Emerging Paradigms: Diachronic Shift in a Distributed Morphology Framework Hezekiah Akiva Bacovcin 1 Introduction Kurylowicz (1945) correctly notes that the existence of morphemes and their phonological content can only be learned in a paradigmatic context. By this, I mean that a word like Latin lup-us “wolf-nom.sg.”, can only be broken into its component parts by a native learner because of the existence of forms like lup-¯ ı “wolf-gen.sg.” and am¯ ıcus “friend-nom.sg.”. It is only because a morpheme occurs in multiple environments that it is possible to separate it from the words in which it occurs. Many theories of morphology (Kurylowicz, 1945; Aronoff, 1994; Maiden, 2005) take this fact about the nature of learning and convert it into a theory about the structure of the grammar, claiming that the distribution of forms (usually as distinct words) is simply the memorisation of the observed paradigmatic patterns. Maiden in particular argues that these paradigmatic patterns have implications for historical change. In this paper, I will argue that analysis of morphological change can be better described by studying changes in the distribution and form of individual morphemes crucially relying on syntactic/semantic and phonological conditioning. I will argue that DM (Distributed Morphology) has the proper grain to be able to both capture the broad generalisations that are made by paradigmatic theories, as well as capturing the particularism inherent in many changes. 1 Diachronic investigation is interested in understanding the perturbations of otherwise stable systems. Since changes arise from the disturbance of the preceding synchronic system, any diachronic theory is necessarily dependent on a synchronic theory to define the structure 1 Information about the version of Distributed Morphology used here comes from (Embick, 2010, 2013b,a) and references, as well as personal communication with Embick. 1