L2/98-292 R
Date: August 19, 1998
Revision date: September 15, 1998
Title: Comments on proposals to add characters from ISO standards developed by ISO/TC 46/SC 4
Source: NCITS/L2 and the Unicode Consortium
Status: Joint Contribution
Action: For the consideration of WG2
This contribution addresses the proposed addition to ISO/IEC 10646:1 of characters from standards for bibliographic data interchange developed by ISO/TC 46/SC 4. The characters are proposed in the following documents:
N 1741: Additional Latin characters for the UCS
N 1743: Additional Greek characters for the UCS
N 1744: Additional Cyrillic characters for the UCS
N 1745: Additional Math characters for the UCS
N 1746: Additional combining characters for the UCS
N 1747: Contraction characters for the UCS
N 1748: Additional signature mark characters for the UCS
N 1749: Additional Hebrew cantillation characters for the UCS
This contribution also comments on the proposed addition to ISO/IEC 10646 of three Cyrillic script signs that are documented in the ALA/LC Romanization Tables published by the Library of Congress.
ISO/TC 46/SC 4/WG1 wanted to ensure that all characters in the ISO/TC 46 bibliographic character sets were represented in ISO/IEC 10646, and so undertook a mapping project. These proposals cover the characters for which WG1 was unable to identify an equivalent in ISO/IEC 10646.
The outcome of the mapping process determined the content of these proposals. Where a mapping was overlooked, a TC 46 character was proposed for addition when it should not have been. Questionable mappings, on the other hand, meant that a TC 46 character was not proposed for addition when perhaps it should have been. Both these cases are addressed in comments on specific proposals.
TC 46 character sets, in general, have not been influenced by the Character/Glyph Model. Indeed, for some of the TC 46 standards, development began around the time that the distinction between character and glyph was being defined. As a result, TC 46 standards include characters which would be considered glyphic variants under the Character/Glyph Model, and so inappropriate for encoding in ISO/IEC 10646.
The Unicode Working Group (predecessor to the Unicode Technical Committee) examined all available TC 46 standards and some DIS versions during the initial development of the Unicode Standard. Some TC 46 characters were not incorporated, because the Unicode Standard (like ISO/IEC 10646) conforms to the Character/Glyph Model.
Since the ISO/TC 46 character sets are used in bibliographic records, reference is made to US cataloging rules where appropriate. Cataloging rules published by the Library of Congress and/or the American Library Association are almost universally used by American libraries. Many libraries outside the United States also follow US practice, because of the availability of cataloging records from the Library of Congress and other US sources.
0220 LATIN CAPITAL LETTER KRA
No proof of the existence of an uppercase form of kra as an independent character in actual use has been presented.
0221 LATIN CAPITAL LETTER YR
This character is already in ISO/IEC 10646:1 as LATIN LETTER YR, encoded at position U+01A6.
LATIN SMALL LETTER YR (encoded at position 79 in ISO 5426-2:1996) has been misidentified as U+01A6 LATIN LETTER YR.
Proof of the existence of a lowercase form of yr as an independent character in actual use should be provided if the lowercase form is to be proposed for addition to ISO/IEC 10646.
NCITS/L2 and the Unicode Consortium note that the British Library Special character set which was a primary source for ISO 5426-2:1996 is consistent with the current repertoire of ISO/IEC 10646. The British Library set (shown in Figure 1) includes only an x-height GREENLANDIC K (code E3) and an h-height SWASH R (code EF), but not the inverse equivalents with respect to height.
03DB GREEK SMALL LETTER STIGMA
03DD GREEK SMALL LETTER DIGAMMA
03DF GREEK SMALL LETTER KOPPA
03E1 GREEK SMALL LETTER SAMPI
The character repertoire of Version 1.0 of The Unicode Standard included these characters. Examples of use have been included in N 1743 for all except small letter sampi.
These characters should be added to ISO/IEC 10646 with the recommended code values and names.
Numeric Signs
0487 CYRILLIC TEN THOUSANDS SIGN
0488 CYRILLIC HUNDRED THOUSANDS SIGN
0489 CYRILLIC MILLIONS SIGN
NCITS/L2 and the Unicode Consortium recommend:
Note that all these characters are enclosing.
Comment on validity of these Cyrillic signs:
These signs are documented in the table for the transliteration of Church Slavic in the ALA/LC Romanization Tables (Library of Congress, 1991), p. 40-42. The ALA/LC tables were developed jointly by the Library of Congress and the American Library Association, and are used by libraries world-wide. Individual tables are evaluated by language experts during development, as well as afterwards, through actual use.
Cyrillic Characters for non-Slavic Languages
04C5 CYRILLIC CAPITAL LETTER CHECHEN KA
through
0519 CYRILLIC SMALL LETTER KOMI TJE.
NCITS/L2 and the Unicode Consortium oppose addition of the characters in the following ranges to ISO/IEC 10646 because they duplicate existing characters: 04C5-04C6, 04C9-04CA, 04CD-04CE, 04EC-04ED, 04F6-04F7, 04FA-04FD, 0508-0515. For details, see Appendix.
