Comparative Analysis of Turkish Translations of Idiomatic Expressions Found in The Movie “Slumberland” Made by Machine Translations Applications
Şahin Sarı, Murat ÖzcanIn this study, how idiomatic expressions in the movie "Slumberland" released by Netflix are translated from English into Turkish by human translators and machine translation tools, and the differences between them are analysed grammatically, contextually, and theoretically with Peter Newmark's Translation Methods and Strategies. It aims to determine to what extent machine translation applications can translate compared to human translators, and whether they can make translations suitable for contextual meaning. As a result, when the findings obtained in the study are evaluated, the dubbing and subtitle translations made by the human translator are grammatically and contextually appropriate and completely target-language oriented. When evaluated theoretically, functional, cultural, and descriptive equivalent procedures are used. When the Turkish translations of machine translation tools are evaluated, Google Translation can make accurate translations in accordance with contextual meaning with a success rate of 63% and other translation applications with a success rate of 54%. Translations suitable for contextual meaning can be explained by functional, cultural, and descriptive equivalent procedures, and translations not suitable for contextual meaning can be explained by the literal translation procedure. As a result, the success rates of machine translation applications are not yet at the desired level when compared to human translators.
Hayaller Diyarı Filminde Bulunan Deyimsel İfadelerin Makine Çevirisi Uygulamalarıyla Yapılan Türkçe Çevirilerinin Karşılaştırmalı Analizi
Şahin Sarı, Murat ÖzcanBu çalışmada, Netflix bünyesinde vizyona giren “Hayaller Diyarı” filminde bulunan deyimsel ifadelerin insan çevirmenler ve makine çevirisi uygulamaları tarafından İngilizceden Türkçeye nasıl çevrildikleri ve aralarındaki farklar dilbilgisel, bağlamsal ve Peter Newmark`ın Çeviri Yöntem ve Stratejileri ile kuramsal açıdan incelenerek, makine çevirisi uygulamalarının insan çevirmen karşısında ne ölçüde çeviri yapabildiklerinin, bağlamsal anlama uygun çeviriler yapıp yapamadıklarının belirlenmesi amaçlanmaktadır. Kaynak kültürde kullanılan deyimsel ifadeler, insan çevirmen tarafından yapılan altyazı ve dublaj çevirilerindeki karşılıkları, makine çevirisi uygulamalarından “Google Çeviri, Yandex Çeviri, Microsoft Bing Çeviri ve Apple Çeviri” çıktıları çalışmanın örneklemini oluşturmaktadır. Çalışmada kullanılan veriler nitel araştırma yöntemleri bünyesinde doküman incelemesi yani karşılaştırmalı ve betimsel analiz yönteminden faydalanılarak incelenmiştir. Sonuç olarak çalışmada elde edilen bulgular değerlendirildiğinde insan çevirmen tarafından yapılan dublaj ve altyazı çevirilerinin dilbilgisel ve bağlamsal açıdan uygun ve tamamen erek dil odaklı oldukları söylenebilir. Kuramsal açıdan değerlendirildiklerinde, çoğunlukla işlevsel ve kültürel eşdeğerlik süreci olmak üzere betimsel eşdeğerlik sürecinden yararlanıldığı da söylenebilir. Makine çevirisi uygulamalarının Türkçe çevirileri değerlendirildiğinde Google Çeviri %63, diğer çeviri uygulamaları ise %54 başarı oranıyla bağlamsal anlama uygun çeviri sunmaktadır. Bağlamsal anlama uygun çeviriler işlevsel, kültürel ve betimsel eşdeğerlik süreçleriyle, bağlamsal anlama uygun olmayan çeviriler ise sözcüğü sözcüğüne çeviri süreciyle açıklanabilir. Sonuç olarak, makine çevirisi uygulamalarının başarı oranları insan çevirmenle kıyaslandığında henüz beklentiyi karşılamamaktadır.
