Message-Id: <9201301821.AA17439@cunixf.cc.columbia.edu>
Date:         Thu, 30 Jan 1992 12:29:00 EST
Reply-To: "61510::GILSON" <cbmvax!uunet!cuvmb.cc.columbia.edu!gilson!61510.decnet>
Sender: Lojban list <cbmvax!uunet!cuvmb.cc.columbia.edu!LOJBAN>
From: "61510::GILSON" <cbmvax!uunet!cuvmb.cc.columbia.edu!gilson!61510.decnet>
Subject:      Allophones of zero in Lojban
To: John Cowan <cowan@snark.thyrsus.com>
Status: RO

And Rosta <ucleaar%UCL.AC.UK@cuvmb.cc.columbia.edu> writes:

>As I have pointed out before, without incurring dissent or contradiction
>from the gurus, Lojban is in fact a CV language.

>According to the current phonological analysis of Lojban, it is permissible
>to insert The Buffer Vowel into the phonological string, as long as it is
>not inserted such that it is adjacent to a vowel (/i e a o u y/). (I would
>count /./ as a consonant - realization = glottal stop.)

>Another way of looking at this is to say that Buffer Vowel slots are
>already present on the skeletal tier. These slots may be left empty or
>filled with the buffer vowel. So, _mlatu_, for instance, looks like this:

>     C V C V C V
>     |   | | | |
>     m   l a t u

>The empty V can be filled with the buffer vowel or left empty.

>Consequently, Lojban has a syllable structure no more complex than the
>simplest syllable structure attested in natural language (i.e. no language
>lacks CV syllables, but some languages have nothing but CV syllables).

>Lojban is to be commended for the simplicity of its syllable structure,
>but not, alas, for the phonology of its buffer vowel. The buffer vowel
>is defined as "any vowel that isn't /a e i o u y/ in the accent of the
>speaker". Therefore, your buffer vowel may be located in the vowel space
>I assign to a e i o u or y. John Cowan, for example, uses [I] for his
>buffer vowel, which would fall into my /i/ space. So if he said
>[mIlatu] I'd hear /milatu/.

From a typological perspective, one would expect the buffer vowel to
>be unrounded, central and middish - i.e. in the space of /y/. I've
>advocated the adoption of [@] as buffer vowel, but this would require
>either altering the vowel space assigned to /y/ (e.g. giving it [y])
>or removing the morphological functions that _y_ performs. The latter
>approach seems to me the best: although I am not sure of this, and,
>indeed am probably wrong, I believe that _y_ could be replaced by
>_r_, _l_ or _n_ (or something like that - I don't have the documents
>handy).

And's feeling about this coincides with what I have long felt, though I
would use a little different terminology. In Lojban, The Buffer Vowel is
treated as an allophone of zero. Now, there are languages with allophones
of zero. French is one. The vowel in French is one that most books I see
characterize as schwa, but when I said thus, Pierre Savoie objected, saying
that to his ear it was a weakened version of the vowel of "boeuf." In any
case, that vowel is the closest thing to a weak central vowel that French
has, and as I said, most French phonology books I've seen use schwa to
represent it.

Lojban's use of schwa (the weakest, least audible vowel) for a non-zero
(written y) while allowing a stronger, more audible vowel (and, in the
usage of Lojbanists I've met, one relatively close to the Lojbanic vowel
i in sound) as an allophone of zero is quite counter-intuitive. However,
I suppose one can say that Lojbanis have no great desire to avoid counter-
intuitive structures, so that this doesn't matter to them. All I can respond
is that _I_ think And's ideas are sensible here.

                                                      Bruce