Computerizing Arabic Morphology


This topic is not novel at all; several similar studies [eg, 1, 3, 4], have been conducted over the past 20 years, but in isolation from one another and without benefiting from each other's achievements. Many theoretical problems in morphology have to be resolved for spell-checking to become an effective feature of Arabic-enabled word processors, yet the very theoretical foundation upon which the morphological parsing of these word processors is based remains enigmatic. Arabic computational linguistics suffers from two main predicaments; one relates to the traditional grammatical description of the language and the other to the lack of cumulative effort on the part of computational linguists themselves. There is no doubt that Arab grammarians' description of the language is extremely sophisticated, yet it has been characterized by a high degree of ratiocination that complicates an otherwise useful description. Therefore, a systematic effort must be made -to render the grammatical description more congenial to computers. Towards this end, several individual attempts have been made [eg, 10, 15]. Arab grammarians need to adopt tools that rely more on empirical techniques and less on introspection. They need to make use of recently developed methods in corpus, functional, and structural linguistics.


SaneM Yagi