Stem alternation is present in the verbal inflection of all documented Kiranti languages, where it ranges from the straightforward phonologically conditioned (e.g. Athpariya and Chintang) to the purely morphological and baroque (e.g. Khaling and Dumi). This paper surveys stem alternation patterns across the whole family. Its main finding is that, unlike the morphological stem alternations of West Kiranti, the phonologically-conditioned stem alternations of East Kiranti are characterized by a very striking distributional similarity (often identity) across languages, even in the presence of quite drastic affixal changes. This and other findings suggest that these stem alternation patterns should be regarded as a (morphomic) grammatical phenomenon of its own right, despite being derivable from the forms of suffixes. Furthermore, comparison with West Kiranti suggests that this coextensiveness with a coherent phonological environment actually enhances some typically morphomic traits such as diachronic resilience and productivity.

Maiden (e.g. 2009a) shows that treating the paradigmatic distribution of root allomorphy in Romance verbs as morphomic, in the sense of Aronoff (1994), provides a coherent explanation for the diachronic behaviour of such allomorphy. The major templates for distribution (‘metamorphomes’, Round 2015) shared by most Romance varieties are also found in early French, but are not well represented in the modern language, which has developed new metamorphomes. By charting the diachronic development of metamorphomes in French, this study investigates the processes responsible for change to such templates. Overall, the French data point to segmental sound change as the central factor in change to metamorphomes: segmental sound change modifies the observable paradigmatic distribution of allomorphs, reducing the number of lexemes in which an existing metamorphomic template could be deduced, and increasing the number of lexemes across which a novel metamorphomic generalisation can be made. The loss of existing metamorphomes, and the rise of new ones, can be considered a single process, of metamorphomic templates changing shape as further paradigm cells attach to or defect from them. This process must be distinguished from changes in metamorphome shape due to the creation or elimination of paradigm categories for independent morphosyntactic reasons.