Return of the minimal pairs; credit where credit is due

[Robert Whiting]

[ moderator edited ]

[Just when you thought it was safe, the ugly spectre of minimal pairs
rears its head again.  I apologize for being so tardy in getting to this,
but since the end of last year I have not been able to keep up with the
list.  I still have not caught up with the past month's postings, but
about all I have found out from the rest of the year's postings is that
Larry Trask is really Larry the Etruscan. :)]

Over a year ago, on Thu, 30 Mar 2000 21:00:45 +0300 (EET DST) I posted a
message in which I cited the minimal pair 'thigh' / 'thy' and then said:

 Most people would not insist on phonemic status for both [th] and [dh] in
 English on the basis of this minimal pair ...

I considered this a fairly unremarkable statement, since it is obvious,
even by the rules of classical phonology, that this is not a phonemic
contrast.  (I will post a couple of very long messages in the next few
days, replies to messages posted last November that will demonstrate
this.)

I was somewhat surprised, therefore, when this was met with a storm of
protest of the type "Of course [T] and [D] are phonemes in English" since
I had never said that they were not.

But then on Tue, 25 Apr 2000 John McLaughlin posted the following
concerning my original post:

 What got my goat in his first post was the comment that (not quoting
 directly), "No one doubts that Modern English [th] and [dh] represent
 allophones of the same phoneme."

Now I have known John for quite some time as a participant in various
lists, and he is not given to flights of fancy and does not often
misconstrue things (at least not that badly :>).  Since John's perception
of what I had said was nowhere close to what I had actually said, I began
to wonder how he could have gone so far astray.

So I searched through the archives for a possible explanation and found
the following message posted by Peter Gray (I quote the entire message
as found in the archives):


-----------------------------------------------------------------------
  Date:         Mon, 3 Apr 2000 20:42:55 +0100
  Reply-To:     Indo-European at xkl.com
  Sender:       The INDO-EUROPEAN mailing list <indo-european at xkl.com>
  From:         petegray <petegray at btinternet.com>
  Subject:      Re: minimal pairs (was: PIE e/o Ablaut)
  Comments: To: Indo-European at xkl.com
  Content-Type: text/plain; charset="iso-8859-1"

  >> Most people would not insist on phonemic status for both [th] and [dh] in
  >> English

  Isn't there another minimal pair in   ether :  either  (at least in some
  dialects)?

  Peter
-----------------------------------------------------------------------

Now Peter did not refer to my posting in his message (in fact that seems
to be sort of a trademark of his), but I don't believe that anyone was
in any doubt about where the statement came from.  I didn't pay any
attention to the message at the time because it had nothing to do with
what I was talking about, but since the crucial qualification of my
original statement was eliminated without even an ellipsis (another
trademark of Peter's), it made it appear that I was claiming something
that I was not.  For those of you who do grammar, the lack of a comma
after "English" shows that what follows is restrictive, i.e., it is
essential to the meaning of the main clause, not merely a modifier.

Now Peter may or may not have misunderstood the intent of my original
posting; it really doesn't matter, because his truncated statement
surely misrepresented it.  If Peter wanted to discuss whether [T] and
[D] are phonemes in English that was fine with me, but it had nothing
to do with the point I was trying to make.

But this explained John's misperception of my statements as well as the
wild rash of postings offering to prove the phonemicity of [T] and [D]
and suggesting that anyone who didn't believe in it was a few bricks
short of a load, and I'm afraid that I replied to the misstatements of
fact in these in much the same vein.  Most of these were, however, based
on a misunderstanding of what my claim was and so there is no point in
pursuing them on that basis (although I may respond to some of them for
other points when I have time).  The misunderstanding was caused by a
careless quotation out of context that I did not notice or correct at
the time.

This misquotation was regrettable in that it completely obscured the
point that I was trying to make, and forced the discussion into a
completely unproductive direction.  Particularly, the little contretemps
with John was regrettable because he actually had evidence of exactly
the point that I was trying to make, but this got lost in the discussion
of points that ware not at issue (or shouldn't have been).

To make it clear what my statement was about:  My entire point was that
[T] and [D] do not contrast in initial position in English.  My
statement was not about whether [T] and [D] are phonemes in English or
not.  The two are not necessarily related ([T] and [D] can still be
phonemes even if they don't contrast in some environments).

