Message-Id: <199511211833.NAA21435@locke.ccil.org>
Date:         Tue, 21 Nov 1995 18:17:09 +0000
Reply-To: ucleaar <ucleaar@UCL.AC.UK>
Sender: Lojban list <LOJBAN@CUVMB.BITNET>
From: ucleaar <ucleaar@UCL.AC.UK>
Subject:      Re: Buffer and Vowel phonology
To: John Cowan <cowan@LOCKE.CCIL.ORG>
Status: OR

Lojbab:
> > The question is: is it better
> >for the language that the phonology be as clear and simple as possible
> >for speaker and hearer or that anything goes?
> Yes. %^)
> Since clear and simple are relative terms, I am free to conmtend that
> what some speakers and listeners would find clearest and simplest would
> be one in which "anything goes" assuming you mean that we tolersate
> inexact values for some phonemes.

"Anything goes" was my caricature of your version of the phonology:
"realize the phonemes however you like, so long as you keep them
distinct from one another". I think we can all agree that in principle
speakers should find Anything-Goes relatively simple and listeners
shd find A-G relatively difficult.

> >If everyone has the same centre/norm, then they probably have much the
> >same map. But why will you not prescribe the centre/norm for the buffer
> >(which was one of the options John suggested)?
> Because  I don't speak a significantly biffered dialect, and until we
> have some such speakers we won't be able to find out which they prefer as
> "simplest" and clearest.

I see no rhyme or reason to which bits of the lg design are done in
advance and which bits are left until lots of people are using it.

I used to "speak" (not easy when there's noone to speak to) a somewhat
buffered form, but I got fed up with the [y] [Y] I was using - it seemed
too marked a phonological object for such a function - so I then
merged my buffer and /y/ phonemes.

> >Is that still allowed? I thought that licence had been rescinded! And
> >there's me doing [h]s when I cd have been doing voiceless bilabial
> >trills!
> I think someone just recently mentioned using unvoiced /th/.
> I would truly like to hear Lojban spoken with your suggested alternate
> buffer.

By "buffer" you here mean /'/? You want to hear me using voiceless [B]
instead of [h]? Okay.

> >If the bv has phonological existence then presumably it is a lojban
> >sound. And if that sound is required to be shorter than other vowels
> >then length is a phonological feature.
> I thought you said that a burp has phological existence but does not map
> to a phoneme.

No. I said "But obviously, if you hear a *burp*, you don't take it to be
a /u/ or whatever". That doesn't say that burps have phonological
existence. They don't.

> >> any consonant phonologically happens in an environment surrounded by
> >> "vowels" of some shortness in any consonant cluster.
> >That's not true of English (at least not in any theory I know) but I
> >had thought until recent discussions that it is true of lojban.
> I think it is true for all languages.  isn't aspiration a "vowel sound"
> for example?

Neither phonetically nor phonologically. What you are saying is that
no language has consonant clusters, phonologically. That is a theoretical
claim, but I think all phonologists would reject it. It is demonstrably
false that phonetically there are no consonant clusters.

> >> I left the word "vowel" out on purpose, but perhaps should have left
> >> the word phoneme out.  The buffer is a vowel sound, but has no phonemic
> >> significance. In effect, its existence serves more to define a limit on
> >> what vowel sounds among possible Lojban ones DO map to phonemes, and
> >> which are simply noises that accompany consonants.
> >If someone other than you thinks this makes sense, they should join in
> >and help get me to see that.
> Trying again - the buffer is a phone that can be heard by Lojbanists, but
> is not mapped to any phoneme in the formal sense of the difinition of
> phoneme which requires a miniml pair to exist.  By definition, there can
> BE no minimal pair that differs only on the presence or absence of a
> buffer vowel.

You have the formal definition wrong. To see why, consider the following
example. suppose in English /N/ occurred only syllable-finally. Then,
it would not be the case that because there is no contrast between
the presence and absence of /N/ - e.g. /baN/ v. */ba/ there is therefore
no /N/ phoneme. The proof that there is an /N/ phoneme is that there
are a range of minimal pairs like "bang" /baN/ v. "back" /bak/, etc.
So by the minimal pair test it is easy to show that there is a buffer
phoneme - e.g. [kalama] [kelama] [kilama] [kolama] [kulama]
[k@lama] [k%lama] (where % is a certain vowel of unclear identity -
[I-], say) all contrast.

> >But the acoustic space between [i] and [u] - i.e. around [y] - is
> >relatively unoccupied, by this standard. And I believe that cross-
> >linguistically, phonemic distinctions near the centre are rarer than
> >distinctions around the periphery, and the same goes for phoneticians'
> >perceptual abilities.
> True, but distinctions of only 2 central vowel sounds is not that
> uncommon.

Could you (or anyone else) give me a sample of such vowel systems
& tell me which lgs they belong to? [I don't have this info on my
bookshelves.]

> >True. I think the normal practise is to hold extensive trials. If
> >Lojban was to do that, there'd be a big redesign after the lg has seen
> >a number of years of heavy use. But at that point everyone who'd
> >learnt it wouldn't tolerate a redesign.
> If it is truly broken in some area, undoubtedly someone would lead an
> effort to creeae a new version correcting that error.

I doubt that. The lg is clearly not so broken that it is unusable, &
that's what counts in impelling reform.

> >> > > That emphatically means that we do NOT want schwa used as a buffer.
> >> >Why not, so long as /y/ maps to something different from the buffer?
> >> Well, I would rather have one non-standard vowel than two.
> >? Which are the one or two non-standard vowels, and which standard to
> >they deviate from?
> What I am saying is that I would aaccept non-standardization of the
> buffer vowel, but not of the hyphen value.

You don't just accept the non-standardization of the buffer vowel - you
insist on it. And you accept the non-standardization of all vowel
phonemes, since by leaving the buffer undefined all other vowels
remain partly undefined.

---
And