LINGUIST List 11.666

Thu Mar 23 2000

Disc: Underlying Schwa?

Editor for this issue: Karen Milligan <karenlinguistlist.org>

Directory

1. Mark_Mandel, relatedness (was: underlying schwa)
2. James L. Fidelholtz, Re: 11.658, Disc: Underlying Schwa?
3. Geoffrey S. Nathan, Re: 11.654, Disc: Underlying Schwa?

In 11-654 Geoff Nathan writes:

>>>>>
 [...] I just graded an
elementary exercise in derivational morphology from _Language Files_. At
least one student, a native speaker, did not recognize the following
morphological relatednesses:

explosion (didn't think of it as made up of 'explode+ion')
active (didn't see 'act' in it)
responsibility (didn't see 'responsible')

One of the problems with clever underlying forms based on relatedness
between words is that there is great individual variation among native
speakers as to whether there IS a common form-meaning pairing. In
Mohanan's old book on lexical phonology, for example, he uses the pair
'native': 'nation' as evidence for a level one rule of palatalization. But
how many would argue for a synchronic semantic connection between those
two? Etymological, sure, and obviously, orthographic. But the core
meanings of 'native' (something like 'aboriginal') and 'nation' (political
entity) are sufficiently distant to argue against a common lexical entry,
which is what 'underlying form' is all about.
<<<<<

I fear that all too often we linguists overgeneralize our own native-speaker
intuitions to the rest of the native speaker population, ignoring two
differences:
 1. More obvious: we and our intuitions are *trained*, not only by our
education as linguists, but by our education itself. Most native speakers do not
have university degrees and corresponding (or prerequisite) degree of literacy
and breadth of exposure to varieties of language beyond daily or broadcast
contact.
 2. Less obvious: to a certain degree we are a self-selected subset of the
population. I know that my love of language and "knack" for languages, and my
desire to study as many as I could, is inextricable from my choice of career,
going back at least as far as junior high school. I suspect that a far greater
proportion of us have a similar bent than is to be found in the general
population. To that extent, our intuitions on relatedness are not reliable
guides to the structure of the language in the mind of the typical speaker.

 Mark A. Mandel : Senior Linguist and Manager of Acoustic Data
 Mark_Mandeldragonsys.com : Dragon Systems, Inc.
 320 Nevada St., Newton, MA 02460, USA : http://www.dragonsys.com/
 (speaking for myself)

Message 2: Re: 11.658, Disc: Underlying Schwa?

On Thu, 23 Mar 2000, The LINGUIST Network wrote:

Jorge Guitart says:

>Why not think that this is what happens in native and nation, even though
>the speaker may not see the semantic relation?

	I say, Bingo! Speaking of hitting the nail on the head. There
seem to be lots of cases in language (folk etymology is an outstandingly
good one) where speakers require AN analysis of a word into shorter,
morpheme-like parts, without caring, apparently, that the semantics is
awry or nonexistent: cf. the hundreds of published discussions on
cran+berry, with very few even noticing that (a) historically, cran
comes from crane; and (b) speakers don't care what cran MEANS, they just
know that it's a 'meaningful part' of the word (the term in quotes is
obviously used metaphorically); and even (c) that such 'obvious' cases
as 'blackberry' are in fact none too obvious. 
	Jim

James L. Fidelholtz			e-mail: jfidelsiu.buap.mx
Posgrado en Ciencias del Lenguaje	tel.: +(52-2)229-5500 x5705
Instituto de Ciencias Sociales y Humanidades	fax: +(01-2) 229-5681
Beneme'rita Universidad Auto'noma de Puebla, ME'XICO

Message 3: Re: 11.654, Disc: Underlying Schwa?

Theo Vennemann wrote:

 >a
 >single-level theory of phonological representations which some people may
 >remember by the name of NGP (Natural Generative Phonology), a theory in
 >which all rules are constraints or, as Joan Bybee Hooper then said, "rules
 >have to be true on the surface". It is quite gratifying to see other
 >linguists arriving there.

	Unfortunately I have to disagree with my distinguished
colleague. Even if we don't permit phonological rules to account for
allomorphic alternations (which I would not admit) we are still left
with radical alternations _within single words_ depending on
formality, speed and so forth. And there are plenty of rules that
aren't 'true on the surface' (which is a problem OT has been wrestling
with for quite a while). Consider, for example, the contrast between
'police' and 'please', which, on the surface contrast in voicing of
the /l/ (ignoring the irrelevant final consonant difference). In
old-fashioned ordering terms, schwa-deletion counterfeeds liquid
devoicing, and no surface-true theory could deal with this situation,
because devoicing of liquids following initial voiceless stops is not
surface-true. yet one may hear [plis] and [plis] in the same
sentence, with native speakers clearly under the impression they said
the same word.

	In response to Jorge's comment on the same subject, however, I
need to point out that this is quite different from the situation of
'native' vs. 'nation', where most native speakers don't recognize
that there IS a suffix in 'nation', let alone that the suffix is
triggering a palatalization--it's very unlikely a native speaker would
figure out that the [S] in 'nation' is in any way different from the
one in 'fashion' , except for the fact (alluded to in an earlier
posting on the subject) that they are spelled differently. But
current linguistic theory has no place for orthographic underlying
forms (if, by underlying form we mean what Baudouin and Sapir meant:
the form that a morpheme takes in long-term storage).

	It is possible, I believe, to defend a middle ground between
the SPE reconstruction of Middle English as underlying form and the
NGP totally surfacy view. It is necessary for a phonologist to
demonstrate that a particular combination of underlying forms and
processes are living, active processes in the language in order to
claim that they belong in a synchronic description, but living
processes can produce radical alterations in underlying forms that are
clearly well-motivated. When someone says [aeZuwElno] for 'As you
well know..' (ae=ash, Z=ezh, E=epsilon) it is clearly being derived
from a combination of underlying /z+j/, and I've collected much more
complex cases that involve longish chains of feeding and
counterfeeding orders, all involving commonly used words with no
historically reconstructed underlying forms necessary. It is not
abstractness per se that is unlikely, but rather abstractness that
only a linguistically-trained phonologist would be likely to posit.

Geoff

Geoffrey S. Nathan
Department of Linguistics
Southern Illinois University at Carbondale
Carbondale, IL, 62901-4517
Phone: (618) 453-3421 (Office) / FAX (618) 453-6527
 (618) 549-0106 (Home)
 geoffnsiu.edu