I had the same problem. I have searched for years to find a font that has glyphs for the tone marks, but could find none. The funny thing is that you can use the tone marks during bopomofo input, but it seems impossible to use them as "characters". I have finally decided to just use the numerals 1 to 5 instead of the standard tonal marks. The result looks as follows: 中國 zhong1 guo2 China Admittedly not very nice, but everybody who has only a basic understanding of Chinese knows what it means. Best regards Ulrich Ruess On Monday, December 8, 2003 23:44, Mike FABIAN wrote:

Philip Amadeo Saeli さんは書きました:

...

What is the standard practice for inputting Chinese pinyin (not bopomofo!) phonetics into a document? I'm writing a document where I'd like to include, for each sentence:

1. Chinese characters 2. pinyin transcription of the Chinese characters 3. English translation

I am having little problem with the Chinese character input. The problem is including the tone marks required by the pinyin transcription. I've tried wierd latin vowels, but have not been able to find a complete set necessary. I've been using "Insert -> Special Character" in OpenOffice.

Maybe you are using an unsuitable font? You can choose a font in this "Insert -> Special Character" dialog in OpenOffice.

Which characters do you need for PinYin?

I guess "FreeSans" or "Luxi Sans" have all you need.

...

Also, wheneven I set the XMODIFIERS to other than "@im=local", the compose key stops working,

That is unfortunately a design limitation in XIM (X Input Method). You can only use one at a time. Either "@im=local" which means compose or something else like "@im=SCIM".

Very few applications are able to switch between such input methods at runtime. For example in mlterm you can switch between such X Input Methods, i.e. you can switch between using compose and SCIM for example.

But this doesn't help you for OpenOffice as OpenOffice doesn't support such input server switching.

...

so I am forced to use cut-n-paste to input any non-English chars.

You can make cut-n-paste somewhat more effective if you create a plain text file with the characters you most frequently need, display this file and cut and paste from there.

This is certainly more efficient than the special symbol input of OpenOffice where you don't have all the characters you need close together and apparently you cannot keep that dialog open either.

...

How can I get the compose key to work together with a Chinese input method?

IIIMF is supposed to solve that problem in the long run.

Maybe it is possible to extend SCIM a little bit in order to input pinyin. SCIM can switch between different input methods, it shouldn't be too difficult to add another input method for PinYin transcription. That is probably the best short solution for the moment.

-- Mike FABIAN http://www.suse.de/~mfabian 睡眠不足はいい仕事の敵だ。

-- To unsubscribe, e-mail: m17n-unsubscribe@suse.com For additional commands, e-mail: m17n-help@suse.com

On Wednesday, December 10, 2003 2:13, Mike FABIAN wrote:

Ulrich Ruess さんは書きました:

...

I had the same problem. I have searched for years to find a font that has glyphs for the tone marks, but could find none.

Have you tried "FreeSans", "FreeSerif", "FreeMono" (from the "freefont" package on SuSE Linux 9.0)?

Yes, I have tried them. The problem is, in traditional Chinese and in the UN transcription (yes, pinyin also applies to traditional Chinese), the tone mark is not put above one of the vowels, but at the end of the word as it's last "letter". You therefore do not need accented characters, but only the five glyphs representing the five tones (normally people talk about four tones only, the fifth is for words that have lost their original tone). Putting the tone marks above the vowels is only a typographical nicety done in the PRC. There is no meaning as of where the stress in pronunciation should be behind it. Chinese consider a syllable a unit. The tone applies to the whole unit as such, not the vowel!

...

The funny thing is that you can use the tone marks during bopomofo input, but it seems impossible to use them as "characters".

I have finally decided to just use the numerals 1 to 5 instead of the standard tonal marks.

The result looks as follows:

中國 zhong1 guo2 China

-- Mike FABIAN http://www.suse.de/~mfabian 睡眠不足はいい仕事の敵だ。

-- To unsubscribe, e-mail: m17n-unsubscribe@suse.com For additional commands, e-mail: m17n-help@suse.com

Philip Amadeo Saeli さんは書きました:

* Mike FABIAN [031208 15:45]:

...

