Counting Unit and Text Length in Measuring L2 Spoken Lexical Diversity: A Pilot Study

Authors:

Abstract
Lexical diversity is widely used as an index of second language proficiency, yet most of the evidence comes from writing, and we still know little about whether the unit chosen to count words, the simple type, the lemma, or the flemma, changes how well lexical diversity reflects spoken proficiency. This pilot study examined spoken presentations from 55 second language learners of English at a United Kingdom university, whose speaking had been rated against an International English Language Testing System (IELTS) based rubric at bands 6.5, 7.0, and 7.5. Six lexical diversity measures were computed under each of the three counting units, across fixed text lengths from 200 to 450 words and across full-length transcripts. Only the type-token ratio distinguished between the proficiency bands, and only at full length, with the difference falling between the lowest and highest bands rather than between adjacent ones. The choice of counting unit made no appreciable difference. Because the effect was only observed when text length was free to vary, our cautious interpretation is that length (rather than diversity) of text is the dominant influence. This points to the need for length-controlled designs and larger samples in spoken lexical diversity research. The contribution is therefore methodological: the study sets out the conditions a design must meet before lexical diversity can be read as an index of spoken proficiency.

This publication has 52 references indexed in Scilit: