minimal pairs (was: PIE e/o Ablaut)

[Robert Whiting]

On Fri, 07 Apr, Ross Clark <r.clark at auckland.ac.nz> wrote:

At 09:00 PM 3/30/00 +0300, Robert Whiting wrote:

>>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/).  This is because
>>otherwise the sounds are in complementary distribution, [dh]
>>occuring in voiced environments and in deictic words and
>>pronouns, [th] otherwise.

>I am astonished that this discussion has proceeded for several
>days without anyone questioning the original statement about
>complementary distribution of [th] and [dh] in modern English,
>which is simply incorrect. Even if one does not have the
>pronunciation which makes "either" and "ether" a minimal pair,
>examples of [th] in voiced environments are not at all hard to
>find: pathology, authority, anathema, mathematics, Gothic,
>Arthur, etc etc.

And I am astonished that anyone would present a list of
loanwords, however long, and claim that it has some bearing on
native English phonology.  Loan words do not necessarily follow
the phonological rules of the borrowing language.  In fact this
is usually one of the first clues that a word is a loan when it
doesn't obey the phonological rules.  This is how you can tell
that 'father' is a native (inherited) word and 'padre' is a loan.

I'm sorry if you got confused, but I thought it was clear that I
was speaking about native English words, not borrowings.  Perhaps
I should have been explicit, but I really thought that everyone
knows that when you are trying to establish the phonology of a
language you should deal with words that are native to that
language.  I'm surprised that you didn't include 'Athens' in your
list.  You can make a list of hundreds of words in English that
have [th] in voiced environments and every one of them will be a
loan.  There are a very few examples where the complementary
distribution of [th] and [dh] does break down, but you haven't
mentioned any of them.

Basically, anything that comes from Greek theta is going to be
pronounced [th] in English.  Apparent exceptions like Thomas or
thyme can be accounted for by tracing the path of the word into
English.  But even these apparent exceptions only show that [th]
opposes [t], not that [th] opposes [dh].  Now if you can make a
similar list of words from Greek or Latin/French where original
theta is pronounced [dh] (I expect that 'rhythm' and 'logarithm'
and the similar but unrelated 'algorithm' [but not 'arithmetic']
are special cases because of the -thm#) or you can show a list of
words of Germanic origin where intervocalic <th> is pronounced
[th] then you would have a good point and something to look at.
If not, not.

Now "foreign word" is a perceptual category (just as "phoneme"
is) and it is how the speaker perceives the word that decides
what phonological (and sometimes morphological) rules he can
expect to apply.  Moreover, the perception of whether a word is
foreign or not is likely to change over time (the longer a word
is in the language the more likely it is to eventually be
regarded as non-foreign).  And once a word is considered native,
then it will treated phonologically as a native word.  So perhaps
you are saying that [th] and [dh] are separate phonemes and
intervocalic [th] is used to mark foreign words since all native
words will have [dh] intervocalically.  This does not sound
particularly convincing to me.

Let's look at what happens with words borrowed from languages
that do have /dh/.  Since [dh] allegedly has phonemic status in
English, one would expect that they would be borrowed as [dh]
just as Greek theta is borrowed as [th] (cf. borrowings with /f/
and /v/ below).  But if one looks at Arabic 'dhow', although the
spelling preserves the Arabic /dh/, the pronunciation is with
[d], not [dh].  And so on.  I know of no example where a word
with /dh/ in the original has been borrowed as [dh] into English
despite the frequent preservation of <dh> in the spelling, but if
there are some then I would consider them as tending to indicate
phonemic status for [dh] in English.  But I wouldn't be convinced
without minimal pairs like 'focal' - 'vocal' or 'file' - 'vile'

>[dh] occurs in word-final position in breathe, bathe, writhe, etc

These are morphophonemic variants.  One method of forming verbs
from nouns in English is by voicing a final unvoiced spirant.

   noun (adj.)          verb

  life                 live
  half                 halve
  house [haws]         house [hawz]
  glass                glaze
  grass                graze
  breath               breathe
  bath                 bathe
  cloth                clothe
  wreath               writhe (wreathe)
  teeth (tooth)        teethe
  loath (loth)         loathe

  [An apparent exception is 'tithe' [noun and verb] but this is
  rather a fossilized form than an exception.  The word
  originally meant "tenth" and indeed was identical with the
  contemporary word for "tenth" ('te:odha') with the proper
  intervocalic voiced [dh].  With the marginalization of the word
  as a special kind of "tenth", it dropped out of its word class
  (ordinal numbers) and did not undergo the same changes as the
  rest of the group, which resulted in the levelling of the
  category to a final [th].  Thus the pronunciation with final
  voiced [dh] was preserved in both noun and verb [and is also
  reflected in the spelling.]

