Abstract
The aim of this chapter is to highlight the complexity and importance of Arabic morphological information in an Arabic Machine Translation (AMT) system, i.e. a system that translates to or from Arabic. We summarize Arabic morphology and introduce the main morphological information that we have found relevant to machine translation to Arabic and categorize it into various types of features. In order to show the impact of these morphological features on machine translation quality, we have adopted an approach whereby we relate each of them to the quality of the translation. This leads us, through a statistical analysis of the test data, to a characterization of which features are more important in terms of their impact on the quality of the translation of a given AMT system (AMTS). The approach has been implemented and applied to evaluating an English-to-Arabic web-based MT system. The results of the evaluation of this system are presented, conclusions are drawn, and recommendations for improving their outputs made.