ANALIZA KONTRASTIVE E PËRKTHIMIT ME MAKINË NGA SHQIPJA NË SERBISHT - GOOGLE TRANSLATE DHE CHATGPT

MARKOVIQ, Angjella and MALA-IMAMI, Naile (2026) ANALIZA KONTRASTIVE E PËRKTHIMIT ME MAKINË NGA SHQIPJA NË SERBISHT - GOOGLE TRANSLATE DHE CHATGPT. ALBANOLOGJIA International Journal of Albanology, 13 (25-26). pp. 372-384.

[img] Text
ALBANOLOGJIA 2026-372-384.pdf
Download (693kB)

Abstract

Building on prior research into machine translation using Google’s service – Google Translate, we conducted a comparative analysis of this translation and the translation generated by another popular model, ChatGPT. The translation of natural languages through artificial intelligence has become an indispensable aspect of modern communication and language processing. This paper aims to examine the advantages and drawbacks of both tools by analysing translation quality, accuracy in conveying meaning, and identifying syntactic and semantic errors. The research was conducted using sentence examples from The Palace of Dreams by Ismail Kadare, a work that has gained international recognition, including significant reception in Serbia due to its profound social and political themes, as well as linguistic and stylistic specificities that pose challenges for translation. In the analysis, translations generated by Google Translate and ChatGPT were compared with the author’s translation by Shkëlzen Maliqi, considered a reference sample as it has undergone some professional editing. The findings indicate that Google Translate offers a more efficient approach due to its extensive database but often produces translations with syntactic inconsistencies and literal rendering of meaning. Additionally, ChatGPT, while more flexible in handling complex structures, shows a tendency to make significant semantic errors, translating certain phrases literally or even incorrectly, which diminishes translation accuracy in contexts that require precise terminology. The conclusion of this study is that both tools can be useful for different types of translation tasks, with the optimal choice depending on text complexity and the purpose of the translation. Further development of these technologies could enhance the quality of machine translation and improve language processing capabilities.

Item Type: Article
Subjects: L Education > L Education (General)
Divisions: Faculty of Law, Arts and Social Sciences > School of Education
Depositing User: Unnamed user with email zshi@unite.edu.mk
Date Deposited: 28 Sep 2026 13:00
Last Modified: 28 Sep 2026 13:00
URI: http://eprints.unite.edu.mk/id/eprint/2546

Actions (login required)

View Item View Item
Universiteti i Tetoves is powered by EPrints 3 which is developed by the School of Electronics and Computer Science at the University of Southampton. More information and software credits.