How Many Ways To Go In Arabic_anonym

How many ways to GO in Arabic? A corpus-based approach to determining polysemy and synonymy of the verbs ḏahaba, maḍā, rāḥa,and inṭalaqa Regardless of which particular linguistic theory one adopts concerning the lexicon, polysemy stands out as a pervasive feature of words (and constructions), and yet it is a feature which presents real challenges to linguistic theory and description. In particular, the methods employed to identify and justify polysemy remain problematic. The criticisms levelled at purely intuition-based approaches to describing polysemy in Sandra and Rice (1995) continue to be as relevant now as they were when first made 14 years ago (though see Vanhove (2008) for various new methods). In this paper, we explore the value of adapting Gries’ Behavioural Profile approach (Gries 2006; Gries & Otani to appear) as an empirical check on intuition-based claims about polysemy and synonomy of the Arabic GO verbs ḏahaba, maḍā, rāḥa,and inṭalaqa. We began with 300 randomly selected concordance lines for each of the verbs ḏahaba, maḍā, rāḥa,and inṭalaqa, taken from arabiCorpus (http://arabicorpus.byu.edu). Each verb use was coded as one of the polysemous sub-senses, as implied by contemporary dictionary practice and intuition. So, for example, GO ‘to go (somewhere)’ uses were distinguished from TAKE ‘to take along’ uses. Each instance of these verbs was then additionally coded for a wide variety of features or “ID-tags”, including morphological, syntactic, and semantic properties, with around 90 such features utilized. The dataset of ID-tags for each verb was then subjected to a clustering algorithm, carried out in R, which results in clusters of the sub-senses. The pvcluster package in R (Suzuki & Shimodaira, 2006) calculates probability values for each cluster which further refines the claims that can be made on the basis of the resulting dendrograms. In our adaptation of the Behavioural Profile method, we explored sub-senses of verbs, for each verb, as well as combining all sub-senses of all four verbs, allowing us to explore clustering properties relating to polysemy of a single verb as well as synonymy between different verbs. Results showed that there are clusters of sub-senses which can be justified in a principled way, as opposed to the simple enumeration of sub-senses found in typical dictionary practice. In the case of the verb maḍā by itself, for example, the sub-senses ‘continue/go on (doing something)’ and ‘advance/progress’ turn out to be closely clustered by our method even though some dictionaries might not group them together. It is particularly interesting to examine the clustering patterns when data from all four verbs are combined in order to examine the degree of relatedness between sub-senses of the four different verbs. Given the computational basis of the method, it is quite easy to alter the number and nature of ID-tags which are used in the calculations of the clusters. We report on the effects of these changes, e.g., relying on syntactic ID-tags only versus semantic ID-tags only. References Gries, Stefan Th. 2006. Corpus-based methods and cognitive semantics: the many meanings of to run. In: Gries, Stefan Th. and Anatol Stefanowitsch (eds.). Corpora in cognitive linguistics: corpus-based approaches to syntax and lexis. Berlin, New York: Mouton de Gruyter. Gries, Stefan Th. and Naoki Otani. to appear. Behavioral profiles: a corpus-based perspective on synonymy and antonymy. ICAME Journal. Sandra, Dominiek and Sally Rice. 1995. Network Analyses of Prepositional Meaning: Mirroring Whose Mind – the Linguist’s or the Language User’s? Cognitive Linguistics, 6, 89-130. Suzuki, Ryota and Hidetoshi Shimodaira. 2006. Pvclust: an R package for assessing the uncertainty in hierarchical clustering. Bioinformatics, 22(12), 1540-1542. Vanhove, Martine (ed.). 2008. From polysemy to semantic change: Towards a typology of lexical semantic associations. Studies in Language Companion Series 106. Amsterdam/Philadelphia: John Benjamins Publishing Company.