This study examines how idiomatic expressions in the Netflix film "Slumberland" are translated from English into Turkish by human translators and machine translation tools, and how the two translation methods differ from one another linguistically, culturally, and conceptually. Idiomatic expressions used in the source culture, their equivalents in subtitle and dubbing translations made by human translators, and the outputs of "Google Translate, Yandex Translate, Microsoft Bing Translate and Apple Translate" from machine translation tools constitute the sample of the study. The study`s data were examined by using document analysis, that is, comparative and descriptive analysis methods, within the scope of qualitative research methods.
In this study, answers to the following questions are sought. 1. How are idiomatic expressions used in the source language translated into Turkish by human translators and machine translation applications, and what are the differences between them?
2. Which methods and strategies of Peter Newmark do human translators and machine translation applications use in the translation of idiomatic expressions?
3. Are translations source language-oriented or target language-oriented?
4. Can machine translation tools make translations that are appropriate for contextual meaning when compared to human translators?
In the study, the online databases of "Cambridge Dictionary, Merriam Webster and Farlex Dictionary of Idioms" for idiomatic expressions in the source language and "Turkish Language Institution" dictionaries for Turkish idiomatic expressions are used. The translations of machine translation applications were accessed in January and February 2023. First, the meanings of idiomatic expressions translated from the source language, namely English to the target language Turkish, are explained by using the online databases of the "Cambridge Dictionary, Merriam Webster and Farlex Dictionary of Idioms" dictionaries. Then, grammatical differences that will affect the translation of idiomatic expressions are evaluated. Afterwards, subtitles and dubbing translations made by human translators and translations made by machine translation applications are analysed comparatively in tables. As a result of the analysis, it is determined whether the translation outputs of machine translation applications could make translations suitable for contextual meaning like human translators, and which methods and strategies they used to translate.
According to the analysis results, it can be said that dubbing translations made by human translators are grammatically and contextually appropriate and completely target language oriented. When evaluated from a theoretical point of view, it can be said that in dubbing translations of eleven idiomatic expressions, the functional equivalence procedure is used five times (45 per cent), the cultural equivalence procedure is used four times (36 per cent), one with the addition procedure (9 per cent), and the descriptive equivalence procedure is used twice (18 per cent). In subtitle translations, it can be said that the functional equivalence process is used five times (45 per cent), the cultural equivalence process is used three times (27 per cent), and the descriptive equivalence process is used three times (27 per cent).
According to the analysis results of eleven idiomatic expression translations made by machine translation applications; Google Translate offers a total of seven translations suitable for contextual meaning (63 per cent). Three of the translations suitable for contextual meaning can be explained by the procedure of cultural equivalence (27 per cent), three of them by functional equivalence (27 per cent) and one of them by descriptive equivalence (9 per cent). All four translations that do not fit the context can be explained by the literal translation process (36 per cent).
Yandex Translate offers a total of six translations suitable for contextual meaning (54 per cent). Three of these can be explained by the procedure of cultural equivalence (27 per cent) and the other three by the process of functional equivalence (27 per cent). All five translations that do not fit the context can be explained by the literal translation process (45 per cent).
Microsoft Bing Translation and Apple Translation offer a total of six translations suitable for contextual meaning (54 per cent). Three of these can be explained by the procedure of functional equivalence (27 per cent), two by cultural equivalence (18 per cent), and the other by the descriptive equivalence procedure (9 per cent). All five translations that do not fit the context can be explained by the literal translation process (45 per cent).
In conclusion, when the Turkish translations of machine translation tools are evaluated, Google Translation can make accurate translations in accordance with contextual meaning with a success rate of 63% and other translation applications with a success rate of 54%. Translations suitable for contextual meaning can be explained by functional, cultural, and descriptive equivalent procedures, and translations not suitable for contextual meaning can be explained by the literal translation procedure. The most important factors affecting the success rate are the idioms usage frequency and the contributions made by the users. It is directly proportional to the successful translations of the applications. In this way, it becomes easier to reach the correct translations in the database and the applications continue to learn. Considering that the number of internet users and the entries in the database are increasing day by day, it will be inevitable for machine translation tools to make much more successful translations in the future.
