This paper explores the notion of analyzing cross-linguistically uncommon morphosyntactic structures in terms of their historical development. What may seem extraordinary in the synchronic ‘snapshot’ of a language can often be clearly accounted for through diachronic considerations. To illustrate this, the current study examines the typologically uncommon phenomenon of multiple exponence, the realization of the same grammatical information in multiple places within an inflected word, in the Kiranti (Tibeto-Burman) languages. Typologically speaking, we do see a strong tendency cross-linguistically towards encoding grammatical information once within an inflected word, and against multiple exponence. Yet the phenomenon of multiple exponence is attested in a number of languages. Notable examples include the gender agreement system of Batsbi (Northeast Caucasian) and person agreement in Hualapai (Yuman), subject agreement in Ibibio (Niger-Congo), as well as in the Skou language of New Guinea. Several of the Kiranti languages of Nepal also exhibit multiple exponence in their agreement and negation morphology. Drawing on comparative Kiranti evidence and well-attested processes of historical language change, I advance the claim that multiple exponence in synthetic verbs in the modern Kiranti languages comes as a result of the interaction between language(family)-specific typology (multiple agreement in periphrastic verbs) and an uncontroversial language change process (coalescence of periphrastic forms into synthetic forms).