| Abstract [eng] |
The aim of this Master Thesis is to analyze collected conceptual metaphors related to human health in Lithuanian and English media and their machine translation. The objectives of the research are: 1) To analyze the theoretical framework of conceptual metaphors and machine translation. 2) To collect a database of Lithuanian and English journalistic articles related to human health. 3) To analyze how machine translation engines translate conceptual metaphors related to human health collected in Lithuanian and English journalistic texts. 4) To compare how machine translation engines translate conceptual metaphors in Lithuanian and English. The research vividly describes 84 examples (42 from each language) that reflect the metaphorization of human health, specifically diseases. The results of the study show that the most prominent way of depicting human health in the Lithuanian media is to compare the illness to a thief, a dictator, a struggle, a beast, death, a companion, a natural disaster, and misery. In the English media, disease is described as a killer, a thief, a struggle, a process, a hungry beast, an irresistible force, a nightmare, and a destroyer, respectively. The main aim of the study was to find out how machine translation (MT) engines deal with the translation of conceptual metaphors related to human health. The quality of the translations provided by the five MT engines (Tilde, ChatGPT, Microsoft Translate, Google Translate, and DeepL) was evaluated according to the following three groups, which were marked with three colours: 1) GREEN colour was used to mark examples of good translations, where conceptual metaphors were translated vividly and correctly; 2) RED colour was used for examples whose translations did not reflect the conceptual metaphor and lost its imagery and meaning; 3) BLUE was used for words and fragments that were distant from the original text, and contained logic, style and grammar errors, that significantly compromised the quality of the translation. According to the study results, The most effective MT engines for translating metaphors from Lithuanian into English are Microsoft Translate and Google Translate, and for translating metaphors from English into Lithuanian are DeepL, Microsoft Translate, and Google Translate. The DeepL and ChatGPT translations from Lithuanian to English had the lowest number of style, logic, grammatical, and lexical errors, while DeepL and Google Translate had the lowest number of errors from English to Lithuanian. Despite the good translation suggestions for conceptual metaphors, Tilde turned out to be the least trustworthy MT engine. The translations provided by this MT engine were riddled with errors in punctuation, logic, grammar, and lexis, which majorly compromised the quality of the translation. |