The Unicode Technical Committee and NCITS/L2 consider that the remaining letters (04FE-04FF, 0500-0507, 0516-0519) are doubtful and should not be added to ISO/IEC 10646 at this time. Further study is needed to determine whether they are truly unique or are glyphic variants of existing letters. The first pair (04FE-04FF) are from the Missionary orthography of the Mordvin-Moksha Dialect and could be the case forms of a typographic digraph; the others are all from the 1919 (Molodtsov) orthography for the Komi language.
Background comments
Many of the characters from ISO 10754 proposed for addition to ISO/IEC 10646 are glyphic variants of existing characters. It should be borne in mind that the initial development of ISO 10754 predates the development of the character/glyph model. ISO 10754 encodes what is seen, not the underlying conceptual character.
The case to add a variant form as a distinctive character in ISO/IEC 10646 must be supported by detailed, printed evidence that there is actual, contrastive use of the different forms. The case must also be made as to how Cyrillic script differs from other scripts where unification across languages applies, e.g., Nastaliq forms are not encoded for Urdu written in Arabic script, and language-specific forms of ideographs are encoded only because of the Source Separation Rule.
The ALA/LC Romanization Tables display the exact form preferred for a particular language (i.e., what is seen) because this is essential for romanization (where a Latin script string is substituted for the original Cyrillic character). The preferred letter form for a language must be shown so that the cataloger can determine the proper Latin script substitute. The fact that the ALA/LC Romanization Tables display various language-specific letter forms should have no bearing on the content of a Cyrillic character repertoire formulated according to the Character/Glyph Model.
22F2 VECTOR OR SUM
22F3 VECTOR PRODUCT
22F4 SUM OR UNION OF CLASSES OR SETS
22F5 PRODUCT OF INTERSECTION OF CLASSES OR SETS
22F6 IS INCLUDED IN SET
22F7 INCLUDES IN SET
These characters are from Table 2, Extension of Basic Set G0 of ISO 6862:1996. ISO DIS 6862 was a source for Version 1.0 of The Unicode Standard. In the opinion of the Unicode Working Group (predecessor to the Unicode Technical Committee), these characters replicated characters in Table 1, Basic Set G0. The Unicode Technical Committee re-examined the characters and came to the same conclusion. NCITS/L2 concurs.
NCITS/L2 and the Unicode Consortium oppose addition of this whole collection to ISO/IEC 10646 because they duplicate existing characters.
Right and Left Descenders
0346 COMBINING RIGHT DESCENDER
0347 COMBINING LEFT DESCENDER
These combining marks are intended for use with Cyrillic letters (as described in Clauses 6.2 and 6.4 of ISO 10754). Table A.2 in ISO 10754 documents combinations of Cyrillic letters and descenders. The ISO 10754 sequence of a combining descender followed by a letter can be mapped to extant characters in ISO/IEC 10646:1, except for one case pair which is proposed for addition in N 1744. The combining descenders are therefore redundant.
NCITS/L2 and the Unicode Consortium also oppose addition of these characters to ISO/IEC 10646, because of the serious implications their addition would have for decomposition, which:
Combining Small Letters Above
0348 COMBINING LATIN SMALL LETTER A ABOVE
0349 COMBINING LATIN SMALL LETTER E ABOVE
034A COMBINING LATIN SMALL LETTER R ABOVE
034B COMBINING LATIN SMALL LETTER Z ABOVE
The only documentation for these characters is ISO 5426-2:1996 itself.
The proposed character COMBINING LATIN SMALL LETTER E ABOVE is an early form of the umlaut. US cataloging practice is to substitute the umlaut (Descriptive Cataloging of Rare Books, p. 69).
NCITS/L2 and the Unicode Consortium oppose addition of a character that is essentially a glyphic variant of the umlaut.
The remaining letters appear to be Latin contractions (see comments on N 1747 below).
Multiple Diacritical Marks
034C COMBINING DOUBLE CARON
034D COMBINING DOUBLE CIRCUMFLEX
034E COMBINING CIRCUMFLEX GRAVE
The only documentation for these characters is ISO 5426-2:1996 itself.
These marks appear to be Latin contractions (see comments on N 1747 below).
00 LATIN CONTRACTION AGUS (et, ond)
through
15 LATIN SMALL CONTRACTION LONG S WITH HOOK
also, 2048 REVERSED SECTION SIGN in N 1748, and possibly many of the combining characters in N 1746.
LATIN CONTRACTION AGUS
Mr. Michael Everson has said that this character is the modern Irish equivalent of the ampersand.
The British Library’s "Special" character set from which ISO 5426-2 is partially derived calls this character "Ampersand form type 2." Descriptive Cataloging of Rare Books (2nd. ed., Library of Congress, 1991) instructs the cataloger (p. 69):
If the Tironian sign (<image of Tironian sign>) cannot be reproduced, treat it as an abbreviation and substitute "[et]" for it.
