There are no special clitics
The hypothesis of Clitic Idiosyncrasy holds that special clitics are neither words nor affixes, but constitute a separate type of object whose behaviour is partly governed by dedicated grammatical mechanisms. In an influential implementation of this idea, Judith L. Klavans and Stephen R. Anderson claim that special clitics are phrasal affixes, introduced by a set of postlexical morphological rules that is separate from stem- and word-level morphology. This paper criticizes the hypothesis of Clitic Idiosyncrasy and its implementation through phrasal affixation. First, we show that the identification of a distinct class of special clitics depends on a concept of ‘special syntax’ that is not well-defined: in many instances, there are syntactically autonomous units that exhibit the same behaviour as putative special clitics. Secondly, we note that the theory of phrasal affixation incorrectly predicts that special clitics will be invisible to lexical morphophonology. Thirdly, we demonstrate that, in certain crucial cases, phrasal affixation cannot place special clitics in the right positions: in Bulgarian, for example, the definiteness marker is suffixed to the head of the first syntactic phrase immediately contained within the NP. We show that this behaviour is straightforwardly handled by a theory of syntactic feature-passing within subtrees that allows phrasal features to be transferred now to heads, now to edges. This theory is independently motivated by phenomena such as the English ’s genitive and Old Georgian <i>Suffixaufnahme</i>.