International Journal of Computational Linguistics Research Wafaa Daffa, Raad Alshahry, Imtiaz Hussain Khan 6 3 2015 http://www.dline.info/jcl/fulltext/v6n3/v6n3_3.pdf Syntactic ambiguity is a common problem in Arabic language. We are exploring the possibility of using corpus based word collocation data to predict different interpretations of a potentially ambiguous sentence in Arabic. As a case study, we address the problem of disambiguating coordination structures in Arabic to determine how the external modifier (adjective) applies to the coordinated words (nouns) like القطط والكلاب السوداء (black cats and dogs). In this paper, we report on an empirical study in which participants were presented with a sequence of trials, each of which consists a potentially ambiguous sentence followed by a comprehension question that relates to the preceding sentence. The study reveals that lexical co-occurrence information, derived using Kilgarriff’s Sketch Engine operated on a ca. 1.7 millions words Arabic Web Corpus, can be used to predict the most likely interpretation of a potentially ambiguous sentence.