Insofar as machine translation is based on computerized natural language processing techniques, it still subscribes to the popular notion that the best translations are not simple word-for-word translations. Consequently, approaches to translation both by humans as well as machines face the same difficulties. The need for analyzing structural similarities between natural languages (e.g., English and Arabic), going beyond the surface structure to analyze the core meaning and translate concepts into other languages , among other things, still holds This paper maps out the pros and cons of machine translation in dealing with problems of contextuality, culture-bound expressions, lexical and structural ambiguity, and idiomatic expressions. The paper concludes that while considering machine translation a step in the right direction, it is premature to announce the birth of a full-fledged and independent approach to translation which can replace human translators . Even by capturing word expressions and building a database of translation phrases, computers cannot perform so well as human translators in most types of translation, despite the computer's ability to save time, cost and effort.
Saleh M. AI-Salman