  [Another form that falls outside the system is 'smooth' (adj.)
  and 'smoothe' (v.).  Here, again, there is no contrast between
  [th] and [dh]; [dh] simply appears in an unexpected place.  And
  while 'smooth' is ancient in English, its origin is unknown.
  Contrast this with 'sooth' (n.) and 'soothe' (v.).]

Morphohonemic variants are generally not considered distinct
phonemes in that environment.  At most they are considered
morphophonemes and at the least simple allophones because the
distribution of sounds is goverened by rule (in this case a
morphological-phonological rule) and hence the value of the sound
is predictable from its environment.  Morphophonemic alternation
is not sufficient to establish phonemic status (although it is
often a prelude to it).  Morphophonemic variants may very well be
phonemically distinct in other environments, but it is the basic
precept of internal reconstruction that morphophonemic variants
can normally be traced back to some archephoneme in the
pre-language.  This is one reason why unrelated words is usually
a requirement for minimal pairs.

>That these two consonants have undergone a split parallel to that
>of /s/-/z/ and /f/-/v/ in the history of English is hardly
>controversial view

Phonemic status is easy to show for /s/ and /z/, somewhat more
difficult for /f/ and /v/ (but it exists), and very difficult for
[th] and [dh].

If you look at a list of English words that begin with [v] you
will find very few native words ('vat' and 'vixen' are about it).
But both of these are dialect borrowings in standard English
replacing earlier 'fat' (vessel) and 'fyxen' (feminine of 'fox'
with '-en' feminine marker and umlaut).  However, modern 'fat'
("fat", cf. G. 'Fett') and 'vat' ("large vessel", cf. G. 'Fass'
[scharfes s]) are clearly a minimal pair since neither is likely
to be recognized as a foreign word.  On the other hand, 'focal'
and 'vocal' are both recognized as foreign words, but the
distinction /f/-/v/ is sufficient to differentiate them.
Speakers now have to recognize inherent [v] sounds in words to
distinguish them from both inherent [f] sounds and from [v] as a
morphophonemic variant of /f/. Therefore /f/ and /v/ are separate
phonemes (almost a borrowed phonemic distinction, but made
possible by the mophophonemic alternation of [f] and [v]).  As
for 'of' and 'off', this is a stress difference.  'Off' is simply
a lexicalized stressed form of 'of' (this is why 'off of' as in
"get off of the grass" is considered substandard).

>("some would doubtless claim") -- I would be most interested to
>hear of any description of modern English (save perhaps from the
>Baroque Period of SPE abstractionism) in which this is not taken
>as a simple fact.

I dare say that most descriptions of modern English do take this
as a simple fact (just as many dictionaries will tell you that
pronominal determiners are pronouns) and therefore not worth
investigating.  But facts are not data.  Data exists in nature.
Facts are observations about data.  Facts are a matter of
interpretation and facts can be wrong.  Sometimes it is necessary
to go beyond surface appearances to see if the facts are correct
interpretations of the data.  What I want to know is what are the
data on which this "simple fact" of the phonemic status of [th]
and [dh] in English is based.  In this case, it would seem that
[th, dh] just ride along on the coattails of [s, z] and [f, v] in
the belief that if [s, z] and [f, v] split then [th, dh] must
have too.

The creation of the allophones of original /s/, /f/, and /th/
happened for all at the same time:  unvoiced spirants between
voiced sounds following a stressed vowel became voiced.  This did
not increase the number of contrasts (phonemes), merely the number
of allophones (/s/ [s, z]; /f/ [f, v]; /th/ [th, dh]).  The
phonemic splits came later and for different reasons.  First came
morphophonemic alternation followed by, for /s/ and /z/, the use
of /z/ to create expressive and imitative words; for /f/ and /v/,
the need to differentiate borrowed words with inherent [f] and
[v]; and, for [th] and [dh] -- well, I just can't think of
anything that compels the use of [th] and [dh] for making
distinctions.  If you can provide something, I'd be glad to
listen to it.  But it will have to be better than loanwords that
don't contrast with anything else and morphophonemic alternations
or parallelism with [s, z] and [f, v].

