Understanding Cyclic Spell-Out

In Chomsky (1998) the interesting suggestion is made that syntactic structures may be phonetically interpreted in successive stages rather than at a single point, and that the input to occurrences of this “cyclic Spell-Out” process are phase constituents of type vP and CP. The cyclic Spell-Out/CSO hypothesis of the relation of syntax to phonology has been considerably influential since its introduction and is now widely adopted in Minimalist approaches to syntax. Notwithstanding this success, certain aspects of CSO still remain poorly understood and there has been little evidence of the right phono-syntactic kind available to throw light on the basic mechanics of such a procedure. The present paper now shows how the interaction of syntax and phonology in tone sandhi patterns found in Taiwanese provides a potentially useful window into the process of CSO and in certain cases allows one to observe very clearly how such an operation applies to a syntactic input form. The patterning observed leads to a number of new insights into CSO and suggests the conclusions that (a) the linear order of elements inside a syntactic unit interpreted by CSO may not be necessarily fixed by the CSO process in any fully final way, and (b) the regular input to CSO may only be fully propositional CP constituents and not smaller vP structures, contrary to what has frequently been assumed. The structure of this paper is as follows. Section 2 provides an overview of relevant aspects of CSO and the notion of phase as introduced in Chomsky (1998). Section 3 then presents Taiwanese data originally considered in Simpson & Wu (2002) which lays the groundwork for a reconsideration of the notion of CSO. This is done in section 4, where the Taiwanese patterns are shown both to provide positive evidence for the idea of CSO and to lead a clearer understanding of what may be involved in such an operation. The section also discusses how the patterns observed interact with related phenomena observed in English whconstructions studied in Bresnan (1971), and notes that the Taiwanese and English constructions both seem to lead to similar conclusions concerning the nature of CSO. Section 5 finally considers what consequences such conclusions may have for the Minimalist system proposed in Chomsky (1998/99).