LINGUIST List 2.113

Wednesday, 3 Apr 1991

Disc: Phonology

Editor for this issue: <>

Directory

1. Mike Hammond, Penults in English
2. , History of Phonology
3. "Grover.Hudson", Phonology
4. John Coleman, Vowel reduction; consonant cooccurence constraints
5. John Coleman, Consonant cooccurrence constraints for roots

Alexis Manaster-Ramer cites a number of interesting apparent
counterexamples to the claim that closed penults in English attract
stress and that long ultimae attract stress.

Closed penult that doesn't attract stress: Orchestra, pOdagra (for
some speakers, apparently), ClArendon, CAvendish, Ogilvi(e),
badminton.

long ultima that doesn't attract stress: diabEtes (with "flapped" t).

Some of these are fairly easy to rule out. Orchestra and podagra
contain open penults, assuming onsets are maximized. Ogilvie can be
analyzed like galaxy--with a final /y/--notwithstanding the
(irrelevant) spelling. Diabetes is probably analyzable as a "plural
disease" with stem-final tensing: /diabete+s/, parallel to mumps,
measles, shingles, etc. (Cf. cities, etc.)

That leaves Clarendon, Cavendish, and badminton. Note that all of
these penults are closed by sonorants (specifically an alveolar nasal
here). A number of possibilities present themselves. One is to admit
them as exceptions to the larger generalization about closed penults
and treat them with some generalization of Sonorant Destressing.
Another (perhaps less desirable) possibility is to maintain that the
sonorant consonants are stray in underlying representation:
/clarndon/, /cavndish/, and /badmnton/. Another possibility in the
case of clarendon and badminton is to maintain that the final sonorant
is stray: /clarendn/ and /badmintn/.

Personally, I don't know if I get the stresses AM-R suggests. For me,
they have the following stress patterns.
 clarendon 1 0 0
 cavendish 1 0 2
 badminton 1 2 0
My most natural pronunciation of badminton has an open penult:
[baedmItn]. This still leaves clarendon as a counterexample though
unless the final sonorant is stray.

Pierre Martin writes:

> There is at least one school of thought in current phonological theory which
> practices a clear distinction between the purely phonic component of a 
> language (=phonology) and the constraints imposed on the distinctive units by
> morphemes and vice-versa (moneme variants= morphology, and not 
> morphonology!) This is the (structuralist) functionalist approach of 
> Andre Martinet and his Paris School.

It is interesting that Martinet was one of the few structuralists who made an
effort to persuade his Prague School colleagues not to conflate archiphonemes
(phonological theory) with morphophonemes (morphonological theory). I am not
very familiar with this more recent school, unfortunately.

Modern phonologists tend not to be very well schooled in the history of the
field. Most (in the West, at least) do not realize that three distinct schools
of phonology evolved out of Baudouin's alternational dichotomy: the Moscow
school, the Leningrad school, and the Prague school. We in the West tend to
view the Prague school as the foundation of modern phonology, although
Baudouin's seminal work had begun roughly half a century earlier. The 
Leningrad school, the more visible Russian school of the two, was dominated
by Shcherba, who completely revised Baudouin's alternational dichotomy. In
Shcherba's approach, the phoneme came to be seen as a perceptual unit, so that
phonemic neutralization (e.g. vowel neutralization, final devoicing)
tended not to be recognized as a phonological phenomenon. Shcherba's ideas
crept into the West and came to be accepted as the basis of what we now call
'classical phonemic theory'. Of the three schools, only the Moscow school
retained the original Baudouinian level of phonemic (a.k.a. phonological)
representation. Here is a rough sketch of how the three major schools might
differ on the levels of representation for the noun 'lives' (although I am
technically fudging the 'levels' to make a neater comparison):

 Prague(Trubetzkoy) Leningrad Moscow

Morphophonological {layF+S} /layf+z/ /layf+z/
& Archiphonemic
Phonemic /layv+z/ /layv+z/ /layv+z/
Phonetic [layvz] [layvz] [layvz]

One major difference here is that the two Russian schools, like Baudouin,
viewed the morphophonological 'level' as comprised completely of phonemes
(which were phonetic segments). (OK. I'm ignoring the Moscow concept of
the 'hyperphoneme', which is another story.) 

