Title
Bruce Morén (Tromsø): Toward constructing a minimalist phonological feature theory

CASTL colloquium

Date:  07.12.2006

Time:  19:15

Venue:  E010 1

Description
The main purpose of this talk is to explore a modular, minimalist and “substance-free” approach to phonological features in which there is no universal set of features connected innately to the auditory system or vocal tract (or visual system and non-vocal anatomy in the case of sign languages). Rather, there is a minimal set of abstract structures available for mapping to the physical world given overt, positive evidence from the language environment. In essence, the number and instantiation of features is a language-particular emergent quality, and the phonological features themselves are abstract and free of phonetic content. While this may seem antithetical to the “received wisdom”, it is in line with more recent work (and a standard Structuralist view) doubting an innate 1:1 mapping of phonological features to phonetics and suggesting that much of what can be functionally motivated might also be functionally, not phonologically, explained.
There are at least two important and widely acknowledged, but often ignored, sources of support for the substance-free hypothesis. First, there is a non-isomorphism between the phonetic characteristics of speech sounds and their behavioral patterns when we compare cross-linguistically. Second, sign language segments differ radically in their physical realizations in comparison with spoken languages, but there is every indication that both types of languages combine and manipulate phonological features in the same way. In fact, signed languages seem to have an articulatory and combinatory richness that exceeds spoken languages (at least as measured by standard feature theory), so it cannot simply be the case that sign languages map exceptional phonetics onto innate “spoken language features” (e.g. index finger ≠ [coronal]).
In this talk, I will present a programmatic view of a particular feature theory and discuss at least two advantages of using this theory. First, it eliminates the duplication of effort problem pointed out by Hale and Reiss (2000) and others in which some phenomena that can be explained via functional phonetic principles are also given separate formal phonological analyses. Second, it allows us to “rein in” OT analyses by 1) simplifying the feature set/segmental structures that need to be evaluated in a given language and 2) providing a limited set of features/segmental structures that are available for use by phonological constraint templates. Essentially, only phonological features for which there is overt evidence in a given language are manipulated by GEN, subject to evaluation, and made use of in constraints in that language. This has a range of implications for OT analyses and requires a re-thinking of some of the core assertions of classic OT.
There are at least two immediately obvious potential downsides to this approach. First, it challenges a core assumption held by many phonologists that phonological features are innately specified with phonetic content and that language-particular sound patterns must be explained in terms of phonological analysis rather than phonetics (or phonetics-phonology mapping). Second, it poses serious practical challenges in that it forces us to look more carefully at a larger range of phonetic details and phonological patterns in specific languages in order make any claims about either the phonological features used by those languages or propose an analysis of particular phenomena. This has the further effect of making it more difficult (at least initially) to make cross-linguistic comparisons and generalizations. Are these really downsides? Or, is this just good linguistics?
This talk is a modified version of my Phonological Bases of Phonological Features Workshop presentation.