on length, syntactic clarity, and referential coherence to ensure high quality. LegISTyr includes a homonym subset, which challenges systems on the selection of the correct homonym where sense disambiguation is deducible from the context. Results show that while generic LLMs achieve the highest raw term insertion rates (approximately 64%), translation-specialized LLMs deliver superior fluency (∆ COMET up to 0.04), reduce incorrect homonym selection by half, and generate more controllable output. We posit that models trained on translation-related data are better able to focus on source-side information, producing more coherent translations.

This paper illustrates the process of training and evaluating NMT systems for a language pair that includes a low-resource language variety.A parallel corpus of legal texts for Italian and South Tyrolean German has been compiled, with South Tyrolean German being the low-resourced language variety. As the size of the compiled corpus is insufficient for the training, we have combined the corpus with several parallel corpora using data weighting at sentence level. We then performed an evaluation of each combination and of two popular commercial systems.

The paper reports on the creation, annotation and curation of the MT@BZ corpus, a bilingual (Italian–South Tyrolean German) corpus of machine-translated legal texts from the officially multilingual Province of Bolzano, Italy. It is the first human error-annotated corpus (using an adapted SCATE taxonomy) of machine-translated legal texts in this language combination that includes a lesser-used standard variety. The data of the project will be made available on GitHub and another repository. The output of the customized engine achieved notably better BLEU, TER and chrF2 scores than the baseline. Over 50% of the segments needed no human revision due to customization. The most frequent error categories were mistranslations and bilingual (legal) terminology errors. Our contribution brings fine-grained insights to Machine translation evaluation research, as it concerns a less common language combination, a lesser-used language variety and a societally relevant specialized domain. Such results are necessary to implement and inform the use of MT in institutional contexts of smaller language communities.