語料庫語言學量化研究的問題及其解決方案

Some Current Quantitative Problems in Corpus Linguistics and a Sketch of Some Solutions

目前迅速發展的語料庫語言學面臨眾多研究方法的挑戰。與語料庫語言學本身相關的方法論問題──資料的散布離差、詞種頻率／亂度、與資料隱含的方向性問題等，直接影響語料相關性指標的計算；忽視語料庫的資料取樣結構也與之後的統計分析結果直接相關。歷史語言學與學習者語料庫研究等領域應用語料庫語言學時，也有方法論的問題。本文詳細討論以上所提到的問題，並具體提出實例演示相對應的解決方法。

This paper surveys a variety of methodological problems in current quantitative corpus linguistics. Some problems discussed are from corpus linguistics in general, such as the impact that dispersion, type frequencies/ entropies, and directionality (should) have on the computation of association measures as well as the impact that neglecting the sampling structure of a corpus can have on a statistical analysis. Others involve more specialized areas in which corpus-linguistic work is currently booming, such as historical linguistics and learner corpus research. For each of the problems, first ideas/pointers as to how these problems can be resolved are provided and exemplified in some detail.