Re: phonology of Plan B

Hallo!

On Fri, 13 Jul 2007 08:19:54 +0100, R A Brown wrote:

> Jörg Rhiemeier wrote:
> > Hallo!
> >
> > On Thu, 12 Jul 2007 10:32:30 +0100, R A Brown wrote:
> [snip]
>
> [Greek accusative singular ending -a ~ -n]
>
> > It is just that this particular morpheme has two completely different
> > *allomorphs* (diachronically, of course, they are connected, both
> > descending from PIE *-m by regular sound changes).
>
> Also correct. Yes, both are derived from PIE /m/. After a vowel the /m/
> was not syllabic and, in ancient Greek, became /n/ in final position;
> but after a consonant the PIE /m/ was syllabic, i.e. [m=], and both
> syllabic [n=] and [m=] regularly became /a/ in ancient Greek.
Exactly.  That's what I found in the handbooks I read, too.

> I have genuinely no idea how this is viewed in generative phonology, but
>   in ancient (and modern) Greek itself, /a/ and /n/ are certainly
> distinct phonemes.
I don't know how generative phonology treats the matter, either,
but AFAIK generative phonology is not concerned with diachronics,
and I doubt that the generative phonologists are crazy enough to
assume a single phoneme here!

> [snip]
> >>I still hold that it is risible to analyze Plan B as having 16 phonemes.
> >
> >
> > I wouldn't say "risible"; I'd rather say "misleading".
>
> You are kinder than me   :)
>
> It really does seem to me that if one arrives at a conclusion that has
> such phonetically disparate pairs of allophones as[E]~[b],  [ej]~[S],
> [I]~[d], [r\E]~[D], [r\I]~[n], [r\ej]~[m] etc., one should question
> whether what one is abstracting are phonemes at all.
Yes.  Such drastic alternations certainly go way beyond allophony,
and it ought to be evident that there is something else than phonemes
in play here.

> > And's analysis
> > is not entirely ludicrous; it's just that the term "phoneme" is poorly
> > chosen.  In *one single aspect*, namely the role of fundamental building
> > blocks of the language, the 16 bit quartets *are* "phonemes",
>
> On this I must disagree. This is giving "phoneme" a meaning which it
> does not normally have.
Note that I put the word "phonemes" in quotes.  I wanted to say that
*in one single aspect*, the 16 bit quartets are what phonemes are in
natural languages, namely the basic building blocks of the language.
I apologize for the misunderstanding.

>       A "phoneme" is normally understood to mean the
> minimal unit in the sound system of a language.
Of course!

>       I believe the term
> 'phoneme' is, by extension, used to denote minimal units in signing
> languages (tho I may be wrong on this); if so, those using the term in
> that context understand what is meant.
Yes.  I have heard that, too.  When discussing signed languages,
this does not cause confusion, because signed languages usually
do not have a phonetic manifestation, and it is understood that
the word "phoneme" refers to gestural elements that play the same
role as phonemes in spoken languages.

> But And, by quoting allophones in square brackets, has been using the
> term to refer to _sounds_, as such one would surely understand the term
> in its traditional sense.
Yes.  And sounds and bit quartets are entirely different things!

> Neither you, nor I nor And at all in disagreement about the 16
> bit-quartets (tho arguably one might hold that the bit itself, 0~1, is
> the _minimal_ underlying unit).
That's a valid analysis as well, especially given the way the length
of a morpheme is calculated from the number of consecutive "1" bits
at the head of it.  Does Plan B have just two phonemes, then?  No,
because bits don't live in the human phonetic space :)

>       None of us disagrees that these quartets
> are fundamental to the language.
>
> The disagreement, which shows no signs of being resolves, is in the
> analysis of the spoken representation that Jeff Prothero happen to give
> to these quartets in his paper of May, 1990.
Yes.  And calls them "phonemes"; you and I do not.

> I have more than once point out that Jeff also wrote: "the particular
> letters and pronunciations chosen don't matter much." This statement in
> itself ought to cause us to wonder how meaningful it is to be talking
> about the _phonology_ of Plan B.
Indeed, it tells us that it is meaningless.

