ARABIC MORPHEMES AND MACHINE TRANSLATION

Öz This article aims to familiarize lecturers and researchers with the ambiguities of the Machine translation and to make understand the importance of the morphemes functions to identify words and later sentences’ syntax, grammar and meanings. This manner to do is also useful to learn or teach Arabic, because it gives another way to approach linguistics or simply languages. Therefore, for the Arabic language, letters and short vowels are important for lexical and syntactic understanding, for that we should not neglect any of them. For the Turkish language, machine translation encounters same difficulties and more, because of its unusual syntax for Arabs and European people, so we need to think differently to resolve teaching and translation problems. We have the lexical ambiguity, which is introducing word into the syntax, which should be able to link between syntax’s words to give the appropriate meaning. This work is trying to open a window to look through it to the language as machine can look and see it.    The Machine Translation is one application among many ones what concern languages’ engineering. All of them use languages’ process. …….      I begin by introducing an example to illustrate this big science. The problem of the Machine Translation or any processing system is to be able to identify words individually in sentence and combine between them to get the possible meanings. Therefore, there are many problems, to reach this goal. We call these problems ambiguities. I had defined in my researches, in Lucien Tesnière researches centre, a typology of ambiguities that we encounter in Machine Translation, with goals to improve translation’s quality. One of these ambiguities is the segmental ambiguity; it means how we should segment a word that we want to identify or we cannot identify, as a unit. We take this lexis to illustrate this ambiguity. This word:  “أقال”  How it should be identified? The system (the computer) will identify it as one word that means “Dismiss”, but depending of its syntax, this identification can be an error. So lexical identifications can be limited by their syntaxes and meanings. If we say “أقال المدير الموظف”, if this sentence is limited to theses vocabulary, our first identification (dismiss =أقال) is right; and in this case, it means, “The director dismissed the employee.” Moreover, this is a right translation. But if the sentence is “أقال الطالب ال

___

Baudouin Joseph-Gabriel, (2011), Morphemes and grammatical functions in Arabic, SOAS LFG meeting, 05 March 2011, London, UK. Baudouin Joseph-Gabriel, (2010), Les ambiguïtés de la langue arabe pour un traitement automatique, Éditions Universitaires Européennes, Sarrebruck, Allemagne. Baudouin Joseph-Gabriel, (2008), Désamiguïsation morphologique de l'arabe (Arabic’s morphological disambiguation), Deuxième colloque international en traductologie et TAL, 7-8 juin 2008, Oran, Algérie. Baudouin Joseph-Gabriel, (2007), La lexie, sa graphie, sa morphologie, ses affixations, sa syntaxe, sa grammaire et sa sémantique (Lexis, its written form, its morphology, its affixations, its syntax, its grammar and its semantics), Colloque international en traductologie et TAL, 9-11.04.2007, Oran, Algérie. Al-Labdī Mohammad, le dictionnaire de la terminologie grammaticale (Mucğam al-muṣṭalaḥāt an-naḥawiyya wa aṣ-ṣarfiyya), Beyrouth, Mu'asasat ar-risala, 1986 Chenfour Noureddine, Harti Mostafa, Tahir Youssef, Modélisation à objets d’une base de données morphologique pour la langue arabe, JEP-TALN 2004, Traitement Automatique de l’Arabe, Fès, 19-22 avril 2004. Aloulou Chafik, Analyse syntaxique de l’Arabe : Le système MASPAR, Récital 2003, Batz-sur-Mer, 11-12 juin 2003. http://www.sciences.univ-nantes.fr/irin/taln2003/articels/aloulou.pdf.