A statistical approach for the induction of a grammar of Arabic

Over the last decade, a lot of research has focused on Arabic Natural Language Processing (ANLP). Various approaches and techniques have been used to develop ANLP tools. Some of these are rule-based while others are statistical or machine-learning-based. However, the development of some ANLP tools depends on the availability of a good Arabic grammar which covers the entire language. It turns out that the Arabic grammar used by most of the developed approaches was hand-crafted and most often extracted from short sentences. This manual development process is painstaking and time consuming while the developed grammar cannot describe the entire Arabic language.