Frequency Studies of English Consonants

Abstract

In this study an explanation was sought for the disagreement among the various frequency counts which have been made of English consonants. The data for a set of ten different frequency counts were converted to IPA symbols and compared by means of the coefficient of linear correlation. It was found that the relative frequency of consonants in English is not seriously affected by the style of literary content or by the dialect of the sample and that a relatively small sample yields typical values. Differences in the general type of corpus (dictionary or running texts) and in transcription, however, cause significant discrepancies among the various studies. It is concluded that higher order frequency data are probably more relevant to mechanical speech recognition than the first order data considered in this paper.