Grammatical Tagging of Arabic Word for Corpus Analysis

Abstract

Word tagging is one of the important processes in computational linguistics studies. It provides detailed information about the grammatical part of speech (PoS) to the researcher. However, tagging process is not an easy process particularly when relating to Arabic text. One of the main reasons that led to this situation is the fact that many Arabic word characters (space between characters) could contain more than one word. Thus, this paper aims to revisit our experience in tagging Arabic words and to highlight critical issues related to the process. It also aims to validate the possibility of using computer analyzer software, namely Wordsmith 5.0 for Arabic texts. Using several Arabic texts from newspapers as examples, this paper reveals how the tagging process was executed using a self-constructed tag set. It also discusses the issues and linguistic factors considered in constructing the tag sets. This includes the differences that exist between Arabic word characteristics in writing and their articulation. This paper finally stresses the more complicated issues in tagging Arabic words in relation to the fact that Arabic has the affixation and diacritics systems. The application of short vowels and orthographic diacritics in Arabic language also contributes to the phenomenon. Despite this fact, this paper also proves that a researcher can make the tagging process simpler by focusing on certain grammatical features, aiming the tagging for word disambiguation and applying an appropriate analyzing software.