Analysis of language variation and word segmentation for a corpus of Vietnamese blogs

Abstract

This dissertation examines issues of units of meaning, word segmentation and language variation in a corpus of Vietnamese language blogs collected from publicly accessible internet sources originating in Viet Nam, the US, and Australia. Research using corpus linguistics techniques for study of the Vietnamese language have begun to proliferate in western sources in the past decade, however, studies using language-in-use data remain rare. Analysis of the corpus as a whole and by comments and blogs and Viet Nam, US, and Australia subcorpora used the Vietnamese syllable, or tiếng, as the basic unit of meaning, with subsequent iterations of one- through 5-tiếng. While results support previous research asserting the Vietnamese syllable as the basic distributional element in Vietnamese discourse, claims about Vietnamese as a monosyllabic language are not supported by results. Tiếng collocate and colligate meaningfully and regularly throughout the corpora in clusters larger than one syllable, indicating that syllable combinations, the union of tiếng (Nguyen, 1984), are also primary distributional patterns for the Vietnamese language. Varieties of Vietnamese by country show similarity in a variety of distributional patterns, including by a-curve (frequency of frequencies), structural, content, and units of meaning analyses. Variations of Vietnamese by country are primarily limited to collocational and colligational content and topical patterns.