Shawalphabet YahooGroup Archive Browser

Hello! I'm new on the Yahoo! ShawAlphabet Group. As I compared Shaw's and Read's original Shaw Red with the new Shaw Blue alphabets, I noticed this: To capitalize a word (like the 1st one in a sentence) or proper names, you must place a raised *namer* character (a high dot in Shaw Red, an asterisk in Shaw Blue). Characters currently encoded within the Shavian block only go from U+10450 to U+1047F (a total of 48 characters); there's currently no provisions for either namer signs nor for additional Shaw Blue characters!! This situation MUST be rectified!

The Shaw Blue (Revised) System has a bunch of brand-new characters for sounds not provided for within the Original Shaw Red System; besides, the forms for a few of the letters differ from those in Shaw Red (Ex.: <p> and <b> have loops in Shaw Blue, but don't in Shaw Red). Also—the word *and* (also the sign <&>) is a slash-crossed Shaw Blue <n> (letter *nash*), while *of* has a dot beside the Shaw Blue <v> (letter *vav*); *the* is written as <dh> (*dhawn*), and *to* (the preposition) with <t> (*tawf*).

If nothing is forthcoming as proposals go for a reserved subblock (U+1F570 - U+1F5CF), then let's utilize that (or—at least—a good portion of it) to house the additional Shaw Blue characters.

Not sure where this is coming from, but one small point of fact, you use the
Namer Dot ONLY for proper names. It is improper to use it at the beginning of
the sentence unless the beginning of the sentence is a proper name.

--Star

==========
"There's a simple solution to this: pistols for two and coffee for one."
--Ralestone Luck, by Andre Norton.


>
>From: rob_wheel12 <rob_wheel12@...>
>To: shawalphabet@yahoogroups.com
>Sent: Mon, October 25, 2010 11:26:36 PM
>Subject: [shawalphabet] Shaw Red (Original) vs. Shaw Blue (Revised)
>
>
>Hello! I'm new on the Yahoo! ShawAlphabet Group. As I compared Shaw's and
>Read's original Shaw Red with the new Shaw Blue alphabets, I noticed this: To
>capitalize a word (like the 1st one in a sentence) or proper names, you must
>place a raised *namer* character (a high dot in Shaw Red, an asterisk in Shaw
>Blue). Characters currently encoded within the Shavian block only go from
>U+10450 to U+1047F (a total of 48 characters); there's currently no provisions
>for either namer signs nor for additional Shaw Blue characters!! This situation
>MUST be rectified!
>
>The Shaw Blue (Revised) System has a bunch of brand-new characters for sounds
>not provided for within the Original Shaw Red System; besides, the forms for a
>few of the letters differ from those in Shaw Red (Ex.: <p> and <b> have loops
>in Shaw Blue, but don't in Shaw Red). Also—the word *and* (also the sign <&>)
>is a slash-crossed Shaw Blue <n> (letter *nash*), while *of* has a dot beside
>the Shaw Blue <v> (letter *vav*); *the* is written as <dh> (*dhawn*), and *to*
>(the preposition) with <t> (*tawf*).
>
>If nothing is forthcoming as proposals go for a reserved subblock (U+1F570 -
>U+1F5CF), then let's utilize that (or—at least—a good portion of it) to house
>the additional Shaw Blue characters.
>
>
>

On Tue, Oct 26, 2010 at 5:26 AM, rob_wheel12 <rob_wheel12@...> wrote:
> Characters currently encoded within the Shavian block only
> go from U+10450 to U+1047F (a total of 48 characters); there's currently no
> provisions for either namer signs nor for additional Shaw Blue characters!!

As far as I know, the Shaw naming dot was explicitly unified with the
existing character U+00B7 MIDDLE DOT "·", so it's not true that
there's no provision for it. (Unicode punctuation tends to be encoded
by shape, not function, so there was see no need for a separate "Shaw
namer dot" in the Shaw block that looks exactly like the existing
middle dot.)

