There can be little doubt that the greatest innovation in phonology of the past two thousand years was the development of the concept of separate tiers. However, current brands of autosegmental phonology have failed to exploit this innovation to the full, so that as a result they still show influences from outmoded structuralist theory. This paper will rectify that failing.
The first major problem of modern autosegmentalism is that it doesn’t have enough tiers. The fact is, every single distinctive feature should have its own tier. The second problem is that the modern timing tier retains the silly structuralist idea that the speech stream is divisible into discrete segments. In truth, the only boundaries in speech are sentence boundaries (Dionysius Thrax and Noam Chomsky have both taught us that a sentence is a complete thought, the only true form of linguistic structure
TIER | |||
Cons | +-+-+-+ | ||
Cor | +-+-+ | ||
Ant | -+- | ||
Obs | +-+-+-+ | ||
Cont | +-+-+- | ||
Liq | -+- | ||
Lat | -+- | ||
Hi | - | ||
Lo | -+- | ||
Fr | -+-+-+- |
Of course, since there is a temporal factor in speech production, it will be necessary to account for how temporal coordination occurs. In keeping with the general nature of solutions in scientific linguistics, I suggest that there exists a filter which rules out all unacceptable phonological strings. An acceptable phonological string would be one whose syntax corresponded to the syntax of the string being generated.
A speaker will need a very long list of possible strings to which to compare his generated utterances. Since he could not possibly hear all possible strings in one lifetime, it stands to reason that this list of acceptable strings must be innate. Neurologically speaking, it must be centered on Broca’s area. Psychologically, it is probably a push-
Morris Goldparsky | American National University |