The concept Muslim: discourses in Bosnian, Croatian, German and Serbian webcorpora

Abstract

The aim of this contribution is to examine the meaning of the word musliman or German Muslim and to answer the question whether the meaning of this word in a dictionary can be mapped by means of classical distributional analysis, and whether further meanings can be captured by this method. In addition, a simple method for a distributional semantic analysis is presented. The basis of the investigation is the assumption common in corpus–based semantics and lexicography that the meaning of words can be interpreted from the context and thus polysemy and synonymy can be recognized. The article examines four web corpora for Bosnian, Croatian, Serbian and German. First, the context is analysed, assuming that the syntactic structures in the context are themselves inherently meaningful. Specifically, adjective attributes are used to analyse properties of the keyword, coordination partners as indicators for relevant categories, verbs with the keyword as subject and accusative object as indicators for typical actions the keyword performs or is target of. In a second step, the lexemes that occur in these four contexts are clustered and the semantic dimensions of the keyword are determined. The results suggest that the distributional method can in principle map dictionary definitions well. However, this is mainly true for contexts which involve attributes and coordination but not for action–related syntactic configurations. Overall, the most salient semantic dimensions of the lexeme are those pertaining to members of a religious community, members of a regional ethnic group, and experienced or induced aggression. Here, the four corpora differ considerably: while German usage almost exclusively focuses on religion–related aspects, the other corpora also contain ethnic/regional components, as well as the discourse of aggression. This aggression is mainly verbal in German, but physical in the other corpora. Neutral meaning components for the word musliman are salient only in the Bosnian corpus.