> This situation MUST be rectified!

I've never heard of "Shaw Blue". But if you would like to submit that
script for encoding, http://www.unicode.org/pending/proposals.html
tells you how you can go about doing that.

(Do note especially the section "Proposal Review Process", in
particular the first and third paragraphs thereof.)

> If nothing is forthcoming as proposals go for a reserved subblock (U+1F570 -
> U+1F5CF), then let's utilize that (or—at least—a good portion of it) to
> house the additional Shaw Blue characters.

U+1F570 - U+1F5CF is not available for random allocation - it's part
of the "Miscellaneous Symbols and Pictographs" block (
http://www.unicode.org/charts/PDF/U1F300.pdf ), so is not appropriate
for letters of any kind. In any event, Unicode (and ISO) determine
what code points go to newly-allocated scripts.

If you want to encode the script before it has been accepted by
Unicode & ISO, the proper region to use is a Private Use Area - so
either somewhere within U+E000 - U+F8FF or on Planes 15 or 16 (U+F0000
.. U+10FFFD).

You may also wish to co-ordinate your allocation work with the
ConScript Unicode Registry ( http://www.evertype.com/standards/csur/
); this may make it easier for users to use such preliminary,
by-private-agreement code points. (For example, you may be able to use
the code point space that was previously taken up by Shavian.)

Cheers,
Philip
--
Philip Newton <philip.newton@...>

The namer dot (·) is very much in unicode, just not part of the shavian
block, and is used every existing translation I've found. It should only be
used for proper nouns, not at the start of a sentence.

If you want more experience reading shavian download a free copy of Battle
for Wesnoth (http://www.wesnoth.org/) which has been mostly translated.
Always looking for volunteers to help with that effort.

I have never heard of either the "Shaw Red" or "Shaw Blue" alphabets, nor
has Google. Perhaps you're confusing the shaw alphabet with Pokemon?



On Mon, Oct 25, 2010 at 11:26 PM, rob_wheel12 <rob_wheel12@...> wrote:

> Hello! I'm new on the Yahoo! ShawAlphabet Group. As I compared Shaw's and
> Read's original Shaw Red with the new Shaw Blue alphabets, I noticed this:
> To capitalize a word (like the 1st one in a sentence) or proper names, you
> must place a raised *namer* character (a high dot in Shaw Red, an asterisk
> in Shaw Blue). Characters currently encoded within the Shavian block only
> go from U+10450 to U+1047F (a total of 48 characters); there's currently no
> provisions for either namer signs nor for additional Shaw Blue characters!!
> This situation MUST be rectified!
>
> The Shaw Blue (Revised) System has a bunch of brand-new characters for
> sounds not provided for within the Original Shaw Red System; besides, the
> forms for a few of the letters differ from those in Shaw Red (Ex.: <p> and
> <b> have loops in Shaw Blue, but don't in Shaw Red). Also—the word *and*
> (also the sign <&>) is a slash-crossed Shaw Blue <n> (letter *nash*), while
> *of* has a dot beside the Shaw Blue <v> (letter *vav*); *the* is written as
> <dh> (*dhawn*), and *to* (the preposition) with <t> (*tawf*).
>
> If nothing is forthcoming as proposals go for a reserved subblock (U+1F570
> - U+1F5CF), then let's utilize that (or—at least—a good portion of it) to
> house the additional Shaw Blue characters.
>
>
>
>
> ------------------------------------
>
> Yahoo! Groups Links
>
>
>
>

You guys who also hang around the Quickscript group will find this familiar, but for those of you who don't, here goes.

I recently started a project on SourceForge called Skribanto (if anyone has a better name in mind, please tell me) that will hopefully combine the process of transliterating texts by hand (high quality but TIME CONSUMING) with automated systems based on phonetic dictionaries (fast but not very high quality). The approach that I thought of is detailed in a PDF I put in the project's files section.

