Binary Branching and Linguistic Theory: Morphological Arguments

One of the fundamental issues in the design of linguistic theory concerns the overall complexity of the grammar. By Occam’s razor, given two competing grammars (or theories) of equal descriptive power, the one that is less complex should be preferred. Since the beginnings of generative linguistics, a number of principles and constraints have been devised in order to keep this complexity to the barest minimum. The Binary Branching Hypothesis (BBH) is, perhaps, the single most important principle limiting the complexity of the grammar in mainstream generative linguistics. Binary branching is generally assumed as the most restrictive hypothesis of phrase structure. Its consequences are far reaching. Since Kayne’s (1984) work on unambiguous paths, it has been commonly assumed that syntactic constituents are mostly binary; from then on, binary branching has been generalized to all lexical and functional projections: it is at the core of the the VP-shell hypothesis (Larson, 1988), it has been incorporated as a fundamental property of X-bar theory, it lies at the heart of Antisymmetry (Kayne, 1994), it forms the basis for the fundamental operation Merge in the Minimalist Program (Chomsky, 1995, henceforth MP), etc. A theory that assumes binary branching as a guiding principle (e.g. MP) has the advantage of minimizing the class of possible structures while ensuring that the relations between their constituents are unambiguous. This advantage does not come for free: it comes at the cost of generating a great amount of structure. For instance, consider the following representations: