Errors and buffers: essays in the economics of syntactic rearrangement

This dissertation draws upon modern Chomskyan theory to address issues surrounding the development of a unified, minimalist account of language as a mental and biological object, both in terms of its generation and historic change. Towards that end, I investigate, apply, and advance the labeling approach to generative syntax. Labeling is a hypothetical process, operating within the confines of phase theory, which is thought to prepare constructed syntactic objects for interpretation at relevant mental interfaces. I argue a number of points applicable to both synchronic and diachronic linguistics: 1) Labeling failures happen as a matter of course during a derivation, forcing re-evaluation of labeled syntactic structures which ultimately leads to a successful derivation. 2) Labeling and its errors do not happen in real-time, but are bounded by phases. This has consequences for how researchers ought to look at notions and limitations of phasal memory. 3) Labeling not only drives an individual’s mature syntax, but has an effect on how children acquire their syntax, causing them in some cases to alter structures and create new categories. This is responsible for many cases of language change, and I support this argument by investigating data from the history of Chinese and Macedonian that are sensitive to labeling-based phenomena. 4) Research into labeling can help us speculate about the evolution of language generally. Although recursion is sometimes thought to be a defining feature of Universal Grammar, labeling in fact is a much more likely candidate in this regard.