But before you get too deeply involved in trying to find
something, consider this also simple fact:  If it is not possible
for English speakers to determine the pronunciation of <th>
as [th] or [dh] entirely by rule, how is it possible for the
graphemic system to get by with only one grapheme for the two
sounds?  Now English does not have the world's greatest fit
between writing and sounds.  The same sounds can be written with
different characters and the same characters can be used to write
different sounds.  In some cases, there is no way to tell how
certain written combinations are to be spoken.  There is no rule
to tell you how a word like 'cough' is to be pronounced.  One
simply has to learn the pronunciation with the word.  But with
[th] and [dh] this is not necessary, even though there is no clue
in the writing (with the exception of <-e> in verbs which marks
the fact that the preceding <th> is voiced).  Otherwise one would
have to learn the pronunciation of every word containing <th>
separately.

When dealing with phonemes, if you know the pronunciation, you
can distinguish a word from all other words with similar
sounds except for the different phonemes.  Phonemes distinguish
between words when there is no other criterion for distinguishing
them and are completely arbitrary.  Thus there is no rule
(phonological, morphological, or syntactical) for distinguishing
between /fat/ and /bat/ except the rule that says that /f/ and
/b/ are different phonemes.  The phonemes /f/ and /b/ don't tell
you anthing about these words except that they are different.  If
/f/ were always used only in verbs and /b/ were always used only
in nouns, then one would have to take another look at the
phonemic status of /f/ and /b/.

When dealing with [th] and [dh] if you recogize the word
category, you will know what the sound is.  You don't have to
learn the pronunciation with the word except in a very few cases
(like 'rhythm').  Otherwise, even for words that you may not be
familiar with like 'thole' or 'wether' or 'heterochthonous' you
will know whether [th] or [dh] is correct.  Words such as
'blithe' which can be pronounced with either [th] or [dh] (not as
a matter of stress, but simply free variation) do not speak
strongly in favor of phonemic status for these two sounds
(athough it doesn't necessarily speak against it either).

So once again, the only native words where [th] and [dh] contrast
is 'thigh' and 'thy'.  'Thy' belongs to a class of words that
always has [dh] in this position.  'Thigh' belongs to a class of
words (not really a significant class, merely the complement of
the other class) that always has [th] in this position.  The
apparent opposition between [th] and [dh] in this example is just
a historical accident (in much the same way that homynyms come
into existence through historical accidents) since the original
distinction between these words did not depend on this
opposition.  Thus this "minimal pair" is not adequate to establish
[th] and [dh] as separate phonemes in my opinion since
distribution of sounds by rule is a more important criterion even
if the rules are not phonological.

And again, if you have evidence for the phonemic status of [th]
and [dh] in English (and I don't mean evidence that they are
recognized as different sounds [phones] or that they are
morphophonemic variants), I will be happy to evaluate it.  I just
haven't found any that I consider convincing despite the fact that
most grammars will list them as separate phonemes and put forth
such "minimal pairs" as 'thigh' - 'thy' or 'ether' - 'either' or
'wreath' - 'wreathe' as evidence.

Now there is no reason why [th] and [dh] can't be separate
phonemes in English.  They might very well be.  Or they might
well be on their way to becoming separate phonemes.  But if they
are, one can reasonably expect there to be some evidence of it.
If [th] and [dh] are different phonemes in English, then it is an
unused phonemic distinction.  And since it is unused, it can't be
proved that they aren't phonemically distinct just because there
is no evidenc that they are.  What can be shown is that [th, dh]
are allophones of /th/ that resulted from the original voicing of
spirants in a voiced environment and that these allophones can be
used as morphophonemic variants.  But without unequivocal
evidence that they are used as separate phonemes, it is safer to
assume that they are not.

In summary, if you find [dh] in initial position in a native
English word it tells you that the word is a pronoun or a deictic
word (this, that, thou, then, there).  If you find [th] in a
voiced environment in English this screams that the word is a
loan ([insert here list of loan words given above]).  If you find
[dh] in final position it tells you that the word is part of a
noun-verb or a singular-plural pair.  What is needed to show that
[th] and [dh] are distinct phonemes is a clear example where they
mark an arbitrary distinction in a non-contrastive environment
that is completely independent of the environment or any rule.

Bob Whiting
whiting at cc.helsinki.fi