minimal pairs (was: PIE e/o Ablaut)

[Robert Whiting]

On Tue, 4 Apr, proto-language <proto-language at email.msn.com> wrote:

<snip>

>[RW]

>I would say that even a minimal pair is not a sufficient
>condition to establish two sounds as separate phonemes.

>[PR]

[ moderator snip ]

>There is no phoneme in any language which has not been
>established as a component of a minimal pair.

So what?  Do you assume therefore that anything that is a
component of a minimal pair is a phoneme?  I know that you have
said that using logic is childish, but you really ought to try it
sometime.

Saying that anything that occurs in a minimal pair is a phoneme
because all phonemes occur in minimal pairs is like saying that
anything that is black is a raven because all ravens are black.

I know you won't understand what I'm talking about, but any
textbook on logic will tell you that for any universal statement
(all S are P) the simple converse (all P are S) is not valid.
The minimum valid conversion of "all S are P" is "some P are S".

[RW]

>>The distribution by rule takes precedence.  Take the English
>>minimal pair

>>         'thigh'  /  'thy'

>>         (the pair 'thistle'  /  'this'll' [contraction of 'this will']
>>         is clearly marginal)

>>Most people would not insist on phonemic status for both [th]
>>and [dh] in English on the basis of this minimal pair (although
>>some would doubtless claim that there has been a phomemic split
>>similar to what occurred with /s/ and /z/).

>That is exactly what I would claim. I would claim phonemic
>status for both.

Yes, of course you would.  I wouldn't expect you to do anything
else.

>>This is because otherwise the sounds are in complementary
>>distribution, [dh] occuring in voiced environments

>[PR]

>What in Heaven's name is a "voiced environment"?

An environment that is voiced.

>What is environmentally voiced in 'bathe' as opposed to 'bath'?

This is a morphophonemic alternation.  You can make as long a
list as you want of such alternations and wherever there is a
distinction between [th] and [dh], [th] will occur in a
substantive and [dh] will appear in a verb that is derived from
it.  Very rarely, [dh] will appear also in the substantive, but
it will always appear in the verb.

[RW]

>>and in deictic words and pronouns, [th] otherwise.

>[PR]

>I think it most illegitimate to suggest non-phonological
>conditioning factors.

And here you would have the full support of Stanley Friesen.  But
rules are rules, whether they are phonological, morphological, or
lexical.  It is just a question of how much one area of language
can affect another.

[RW]

>>Thus it is not only as Larry says "If the distribution of two
>>sounds cannot be stated by rule, then they can't be assigned to a
>>single phoneme," but also 'If the distribution of similar sounds
>>can be stated by rule, then they can't be assigned to separate
>>phonemes.'

>[PR]

>I am claiming that the *e/*o-Ablaut can be described by a rule.

Which is your strongest argument.  Which is why you should be
supporting my argument instead of claiming that distribution by
rule is less important than minimal pairs in determining
phonemicity.  You don't seem to realize that, while what I have
said does not directly support your position, it does give you
a stronger position to argue from.  I know that thinking things
through is not your forte, but you really should learn to think
in terms of how the various parts of a problem relate to each
other.  You say that the distribution of *e and *o is governed
by rule and that only a minimal pair would prove their phonemic
status.  And I say that even a minimal pair wouldn't necessarily
demonstrate their phonemicity so long as they can still be
predicted by rule and then you say no, all phonemes occur in
minimal pairs so anything that occurs in a minimal pair must be a
phoneme.  And you claim that this is a rational line of thought?

But if you claim that [th] and [dh] must be phonemes because they
occur as morphophonemic variants as you did above (bath, bathe),
then you have lost your argument about *e and *o because they are
also morphophonemic variants.  Indeed, it was in part their
morphophonemic alternation that led, through internal
reconstruction, to the laryngeal theory.  So if morphophonemic
alternation (conditioned variance) demonstrates phonemicity, then
*e and *o are separate phonemes.  Which position do you want to
support?

>[RW]

>>Minimal pairs are a shortcut to finding phonemes, but
>>contrastive environments are a clincher.

>[PR]

>I find this totally unacceptable.

Yes, I can see why it would confuse you.  The second part of the
statement is not entirely clear.  So I will try to explain it in
more detail and write very slowly and distinctly.

