A corpus linguistic analysis of English and Arabic parallel business discourse domains

This thesis investigates the linguistic properties of the English and Arabic business discourse domains, with respect to their co-occurring linguistic features, the meanings which these features index and construct, and finally the relationships among individual genres with respect to these co-occurring features, and their associated meanings. For this purpose, the thesis adopts a corpus linguistic methodology, more specifically, the multidimensional methodology (Biber, 1988, 1995). The English and Arabic business discourse domains are represented by two electronic corpora representing five parallel genres: business/economics academic articles, business editorials, stock reports, brochures, and Internet based corporate "about-us" genre. The corpora are analyzed with respect to a range of linguistic features, which represent six areas of the English and Arabic grammars: representation of agency/ causation, reporting, interaction among the participants, modality, metadiscourse, and spatio-temporal indexing. Factor analysis, a multivariate statistical tool, is used to isolate the major co-occurrence patterns among the linguistic features in each corpus. Four major co-occurrence patterns are identified, which are interpreted as dimensions of variation (i.e., functional parameters of variation). Two perspectives are adopted to compare the English and Arabic discourse domains: dimensions and genres. The dimensions are compared cross-linguistically with respect to their underlying functions and their linguistic realizations, while the genres are compared with respect to their multidimensional profiles. The analysis shows that the functions underlying dimensions and the linguistic features realizing them are strikingly similar cross-linguistically. The analysis shows also that parallel genres are remarkably similar cross-linguistically. In fact, one of the major findings of this thesis is that English and Arabic parallel genres are more similar functionally/structurally than are different genres of the same language. The thesis concludes by outlining the implications of these findings for both computational and applied linguistics.