Other approaches to phonology which accept the Baudouin/Moscow level are/were
Stampean Natural Phonology and Sapir's approach. I am unclear as to whether
lexical phonology should be counted in this camp, although it has done 
something to repair the damage to phonological theory caused by the SPE model. 
(BTW, I claim that the SPE model of phonology does little, if anything, to 
resurrect Sapir's view of phonology. If anything, it completes the process of
burying it.) SPE effectively threw out the baby with the bathwater when it
trashed the Scherbeme level of phonemic representation without considering
the Baudouin/Sapir/Moscow level as a possible alternative. The result was the
mistaken position that there was no basic difference between morphonology
and phonology. IMHO, an incredible setback for linguistic theory.

 -Rick Wojcik (rwojcikatc.boeing.com)

Emmon Bach properly provides clarification regarding the famous quote,
attributed to Martin Joos that "languages can differ without limit...". On
page 96 of the same book in another footnote, Joos attributes this idea
not just to the "Boas tradition" but to the "American (Boas) tradition".

Rick Wojcik passed the following comment on my "title/titular" suggestion:

> Vowel Reduction is a phenomenon that bears on how we pronounce sounds in
> prosodic environments. That is why stress is central to the question. The
> 'title/titular' example bears on the question of what phonological forms we
> assign to morphemes, not how we pronounce vowels. I wish that modern 
> phonologists would try harder to understand that fundamental dichotomy 
> in alternations that Baudouin de Courtenay observed when he developed
> the foundations of modern phonology. He would have called vowel reduction
> alternations 'physiophonetic' and the 'title/titular' vowel alternation
> 'psychophonetic'. ... Alas, modern phonology
> conflates the two very different phenomena under the rubric of 'phonology'.

Whether a phenomenon such as this is morphophonological or just an
automatic alternation depends on other considerations, especially the
form of morphophonological representations and the theory of phonetic 
interpretation which is being assumed. Wojcik's comment seems to me to 
assume that both of these are settled, and that there can therefore be
no redrawing of the lines between phonetics and phonology, even if such
realignment allows data previously regarded as independent to be handled
by generalisations already proposed for some other set of examples.

To demonstrate my point about the theory-internal nature of the dividing
line between (morpho)phonology and (physio)phonetics, EVEN IF THE TWO
LEVELS ARE KEPT SEPARATE, consider the following two observations:
1) before the use of feature-based representations became commonplace,
nonsegmental alternations were treated in segmental phonology 
essentially as suppletion i.e. there were two forms in the dictionary.
2) much recent work in Laboratory Phonology and speech synthesis,
such as Pierrehumbert and Beckman's, Browman and Goldstein's (perhaps)
and ours at York, has demonstrated that `systematic phonetic representations'
can be far more `abstract' e.g. less specific about `allophonic'
detail etc. than is usually assumed by phonologists working without
an explicit model of phonetic interpretation. The "title/titular"
case is just such. It is perfectly reasonable to propose a morphophonolical
representation of the /titUl(ar)/ form and attribute the different
phonetic exponents of /U/ to the different contexts with which it cooccurs.

> But from a historical point of view, it is J. Coleman who is making the 
> theory-internal assumption about the relevance of certain vowel alternations.

It's not an assumption. It was just a suggestion, backed up by a story about
vowel "alternations" which is perhaps a bit new (and consequently, I had 
hoped, might have been of interest to some people).

Pierre Martin <PMARTINLAVALVM1> comments:

> There is at least one school of thought in current phonological theory which
> practices a clear distinction between the purely phonic component of a 
> language (=phonology) and the constraints imposed on the distinctive units 
> by morphemes and vice-versa (moneme variants= morphology, and not 
> morphonology!) This is the (structuralist) functionalist approach of Andre 
> Martinet and his Paris School.

There are two others I know of. The Copenhagen school, I believe still 
has adherents, and also Firthian Prosodic Analysis, which is still very
much alive. That's my background, which is why I found Wojcik's
piece a bit ironic.

Keep it up, Rick!


--- John Coleman

How about the English *sCXsC constraint that prohibits *stask, *spast,
*skusp, *strask, *splast, *skrusp, or the *sC1XC1 that prohibits
*spup, *skak, *snon, *smam, *splup, *skrak (C1:place \= [+coronal]),
or the *CGXG constraint that prohibits *klilt, *krark, and *klul*.
(I got these from C. E. Cairns (1988) "Phonotactics, markedness and
lexical representation". Phonology 5, 209-236.)
Also what about the distribution of /h/ in English? isn't that restricted
to one occurence per root (excluding obvious loans such as "jojoba")?
Come to think of it, what about /h/ in Classical Greek? or many other
classical prosodic phenomena, such as Aspiration in Harauti, Nasalisation
in Sundanese ... all these induce consonant cooccurence constraints
when viewed segmentally rather than prosodically.

--- John Coleman

[End Linguist List, Vol. 2, No. 113]