Якість машинного перекладу в парі українська — польська: Google Translate (2018 р. проти 2025 р.) та GPT-5

2026, 14, Том 14

Publishing model

open access

License type



Language of publication

Polish, Ukrainian, English

Number of views:1

Number of downloads:0

Crossref citations:—


Abstract

The article examines the quality of machine translation in the Ukrainian–Polish language pair. With the growing importance of this pair and the wider availability of linguistic resources, the quality of translations has improved noticeably, as demonstrated in this study. The analysis focuses on syntactic and lexical aspects, with particular attention to difficulties caused by the flexible word order present in both languages. The study begins with a comparison of translations generated by Google Translate in 2018 and the results of current translations (2025) produced by the same system working with neural machine translation (NMT). These outcomes are then compared with translations of the same text provided by GPT-5, which employs large language models (LLMs). This comparison makes it possible to trace the development of translation technologies and shows a clear improvement in translation quality within the examined period.