Re: brz, or Plan B revisited (LONG)

Hallo!

Here's a second reply after the whole matter undeservedly got
short shrift in my first one.

R A Brown wrote:

> Recently there were a few mails about Jeff Prothero's "Plan B". Jörg and
> I both agreed his article "design and Implementation of a Near-optimal
> Loglan Syntax" made many good points, but that his phonology for Plan B
> is naive. It is arbitrary and American-centered.
Yes.

> Jörg suggested dropping the dual consonant ~ vowel pronunciation of each
> of the 16 letters, pronouncing them always as consonants (since
> consonant symbols are used for the 16 letters),and inserting epenthetic
> vowels where needed.
My original plan wasn't all too creative - just slip in [a] after
every other consonant, starting with the first, thus breaking the
whole thing up into CVC syllables, with a CV syllable at the end
when the number of consonants is odd.  In that system, Jeff's
|cjbjt| would yield something like [Sajbajta].

But Ray has a better plan...

> In the meantime I had been thinking along similar lines, and suggested
> that as Jeff's 16 consonant letters mapped to unique four-bit patterns,
> the last bit or least significant (LSB) of the first consonant & the
> first or most significant bit (MSB) of the second consonant would
> determine the value of the vowel (see my mail of 11th Sep).
That's a good idea.

> That may have reminded one or two of an email I posted on 9th Nov. 2003
> in which I proposed a four-vowel scheme for BrScB in which the vowels
> were determined from the bit patterns: 00, 01, 10, 11.
>
> At that time, BrScB (a name which is now deprecated!) used a sort of
> syllabary based on an idea Dirk Elzinga had suggested way back in May
> 1999 (on another list). I quote a snippet from that mail:
> {quote}
> ... where each syllable has the same consonant but varies in the vowel
> quality. The vowel qualities are "resting" (central unrounded), rounded,
> and fronted.
>
> p [p|, pu, pi]  b [pa, po, pe] ("|" is "barred-i")
> t [t|, tu, ti]  d [ta, to, te]
> {/quote}
>
> Returning to these two BrScB ideas of Nov. 2003, and combining them with
> Jeff's four-bit (hex digit) idea, I have come 'discovered' this strange
> little language called _brz_ /pElusi/
> ---------------------------------
> VOWELS
>
> The language has 4 has four vowels:
>       Front  Back
> High  /i/    /u/
> Low   /E/    /O/
This is a nice inventory.

> My mail of 9th Nov. 2003 mapped these in a somewhat arbitrary way, which
> showed its anglocentricity with Lo=0 and Hi=1. Now, it seems to me
> preferable that:
> - final (probably unstressed) vowel in a string be a high vowel (there
> are many natlang precedents for this; also unstressed low vowels tend
> towards [@] in many languages);
Yes; high vowels are phonetically much more salient than low ones.
There are languages in which unstressed low vowels went to
cloudcuckooland and unstressed high vowels didn't.

> - the vowel of the last syllable will have only a distinct LSB; it seems
> more reasonable to assume he 'missing' MSB is a 0 rather than a 1;
> therefore high will be the unmarked (0) value, and low the marked (1) value.
I'd say so, too.  This makes perfect sense.

> As for the other dimension, one tends to work from front to back. Thus 0
> as LSB will trigger a front vowel, and 1 will trigger a back vowel.Thus
> we have:
>       Front  Back
> High  00    10
> Low   01    11
So we have: 00 -> /i/, 01 -> /E/, 10 -> /u/, 11 -> /O/.

> --------------------------------
> CONSONANTS
>
> The language has eight consonants, arranged in four series (0 to 3) of
> two grades,
> thus:
>               #0   #1   #2   #3
>   Sonorant: (zero) /l/  /n/  /m/
> Obstruent:   /k/  /s/  /t/  /p/
Nice.  AFAIK, these are the seven most frequent consonants in the
world's languages.

