Abstract
<jats:p>This chapter describes the process of morphological reanalysis that yields grammatical morphemes by means of bleaching, metonymy, and metaphor. While this process does not itself create new forms, the resulting morphemes are often subsequently spread by analogy. Since both lexemes and grammatical morphemes can be subject to morphological reanalysis, it is theoretically possible for this process to take place recursively and to be unrestricted in direction. However, the maximum length of a morphological reanalysis chain seems to be just two steps, for example from body part (like ‘backside’) to location/trajectory (like ‘backward’) and from location/trajectory to temporal marker (like ‘ago’). Morphological reanalysis is furthermore strongly directional: examples of bi‐directional pathways for semantic shift are rare. In order to give some indication of the range of morphological reanalysis, I suggest viewing the phenomenon from two dimensions: the semantic shift as typologically common vs. uncommon, and the status within the given language as solitary (an isolated change) vs. convergent (resulting in a grammatical system). These dimensions are not envisioned as discrete, rather they are devices that highlight the variety of morphological reanalysis. I present three types: the typologically common solitary type is illustrated by isolated changes that are commonly found across languages, the typologically common convergent type is illustrated by the evolution of classifier systems for both nouns and verbs, and the uncommon solitary type is illustrated by examples of exaptation in which morphemes left behind by language change shift their meanings across grammatical categories.</jats:p>