As I said, I don't consider this statement to be particularly
remarkable.  I had rather thought it was common knowledge (silly me;
perhaps there is no such thing as common knowledge in linguistics, or at
least not in phonology).  A good summary of the situation can be had
from Edward Finegan, "English" in Bernard Comrie (ed.), _The World's
Major Languages_ (London & Sydney: Croom Helm 1987), 77-109:

  Several notable differences between the consonant systems of Old
  English ... and Modern English can be mentioned.  The members of the
  three Modern English voiced and voiceless fricative pairs (/f/-/v/,
  /<theta>/-/<edh>/, /s/-/z/) were allophones of single phonemes in Old
  English, the voiced phones occuring between other voiced sounds, the
  voiceless phones occurring initially, finally and in clusters with
  voiceless obstruents.  Relics of the Old English allophonic
  distribution remain in the morphophonemic alternants _wife/wives_,
  _breath/breathe_ and _house/houses_, where the second word in each
  pair, disyllabic in Old English, voiced the intervocalic fricative.
  Significantly, initial /<edh>/ in Modern English is limited to the
  function words _the_, _this_, _that_, _these_, _those_, _they_ and
  _them_, _there_ and _then_, _thus_, _thence_, _though_ and _thither_,
  with initial voiceless /<theta>/ in Old English later becoming voiced
  by assimilation when unstressed, as these words often are.  Similarly,
  /<theta>/ does not occur medially in any native words, though it can
  be found in borrowings.  During the Middle English period, with the
  baring of the voiced phones word-finally when the syncreted
  inflections disappeared, the allophones achieved phonemic status,
  contrasting in most environments; there may have also been some
  Anglo-Norman influence, though not so much as is sometimes claimed.
  (pp. 90-91)

I find this to be an adequate summary of the situation, although it
skips over the difficult bits (but in a general work like this with
limited space and restrictions on the technical level of the discussion
one has to).  In general, it is descriptively adequate but not
necessarily explanatorily adequate.  Finegan has left himself an out,
however, when he says "contrasting in most environments."  A little
thought will show that lack of contrasts can only refer to [T D] since
[s z] and [f v] clearly contrast in all environments (initial, medial,
and final).  Thus room is left for a lack of contrast of [T D] in
initial position.  But, again, considering the type of publication, I
would have thought that this information was common knowledge, not some
outpost of hocus-pocus linguistics.

A. Sommerstein (_Modern Phonology_, Theoretical Linguistics 2 [London
1977]), in doing a classical phonological analysis of English (as an
example), puts the situation this way:

  There is no doubt that /<theta>/ and /<edh>/ contrast phonemically;
  but one might nevertheless feel that such a statement conceals an
  important fact.  For /<edh>/ occurs in initial position only in
  _grammatical_ morphemes:  the archaic second person pronoun _thou_
  (_thee_, _thy_, _thine_), the definite article, the root of a
  demonstrative (_that_, _there_, _then_, etc.):  /<theta>/, on the
  other hand, occurs initially only in _lexical_ morphemes.  This is
  complementary distribution, but not of a kind that a classical
  phonemic analysis can recognize; nor could the regularity be stated in
  the morphophonemic rules, since no alternation is involved.  If the
  regularity is to be stated at all, it must be as part of a set of
  principles governing the phonemic makeup of morphemes.

There are two points where this statement is at variance with my
observations.  First, no one (else) on this list seems to feel that
a statement about the phonemicity of [T] and [D] hides any facts at all,
important or not.  Finegan (cited above) points out the distributional
peculiarities of English [T] and [D], but these seem to be generally
unknown or, if known, considered unimportant, while Finegan considers
them "significant" (but without committing himself to why).

The second point is that, while it is true that the distribution of
initial [T D] in English is dependent on the morphemic structure of the
words involved, it is quite possible to capture this generalization
using the rules of classical phonology, so long as morphological
conditions to the extent of morpheme boundaries are permitted as
phonological conditions.  It is actually quite easy by using a
morphemic analysis similar to that already done by Z. Harris, _Methods
in Structural Linguistics_ (Chicago 1951), 192-93 or G. Trager,
_Language and Languages_, (1972), 76-79, and I am mildly surprised that
it does not seem to be obvious.  As I threatened earlier, I will have
a long, tedious, and doubtless boring posting on this (um der Moderator
es willen) very soon.

I know this kind of thing takes the fun out of the list which functions
best with lightning banter on which no great thought is expended <joke>,
but I haven't had a chance to do anything for 5 months, so please bear
with me.

Bob Whiting
whiting at cc.helsinki.fi