Introduction

V_Size is part of our ongoing research into ways of assessing the productive vocabulary of L2 speakers. Most research in this area assesses the texts that L2 speakers produce in terms of their lexical richness – usually measured by looking at the type/token ratio of the text, or some variation of this ratio such as Malvern and Richard's vocd measure (e.g. Malvern et al., 2004). Alternatively, they use profiles such as the lexical frequency profiles developed by Laufer and Nation (1995). V_Size belongs to the second category – it works with profiles, rather than with type/token ratios. Laufer and Nation's profiles are not particularly easy to work with, however, and it is difficult to summarise the data that they encapsulate in an economical and transparent way. V_Size is an attempt to do just that: it goes beyond the mere shape of the profile generated by a text, and asks what the profile tells us about the size of the productive vocabulary of the person who produced that text. Basically, the program allows us to look at a profile and make statements such as: 'this text looks as though it was generated by a person whose productive vocabulary is about 5500 words.' Obviously, this is a rather strong claim, and it requires some justification.

Briefly, the program makes use of a number of results that derive from Zipf's Law (Zipf, 1935). Zipf argued that language was one of very many things where a simple relationship could be found between the rank order of an event and the size of the event. For language, Zipf argued that the pattern of frequencies exhibited by words followed this law, and he claimed that there was a straightforward relationship between the number of times a word occurred in a corpus and its rank order in a frequency list generated from the corpus.