Dan Everett: taxonomic phonemics (reply to Martha McGinnis)

[Martha McGinnis]

On Tue, 20 Nov 2001, Martha McGinnis wrote:

> Dear Mike,
>
> There is, or should be, a relationship between lexical phonology and
> DM.  Under DM there is no phonology in the pre-syntactic lexicon, so
> there's an issue as to how the insights of LP should be recaptured.
> For example, if there really is a highly local morphosyntactic domain
> within which phonological rules must be 'structure-preserving', this
> is an important insight no matter what theory we adopt. I think we
> could stand more postings on LP, if people want to continue the
> discussion here.

Yes, I agree that it seems quite relevant to DM. And I think that cases of
so-called Structure Preservation outside the lexicon (by LP's own
definition) are worth pursuing because they show that the insight of LP in
this regard, a genuine insight as Martha points out, is not a result of
the lexicon per se but of a constraint or constraints that are applicable
at different levels in the grammar, so that theories which allow
word-formation access to syntactic information might be in a privileged
position wrt these phenomena.
>
> However, no more postings about 'taxonomic phonemes' will be
> forwarded until someone defines the term.

Hmm, I am surprised this is a problem (though I also realize that some
people ask for definitions when they don't think there is one to drive
home their point subtly). A phoneme, still controversial in many parts of
Europe, and among many phoneticians internationally, is a set of sounds,
not features, which the native speaker hears as a single sound (the
general line things took after Daniel Jones at UCL made the intial
proposals). The label given to that set, as Syd Lamb pointed out in the
early 60s is irrelevant (you could use the number 4 to represent p for
example). Paul Postal replied that the label *is* relevant and introduced
the Naturalness Constraint to make this explicit. These perceptual
groupings are determined, however, by analysis at a single 'level' of the
grammar, the level which involves allophonic rules but not morphophonemic
rules. Postal used the term 'taxonomic phoneme' as a perjorative
description of these units. The idea was, as I understood it at the time
of my initial reading, that the phoneme as pursued in these theories was
not genuinely scientific, like Gen. Phon., say, because (i) they didn't
look for the systematic relations uniting deep and surface structures and
obliterating the allophonic/morphophonemic distinction and therefore (ii)
the phonemes of the structuralists were apparently just like the labels
given by naturalists in the early days of biological survey when the goal
was to classify rather than to explain. Postal's (1968) Aspects book is
still the best read on this, though Anderson's Phon in the 20th Century
gives a more detailed, less emotional, more objective survey of the ideas
(as Shane does briefly in his textbook).

Having given my version of history (not really having lived through it,
but having studied linguistics first with Pike, before moving on to
Generative and then post-Chomskyan views), I am under no illusion that
there will be widespread consensus with what I have just said. These
notions were rarely agreed upon and revisionism for various reasons is
common in scientific disciplines.

I would suspect that even now, certainly in a few years, someone on some
list is going to suspend discussion of Distributed Morphology until
someone defines the term. This is of course quite reasonable. We should
always rethink through the terms.

Cheers,

-- Dan Everett