LINGUIST List 11.511

Thu Mar 9 2000

Disc: Underlying Shwa?

Editor for this issue: Karen Milligan <>


  1. James L. Fidelholtz, Re: 11.499, Disc: New: Underlying Shwa?

Message 1: Re: 11.499, Disc: New: Underlying Shwa?

Date: Wed, 8 Mar 2000 19:03:15 -0600 (CST)
From: James L. Fidelholtz <>
Subject: Re: 11.499, Disc: New: Underlying Shwa?

On Wed, 8 Mar 2000, wrote:

>No-one seems to have mentioned stress so far. In "cactus",
>"abacus", "syllabus","phosphorous" (N), and "opus" the schwa
>syllables are all unstressed. The invented word "cactusia" like the
>more or less real medical words ending in "'-ia" and "-ic" would in
>my view be stressed on the penultimate syllable as in "amnesia"
>and "ambrosia" (sorry but I couldn't think of any words ending in "-
>usia") which would prevent schwa-ing.
>This raises another point. Presumably the first vowel of "ambrosia"
>would never be schwa-d but what about "amnesia" ? Could the "a"
>become schwa in certain contexts, eg "He is suffering from
>amnesia"? Or does it depend more on the number of consonants
>following the vowel, eg "ammonia" with schwa but "amnesia"
>If you combine stress and the number of consonants, do you have
>the conditions for an underlying schwa?

Well, reduction facts have to do with both, of course. Generally, 2 or
more consonants impede reduction, except in 'frequent' words (in my
data, > about 5 occurrences/M[illion words]. Note that, as per the
arguments in my 1976 article, it does not matter here whether we are
considering spoken or written language, since this would be precisely
the frequency where they would generally agree -- of course, frequent
words are more frequent in spoken language, while infrequent words are
even more infrequent, with 5/M, apparently, being the crossover point.
The article addresses a few of the exceptions to this statement.

>From: "Douglas G. Wilson" <>

>(2) The underlying vowel can be coaxed out by asking "What did you
>say?" repeatedly until the speaker elongates and stresses all
>syllables. Thus 'calculus' will be rendered 'Kal! Kyu! Luss!' or so.
>Now suppose that in the speaker's pronunciation 'rabbit' and 'abbot'
>rhyme exactly. The stressed forms will be 'Rab! Bit!' and 'Ab! Bot!'
>(or 'Ab! But!') probably. But suppose the speaker is entirely
>illiterate: now we might expect that the written vowel will lose its
>influence and probably he'll say 'Rab! But!' and 'Ab! But!', with the
>schwa expressed as IPA inverted-v (the usual stressed version of schwa
>in English), I think. What about 'rhythm'? I suppose some speakers
>will make the second syllable 'Mm!', others 'Um!' (with IPA
>inverted-v); others might utter a central vowel, a stressed schwa.
This is just about right--shwa IS a central vowel (high-central in its
mostly predictable 'barred-i' variety), so when it does get stressed, or
restressed for any reason, the natural tendency, if there are no
alternations to guide one, would be to pronounce a central vowel. I
quote from a private email sent to Jorge:

[I'd like to]
mention a couple of historical facts: in fact, there ARE underlying
shwas, or at least there were. Note words like 'comfort', which, if my
feeble brain is remembering correctly, were at some point accented on
the final syllable. The historical source of the first vowel is
unequivocably [o] or open-o or some such. Obviously, at the point when
the accent was on the second syllable, the first vowel was shwa. But
when the stress shifted back to the first syllable (or shifted to the
first syllable), the result was 'upside-down v' or 'uh', which is how it
is still pronounced. In other words, the same thing happened, and in
some of them there was later pronunciation shift by analogy with other
forms, but we still have lots of such pronunciations. Also, there is a
strong tendency, when we do stress syllables which have what seems to be
a shwa, to use this same vowel (cf. I said 'sofA', not 'so far').


Going back to Jorge's original query:

>1. Given that shwa appears predictably in weakly stressed syllables in
>AE, is there a ***class*** of vowels that are pronounced shwa under
>weak stress,

Yes. ALL vowels (see below) [Oh, well, SOME nonlow front V reduce to
'barred-i' instead of shwa, generally due to environmental factors (eg
j_kt, [in 'object'] ie palatals)] 

>2. [snip] Is it the case, then, that swha alternates only with vowels
>other than front nonlow or are there cases of alternation between front
>nonlow and shwa?

Well, we have the following examples:
lEmur, lmyUrin
salIva, salvAt
VEns, Vn(j)Uzin
judgmnt, judgm_e_ntl (also barred-i in judgment)
The only vowel I don't know any shwa alternations for is [oy] (see
Fidelholtz w/ Browne ca. 1974 (Gtown paper 'Oy, oy, oy' in volume edited
by Shuy & Bailey); compare, however, 'destroy/destr_u_ction' By the
way, see Fidelholtz 1976, pp. 200-213 in the CLS vol. 10, for a fairly
thorough treatment of VR in English.

>2. Are there underlying shwas? Calculus, ...[snip]. Is there a
>pronunciation of calcul- that would show that the underlying
>vowel is other than shwa?

 Yes, all of the above. The only source (not quite, but after
any other C) for [j] is in certain environments before lax [u] (when it
gets tensed, basically before CV, with a few more details covered in
Fidelholtz 1967 (_MITRLEQPR_, I forget which number right now).
 And what about nonalternating shwas? You'd have to torture me
to get me to admit that they're underlying. If pushed, I'd derive them
from an unspecified vowel, which, being unstressed, simply reduces.

OK, the facts, as always, get a little messy around the edges, but it
seems to me that the general situation is quite clear: unstressed vowels
reduce to shwa (or one of its contextual variants, depending on the
dialect, etc.), except for:
	'frequent' words if the vowel is before two consonants and a
stressed vowel and in the first syllable of the word (this is actually
somewhat more complicated); frequency or rarity has no effect in
unstressed syllables surrounded by stressed ones: all vowels reduce
always (cf. 'sal_i_vate', where sal_I_va shows the vowel is underlyingly
tense, and 'comp_e_nsate').

James L. Fidelholtz			e-mail:
Maestr�a en Ciencias del Lenguaje
Instituto de Ciencias Sociales y Humanidades
Benem�rita Universidad Aut�noma de Puebla, M�XICO

Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue