Human Translators and Artificial Intelligence in the Translation of Onomatopoeias and Interjections Based on the Problematic of the Arbitrariness/Motivation of the Sign
Dilber ZeytinkayaSounds that imitate and depict human voices, animal sounds, sounds of nature, and sounds emanating from objects can be defined as onomatopoeia. Interjections are words that convey emotions such as joy, pain, surprise, greeting, approval, disapproval, fear, and hesitation. As an interdisciplinary field of study for linguistics and translation studies, it is important to elucidate how onomatopoeias and interjections are written, how they are translated, and which translation strategies are used. Can an artificial intelligence tool translate onomatopoeias and interjections? In this study, the research object was the French comic strip Lucky Luke–Daisy Town. Published in 1989 by Milliyet Publishing House and in 2000 by Yapı Kredi Publishing House, the Turkish translations of Red Kit–Papatya Kasabası and the English translations of A Lucky Luke Adventure–Daisy Town published in 2016 by Cinebook Publishing House were analyzed. Translations by human translators and translation suggestions of the artificial intelligence tool ChatGPT3.5 were compared and interpreted within the scope of linguistics and translation studies. Because the emergence of onomatopoeias and interjections is associated with the emergence of language, assumptions about the emergence of language are included. Sound symbolism, in onomatopoeia is one of the main subject areas, was emphasized, and the relationship between sound and meaning was analyzed. The concept of language sign was clarified, and the arbitrariness/motivation of the sign was discussed. The study concluded that human translators preferred the substitution strategy the most, used the borrowing strategy and perspective shifting strategies at the same rate, and preferred the word-for-word translation strategy the least. The results revealed that artificial intelligence cannot provide perfect results in every field and under every condition. Thus, human translators cannot be replaced for a long time, particularly in the translation of works containing humor, metathesis, onomatopoeias, and interjections.
Göstergenin Nedensizliği/Nedenliliği Sorunsalından Hareketle Yansıma Seslerin ve Ünlemlerin Çevirisinde İnsan Çevirmenler ve Yapay Zekâ
Dilber Zeytinkayaİnsan seslerini, hayvan seslerini, doğa seslerini, nesnelerden çıkan sesleri taklit ve tasvir eden sesler, yansıma sesler olarak tanımlanabilir. Ünlemler ise sevinç, acı, şaşkınlık, selamlama, onaylama, onaylamama, korku, tereddüt gibi duyguları ileten sözcüklerdir. Hem dil biliminin hem de çeviri biliminin çalışma alanı olarak disiplinlerarası bir konumda yer alan yansıma seslerin ve ünlemlerin nasıl yazıldıkları, nasıl çevrildikleri ve hangi çeviri stratejilerine başvurulduğu önem teşkil eder. Yapay zekâ aracı yansıma sesleri ve ünlemleri çevirebilir mi? Bu çalışmada Fransızca çizgi roman Lucky Luke - Daisy Town araştırma nesnesi olarak belirlenmiştir. 1989 yılında Milliyet Yayınları ve 2000 yılında Yapı Kredi Yayınları tarafından yayımlanan Türkçe Red Kit - Papatya Kasabası ile 2016 yılında Cinebook yayınevi tarafından yayımlanan İngilizce A Lucky Luke Adventure - Daisy Town adlı çeviriler incelenmiştir. İnsan çevirmenlerin çevirileriyle, yapay zekâ aracı ChatGPT3.5’in çeviri önerileri karşılaştırılarak dil bilimi ve çeviri bilimi kapsamında yorumlanmıştır. Yansıma seslerin ve ünlemlerin ortaya çıkışı dilin doğuşuyla ilişkilendirildiği için dilin doğuşuna dair varsayımlara yer verilmiştir. Temel konu alanlarından biri yansıma sesler olan ses sembolizmi üzerinde durulmuş, ses-anlam ilişkisi irdelenmiştir. Dil göstergesi kavramına açıklık getirilmiş, göstergenin nedensizliği/nedenliliği tartışılmıştır. İnsan çevirmenlerin en çok ikame stratejisini tercih ettikleri saptanırken, ödünçleme stratejisi ile bakış açısı kaydırma stratejilerini aynı oranda kullandıkları, en az sıklıkla ise sözcüğü sözcüğüne çeviri stratejisini tercih ettikleri sonucuna varılmıştır. Yapay zekânın her alanda, her koşulda mükemmel sonuçlar veremeyeceği ortaya konmuştur. Özellikle mizah, göçüşme, yansıma ses ve ünlem içeren eserlerin çevirilerinde insan çevirmenlerin yerinin uzun bir süre doldurulamayacağı anlaşılmıştır.
