Analysis of Translations of Idiomatic Expressions by Google Translate Service: a Suggestion for Improvement

Automated translation tools are widely used in many different languages. However, the capability of these tools in translation of idiomatic expression is questionable. This is because the meaning of idioms is non-compositional and expressed in an imaginative way, and if they are rendered literally into the receptor language, as the machine translation does in most cases, the equivalent is usually erroneous and ambiguous. The current study attempts to show to what extent the aforementioned automated tool can properly render different types of idiomatic expressions from English into Persian. To do so, the researchers provide a typology of idioms, namely literal, semi literal, and pure. The equivalent translations of different types of English idioms are investigated to see whether the tool has accurately translated them in the receptor language. The findings of this study show that although Google Translate Service is of high capability in rendering just literal idioms into Persian, it has a poor performance on translation of semi, pure ones. In order to improve the tool performance in rendering idioms from English to Persian, a method named blacklist was implemented which is based on the automatic detection of literal translation errors. The word for word translation of non-compositional elements comprises a list whose function is to signal the presence of these figurative languages in a text, and their meaning needs a transfer method rather than word for word one before translating them via the automated translation tool.

کلیدواژه ها:

automated translation tool ، literal translation ، idiom ، blacklist method