摘要
本文设计了传统的词频分布(Lexical Frequency Profile,LFP)、改进的P_Lex和随机取样等三种词汇复杂度测量方法的计算机程序,通过分析三种方法的信度和效度并进行比较,试图找出测量词汇复杂度的最佳方法。研究发现,虽然三种方法都表现出一定程度的信度和效度,但是LFP明显受长度影响,改进的P_Lex方法和随机取样方法受文本长度影响小,特别是随机取样方法可以用于分析较短文本和比较不同长度的文本,因此更适用于分析二语学习者的产出性文本。
A computer program,which integrates three measures of lexical sophistication,i. e. the LFP,the Modified P_Lex and the Random Sampling Based Measure,was used to compare the three measures in terms of their reliability and validity. The result indicate that although the three measures demonstrate a certain degree of reliability and validity,the LFP is most likely to be affected by text length,while the Random Sampling Based Measure and the Modified P_Lex Measure are less likely so. The Random Sampling Based measure is the best in that it works better with the shorter texts and can be used to compare texts of different lengths. Therefore,the Random Sampling Based measure is considered as the most suitable measure for L2 learners' essays.
出处
《解放军外国语学院学报》
CSSCI
北大核心
2013年第5期44-48,127,共5页
Journal of PLA University of Foreign Languages
基金
国家社会科学基金项目"英语语言学研究生实验研究能力系统评价研究"(13BYY079)阶段性成果