This will be an open source project from the start and anyone wanting to help is welcome to join. Right now what is needed is coders (I am a horrible one myself) and once the base is in place, transcribers who know Shavian well. With a bit of work, we can have the classics in Shavian very quickly. And that, we can all agree, is awesome.

Link:
http://sourceforge.net/projects/skribanto/

On Tue, Nov 16, 2010 at 23:25, tim_rice09 <tim_rice09@...> wrote:
> I recently started a project on SourceForge called Skribanto (if anyone has a better name in mind, please tell me) that will hopefully combine the process of transliterating texts by hand (high quality but TIME CONSUMING) with automated systems based on phonetic dictionaries (fast but not very high quality). The approach that I thought of is detailed in a PDF I put in the project's files section.

Are you aware of Thomas Thurman's efforts in this area? He already
produced a program that uses part-of-speech information and prediction
for better match of homographs (e.g. differentiating between "perfect"
the adjective and the verb).

You may be re-inventing a wheel here.

Cheers,
Philip
--
Philip Newton <philip.newton@...>

--- In shawalphabet@yahoogroups.com, Philip Newton <philip.newton@...> wrote:
>
> On Tue, Nov 16, 2010 at 23:25, tim_rice09 <tim_rice09@...> wrote:
> > I recently started a project on SourceForge called Skribanto (if anyone has a better name in mind, please tell me) that will hopefully combine the process of transliterating texts by hand (high quality but TIME CONSUMING) with automated systems based on phonetic dictionaries (fast but not very high quality). The approach that I thought of is detailed in a PDF I put in the project's files section.
>
> Are you aware of Thomas Thurman's efforts in this area? He already
> produced a program that uses part-of-speech information and prediction
> for better match of homographs (e.g. differentiating between "perfect"
> the adjective and the verb).
>
> You may be re-inventing a wheel here.
>
> Cheers,
> Philip
> --
> Philip Newton <philip.newton@...>
>
Thanks for brining this up. I am quite aware of Thomas' efforts, and incorporating that functionality would be wonderful. The thing is, at the moment, his software relies on a phonetic dictionary to convert words that are not in Androcles into Shavian. This leads to quite a few Shavian misspellings, especially in regards to the final unstressed /i/ sound. What Skribanto aims to do is refine the process further, and give it an easy to use frontend.

On a side note, which license is Bernard under? If it's GPL compatible, reinventing the wheel isn't in order, figuring how to attach it to a new car is.

Thanks for the help,
Tim

On 17 November 2010 12:38, tim_rice09 <tim_rice09@...> wrote:

> Thanks for brining this up. I am quite aware of Thomas' efforts, and
> incorporating that functionality would be wonderful. The thing is, at the
> moment, his software relies on a phonetic dictionary to convert words that
> are not in Androcles into Shavian. This leads to quite a few Shavian
> misspellings, especially in regards to the final unstressed /i/ sound. What
> Skribanto aims to do is refine the process further, and give it an easy to
> use frontend.
>

This is partly true, but not entirely. The data Bernard etc uses comes from
three sources:

1. 2,429 words from *Androcles and the Lion*; these are canonically
correct
2. 12,553 words from http://shavian.org/wiki; these have been entered by
humans, mostly me and Philip, and then hand-checked, and may reasonably be
assumed to be correct;
3. 110,348 words from CMUdict. These are of low quality (notably because
they incorporate the cot/caught merger). There is no publicly-available
electronic pronouncing dictionary which is of higher quality, unfortunately.

The only part that "relies on a phonetic dictionary" is the CMUdict
portion. This contains by far the largest number of words, but all the
words it contains are fairly uncommon: almost all common words are found in
the first two sections.

Solutions to the problem of the quality of the CMUdict data may involve

- turning off CMUdict support, or at least allowing it to be turned off
- expanding the coverage of the Shavian wiki so that more words are
human-entered and fewer are taken from the phonetic dictionary. On this
note, there should certainly be more ways to enter such words than just
having to go to the website and type them in.

