Classification of Formosan Languages: Lexical Evidence

Classification of Formosan Languages: Lexical Evidence

The purpose of this paper is to present some new lexical evidence for the classification of Formosan languages, as based on all Formosan cognates that have been identified. Fourteen extant and five extinct Formosan languages are compared in this study. Based on the number of cognates shared by each pair of languages, the distance between each pair of languages is calculated. A tree diagram for Formosan languages is then constructed by adopting the procedure developed for quantitative studies by Cavalli-Sforza and Edwards (1967), and Fitch and Margoliash (1967). Since the number of lexical items (about 400) compared for each of the five extinct languages is much smaller than the number (about 1,000) compared for the other fourteen extant languages, some adjustment has been made with regard to the number of cognates shared by each pair of languages by adopting Jaccard's coefficient principle. Furthermore, cognates shared exclusively by a pair of languages are regarded as providing much stronger evidence for a close genetic relationship. The results of this study roughly agree with classifications for these languages based on other types of evidence, including phonological and syntactic evidence. Slightly different results are found when each language is represented by only one dialect, as opposed to being represented by more than one dialect. All these results are more suggestive than conclusive.