THEORY: no more URs!

At 3:42 PM +0100 05/25/02, And Rosta wrote:
>Dirk:
>>  At 2:43 AM -0400 05/24/02, Roger Mills wrote:
>>  >I realize that, but hope you agree that some sort of underlying level is an
>>  >improvement...  Of course, Chomsky/Halle and Classical Phonemics
>>are extreme
>>  >and opposite viewpoints,  and the answer, if there is one, probably lies
>>  >somewhere between the two....:-)
>>
>>  Well, I'm not convinced anymore that a distinction between underlying
>>  and surface representations is necessary (or desireable). Recent work
>>  in phonology and morphology being done by Luigi Burzio (yes, that
>>  Luigi Burzio) and others claims that anything that URs do can be done
>>  by balancing phonotactic requirements with compulsory identity
>>  relations within networks of similar forms. That is, 'parent' and
>>  'parental' show segmental consistency not because they both have
>>  /parent/ in UR, but because identity constraints hold between the
>>  surface forms. The substring {parent} fails to show metrical
>>  consistency across the forms (i.e., stress on pá in 'parent', stress
>>  on rént in 'parental') since the regular phonotactic pattern of
>>  penultimate stress takes precedence over metrical consistency.
>
>I don't know this work by Burzio, and nor, I confess, do I understand
>from your explanation exactly what the two sides of the debate are.
>
>However, in debating the issue of URs, we need to make a distinction
>between:
>
>A. How to get from (I) representations of word forms that encode (only)
>what is lexically contrastive to (II) representations of grammatically
>determinate pronunciation (including allegro processes, etc.)
>
>B. How to capture alternations between the type (I) forms a putative
>single morpheme takes when it occurs in different word forms.
>
>You seem to be talking about (B), but I had the sense that the battle
>against handling these alternations by pure phonology had long ago
>been won. OTOH, I also had the sense that (A) has been more of a
>live issue.
>
>Maybe you could explain a bit more what the debate is about?
You're absolutely right in distinguishing issues A and B; what I've
read of Burzio's work addresses B. Most work in Optimality Theory
assumes some version of Hockett's Structuralist Item-and-Arrangement
morphology without question -- John McCarthy has built his career on
being the Generative phonology (including OT) Item-and-Arrangement
poster boy. Ever since I was made aware of alternatives like
Matthews' Word and Paradigm, or Anderson's A-Morphous Morphology, and
more recently Aronoff's work, I've thought the Item-and-Arrangement
view to be a drastic oversimplification of the actual state of
affairs. A model which seeks to explain phonological similarities
among word forms by direct appeal to something like a paradigm is
more in keeping with Matthews', Anderson's, and Aronoff's ideas, and
that's the kind of model which Burzio is trying to develop in OT
terms.

As far as issue A is concerned, the typical Optimality Theory stance
is that representational economy is not valued for underlying
representations since constraints apply properly to the output (=
surface) rather than the input. I also subscribe to this view, but
I'd like to take it a step further by eliminating URs altogether. As
I said in another post, the idea wasn't directly relevant to the
research for my dissertation, so I dropped it. Well, mostly; there
are hints scattered throughout the dissertation that URs in Shoshoni
are indeterminate; I state in several places that several URs can
converge to the same surface form, so that there is no way to
uniquely identify an input for a given output. Now this isn't the
same as saying that there are no URs, but it does weaken the role
which URs play in the grammar.

>  > I am very sympathetic to this idea (no URs); I tried doing something
>>  like this in grad school, but I was basically "laughed off the stage"
>>  and didn't have the courage to pursue it then.
>
>I wonder if that indicates a cultural difference between US and
>British academia: I can't imagine you getting laughed off the stage
>in Britain in such a circumstance.
Perhaps. The British have a reputation for prizing eccentricity (in
my mind at least), and in 1995, UR-less phonology was considered an
eccentric position in American phonological theory. Declarative
Phonology was in the air (an explicit, formal theory of phonology
which also did without a distinction between underlying and surface
form), but the only people actively pursuing it were British (James
Scobbie, Stephen Bird, John Coleman, Michael Broe and Richard Ogden).
From what I understand of Government Phonology, it also has a
declarative flavor, so it doesn't surprise me that it's much more
popular in British circles than in American.

>--And.
Dirk
--
Dirk Elzinga                  Dirk_Elzinga@byu.edu

Man deth swa he byth thonne he mot swa he wile.
'A man does as he is when he can do what he wants.'

- Old English Proverb