The symmetry of syntactic relations

2011-07-06In generative syntax, the theory of asymmetry in syntax has gained much attention due to the influential work of Kayne (1994), who adopts the null hypothesis that syntax is inherently asymmetric. However, such a direction does not seem fully compatible with the general assumptions in Galilean style scientific theories, all of which aspire to uncover hidden symmetry (in theoretical models) from observed asymmetries (see also Brody 2003, Stewart & Golubitsky 1992, Stewart 1998). In view of this tension, it will be argued in this dissertation that the base syntactic representation should be symmetric, while surface phrase structures are actually derived from symmetric representations by the symmetry-breaking mapping rules that are induced by the asymmetric natures of the interfaces (Prinzhorn & Vergnaud 2004 et seq.). The hidden invariance/symmetry of syntax can be revealed by studying recursive parallel patterns, which can be identified in three types of primitive syntactic coupling relations. The three types of fundamental relations are the N-V relation, the substantive-functional (S-F) relation, and the k-k' (generalized connectives) relation. Following Prinzhorn & Vergnaud (2004 et seq.), it is hypothesized that the basic unit characterizing syntactic merger (a base structure in the sense of Chomsky 1964) can be viewed as a Cartesian product that simultaneously encodes primitive syntactic relations (i.e., a base structure = {N,V}*{S,F}*{k,k'}), which can be represented in a graph-theoretical way as a three dimensional cube, and each node of the cube is a certain combination of syntactic relations that defines the syntactic role played by the lexical item inserted in that node (e.g. a noun (N,S,k) = a nominal substantive root coupled with a classifier (N,S,k')). Each instance of merge, then, is a reflection of a syntactic relation (Collins 2002). The symmetry of syntax, eventually, arises from the recursive parallelisms generated by the primitive syntactic relations. Considering the asymmetry of phrase structures (in the sense of Kayne 1994), it is argued that the asymmetry is imposed by the asymmetric nature of the interfaces. Thus, the translation from a highly symmetric syntactic representation (i.e. a Cartesian product of primitive syntactic relations) to asymmetric phrase structures is identified as a symmetry-breaking process in syntax. Generalizing Moro (2000), labeling can be taken as an asymmetric direction-marking mechanism of symmetric Merge. Following Prinzhorn & Vergnaud (2004), it is argued that different orderings among the primitive syntactic relations may be responsible for the macroparameters found in the cross-linguistic phrase structures. With the primitive syntactic relations remaining invariant, each possible type of phrase structures can therefore be viewed as a member of the symmetry group of the underlying symmetric syntax. Evidence from classifier constructions and indefinite expressions (with special reference to the Chinese languages) will be discussed in favor of the symmetric syntax