On a side note, which license is Bernard under? If it's GPL compatible,
> reinventing the wheel isn't in order, figuring how to attach it to a new car
> is.
>
Same as Perl (bi-licence Artistic and GPL), so yes. All the rest of my
released Shavian software is GPL.

I haven't yet read your PDF, but I'd love to help out. I'd like to find
ways in which the existing software can help with it, as well.

T

Thanks for clearing that up, I should have done a bit more research. Speaking of entering words in the dictionary, this may be a very good platform to help with that. As the plan stands, when a new word is encountered in the source text, the program will prompt the user to give an equivalent. Add a network layer and presto, more people improving the dictionary at a much faster pace. But that is way out of my league at the moment. Back to the subject at hand, I want to see Skribanto (seriously, it needs a better name) to become a great project that brings people from around the world to make an awesome tool.

Thank you so much for your help,

Tim

PS: What books do you guys want to see transferred first?

--- In shawalphabet@yahoogroups.com, Thomas Thurman <tthurman@...> wrote:
>
> On 17 November 2010 12:38, tim_rice09 <tim_rice09@...> wrote:
>
> > Thanks for brining this up. I am quite aware of Thomas' efforts, and
> > incorporating that functionality would be wonderful. The thing is, at the
> > moment, his software relies on a phonetic dictionary to convert words that
> > are not in Androcles into Shavian. This leads to quite a few Shavian
> > misspellings, especially in regards to the final unstressed /i/ sound. What
> > Skribanto aims to do is refine the process further, and give it an easy to
> > use frontend.
> >
>
> This is partly true, but not entirely. The data Bernard etc uses comes from
> three sources:
>
> 1. 2,429 words from *Androcles and the Lion*; these are canonically
> correct
> 2. 12,553 words from http://shavian.org/wiki; these have been entered by
> humans, mostly me and Philip, and then hand-checked, and may reasonably be
> assumed to be correct;
> 3. 110,348 words from CMUdict. These are of low quality (notably because
> they incorporate the cot/caught merger). There is no publicly-available
> electronic pronouncing dictionary which is of higher quality, unfortunately.
>
> The only part that "relies on a phonetic dictionary" is the CMUdict
> portion. This contains by far the largest number of words, but all the
> words it contains are fairly uncommon: almost all common words are found in
> the first two sections.
>
> Solutions to the problem of the quality of the CMUdict data may involve
>
> - turning off CMUdict support, or at least allowing it to be turned off
> - expanding the coverage of the Shavian wiki so that more words are
> human-entered and fewer are taken from the phonetic dictionary. On this
> note, there should certainly be more ways to enter such words than just
> having to go to the website and type them in.
>
> On a side note, which license is Bernard under? If it's GPL compatible,
> > reinventing the wheel isn't in order, figuring how to attach it to a new car
> > is.
> >
> Same as Perl (bi-licence Artistic and GPL), so yes. All the rest of my
> released Shavian software is GPL.
>
> I haven't yet read your PDF, but I'd love to help out. I'd like to find
> ways in which the existing software can help with it, as well.
>
> T
>

Editing the Shavian wiki has always been rather an involved matter. I put an AJAX interface together last night which should help things along, and I thought you might like to test it:

http://shavian.org/temp/

The demo makes the following assumptions:
- the word "twenty" isn't in the lexicon (of course it is really)
- the word "guineas" isn't in the lexicon, but the word "guinea" is
- "use" can be a noun or a verb.

This really does submit your results to the server, but they don't update the wiki. In the real version they will.

Note that you can type into the dialogue box using the Latin alphabet and it will appear as if you'd been using a Shavian keyboard. There is no need for a separate input method.

If you have problems or suggestions, please let me know! Let me know what browser you're using, as well. I haven't yet tested this on Internet Explorer, so that would be particularly helpful, but all reports are useful.

Thomas