A discourse on Phonemics (was: Re: E and e (was: A break in the evils of English (or, Sturnan is beautiful))
|From:||Roger Mills <romilly@...>|
|Date:||Wednesday, May 1, 2002, 23:18|
>On Wed, 2002-05-01 at 16:04, Raymond Brown wrote:
>> But the phonemic theory (and _theory_ it is) is primarily concerned with
>> contrasts and distribution of sounds.
And it's well to keep in mind too that phonemes are something of an
abstraction. One could just as well use funny symbols or numbers to
represent them-- say, / 1 / instead of /i/ and /2/ for /I/; then you would
need a list of rules stating how /1/, /2/ are realized in their various
environments. (But all would agree that's unnecessarily abstract, though
you still need the list of rules to describe "/i/" et al.)
And keep in mind the definition of "phoneme"-- the minimal unit of sound
that distinguishes meaning. So Engl. "beat, bit, bait, bet, bat, boot, boat,
bought, but and (rare) bot(-fly)" are distinguished in meaning only by their
vowel sounds; ergo, those vowels are phonemic.
But the fact is that English distinguishes two types of vowels at
high/high-ish position (i,I, u,U), and two types at mid position (e,E,--
though o,O are another matter). As Ray says, phonemics is concerned with
how these contrast and are distributed in the system; and very little
concerned with their actual phonetic values-- that's a matter of production,
and certainly varies individually, and can vary (considerably) regionally.
Phonemics as developed in the 30s/40s was frankly (arrogantly,
optimistically) based on Standard American and RP speech. (Sorry to say,
American linguists at that time were barely aware of Australian speech, and
the Brits don't seem to have paid it much attention either.) It turned out
that with only minor tinkering with the rules of production, the same system
could describe both varieties-- presence or absence of [r] being one of the
problems-- but SA "beer" [bIr] vs. r-less and RP [bI@] could still be
phonemicized /bir/ (in the iy:i system; /bIr/ in the i/I system), since the
/-r/ comes back in phrasal "beer is.."; similarly with the diphthongs /ay,
aw/, where the /a/ is [a] or [@] depending on dialect-- it's still phonemic
/a/. Generally speaking, the standard phonemic system could even describe
Southern US, though with a lot more tinkering-- and southern speakers tended
not to appreciate this "Yankee" idea.
>> Maybe so. But phonemically /i:/ ~ /I/ surely must suggest two components
>> of contrast: hight of tongue & length, i.e. it implies four phonemes /i/~
>> /i:/ ~ /I/ ~ /I:/, which is not correct.Tristan:
>No, it implies that both length and quality are distinguishing factors
>in /i:/ ~ /I/.Yes, and that's not true. Personally I've always considered the /i:/ vs /i/
or /I/ system deceptive, for that reason; it simply substitutes the length
mark for the glide; they both meant only that the higher vowel differed from
the lower vowel _in some way_ -- perhaps this is wehre the tense/lax term
came into being. And the vowel of "beat" ?/bi:t/ [bit] is in fact
phonetically shorter than that of "bid" [bI:d].
>> While phonetically the difference
>> may be [i:] ~ [I] (tho some English dialects tend to diphthongize thefirst
>> as [ij] or [Ij]), phonemically we have, as I see it, to choose between/ij/
>> ~ /i/, /i/ ~ /I/ or even, possibly, /i:/ ~ /i/.Tristan:
>Okay, maybe you have to look at it from my point of view. Which may be
>entirely flawed, but still. [i:] is about as long as the [A:] in
>'heart'. The vowel in 'heart' is distinguished from the vowel in 'hut'
>by one thing: length.
Not in quality too?? hut [hat] vs. heart [hA:t]?? Just guessing. If so, then
phonemically you are only contrasting [low central] with [low back], with
the length of the latter being predictable (compensatory length due to loss
of the _r_). But if not, if you truly have [hAt] vs. [hA:t], then there has
been a merger, and a real change in your phonemic system.
