The development of new morphology from originally independent forms has long been acknowledged as a common diachronic process. Typically this process is discussed within a grammaticalisation-based framework, and as such is primarily concerned with the question of how an originally lexical item becomes a grammatical one and the syntactic consequences that result. However, by contrast less attention has been paid to the effects such forms have once they enter the morphology. In particular, the relationship of such historical processes to word-and-paradigm type conceptions of morphological structure has yet to be fully explored.

In this thesis, I provide an account for the process of univerbation, founding it upon a learner-oriented approach centred around the problem of segmentation of the speech stream. On this basis, I present a typology of the effects that the creation of new wordforms by univerbation may have in terms of their impact upon inflectional paradigms. I discuss the various kinds of factors that may influence when univerbation is likely to occur. I then discuss in depth the range of different effects the entry of forms produced by univerbation can have upon morphological structure.

I exemplify this model by focussing on a pair of case-study languages. Firstly, I discuss the languages of the Bengali-Assamese group within Indo-Aryan as examples both of how new forms arising by univerbation interact with an existing inflectional system, with both sets of forms mutually influencing each other. Secondly, I discuss Scottish Gaelic as an example of how the non-teleological production of new realisations then feeds into the creation of a new inflectional paradigm from the arrangement of a subset of these single wordforms into paradigmatic opposition with each other.