Effects of text length on lexical diversity measures: Using short texts with less than 200 tokens

Abstract

Despite the importance of lexical diversity (LD) in L2 speaking and writing performance, LD assessment measures are known to be affected by the number of words analyzed in the text. This study aims to identify LD measures that are least affected by text length and can be used for the analysis of short L2 texts (50–200 tokens). We compared the type–token ratio, Guiraud index, Maas, measure of textual lexical diversity (MTLD), D, and HD–D to assess their robustness in relation to text length variation. Spoken texts of 200 tokens from 38 L2 English learners at the lower–intermediate-level were divided into segments of 50–200 tokens and the text length impact was examined. We found that MTLD was least affected by text length across most ranges, but was somewhat affected across 50–150 and 50–200 tokens. We further observed low correlations between equal-sized texts for up to 100 tokens. These results suggest that MTLD can be used with texts of more than 100 tokens and MTLD values can be compared between texts across 100 and 200 tokens. We also showed that D and HD–D produced similar results for texts; this indicates that D and HD–D are comparable.