<![CDATA[Song lyric translation in animated films must preserve narrative and emotional cues while remaining natural and readable. The objective of this study was to examine how translation techniques used in a machine-generated Indonesian version of “Remember Me” (Coco) related to translation quality. A bilingual corpus of 13 English lyric lines was compiled from LyricsTranslate and was paired with unedited Indonesian outputs produced by DeepL in September–October 2025. Each line pair was coded using Molina and Albir’s 18-technique taxonomy through investigator triangulation, and translation quality was assessed by two expert raters using Nababan and Nuraeni’s rubric for accuracy, acceptability, and readability (1–3 scale). Technique mapping showed a refrain-driven profile: established equivalence dominated (38.46%), especially for the repeated line “Remember me,” followed by modulation (23.08%) and literal translation (15.38%), with amplification, generalization, and calque occurring once each. Quality scores were generally high (mean accuracy 98.72%, acceptability 96.15%, readability 97.44%), but reductions clustered where calque-like collocations and connective-heavy modulation made the Indonesian phrasing less idiomatic and harder to process, particularly in one dense line. The findings suggested that MT lyric output could maintain meaning reliably in short repeated segments, while naturalness remained the main vulnerability in culturally and emotionally loaded lines.]]>
Copyrights © 2025