Standard US is [hVt] (IPA inverted-v) vs. [hart], maybe [hArt], r-less
[ha:t] or [hA:t]--( frankly I'm vague on the US contrast a:A; I'm not sure I
have it, or hear it well. [hat] or [hAt] for me is "hot").
>(This argument here might bring my entire argument down if I'm wrong. If
>I am, I'll admit it.) Okay, so we've established we want to do /i:/ or
>/ij/ versus blurp. According to your above examples, We must now choose
>/i/. But the first part of the diphthong /I@/, in such words as 'beer',
>has the same quality as that of 'bit' (at least to my ear). There exists
>an allophone of /I@/ in contexts like 'beer is': [I:].
Aha-- you've lost /r/ even in "beer is..."?? Well, not totally lost-- it's
been replaced with the glide [@] (or length in the case of 'heart'). There
are two problems going on here: First, the neutralization of vowel
contrasts before /r/ in all (?) varieties of Engl.-- in that case the choice
of phonemic symbol is arbitrary, though most would use "phonetic similarity"
as the criterion, and in this case the vowel before /r/ is "closer"to [I]
/I/ or /i/ depending on system than to [i] /iy ~i:/, though phonetically it
seems to be somewhere between [i] and [I]. And second, the subsequent
replacement ("loss") of /r/ with a glide-- from a strictly phonemic POV,
however, there is _still_ something there, whatever it's called.
(You may recall, a while back we went round and round on the nature of Engl.
vowels before final /N/, where neutralization also occurs. Some opted for
tense /siyN/, others for lax /siN/ 'sing'; some for /hEN/, others /h&N/
'hang', etc.-- phonetically these are either "lowered [i]" or "raised [I]"
for ex., though everyone agreed there was no contrast /siyN/ ~/siN/ i.e.2
words with different meanings.)
So we could say "In Aust. Engl., /r/ is realized as a glide [@] in certain
(specifiable) environments, as vowel length in certain others". A purely
Australo-centric analysis (ignoring other varieties of Engl.) however, might
analyze/symbolize it as something else. That would be valid; but the analyst
might still want to mention how it relates to /r/ in the other varieties.
(IIRC some Southern US dialects replace /r/ with a glottal stop in the
pre-vocalic environment-- [bI:?iz] for 'beer is...'-- a somewhat stigmatized
practice I think.)
>What alternatives are there to phonemic theory? Phonemic theory seems
>pretty well accepted; in my relatively limited research (i.e. what I can
>do in my free time on the Net when it's of interest to me), I don't
>believe I've ever seen one.
There's distinctive feature analysis, which I think is not much used
anymore, though it has its points. Then there's Optimality Theory (google
for that)-- but even it needs some way to symbolize the systematic contrasts
of Engl. or any other language. And there are no doubt others, with which
I'm not familiar.
As I said, phonemics proved able to provide a unified explanation of
Standard US and RP speech; it probably would not work for a highly divergent
dialect (Yorkshire perhaps?) where there may have been mergers in the vowel
system (a real change in the phonemic system), or really odd phonetic
realizations of phonemes. It's possible some varieties of Australian are
heading off in that direction, but the fact that other Engl. speakers still
understand most of you suggests you basically retain the same set of
contrasts as the rest of us. Ergo, Australian Engl. can still be described
as _phonemically_ the same as US/RP. Maybe! :-)
Consider Indian English, where they consistently substitute their retroflex
/t., d./ for
our alveolar /t,d/; that's irrelevant to a phonemic description of Indian
Or consider Spanish, whose most widely spoken dialects can be described with
a single phonemic system. You need only adjust for /T/ ~/s/ and /L/ ~/y/.
>If I'm wrong, do tell.You are not wrong. It's just that Australian Engl. is a rather exotic beast
to some of us....and we don't hear a lot of it (save for that strange man
who mauls crocodiles on TV).
>Always tell. I'm never trying to be stubborn, I'm
>trying to help you help me.)
I hope so. By replying just to certain aspects, it's turned out to be sort
of a hit-and-miss discussion. Entire books have been written on the