Translation Shift, Accuracy and Acceptability of Illocutionary Acts in Captions of @merche.id

This study examines Instagram's automatic translation of English-language captions from the local fashion brand @merche.id, focusing on the types of illocutionary acts, the translation shifts that occur during the translation process, and the resulting quality in terms of accuracy and acceptability. A descriptive qualitative approach was employed, drawing on Searle's (1976) taxonomy of illocutionary acts, Catford's (1965) translation shift theory, and Nababan et al.'s (2012) translation quality assessment model. The data consisted of 115 sentences drawn from 40 Instagram captions published between December 2025 and March 2026, along with their automatic Indonesian translations generated by the platform. A Focus Group Discussion involving qualified bilingual raters was conducted to assess translation quality. The findings revealed that Directive acts were the most dominant illocutionary type, reflecting the persuasive orientation of marketing captions. In terms of translation shifts, Structure Shift was the most frequent type across all illocutionary act categories, with Double Translation Shift as the most common pattern. Regarding translation quality, the average accuracy score was 2.55 and the average acceptability score was 2.52, both placing the translations at the borderline between less accurate and accurate categories. The findings demonstrate that Structure Shift is the most translation-neutral type, while Level Shift consistently produces the most damaging quality outcomes, particularly when applied to idiomatic and brand-specific expressions. The type of illocutionary act involved further moderates the impact of each shift on translation quality.

Keywords: illocutionary acts, translation shift, translation accuracy, translation acceptability, Instagram automatic translation @merche.id account.