Philip Amadeo Saeli さんは書きました:

...

I am having little problem with the Chinese character input. The problem is including the tone marks required by the pinyin transcription. I've tried wierd latin vowels, but have not been able to find a complete set necessary. I've been using "Insert -> Special Character" in OpenOffice.

Maybe you are using an unsuitable font? You can choose a font in this "Insert -> Special Character" dialog in OpenOffice.

Which characters do you need for PinYin?

Specifically, I needed the vowels [aeiou] with the first and third tone marks above them, which are basically a dash and an upside-down caret respectively. The vowels with second and fourth tone marks could be represented by existing latin-1 chars.

I.e. you need: U+01CD: LATIN CAPITAL LETTER A WITH CARON U+01CE: LATIN SMALL LETTER A WITH CARON U+01CF: LATIN CAPITAL LETTER I WITH CARON U+01D0: LATIN SMALL LETTER I WITH CARON U+01D1: LATIN CAPITAL LETTER O WITH CARON U+01D2: LATIN SMALL LETTER O WITH CARON U+01D3: LATIN CAPITAL LETTER U WITH CARON U+01D4: LATIN SMALL LETTER U WITH CARON U+011A: LATIN CAPITAL LETTER E WITH CARON U+011B: LATIN SMALL LETTER E WITH CARON U+0100: LATIN CAPITAL LETTER A WITH MACRON U+0101: LATIN SMALL LETTER A WITH MACRON U+012A: LATIN CAPITAL LETTER I WITH MACRON U+012B: LATIN SMALL LETTER I WITH MACRON U+014C: LATIN CAPITAL LETTER O WITH MACRON U+014D: LATIN SMALL LETTER O WITH MACRON U+016A: LATIN CAPITAL LETTER U WITH MACRON U+016B: LATIN SMALL LETTER U WITH MACRON U+0112: LATIN CAPITAL LETTER E WITH MACRON U+0113: LATIN SMALL LETTER E WITH MACRON is that right? I think it is quite easy to add a new input method to SCIM which enables you to input these characters. That would probably be the most efficient way for you to input these characters as you are using SCIM anyway. I have to update the SuSE SCIM packages anyway, I'll have a look whether I can add something like that.

...

I guess "FreeSans" or "Luxi Sans" have all you need.

Don't know about "FreeSans", but "Luxi Sans" did -not- have all the needed chars.

I just checks, you are right, "Luxi Sans" lacks a few glyphs for the above characters but "FreeSans" has them all. "FreeSans" is in the "freefont" rpm-package on SuSE Linux 9.0.

I finally found several fonts (in addition to the Arphic GB TTF fonts) that included the needed chars, which solved part of my problem. The Arphic font I was using ("AR PL KaitiM GB", in OpenOffice) had the needed chars, but they were double wide and hence unsuitable, the rest of the base Latin chars being standard width.

For anyone who is interested, the fonts which included the needed chars were (names from the OpenOffice font selection menu):

"AR PL KaitiM GB" (full char cell width) "AR PL SungtiL GB" (full char cell width) "Caslon" "Caslon RomanSmallcaps" "Courier New" "Gentium" "Gentium Alt" "New Century Schoolbook" "Times New Roman"

"Courier New", "Times New Roman" are commercial fonts (from the Microsoft "Webfonts") "New Century Schoolbook" is one of the classic X11 Bitmap fonts, you probably don't want to use a bitmap font in OpenOffice. "Gentium" is not distributable without asking the author. I wrote an e-mail to the author asking whether it is OK to distribute "Gentium" with SuSE Linux but never received an reply, therfore this font isn't included with SuSE Linux. "Caslon" is free and included with SuSE Linux. But "FreeSans" might be better, a serif version ("FreeSerif") and a monospaced version "FreeMono" are available as well. All of them in regular, bold, oblique, and bold-oblique. And apparently these fonts are actively developed, if you miss some characters, I suggest to ask the author to add it. The home page of the freefont project is: http://savannah.gnu.org/projects/freefont/

...

...

How can I get the compose key to work together with a Chinese input method?

