Re: BBC does it again

Brian M. Scott wrote:
On Wed, 11 Jun 2008 12:05:23 -0400, Harlan Messinger
<hmessinger.removethis@xxxxxxxxxxx> wrote in
<news:6babe2F3a67grU1@xxxxxxxxxxxxxxxxxx> in sci.lang:

John Atkinson wrote:

"Harlan Messinger" <hmessinger.removethis@xxxxxxxxxxx> wrote...

Brian M. Scott wrote:

Harlan Messinger <hmessinger.removethis@xxxxxxxxxxx> wrote:

Brian M. Scott wrote:

Harlan Messinger wrote:

[...]

It does seem (to me) reasonable to suggest that whatever
phonemic representation one chooses to use ought to
express the same thing as the phonemic model the same
person uses to describe a language. [...]

I don't agree,

Why? Why *would* one use a representation that doesn't
correspond to the model? [...]

John's given one reason. Another is to avoid throwing away
information that may be useful, either to someone else or to
me on another occasion. I'm not much interested in abstract
phonology; I *am* interested in useful description.

Well--the phonetic representation conveys *that*!

Quite possibly not: it may convey far too much detail.

I'm talking about the phonemic representation, so by
definition we're talking about the abstract phonology.

There is no reason to pretend while doing so that nothing
else exists.

Are you saying that a phonemic representation isn't "useful description", or anyway that it's is less useful than a phonetic one?

No, I'm saying it serves a different purpose. Since the
whole point is that one phoneme can be realized in a
given language as all kinds of sounds with different
phonetic representations,

No. In principle it would be possible to have a 1-1 match
between phonemes and phones (albeit probably not in a human
language).

How does noting that in theory (and, moreover) not in fact can there be a pure one-to-one match contradict my (correct) observation that phonemes frequently *don't* correspond to a single sound?


there isn't any constraint calling for the phonemic
representation to look like any one of the phonetic
ones, or like any of them at all,

There is: it's a usability constraint.

In my opinion it goes the other way. Unless told otherwise, I infer from seeing word samples phonemically transcribed with /a/ and /A:/ that /a:/ and/or /A/ also exist--that there's more than a two-state variation.


especially if attempting to copy a phonetic
representation hides a detail of the phonemic model that
it would be useful to convey.

I'm not talking about hiding anything; I'm talking about
showing a little *more*.

While at the same time obscuring something that's important at the level of the abstract phonology.
.

Relevant Pages

  • Re: BBC does it again
    ... One reason might be so that everyone's notation for a given language variety is the same, independent of what model they espouse. ... I understand that some people think length is the distinctive feature in those pairs of Dutch vowels, others claim it's lax vs tense, or amount of openness and/or frontness. ... Most linguists involved with writing Dutch phonemically aren't super-interested in which model is "correct", but they would find it convenient if there was one accepted notation for the phonemes of Dutch. ... distinguished by shape as well as by length as in Ruud's representation of Dutch sounds, so I accept that this is done, and that perhaps the point is that it's a model used *consistently* for Dutch--and it certainly is one more demonstration of the unfounded nature of Peter's contention that this is an unused practice and one that won't be understood. ...
    (sci.lang)
  • Re: BBC does it again
    ... One reason might be so that everyone's notation for a given language variety is the same, independent of what model they espouse. ... I understand that some people think length is the distinctive feature in those pairs of Dutch vowels, others claim it's lax vs tense, or amount of openness and/or frontness. ... Most linguists involved with writing Dutch phonemically aren't super-interested in which model is "correct", but they would find it convenient if there was one accepted notation for the phonemes of Dutch. ... distinguished by shape as well as by length as in Ruud's representation of Dutch sounds, so I accept that this is done, and that perhaps the point is that it's a model used *consistently* for Dutch--and it certainly is one more demonstration of the unfounded nature of Peter's contention that this is an unused practice and one that won't be understood. ...
    (sci.lang)
  • Re: All languages are equally fit
    ... individual phonemes are not regularly ... in Chinese writing. ... representation of any typical Chinese text? ... I was more concerned with intent rather than nit-picking word choice. ...
    (sci.lang)
  • Re: BBC does it again
    ... with different phonetic representations, ... between phonemes and phones (albeit probably not in a human ... representation to look like any one of the phonetic ...
    (sci.lang)
  • Re: BBC does it again
    ... don't restrict the scope of your phonemic model unnecessarily. ... use /eI/ for English "long E", then you're limiting the utility of your ... Englishes the same phonemic system with different phonological rules, ... Still, when two varieties have the same number of similar phonemes, it does make sense to use the same phonemic representation for both, for convenience if nothing else. ...
    (sci.lang)
Loading