Machine Translation of Arabic Language: Challenges and Keys

The morphologically rich Arabic language continues to present challenges to Arabic Word Sense Disambiguation researchers. The challenges are quite clear when one compares the gradually emerging Arabic to English Machine Translation with the rapidly developing Machine Translations available today. The lagging of the former may be attributed to two reasons. The first is the impenetrability of some Arabic words, which puzzles their decomposition into morphemes. The second is the incompatibility of existing machine-translation techniques with the Arabic language. The degree of difficulty can easily be observed by checking the results of any artless round-trip machine translation of arbitrary Arabic texts. Even the most advanced technologies that are available today are challenged. Regardless of whether the round trip is Arabicto- English-to-Arabic or English-to-Arabic-to-English, the inevitable outcome will be disenchanting. This leads to an even more disappointing conclusion. That is, as long as no drastic improvements are made, one must not take output of any machine translation of Arabic as orthodox. Yet another disappointing fact is that most of the intelligent machines are built to produce woodenly literal or paraphrased translations. This paper proposes a framework for a biphasic Arabic translation process. Besides, the paper recommends the all-inclusive Holey Quran's English translation as the most reliable benchmark for Arabic to English Translations.