...

IIIMF is supposed to solve that problem in the long run.

What is IIIMF?

The designated successor of XIM (X Input Method). XIM has quite a few design limitations, one of them is that you cannot easily switch input methods on the fly, usually you have to decide which input method to use before starting an application and cannot change it later. This is the reason why you cannot use compose together with SCIM in OpenOffice. IIIMF (Internet Intranet Input Method Framework) is a redesign which supposedly does not have many of the limitations of XIM. Being able to switch between different input methods at random is on of the design features of IIIMF. IIIMF is not yet included in SuSE Linux but may be included in future. -- Mike FABIAN http://www.suse.de/~mfabian 睡眠不足はいい仕事の敵だ。

Ulrich Ruess さんは書きました:

...

I think it is quite easy to add a new input method to SCIM which enables you to input these characters. That would probably be the most efficient way for you to input these characters as you are using SCIM anyway. I have to update the SuSE SCIM packages anyway, I'll have a look whether I can add something like that.

Before you do this, please consider that for pinyin input you do not need these characters, since the "tone" is already mapped to some characters on the keyboard (mostly the keys 1 to 5). The result of a pinyin input on the keyboard is a Chinese character and not it's pinyin representation. If you want the pinyin (with the tone marks) appear on the screen, you use your normal alphabet input. These accented characters are only for typesetting of pinyin in the Latin alphabet.

If I understood Philip right, this is exactly what he wants to do, he wants to typeset pinyin in the Latin alphabet. But his keyboard doesn't have these characters, therefore he needs an input method. "Compose" would qualify as a suitable input method according to Philip, but unfortunately "Compose" cannot be used together with any other XIM like SCIM.

Tone input for pinyin becomes less and less important as the quality of sentence analyzing programs increases.

-- Mike FABIAN http://www.suse.de/~mfabian 睡眠不足はいい仕事の敵だ。

Ulrich Ruess さんは書きました:

Would it not be easier and more convenient to create a special keyboard mapping which contains the necessary accented characters (e.g. by using Alt-Gr switching, or hot-keys)?

For a few characters only, yes. I use my own AltGr mappings for the German special characters äüöÄÜÖß because - I use US-keyboard layout - <AltGr><u> is only two keys but <"><u> is three keys - I can't use compose anyway because I usually use kinput2. But I think such individual mappings make only sense for a few very frequently used characters. It would be very nice to have all the other special characters easily available as well. The "Compose" mechanism isn't that bad. I was thinking about adding an input method to SCIM which gives access to the complete Compose table (which contains much more than just the few characters needed for pinyin).

Switching between different keyboard layouts while using XIM is painless, I do it all the time when I have to switch between English, German and Chinese.

Switching keyboard layouts is of course possible, but it drives me nuts to switch e.g. between US and German keyboard layout. Most special characters (e.g. parenthesis, ...) are on different keys and even alphabetical characters (like 'z' and 'y') change position. Switching between such different layouts isn't good for typing speed, sticking to one layout and using input methods for everything not accessible via that layout is faster in my opinion. -- Mike FABIAN http://www.suse.de/~mfabian 睡眠不足はいい仕事の敵だ。

Ulrich Ruess さんは書きました:

On Wednesday, December 10, 2003 23:51, Mike FABIAN wrote:

...

Ulrich Ruess さんは書きました:

...

Would it not be easier and more convenient to create a special keyboard mapping which contains the necessary accented characters (e.g. by using Alt-Gr switching, or hot-keys)?

For a few characters only, yes.

I use my own AltGr mappings for the German special characters äüöÄÜÖß because

Could you please tell me how to make such a mapping for a US keyboard? I would definitely prefer that to layout switching.

Create a file ~/.Xmodmap with the following contents (or append to your ~/.Xmodmap if you already have one): keysym a = a NoSymbol adiaeresis NoSymbol keysym o = o NoSymbol odiaeresis NoSymbol keysym u = u NoSymbol udiaeresis NoSymbol keysym s = s NoSymbol ssharp NoSymbol keysym e = e NoSymbol EuroSign NoSymbol keysym c = c NoSymbol currency NoSymbol keysym g = g NoSymbol sterling NoSymbol keysym i = i NoSymbol idiaeresis NoSymbol ~/.Xmodmap is automatically read when restarting your X session.

