Phonotactics in morphological similarity metrics

This study explores the contribution of phonotactic regularities to the way we measure morphological similarity between words. I focus on how morphological similarity fits with conventional morphemic analysis and present a method to exploit the latter to create two reference resources from the English part of the CELEX database. The resources are used for the comparison and evaluation of morphological similarity metrics. I also compare four measures that estimate morphological similarity: the standard edition distance of Levenshtein; the measure of De Pauw and Wagacha; Proxinette; and PHACTS. The study provides an idea of the contribution of phonotactic regularities to morphological similarity.