From: Michael Everson (everson@evertype.com)
Date: Wed Feb 12 2003 - 20:14:34 EST
At 23:47 +0000 2003-02-12, Andy White wrote:
>Michael Everson wrote:
>
> > These are A-VIRAMA-YA-AA and E-VIRAMA-Y-AA, which are sequences of an
>> independent vowel plus a subjoined consonant plus a dependent vowel.
>> Those sequences are used to represent foreign sounds in Bengali.
>> Since the ya-phalaa is a common glyph that can also follow consonants
>> it makes sense not to treat the use of it with independent vowels
>> differently.
>
>Oriya Ba(Va), when subscript, is known as Ba-phalaa (bophola). When in
>this position it represents the original Va/Wa consonant.
>Since Ba-phalaa is a common glyph that can also follow consonants, using
>the above logic, you have reasoned that it should not have been included
>in the standard!
Not so. While in Bengali it is common practice to attach ya-phalaa to
both consonants and independent vowels, in Oriya it is only common
practice to attach ba-phalaa to consonants. Only in the rare case of
WA was it attached (by whomever invented the glyph) to an independent
vowel. A book I have on Oriya shows KA + VIRAMA + BA yielding <kwa>,
JA + VIRAMA + BA yielding <jva>, DHA + VIRAMA + BA yielding <dhwa>,
MA + VIRAMA + BA yielding <mba>, and RA + VIRAMA + BA yielding <rba>
(with repha), and SHA + VIRAMA + BA yielding <shba>. In each of these
a consonant cluster is formed where the vowel of the first consonant
is killed and the second consonant takes one of three different
sounds. Since WA is a consonant on its own, and since it is unusually
attached to an independent vowel, it did not seem sensible to encode
O + VIRAMA + BA which ought to yield <oba>, <owa>, or <ova>. The
vowel killed is normally -a, and O doesn't contain one.
Ken Whistler pointed out that Oriya independent AA could have been
decomposed to independent A + vowel sign A. We didn't do that, and WA
is quite similar.
> > Yes, Oriya O-VIRAMA-BA could be considered structurally similar, and
> > it could even be said that Devanagari K-VIRAMA-SSA which is thought
>> of as a letter in Marathi could considered similar.
>
>Not really; KSSA does not include a vowel.
Neither does OBA. It's [wa] not [owa].
> > In the case of
>> Oriya, however, there were two issues. (Did you read my paper, N2525?)
>
>Yes, I read it yesterday, but as I am already very familiar with the
>history of this character, I only skim read it, sorry!
You missed something, perhaps. But do you know when and/or by whom
the characters were introduced? Or are you just analyzing the glyph?
> > The original consonant [va] was lost in Oriya, merging with [ba].
>> Later, a need to represent the foreign sound [va] and the foreign
>> sound [wa] was perceived. Taylor 1883 showed a shape for this [va]
>> which is rather unusual, but in any case what came to be used was a
>> BA with a dot in or above its head.
>
>This was not unusual at all but common practise.
I said that Taylor 1883 gives an unusual shape, which is not BA with dot.
>In both Bengali and Oriya the letter Va looked identical to Ba, It
>is a common theme in Indic scripts for letters that have become
>ambiguous to be marked by a dot.
Taylor 1883 is different.
> > To represent [wa] the consonant
>> BA was, unusually subscripted to the initial vowel O.
>>
>> Two new, rare, foreign consonants were born. We chose to encode them.
>
>No, letter Va was always there - just that it was hard to see at times
>due to the lack of a dot.
It seems to me that historical VA was lost having merged with BA, and
something new had to be introduced.
> > >Another is 'Bengali Letter Central A' used to transcribe English 'a'
> > >as in ball. (Comparable to Devanagri Chandra A). It is visibly a
>> >Bengali letter A with postfix letter Ya (Bengali Letter A with
>> >Ya-phalaa). I think that this letter, among with a few others not
>> >mentioned, should be included for compatibility with the Devanagri
>> >code block. But what do you think?
>>
>> I don't see how Bengali is incompatible with Devanagari.
>I never said it was, my point was that the Bengali block currently has
>no parallel letter to Chandra A.
BENGALI LETTER A + VIRAMA + BENGALI LETTER YA represents this, and
this can be mapped in transliteration to CANDRA A if necessary.
-- Michael Everson * * Everson Typography * * http://www.evertype.com
This archive was generated by hypermail 2.1.5 : Wed Feb 12 2003 - 21:12:05 EST