Minimal pairs is a heuristic.  A heuristic is a way of finding
things that might prove significant.  Minimal pairs is a useful
way of looking for phonemes.  Now if a heuristic works regularly,
people tend to start thinking of it as a law.  Minimal pairs is
so successful at predicting phonemes that often no further
investigation is done and a minimal pair is considered a
sufficient condition for phonemicity.  This is the position taken
by you and Stanley Friesen (and probably a lot of other people).

But a lot of people also realize that minimal pairs is not a
necessary condition for establishing phonemicity (I realize that
you probably don't understand this stuff about "necessary" and
"sufficient" conditions because it has to do with logic, but bear
with me) and this is where you and Stanley part company.

Now I believe that minimal pairs is not even a sufficient
condition for establishing phonemicity, particularly when the
number of minimal pairs is minimal.  I believe that distribution
of sounds by rule is more important than what may appear to be
minimal pairs in determining whether two sounds are distinct
phonemes or not.  I also believe that the rule that determines
the distribution of the sounds does not have to be phonological.
This is where you and Stanley join up again, in the belief that
the conditioning environment must be phonetic.

What I meant by "contrastive environments" is, as I said, not
entirely clear from the statement (but I'm not entirely sure that
saying "non-contrasting" would have been better).  What I had in
mind is that the different phonemes provide the only contrast,
and this contrast must be completely arbitrary for the two sounds
to be separate phonemes.  That is to say that the phonemes
themselves can tell you nothing about the words involved except
that they are different.  If a sound regularly occurs only in a
certain class of words and a similar but different sound
regularly occurs elsewhere, then this is complementary
distribution.  And complementary distribution of similar sounds
points to no phonemic distinction between them.  If you can
always predict which sound will be present from the environment,
then the sounds are not different phonemes (in that environment).
Phonemes should tell you nothing about words except that they are
different (i.e., they shouldn't tell you that one word is a noun
and the other is a verb or that one word is a pronoun and the
other is not, or that one is singular and the other plural, etc.).

>Show me contrastive phonological environments.

Sorry, "phonological" wasn't mentioned in my statement, so you
will have to provide your own.

>[RW]

>>As in the comparative method and internal reconstruction,
>>similar items that are in complementary distribution are usually
>>aspects of the same thing.  But believe it or not, linguists will
>>still disagree on the phonemic status of sounds and different
>>analyses may result in different numbers of phonemes claimed for
>>a particular language.

>[PR]

>Apparently, it is fated for you and me to never agree.

Well, so long as you disagree with everything I say, even if you
have to destroy your own arguments to do it, just because I have
said it, this is doubtless true.  But I wouldn't call it fate,
I'd call it a conditioned reflex.  I myself would have phrased it
differently, and would have said that I will agree with you when
you are right, but perhaps this is actually exactly the same
thing that you have said.

>I will state that in private correspondence, a second professional
>linguist has affirmed the non-phonemic status of IE *o.

I will let you answer this one yourself because:

 On Thu, 27 May 1999, Patrick C. Ryan wrote:

 >Whether any given linguist did or did not accept the validity of
 >my studies is not a proof or disproof of my work.

Now here is something that I can agree with you on.  Who agrees
or disagrees with a theory is not proof pro or con.  Who makes a
statement does not affect its validity.  Who proposes or accepts
the theory is immaterial.  It matters not whether the proponent
of a theory is well or poorly educated, is a "professional
linguist" (PL) or a known crackpot (not necessarily different
things), has been convicted of income tax evasion, or wets the
bed.  All that matters is the evidence and the argumentation.

A theory is not automatically wrong because it is proposed by
Patrick Ryan, and it is not automatically correct because it is
proposed by a PL.  So whether a PL agrees with your ideas or not
is irrelevant, as you yourself pointed out on 27 May 1999.
Obviously PL's are not automatically correct in your mind or you
wouldn't be telling PL's that they are wrong on a daily basis.  If
PL's are always right, then citing a PL is a good argument, but
if they are only right when they agree with you and wrong
otherwise, it rather vitiates your appeal to the authority of an
unnamed "professional linguist".  This doesn't mean that you are
wrong though; it just means that it isn't a valid argument.  And
all that matters is the evidence and the argumentation.

Bob Whiting
whiting at cc.helsinki.fi