...

But I think such individual mappings make only sense for a few very frequently used characters. It would be very nice to have all the other special characters easily available as well. The "Compose" mechanism isn't that bad. I was thinking about adding an input method to SCIM which gives access to the complete Compose table (which contains much more than just the few characters needed for pinyin).

If you look at it in the context of general typesetting it is certainly a good idea. In the context of Chinese input I cannot see an advantage.

Not only Chinese. There is no reason why SCIM could not talk to the Canna server for example to input Japanese. It is much nicer to have *one* input server which can be switched at random between many input methods than a different input server for each method. I think for that reason the GTK2 input modules are so popular, you can switch easily between many input methods. But the GTK2 input modules can only be used in GTK2 applications. I heard that a similar feature is planned for Qt but this will be limited to Qt applications then. An XIM input server which can switch between many methods could be used to give similar functionality to all (XIM enabled) X11 applications. SCIM appears to be useful for that purpose, it appears to be easily extendable. Currently it mostly supports Chinese, but support for other languages could be added. Another interesting project is http://www.freedesktop.org/Software/uim Currently it supports mostly Japanese input methods, it can apparently be extended for other languages as well. IIIMF is supposed to be a general solution to the problem of switching between many input methods, not only for specific toolkits like GTK2 or Qt and not only for X11.

If you have time, you could consider to add an input method to XIM, which uses only the eight basic strokes.

I have not yet heard about the eight basic strokes and have no idea how that works.

Such method was available for MS software from an Australian company, but was not successfull (I guess because they had too many mistakes in the mapping table). If you could help me to understand how to do it, I would be willing to do it myself and later on give it back to the community.

-- Mike FABIAN http://www.suse.de/~mfabian 睡眠不足はいい仕事の敵だ。

* Mike FABIAN [031209 18:12]:

Philip Amadeo Saeli さんは書きました:

...

* Mike FABIAN [031208 15:45]:

...

Philip Amadeo Saeli さんは書きました:

...

I am having little problem with the Chinese character input. The problem is including the tone marks required by the pinyin transcription. I've tried wierd latin vowels, but have not been able to find a complete set necessary. I've been using "Insert -> Special Character" in OpenOffice.

Which characters do you need for PinYin?

Specifically, I needed the vowels [aeiou] with the first and third tone marks above them, which are basically a dash and an upside-down caret respectively. The vowels with second and fourth tone marks could be represented by existing latin-1 chars.

I.e. you need:

U+01CD: LATIN CAPITAL LETTER A WITH CARON U+01CE: LATIN SMALL LETTER A WITH CARON U+01CF: LATIN CAPITAL LETTER I WITH CARON U+01D0: LATIN SMALL LETTER I WITH CARON U+01D1: LATIN CAPITAL LETTER O WITH CARON U+01D2: LATIN SMALL LETTER O WITH CARON U+01D3: LATIN CAPITAL LETTER U WITH CARON U+01D4: LATIN SMALL LETTER U WITH CARON U+011A: LATIN CAPITAL LETTER E WITH CARON U+011B: LATIN SMALL LETTER E WITH CARON

U+0100: LATIN CAPITAL LETTER A WITH MACRON U+0101: LATIN SMALL LETTER A WITH MACRON U+012A: LATIN CAPITAL LETTER I WITH MACRON U+012B: LATIN SMALL LETTER I WITH MACRON U+014C: LATIN CAPITAL LETTER O WITH MACRON U+014D: LATIN SMALL LETTER O WITH MACRON U+016A: LATIN CAPITAL LETTER U WITH MACRON U+016B: LATIN SMALL LETTER U WITH MACRON U+0112: LATIN CAPITAL LETTER E WITH MACRON U+0113: LATIN SMALL LETTER E WITH MACRON

is that right?