> [snip]
> >>Therefore, I now conclude that _both_ my analysis of Plan B as having 16
> >>consonant phonemes, eight vowel phonemes, and the 'glide' /r/ is
> >>mistaken, _and_ you analysis of 16-phonemes is equally mistaken. IMO we
> >>are both analyzing a chimera    ;)
> >
> >
> > Yes.  And that's why this whole thread is running in circles, with
> > the same arguments being exchanged over and over again.  We should
> > agree that Plan B is fundamentally a language of bit quartets, and
> > that any phonological analysis of it (in the traditional meaning of
> > the term "phonology", of course) depends on arbitrary secondary
> > factors and is thus *meaningless*.
>
> Agreed. One can only discuss the merits of one particular way of mapping
> Plan B's quartets to human speech over another way of doing it. Plan B
> has a 'phonology' only in the sense in which English (or any other
> Natlang) has a bit-pattern representation. Its as tho you, I and And
> were arguing over the phonemic or graphemic inventory of EBCDIC or ASCII!
>
> It has no meaning.
Absolutely concurred.  The pronunciation of the letters used to write
Plan B in human-readable form is as peripheral and secondary to Plan B
as the binary encoding of the spelling of English is to English.
You can encode conventionally spelled English in ASCII and get
a bit stream.  You can transcribe English in Cyrillic, encode that
in Unicode, and get a completely different bit stream.  But it is still
English.

> [snip]
> >>The basic units of Plan B are the quartets of bits. One can discuss
> >>whether the ad_hoc system of mapping the bits to writing and to spoken
> >>sound is a sensible one or not, but to discuss the phonology of Plan B
> >>does not make much sense because, in the strict sense, it has no
> >>phonology.
> >
> >
> > Yes.  Plan B has no phonology.  It *can* be represented by vowels
> > and consonants, the same way natural languages can be encoded in bit
> > streams, but the same way there are many different ways of encoding
> > a particular natural language in a bit stream, there are many different
> > ways of encoding Plan B in human speech sounds.
>
> Precisely - and what is the system over which we have been arguing?
>
> The clearly ad_hoc system, in which he chose 16 letters normally used to
> denote consonants, given in Jeff's paper of May, 1990 depends upon:
> - the traditional ordering of letters in the Roman alphabet;
> - the normal consonant pronunciation of those letters in English, with
> the exception of |c| which is given the sound [S] (as in Ro, Lojban, and
> many Conlangs of American origin), |h| the sound [T] (a sound ' may have
> in Lojban] and |l| the sound [D] (presumably because keeping its
> tradition sound would otherwise have given rise the combination [lr].
> - allotting vowel values, not by any scheme, other than that the second
> group of eight should be the same as the first, but prefixed with [r\].
> The vowels are allotted simply from the American English pronunciations
> monosyllabic words beginning with the particular letter in question.
Yes; the whole scheme is completely arbitrary.  Each of the three
schemes you proposed is more elegant, systematic and consistent.

> It does seem strange to me to hold that pairs of disparate consonants
> and vowels, assigned values in such an ad_hoc way, are then allophones
> of the same phoneme. Indeed, it does seem risible to me in that IMO it
> is meaningless.
It is.  Meaningless.  It is.

> To my mind the only meaningful discussion would be whether this is the
> best way to map quartets to spoken language or not. It seems to me quite
> clear from Jeff's paper that he was not exactly over enthusiastic about
> the scheme he outlined. It was surely just a ad_hoc scheme so that it
> could be pronounced.
Indeed.

> [snip]
> >
> >>       A phonological
> >>examination might be made when comparing different mappings of the its
> >>quartets to spoken form for human use, but that IMO is another matter.
> >
> >
> > Yes.
>
> Yep - and that IMO is the only valid discussion involving Plan B and
> phonology.
I agree.  Plan B in itself has no phonology; the phonology is a feature
of the *pronunciation scheme* used.  And Jeff's pronunciation scheme
does *not* have 16 phonemes as And claims it had.

> [snip]
> >>
> >>They do indeed. Also as I have shown, Plan B could have been implemented
> >>in such a way that they were no phonetically unrelated allomorphs.
> >
> > Yes.  There is no "phonology of Plan B", as there is not "the bit stream
> > representation of English";
>
> AMEN! AMEN!
>
> > there are only phonologies of particular
> > spoken representations of Plan B.  Your system has a different phonology
> > than Jeff's, and mine (X-1) yet another one.
>
> Exactly so. In fact over the past two years, I come up with three
> different schemes for mapping Plan B's quartets to human speech (and
> writing). I have no doubt there are other ways in which it can be done.
Indeed.  There are endless possibilities.

... brought to you by the Weeping Elf