Abstract
This article investigates the palatalization of Latin /ks/, /kt/, /gn/, and Romance /k'l/ and /g'l/ in Western Romance and its relationship to the appearance of a palatal semivowel. The relative merits of two major processes of palatalization, the yod-based approach and the consonant-based approach, are discussed, and a combination of these two views is proposed to account for the palatalization of the five consonant sequences. This combined approach then permits two related yet distinct origins for the palatal semivowel: either yod formation during palatalization or glide epenthesis at a point subsequent to the completion of palatalization. The article concludes with a discussion of criteria that one may use to determine which of the two glide-producing processes was responsible for those instances in which a palatal glide was maintained in Western Romance.