時代中的概念：語料庫為本之方法探究台灣戰後初期文學場域及其文藝社會學應用

This study examines the semantic development of key terms in Taiwan from a corpus-based perspective. Keywords in a particular context map an implicit characteristic way of thinking in a community onto an explicit observable linguistic representation. Specifically, we investigated the lexical patterns in Taiwan Early Post-war Corpus, a small-size corpus consisting of opinion articles and prosaic writings produced in 1945–1949. Based on the lexical distribution of the unigrams and bigrams, we used hierarchical cluster analysis to identify the sub-groupings of the texts in the corpus and conducted lexical network analysis with keywords from each cluster to further examine the within-cluster homogeneity and between-cluster heterogeneity. Distinctive keywords associated to text clusters were statistically defined using the distinctiveness values from the multiple distinctive collexeme analysis, covering both unigrams and bigrams from the texts. Moreover, we also examined the networks associated with the authors and periodicals included in each cluster based on their commonalities in keyword usages. Having identified the semantic clusters which represent ideological preferences, we conducted a prosopographic study on the authors to determine predispositions of each cluster/position on an ideological map. Biographical information and organizations’ classification are stored in Taiwan Biographical Ontology (TBIO). The combination of these analyses will create a multi-layered network system, which will not only help identify ideological camps in post-war Taiwan and illuminate their respective ideologies but also characterize their members.