Appendix: Simple Measures of Variation in Lexical Fields

Extract

This appendix suggests a number of measures which may be readily calculated in respect of data on lexical usage to provide an indication of the degree of variation in such usage. It introduces and briefly discusses five straightforward measures of variation or consensus which apply to the case where any informant has been constrained to give only one response per stimulus item. None of these measures is, strictly speaking, a statistical measure, although p is almost so in terms of its logic and the D-measures have a form familiar to that of a wide class of elementary statistics. Each measure in fact generates a numerical index which relates to an arithmetical property of the data in question: the measures are not in general estimates of population parameters. The use is heuristic, exploratory and comparative: by systematically reducing the information which comprises the data, the measures should help the investigator to become more aware of the salient characteristics of the data to hand.