The Arabic traditional grammar as well as Chomsky's mainstream theory may not be able to provide a good analysis of some fixed Arabic phrases. The challenge of such data directly stems from the fact that the general syntactic rules assumed by the two opposing theories cannot explain the syntactic and the semantic aspects of the fixed Arabic data. I argue that the Construction Grammar provides an adequate account that does not rely on syntactic structure alone, as assumed by the mainstream theory or the Arabic traditional grammar, but rather it links phonological, syntactic, and semantic information together in one basic construction by means of some correspondence rules. The Arabic data proves that there is a strong need for a linguistic theory that takes into consideration all data of different range of productivity.