|
A Comparison of the Performance of Google Translate in 2018 and 2023 |
|---|---|
| รหัสดีโอไอ | |
| Creator | Pugpimon Panyakham |
| Title | A Comparison of the Performance of Google Translate in 2018 and 2023 |
| Publisher | Faculty of Liberal Arts, Prince of Songkla University |
| Publication Year | 2567 |
| Journal Title | Journal of Liberal Arts, Prince of Songkla University |
| Journal Vol. | 16 |
| Journal No. | 1 |
| Page no. | 271938, pp.1-51 |
| Keyword | Google Translate, Translation, Lexical errors, Syntactic errors, Discourse errors |
| URL Website | https://so03.tci-thaijo.org/index.php/journal-la/about |
| Website title | Journal of Liberal Arts, Prince of Songkla University |
| ISSN | 2651-1126 |
| Abstract | Google Translate has evolved into an indispensable tool for Thai readers seeking to comprehend English texts. While it may not be flawless, it offers remarkable features that facilitate readers in grasping the overall meaning. Furthermore, its continuous annual progress necessitates ongoing studies. Therefore, this article sets out to compare Google Translate's machine translation errors in two online news articles retrieved from both 2018 and 2023, from an English-to-Thai perspective. One example from beginner-level reading comprehension materials was also included in the analysis. These texts underwent meticulous qualitative and quantitative analyses to identify errors introduced by Google Translate. The findings of this study unveiled the inevitability of errors in Google Translate's translations. These errors predominantly fell into three major categories: lexical, syntactic, and discourse. Notably, Google Translate exhibited a penchant for making lexical errors in the translated texts in both 2018 and 2023. The frequency of errors in Google Translate was 87% in 2018 and decreased to 39% in 2023. From the total errors, Google Translate made lexical errors in 2018 for 55%, syntactic errors for 30%, and discourse errors for 20%. In contrast, the error rate improved in 2023: lexical errors decreased to 25%, syntactical errors to 10%, and discourse errors to 10%, indicating advancements over the past half-decade. Despite the prevalence of errors, this study aims to provide explanations and practical implications to enhance future use. While Google Translate's errors may occasionally hinder a reader's comprehension, the software still holds the potential to offer a general understanding of a text. Recognizing the reliance on translation tools and understanding the types of errors are critical steps for readers to employ these tools more effectively. |
