Return of the minimal pairs

[Paul S. Cohen]

On Thu, 17 May 2001 14:17:07 +0300, Robert Whiting <whiting at cc.helsinki.fi>
wrote:

>On Thu, 10 May 2001, Larry Trask wrote:

[ moderator snip ]

>> Finally, I note that /n/ and [eng] *really* do not contrast in word-initial
>> position in English, even though they contrast elsewhere.  The case of
>> theta and eth does not appear to me to be similar.

>No, it is not similar at all.  [eng] is prohibited phonotactically from
>appearing in initial position in English (except in loan words; we've done
>this before).  In this instance, [eng] behaves like a cluster rather than
>a single segment.

Actually, there's another possible view of this situation that's completely
consistent with classical phonemics:  since [[eng]] and [h] are in
complementary distribution in English (yes, I know about the Trager&Smith
tradition of talking about [h] and [[schwa]] being allophones), and since
they are, in some sense, phonetically similar (they comprise a class that
can be characterized as "the back, continuant consonants"), why not put
them together as allophones of /[eng]/ or /h/ (whichever you like)?
Occam's Razor would seem to demand doing so.  This, I would hope, shows off
some off the inadequacies of classical phonemics.

Bob Whiting wants to allow morpheme boundaries as part of the allowable
phonological conditioners in deciding whether it's "one phoneme or two",
and I'm in full agreement.  In the American school of phonemics, this
position goes back at least to Kenneth Pike in the 1940's, despite the fear
of "mixing levels" in that school.  It's clear that perspicuous analysis of
phonological systems often requires it.  In fact, I would go a step
farther:  There are times when perspicuous phonemic/phonological analysis
requires the use of other phonetic, morphological, and syntactic
characterizations as well.  One may choose to draw the line in various
places in deciding phonemicity, but being too restrictive will often
eliminate generalizations--even some that naive speakers are making--
whether consciously or unconsciously.  An example that I am very familiar
with concerns the tensing and raising of [ae] (that is "ash") in the New
York City area.  In broad outline (and I'll omit irrelevant details) [ae]
is tensed and raised in front of voiced stops, /m/, /n/, and fricatives
when these are followed by an obstruent or a major morpheme boundary, so
that, e.g., the stressed vowel of _adder_ 'snake' is not tensed and raised,
but that of _adder_ 'adding machine; one who or that which adds' is.  If
morpheme boundaries are not allowed, we have to posit two phonemes.  Not a
very satisfying solution.  But wait; there's more.  _I can fish._ (with
emphatic or contrastive stress on _can_) has no tensing and raising of the
[ae] if it means 'I am able to fish', but has tensing and raising if it
means 'I work in a fish cannery'.  Pretty straightforward then:  must be a
phonemic split.  Not so fast:  It turns out that all words that can have
[schwa] as their only vowel, *always* have untensed and unraised [ae] in
their stressed form, at least in one major subdialect.  (This list, which I
have termed "weak words", includes _am, as, can, had, has, have, than_.)
Wait, I hear you say; why not talk about "function words" or "closed-class
words"?  Because, _can't_ *is* always tensed and raised.  What sets it
apart is that it can never have [schwa] as its only vowel--i.e., it is
never completely unstressed.

Anyhow, one phoneme or two?  You be the judge.

Paul S. Cohen