Translation Error Analysis in Instructions of Virtual Assistants on Mobile Devices

Abstract

Abstract

This study aimed to identify translation errors of translated instructions of virtual assistants (IVA) on mobile devices, compare the translation errors of IVA on mobile devices and report the direction of errors occurring in each version of IVA on mobile devices. The research was divided into two phases. The first phase was to identify the popularity of IVA on mobile devices in the market and the common functions used by mobile users to give instruction, and the second phase was to analyze translation errors. The instruments used were a questionnaire and text analysis. The questionnaire was used in a survey and the reliability of the questionnaire (a - Coefficient) was 0.86. The results from the text analysis were verified by three English teachers acting as interraters. The findings were that semantic errors were found in all IVAs on mobile devices and had the highest number of errors, followed by transliteration and grammatical errors. Siri had the least number of errors compared to Google Assistant and Bixby. The results from the direction of the error in each IVA revealed that Siri had no error in the later version while the same number of errors occurred in Bixby and no comparison had been conducted in Google Assistant due to online availability restriction.