ANALYSIS OF TRANSLATION TECHNIQUES AND ACCURACY OF NOUN PHRASES ON THE @KAI121 ACCOUNT ON X

This study examines the translation techniques and accuracy of noun phrases found in the automatic translations of customer information published on the official X account of PT Kereta Api Indonesia (@KAI121). The data consists of 60 noun phrases collected from posts published between January 19 and April 15, 2026. Noun phrase classification follows Quirk et al.'s (1985) framework, translation technique analysis applies Molina & Albir's (2002) taxonomy, and translation accuracy is assessed using Nababan et al.'s (2012) model through a Focus Group Discussion involving one rater. The findings reveal that post-modifier noun phrases are the most dominant structural type, accounting for 90.00% of the total data, followed by pre-modifier noun phrases at 8.33%, and pre and post-modifier noun phrases at 1.67%. In terms of translation techniques, literal translation is applied most frequently at (83.08%), followed by borrowing (9.23%), modulation (3.08%), Established equivalent (3.08%), and amplification (1.54%). The accuracy assessment produces a mean score of 2.38, placing the overall translations in the less accurate category. Among the 60 noun phrases assessed, 56.67% are rated accurate, 25.00% less accurate and 18.33% inaccurate. These findings indicate that while the automatic translation feature on X is capable of preserving basic meaning in many instances, it consistently struggles with domain-specific terminology, culturally embedded expressions, and grammatical naturalness in the target language.

Keywords: noun phrases, translation techniques, translation accuracy, automatic translation, @KAI121