FORMULA FIDELITY VS. POETIC FLEXIBILITY: A QUALITATIVE COMPARISON OF HUMAN AND AI TURKISH TRANSLATIONS OF POE’S “THE RAVEN”
Öz
Translating poetry entails challenges that extend beyond the mere transfer of meaning into the target language. In poetry, meaning is constructed through rhythm, repetition, sound patterning, semantic layering across lines, and culturally specific imagery. These features interact with syntax to shape both perception and the manner in which the text is read. When a poem is translated into another language, readers expect not only the preservation of semantic relations but also the maintenance of the musicality of the lines and the coherence of the scene. The present study sets out to compare four human translations of Edgar Allan Poe’s The Raven into Turkish, namely those by Ahmet Bağışgil, Ülkü Tamer, Burçak Özlüdil, and Oğuz Baykara, with four AI outputs produced by ChatGPT, Gemini, DeepL and Google Translate. The close reading of stanza-aligned poetry is applied to five dimensions: meaning, tone and style, sound patterning, cultural allusion, and formulaic closures. The set of examples has been meticulously curated and numbered and contains exclusively lines that are either striking or problematic. For each example, the translation strategy is identified, such as foreignization, domestication, calque, explicitation, idiomatic paraphrase, lexical shift, prosodic compensation, or register shift, and its effect on narrative coherence, voice and intertextual signalling is briefly explained. Across the corpus, AI translations have been shown to reliably preserve iconic tokens and scene framework. However, they often normalize cadence by flattening echo pairs and triads. The application of human translations is known to sustain musicality and rhetorical nuance through careful line breaks, recomposed echoes, and idiomatic address. It has been observed that such translations can, on occasion, domesticate symbols or props in a skilful manner. The objective is to eschew the use of rankings, and instead to support editorial practices through the implementation of a descriptive method and text-based observations. A complementary perspective on the roles of humans and AI is proposed, in which referential clarity is anchored by machines and prosody and voice are reanimated by humans.
Anahtar Kelimeler
poetry translation, comparison of human and AI translations, prosody, cultural allusion, "The Raven"
Kaynakça
- Attridge, D. (1995). Poetic rhythm: An introduction. Cambridge University Press.
- Castilho, S., Moorkens, J., Gaspari, F., Calixto, I., Tinsley, J., & Way, A. (2017). Is neural machine translation the new state of the art? The Prague Bulletin of Mathematical Linguistics, 108, 107–120. https://core.ac.uk/download/pdf/195384513.pdf
- DeepL GmbH. (2023). DeepL Translator (Online translation tool) Retrieved May 23, 2025, from https://www.deepl.com/translator
- Even-Zohar, I. (1990). The position of translated literature within the literary polysystem. In L. Venuti (Ed.), The translation studies reader (pp.192-197). Routledge.
- Fabb, N., & Halle, M. (2008). Meter in poetry: A new theory. Cambridge University Press.
- Google DeepMind. (2023). Gemini (large language model). Retrieved May 17, 2025, from https://www.deepmind.com/
- Google LLC. (2023). Google Translate (Online translation tool). Retrieved May 17, 2025, from https://translate.google.com/
- Holmes, J. S. (2000). The name and nature of translation studies. In L. Venuti (Ed.), The translation studies reader. (pp. 172-185). Routledge.
- Jakobson, R. (2000). On linguistic aspects of translation. In L. Venuti (Ed.), The translation studies reader. (pp. 113-118). Routledge.
- Läubli, S., Orrego-Carmona, D., & Toral, A. (2018). Has machine translation achieved human parity? a case for document-level evaluation. In: Proceedings of the 2018 Conference on Empirical Methods in Natural Language Proceeding (pp. 4791-4796). Association for Computational Linguistics. https://aclanthology.org/D18-1512/