Onomatopoeic words are the sounds that imitate and depict the sounds in nature, sounds made by humans and animals, and sounds resulting from actions such as falling and hitting. The sounds that people make are universal, whereas onomatopoeic words are pronounced and written differently in each language. A human being or an animal does not make different sounds in different languages. They make the same sounds; however, the way each society perceives and makes sense of sounds differently.
Artificial intelligence technologies are shaping today’s conditions. The emergence of real-time translation systems has brought debates, and artificial intelligence applications in translation are being recognized as a threat to the translation profession. Accordingly, translators must keep up with technology and cooperate with artificial intelligence. The lack of studies in Turkey to date that relate onomatopoeias and interjections to artificial intelligence supports the importance of this study.
In this study, the research object was the French comic book Lucky Luke–Daisy Town, published by Dargaud Publishing House in 1983, because it contains numerous (approximately 95) onomatopoeias and interjections. Published in 1989 by Milliyet Publishing House and in 2000 by Yapı Kredi Publishing House, the Turkish translations of Red Kit–Papatya Kasabası and the English translations of A Lucky Luke Adventure–Daisy Town published in 2016 by Cinebook Publishing House were evaluated. Translations by human translators and translation suggestions of the artificial intelligence tool ChatGPT3.5 were compared and interpreted within the scope of linguistics and translation studies.
Based on the results of the analysis of examples, Turkish onomatopoeias and interjections can be evaluated within the scope of the principle of motivation of signs because they are perceived in the same way and met with the same phonemes. Because the English and French equivalents of Turkish onomatopoeias and interjections are perceived with different phonemes, they can be evaluated within the scope of the principle of arbitrariness of signs.
In the translation of onomatopoeias, human translators used substitution, borrowing, perspective-shifting, and word-for-word translation strategies. Numerous examples were evaluated under the substitution strategy because substitution includes adaptation, equivalence, and domesticating. Given that onomatopoeias and interjections are specific to each language and culture, the substitution strategy basically involves the replacement of the onomatopoeias and interjections in the source text with their equivalents in the target text. In addition to translation strategies, the figure of speech “intak” was used. The concept of metathesis, which is one of the important sound events in the comic book as an element of humor, was explored, and the results revealed that the artificial intelligence tool could not translate the metathesis example. In the 1989 translation, most of the onomatopoeic words were not translated and left as they were. The probable reason was that the publishing house did not want to spend time on graphic editing due to reasons such as time and financial loss. Human translators have a good command of the source and target languages and culture. Conversely, the artificial intelligence tool that was involved in the translation process was not aware of information such as text type, source text author, and target audience. Thus, it sometimes fails to suggest accurate translations. The accuracy rate is higher in English than in Turkish. This may be because the voices of the dominant culture, such as English, have become universally accepted. The artificial intelligence tool provided two suggestions in most cases. At this point, human translators will choose the correct translation suggestion. The judgment of human translators is currently superior to that of artificial intelligence tools.
Faculty members working in Turkish language and literature and faculty members working in the field of translation studies can conduct joint studies. Multilingual dictionaries of onomatopoeias can be prepared. In the future, the same artificial intelligence tool can be prompted to translate the same source text. This can determine whether the artificial intelligence tool has renewed itself over the years and whether it can provide accurate suggestions. Research reproducibility is important for the continuity of translation studies. From the results, human translators cannot be replaced for a long time, particularly in the translation of works containing onomatopoeias and interjections.