Grammatical Gender Assignment on Arabic Nouns: Saudi Dialects

Abstract

Gender assignment has been studied in many languages and the assignment system is considered important particularly when it is related to determining the structure of the lexicon. Semitic languages show two genders, namely masculine and feminine. Nevertheless, it has been claimed that some nouns in Semitic languages can have common gender. Some linguists suggest number of criteria based on the classifications for number and gender in Semitic languages: masculine/feminine, animate/inanimate, human/non-human, individual/collective, concrete/abstract, singular/plural, and major/minor. Arabic, like other Semitic languages, determines gender of its nouns either through meaning or through form. So, grammatical gender is a way of analyzing nouns classes with relation to semantic features, like masculine, feminine, animacy, etc. It is a property of individual nouns regardless of their existing referents (if there are). There have been efforts by many researchers, trying to explain the processing of gender in comprehension. This paper discusses how gender is assigned to nouns in Saudi dialects where a supervised machine learning algorithm is used to test the predictiveness of gender in Arabic. The data show interestingly that the presence of some semantic and morphological features helps to predict the gender of nouns. It is not fully predictable but it is the start point for further investigations.