從［紅樓夢］論齊普夫律中文之適用性

何光國

《圖書館學刊》 9期 (1994/12) Pp. 115-139

摘要

齊普夫律(Zipf Law)的主要特點有二：一爲文獻中的單字出現頻率與其遞減排名次序之積爲一常數。而且該常數等於全文字數的1/10。另一爲由排名與單宇出現頻率之對數值所獲得的曲線是一條自左向右，斜率等於－1的直線。歷年來，經過各種不同的研究和試驗，證實英語確實具有這二個特點。齊普夫還根據不完全的資料統計所獲得結論，認爲該律也適用於中文。本文特以［紅樓夢］第四十回爲樣本，對齊普夫律的特點作一比較性的探討，以鑑定它對中文的適用性。研究發現，該律基本上並不適用於中文。

並列摘要

Zipf Law has two foundamental characteristics (1) the product of the frequency of a word's appearance in a text and its relative ranking is a constant. It equals to 0.1 of the total number of words of the text (2) the slope of its curve equals to -1. Many studies have proven that Zipf Law does fit English pattern. Using an imcomplete statistical data, Zipf concluded that the law is also applicable to Chinese. This study uses Chapter Forty of the Dream of the Red Chamber as a sample to test whether the law is truly applicable to Chinese language. Based on the findings of this study, it is found that Zipf Law is not applicable to Chinese language.