Weaknesses of Google Translation; A Study Based on Sinhalese and English Languages

Abstract

The process of translating words or text from one language into another is known as translation. It communicates the meaning of a source language text into a target language. It is used for legal, official, literary works ... etc. Nowadays there is a popular systematic service for translation called "Google Translation". Being an easier and quicker method than manual translation, Google Translation plays a major role in the translation field. It is a multilingual statistical machine translation service provided by Google to translate texts or speeches from one language to another. The aim of this study is to identifY the problems that can be seen in this translation service and the study is based on translating from Sinhalese to English and vice versa. Data were collected by analysing a set of Sinhalese and English sentences translated using the service. When translating trom Sinhalese into English, errors are seen in sentence structures, major reason for which is the structural differences between the two languages. The syntactic structure of the Sinhalese language is SOY, whereas the structure of English is SVO. It is apparent that this structural difference cannot always be properly comprehended by this program. Furthermore, it is identified that synonyms, homonyms and homophones used in the Sinhalese language cannot be properly translated using this service. The diglossic characters of the Sinhalcse language is yet another aspect which creates problems in "Google Translate". Translating a complex sentence with a main clause and a subordinate clause also appears problematic since identifYing the subordinate clause is not easy for the program. Accordingly, this study points out that in order to provide a better service, these issues have to be properly addressed.