An explanation of causal-noncausal verb alternations in terms of frequency of use

To account for the cross-linguistic trends in the coding of causal-noncausal verb pairs, such as break (tr.)/break (intr.), Haspelmath et al. (2014) proposed, and provided corpus-based support for, an explanation in terms of usage frequency (Prediction 1, hereafter P1). However, this explanation cannot be applied to English causal-noncausal verb pairs. Since English mostly uses the same verb form for the causal and noncausal verb use, the explanation does not follow the form–frequency correspondence principle (hereafter FFCP) that Prediction 1 (hereafter P1) is based upon. In this article I argue, by using causative-affixed verbs, that for a linguistic pair there is a strong correlation in terms of frequency between form and earlier occurrence. Therefore, in place of the FFCP and P1, I propose the earlier occurrence – frequency correspondence principle (hereafter EOFCP) and the Revised Prediction 1 (hereafter RP1) based on it. This study tests the validity of the EOFCP and RP1 by applying them to 20 English verb pairs of sound emission. The investigation of frequency was made not only in present-day English but also at the time when the intransitive or transitive use came to be paired with its counterpart. The analysis shows that by applying the RP1 to the frequency data at the two time periods, we can obtain considerably high scores of matching rate, with a significant Spearman rank order correlation between the two groups.