Dan Everett: phonological arguments for late insertion? (reply to Martha McGinnis)

[Martha McGinnis]

Martha and all,

One other thing on LP and Structure Preservation. The really neat idea
behind SP in Lexical Phonology, as I understand it, is that lexical rules
will be structure-preserving *because* in the lexicon there are no
allophones, only phonemes of one sort or another, depending at the level
you are at. (Now, the output of the lexicon, according to Mohanon and
others, turns out to be the so-called taxonomic phoneme of structuralism.
This is a good result, since it turns out, according to some LP research
and a lot of structuralist work, that people have intuitions not about
abstract or systematic phonemes, but about taxonomic phonemes (the crucial
structuralist and descriptivist evidence for this comes from reports of
speaker reactions in learning to use an orthography based on taxonomic
phonemes).)

LP can, of course, allow access in the post-lexical phonology to the
inventory of phonemes. That information is not lost. And this access will
serve to account for post-lexical structure preservation of the kind I am
claiming to exist. On the other hand, as I noted, this kind of structure
preservation no longer follows from the architecture of the model.
Post-lexical structure preservation cannot be due to the absence of
allophones/variation but to some other reason. Whatever that reason turns
out to be, it does not follow, as does lexical structure preservation,
from being bounded by the lexicon. This significantly weakens the entire
rationale behind lexical structure preservation because, as I stated, we
now need a second mechanism/explanation for this new kind of structure
preservation. Morris Halle raised similar objections to the necessity to
state twice idential distributional rules in the phonemic and
morphophonemic 'levels' of structuralist phonemics.

A model like DM would be able, I should think, to offer an account of this
post-lexical structure preservation which would also handle lexical st.
preservation, one of Lexical Phonology's core results. Do that and the
appeal of LP is significantly weakened. On the other hand, DM ought also
to be able to account for the remaining, very interesting LP result in
this regard, namely, the co-existence of both taxonomic and systematic
phonemes in a single grammar.

Dan Everett

On Sat, 17 Nov 2001, Martha McGinnis wrote:

> Hi Dan,
>
> Thanks for the information... sounds like an interesting project.
> But is this really an argument against Lexical Phonology?  LP doesn't
> necessitate structure-preserving postlexical rules, but I don't think
> it rules them out either.  The situation might be different if we
> found a 'lexical' phonological rule that ISN'T structure-preserving.
> Anyway, the facts you describe are interesting -- thanks for the
> suggestion.
>
> -Martha
>
>
> >There is an interesting fact, not quite like what you asked for, but
> >leaning in the same direction. In recent field work, a colleague of mine
> >and I have discovered an interesting process of post-lexical structure
> >preservation in the Ge language, Suya (no other studies of this language
> >exist). There is a rule of phrase-final lenition of voicless occlusives
> >which is blocked just in case there is no lexical phoneme corresponding to
> >the lenited form. So, for example, p --> w and t --> r, but nothing
> >happens to k. This is because w and r are independently needed phonemes in
> >the language, but there is no corresponding continuant in the velar
> >position.
> >
> >What this means is that one of the basic tenets of Lexical Phonology, that
> >structure-preservation is evidence for phonology in the lexicon, becomes
> >much less secure.
> >
> >Or so it seemeth to me. The paper is in progress. Hope to have it done in
> >a couple of months, when I get out from under some other things.
> >
> >Cheers,
> >
> >Dan Everett
>
>
> mcginnis at ucalgary.ca
>