The LabPhon paper presented several different types of estimations of the significance of the results of Experiment 1. This page presents more details about two of those approaches. Binomial probabilities show that speakers prefer both High-frequency syllables in a quadruplet more than both of the Low-frequency syllables. Linear regressions show that the rime makes a large contribution to a multiple regression analysis, over and above the contributions of the vowel and the coda.
