TÜRKÇEDE MAKİNE ÇEVİRİSİNİN SÖZ DİZİMSEL VE ANLAMSAL HATALARI: GOOGLE TRANSLATE VE DEEPL ÖRNEKLERİ
SYNTACTIC AND SEMANTIC ERRORS IN MACHINE TRANSLATION INTO TURKISH: A COMPARATIVE ANALYSIS OF GOOGLE TRANSLATE AND DEEPL

Summary/Abstract: This study investigates how well machine translation systems-specifically DeepL and Google Translate—can grasp and reproduce the flexible syntactic structure of the Turkish language. Given that these translation tools are primarily developed based on English, the study analyzes the challenges they face when processing Turkish. The analysis compares naturally produced sentences by native Turkish speakers with machine-translated sentences generated from English sources. The sample consists of 20 sentences that gradually increase in semantic complexity. These sentences are analyzed at both semantic and structural levels to evaluate the syntactic and semantic errors made by the translation engines in terms of morphological, lexical, syntactic, and semantic parameters. The study is divided into four main sections: The first section discusses the concept of the sentence in relation to its social and societal dimensions, emphasizing the semantic variations of recurring and shifting syntactic phrases that depend on the user. The second section examines the reasons for syntactic errors specific to Turkish in machine translations, focusing on the pragmatic functions of sentence elements, including emphasis, theme, and rheme. The third section presents a qualitative analysis of how the selected 20 sentences are processed by the translation engines in terms of both function and form. Finally, the conclusion identifies, through quantitative data, the instances in which machine translations fail to capture modality, tense perception, emphasis, and metaphorical meaning in Turkish, as well as their shortcomings in aligning linguistic structure with social context.

• Details
• Contents