<
This research belongs to the field of Natural Language Processing (NLP). It aims to build a standard model for the dictionary used in NLP systems. It concentrates on the field of Machine Translation (MT) , as an application for the dictionary to be designed . The research involves a study of the general requirements of the dictionary system and how it is strongly related to the other parts of the language systems ( Morphology, Syntax, and Semantics). Since the designed system focuses on MT purposes , some of the English and Arabic words phenomena are described and analysed. Also the way in which they are related. The designed system involves the following main stages : 1- Morphological analysis stage for English words , in order to reach the stem stored in the dictionary. This process based on affixes seperation. 2- Main representation stage for English entries ( with relevant syntactical and semantical informations). In the other side, the root is used for the corresponding Arabic words . Together with the word class and frame number which determines the generation behaviour. In addition to the separated representation for verbs structures, morphological patterns, semantic markers, and selectional restrictions. 3-Morphological generation stage for Arabic words. These words are determined by the English entries coming from the analysis stage. The generation process in Arabic is based on the interlocking according to the morphological patterns. The research reaches certain results , the most important of which are: morphological analysis do not always lead to increasing the efficiency of the system , the importance of using the English stem against the Arabic root together with word class code , the great advantage of using verb frames , morphological patterns for Arabic , and the semantic markers with the selectional restrictions. The system is implemented using (TPROLOO ver.2).