I forgot to mention the versions for the letter 'u' with diaeresis: U+01D5: LATIN CAPITAL LETTER U WITH DIAERESIS AND MACRON U+01D6: LATIN SMALL LETTER U WITH DIAERESIS AND MACRON U+01D9: LATIN CAPITAL LETTER U WITH DIAERESIS AND CARON U+01DA: LATIN SMALL LETTER U WITH DIAERESIS AND CARON

I think it is quite easy to add a new input method to SCIM which enables you to input these characters. That would probably be the most efficient way for you to input these characters as you are using SCIM anyway. I have to update the SuSE SCIM packages anyway, I'll have a look whether I can add something like that.

WRT entering the pinyin phonetics with tone marks, I finally decided that the most efficient way for me to finish my paper would be to make two passes: one to enter the Latin chars and another to enter the Han chars. I could use the compose key for the former and then switch locales to activate SCIM for the latter. It -would- be nice, esp for editing a complete document, to be able to use the same input method for -all- input, though. BTW, I looked up the compose sequences in the file: /usr/X11R6/lib/X11/locale/en_US.UTF-8/Compose and all -except- the "U+00F3: LATIN SMALL LETTER O WITH ACUTE" work fine. Interestingly, "U+00D3: LATIN CAPITAL LETTER O WITH ACUTE" works just fine! Note that I'm using a US keyboard and so am using the "Multi_key" (compose) sequences exclusively, having no dead keys.

...

I finally found several fonts (in addition to the Arphic GB TTF fonts) that included the needed chars, which solved part of my problem. The Arphic font I was using ("AR PL KaitiM GB", in OpenOffice) had the needed chars, but they were double wide and hence unsuitable, the rest of the base Latin chars being standard width.

For anyone who is interested, the fonts which included the needed chars were (names from the OpenOffice font selection menu):

"AR PL KaitiM GB" (full char cell width) "AR PL SungtiL GB" (full char cell width) "Caslon" "Caslon RomanSmallcaps" "Courier New" "Gentium" "Gentium Alt" "New Century Schoolbook" "Times New Roman"

"Courier New", "Times New Roman" are commercial fonts (from the Microsoft "Webfonts")

"New Century Schoolbook" is one of the classic X11 Bitmap fonts, you probably don't want to use a bitmap font in OpenOffice.

Yup! Agree.

"Gentium" is not distributable without asking the author. I wrote an e-mail to the author asking whether it is OK to distribute "Gentium" with SuSE Linux but never received an reply, therfore this font isn't included with SuSE Linux.

This one has been my favorite, with the MS "Times New Roman" coming in second.

"Caslon" is free and included with SuSE Linux.

But "FreeSans" might be better, a serif version ("FreeSerif") and a monospaced version "FreeMono" are available as well. All of them in regular, bold, oblique, and bold-oblique. And apparently these fonts are actively developed, if you miss some characters, I suggest to ask the author to add it. The home page of the freefont project is:

http://savannah.gnu.org/projects/freefont/

Thanks for the font info. I always have trouble mapping font names as they appear in applications to the actual files in the X11 fonts dirs and then to their respective RPM files and being able to find out their sources, licensing, etc. BTW, I decided to install the freefont RPM from SuSE 9.0 and give it a try. Indeed it does have all the necessary chars. I don't particularly like the tall vertical line spacing, though. Makes things look double spaced (or, at least 1.5 spaced) though they are single spaced. It -is- a very sharp looking font, though. BTW, how is m17n support in SuSE 9.0 as compared to 8.2? I added quite a number of RPMs from the SuSE ftp site (/pub/people/mfabian/8.2/) to my 8.2 system to get it up to snuff and I notice that there are no packages for 9.0 there. Does that indicate that they are not needed or simply that none have been prepared yet. Thanks again for all the useful info! I've just recently switched over to a UTF-8 locale and have been coming up to speed with Unicode as well. I'm encouraged by the way things seem to be coming together now for multilingual support under Linux, esp support for multiple languages in a single document. Some Chinese people who saw a sample of my paper were impressed that it had been done without expensive payware! Phil -- Philip Amadeo Saeli SuSE Linux 8.2 psaeli@zorodyne.com