In this paper, we present a localized and culturally adapted Estonian translation of the test set from the widely used commonsense reasoning benchmark, WinoGrande. We detail the translation and adaptation process carried out by translation specialists and evaluate the performance of both proprietary and open source models on the human translated benchmark. Additionally, we explore the feasibility of achieving high-quality machine translation by incorporating insights from the manual translation process into the design of a detailed prompt. This prompt is specifically tailored to address both the linguistic characteristics of Estonian and the unique translation challenges posed by the WinoGrande dataset. Our findings show that model performance on the human translated Estonian dataset is slightly lower than on the original English test set, while performance on machine-translated data is notably worse. Additionally, our experiments indicate that prompt engineering offers limited improvement in translation quality or model accuracy, and highlight the importance of involving language specialists in dataset translation and adaptation to ensure reliable and interpretable evaluations of language competency and reasoning in large language models.

The goal of the EstNLTK Python library is to provide a unified programming interface for natural language processing in Estonian. As such, previous versions of the library have been immensely successful both in academic and industrial circles. However, they also contained serious structural limitations – it was hard to add new components and there was a lack of fine-grained control needed for back-end programming. These issues have been explicitly addressed in the EstNLTK library while preserving the intuitive interface for novices. We have remastered the basic NLP pipeline by adding many data cleaning steps that are necessary for analyzing real-life texts, and state of the art components for morphological analysis and fact extraction. Our evaluation on unlabelled data shows that the remastered basic NLP pipeline outperforms both the previous version of the toolkit, as well as neural models of StanfordNLP. In addition, EstNLTK contains a new interface for storing, processing and querying text objects in Postgres database which greatly simplifies processing of large text collections. EstNLTK is freely available under the GNU GPL version 2 license, which is standard for academic software.