> The 'zero consonant' is just that, i.e. phonologically the syllable is
> V, while all the others are CV. Phonetically it may have a glottal stop
> onset or a glide onset ([j] before front vowels and [w] before back vowels).
>
> This set of 8 consonants has a _complementary_ set: one occurring after
> high vowels and the other after low vowels. (This is the opposite of
> Dirk's 1999 idea, but doing it this way means all word final vowels are
> high - see above).
>
> This in effect gives us four rows or 'grades': (a) sonorants after high
> vowels; (b) obstruents after high vowels; a) sonorants after low vowels;
> (b) obstruents after low vowels. We have therefore a script of 16
> consonant letters. This could well map into Tengwar 1 to 16
>    :-)
A nice idea, for those who like Tengwar (which I do).

> But I must map them into quartets of bits and into modern Roman
> alphabetic symbols.
> ----------------------------------
> MAPPING
>
> We have already established that the 8 consonants have 0 as their MSB.
> The next two bits map the four series, #0 to #3, thus (n the 1st set):
> 00, 01, 10, 11.
>
> It seems reasonable that 0000 should map the 'zero' consonant; thus the
> LSB in consonants of the first grade is 0, thus: 0000, 0010, 0100, 0110.
> We can map #1, #2 and #3 simply as |l|, |n| and |m|. I suggest the zero
> consonant is denoted by the apostrophe.
>
> This means that second grade is distinguished from the first by having
> its LSB set to 1, thus: 0001, 0011, 0101, 0111. They are mapped in Roman
> script as |k|, |s|, |t| and |p|.
So we get:

0000  /0/   0100  /k/
0001  /l/   0101  /s/
0010  /n/   0110  /t/
0011  /m/   0111  /p/

Good.

> The second set is the _complement_ of the first set, so the obvious
> thing to do is set the MSB to 1 and use _one's complement_ of the 1st
> set.
Not simply set the MSB to 1 - complement the whole bit pattern!
That rocks!

>       This means that the sonorant grade is 1111, 1101, 1011, 1001. The
> 'zero' consonant can be mapped as |h| which is zero consonant in the
> Romance languages, and #2 by |r|. The next two are not so obvious. If we
> are confined to ASCII, then I guess we have to have |N| and |M|. But if
> Unicode is available (as it should be) we can use |É´| (U+0274 IPA small
> capital N) and |μ| (U+03BC Greek small letter mu).
>
> Finally we have the obstruent grade of the second set: 1110, 1100, 1010,
> 1000. These we can readily represent as |g|, |z|, |d| and |b|.
> ---------------------------------------
> PUTTING IT ALL TOGETHER
>
> Quartet Letter  Value
>   0000     '      front vowel with optional non-phonemic onset
>   0001     k      /k/ followed by a back vowel
>   0010     l      /l/ followed by a front vowel
>   0011     s      /s/ followed by a back vowel
>   0100     n      /n/ followed by a front vowel
>   0101     t      /t/ followed by a back vowel
>   0110     m      /m/ followed by a front vowel
>   0111     p      /p/ followed by a back vowel
>
>   1000     b      /p/ followed by a front vowel
>   1001     M      /m/ followed by a back vowel
>   1010     d      /t/ followed by a front vowel
>   1011     N      /n/ followed by a back vowel
>   1100     z      /s/ followed by a front vowel
>   1101     r      /l/ followed by a back vowel
>   1110     g      /k/ followed by a front vowel
>   1111     h      back vowel with optional non-phonemic onset
A very nice mapping of bit patterns to phonemes.  I like it!

> The above is given in bit value order. The alphabet could be set out in
> series and grade order, thus: 'lnm kstp hrNM gzdb
> -------------------------------
> EXAMPLES
>
> (a) Using the two-letter combo given in my mail of 9th Nov. 2003:
> bz = 1000 1100 /pEsi/
> pz = 0111 1100 /p0si
> bs = 1000 0011 /pisu/
> ps = 0111 0011 /pusu/
>
> (c) Using Jeff Prothero's example in which _cjbjt_ is pronounced
> "showboat", and by keeping his bit values, we have:
> 0001 0110 0000 0110 1101 = km'mr = /kumi.imElu/
Yes, this is much nicer than Jeff Prothero's naive phonology,
even if there are more syllables.

Greetings,

Jörg.

Reply