Errors made by google translate and its rectification by human translators / Farah Hana Amanah

The use of bilingual versions of public communication such as notices letters, pamphlets or legal documents are the norm and often times are expected as Malaysia is a multiracial country. Having a bilingual version is said to cost a certain amount of money depending on the length of a text. This study aimed at investigating and evaluating the output of Google Translate which requires human editing and thus, to find out to what extent does Google Translate help in assisting human translation in particular with Bahasa Melayu (BM) into English translation. In this study, four texts namely notices and pamphlets were used. Three female translators who edited the output of Google Translate were tasked to check the translations. A hybrid taxonomy on error analysis proposed by Keshavarz (1999) and Vilar et al. (2006) in terms of lexical, grammatical (syntactic) and semantic errors is used to identify the errors in Google Translate output. The findings indicated that most of the errors were in the grammatical category such as missing prepositions, missing possessive markers, omission of pronouns and/or incorrect use of lexical words. It is also revealed that using a different choice of word will affect the meaning as well. Hence, the three human translators were needed to rectify the errors, although the three of them had different approaches in correcting the errors. Despite that, it can be said that Google Translate is able to produce a good translation output when the text uses simple sentence structures and thus, Google Translate can aid or assist students or language practitioners in getting a general idea of the text but human evaluation is still required especially in keeping true to the meaning of the original text