This existence of this character is documented by three sources. The US rules for rare book cataloging prescribe its use. It should be added to ISO/IEC 10646, but as a General Punctuation character with the proposed name TIRONIAN SIGN.
Mappable Latin Contractions
06 LATIN CONTRACTION REVERSED US
07 LATIN CONTRACTION IS
08 LATIN CONTRACTION SMALL IS
09 LATIN CONTRACTION UM
These characters can be mapped to existing UCS characters and so should be eliminated from the proposal.
|
Prop. No. |
ISO 5426-2 |
ISO/IEC 10646:1 |
||
|
Code |
Name |
Code |
Name |
|
|
09 |
02/14 |
CONTRACTION MARK LATIN CAPITAL LETTER SCRIPT I |
U+2110 |
SCRIPT CAPITAL I |
|
08 |
02/15 |
CONTRACTION MARK HEAVY APOSTROPHE |
U+02BC |
MODIFIER LETTER APOSTROPHE |
|
07 |
03/12 |
CONTRACTION MARK LATIN SMALL LETTER SCRIPT OPEN E |
U+025C |
LATIN SMALL LETTER REVERSED OPEN E |
|
06 |
03/13 |
CONTRACTION MARK LATIN SMALL LETTER SCRIPT E |
U+212F |
SCRIPT SMALL E |
General Comments about Latin Contractions
US practice for rare book cataloging is to substitute the spelled-out equivalent for the contraction. Rule 0J2 of Descriptive Cataloging of Rare Books (p. 7) states:
When special marks of contraction have been used by the printer in continuance of the manuscript tradition, expand affected words to their full form and enclose supplied letters in square brackets. When an abbreviation standing for an entire word appears in the source, record instead the word itself, and enclose it in square brackets.
(The exception to this rule, noted above, is the Tironian sign.)
The US position is that these contraction signs should not be added to ISO/IEC 10646. Contraction signs of this general sort might be useful additions as glyphic forms for fonts used in the facsimile reproduction of manuscripts and early printed books, but their encoding as characters for the representation of textual content is counterproductive. Their use for the representation of text would cause searching and comparison problems in electronic texts and cataloging records so encoded.
Furthermore, the Latin contractions in ISO 5426-2 are only a small subset of the contractions used over the centuries in manuscripts and later included in some early printed works. Contractions are not exclusive to Latin manuscripts. Examples of Cyrillic contractions appear in the documentation accompanying N 1744. There is no good rationale to merely pick out the small subset of such manuscript contraction forms present in ISO 5426-2 and propose them for encoding, as opposed to any other set which could be brought forward.
While the Unicode Consortium and NCITS/L2 do not recommend the encoding of any of these contraction signs (other than the agus/Tironian sign) in ISO/IEC 10646, if WG2 chooses to accept any of the remaining characters proposed in N 1747 (i.e., 01-15 and 0A-15), then they should:
The names proposed in N 1747 are generally unacceptable.
2048 REVERSED SECTION SIGN
2049 REVERSED PILCROW SIGN
2139 LATIN CAPITAL LETTER ROTATED Q
2183 ROMAN NUMERAL REVERSED ONE HUNDRED
2614 SIX-SPOKED ASTERISK
2615 BLACK LEFTWARDS BULLET
2616 BLACK RIGHTWARDS BULLET
2768 REVERSED ROTATED FLORAL HEART BULLET
Editorial: The first two cells in the code chart extract are misnumbered.
Problematic Characters
2048 REVERSED SECTION SIGN
This character is a Latin contraction, not a signature mark.
The image of this character is incorrect (as is its name, which is based on the incorrect image). ISO 5426-2:1996 says that this is ‘Used for the Latin suffix "orum".’ The British Library’s "Special" character set includes a character "-RUM WORD ENDING TYPE 2" which may be the origin of this character.
If characters from ISO 5426-2 representing Latin contractions are added to ISO/IEC 10646:1, this character should not be added until it has been properly identified.
2614 SIX-SPOKED ASTERISK
CHASE is an EU-funded project to develop Unicode/UCS mappings for character sets used in European libraries. CHASE has mapped the British Library’s equivalent of this character to U+2736, SIX POINTED BLACK STAR. This character should be eliminated from the proposal.
Acceptable Characters
The following characters should be added to ISO/IEC 10646:1, each with the proposed name and code value:
2049 REVERSED PILCROW SIGN
2139 LATIN CAPITAL LETTER ROTATED Q
The image of this character should more closely resemble to the British Library source character.
2183 ROMAN NUMERAL REVERSED ONE HUNDRED
This character is not a section symbol but a Roman numeral. Mr. Michael Everson provided convincing examples of use.
The following characters should be added to ISO/IEC 10646:1, with the proposed names but in the General Punctuation block.
2615 BLACK LEFTWARDS BULLET
2616 BLACK RIGHTWARDS BULLET
2768 REVERSED ROTATED FLORAL HEART BULLET
General Comments about Signature Marks
The marks used by printers to identify signatures (also called collations or gatherings) are important for the study of early printing. The repertoire of signature marks in ISO 5426-2 is only a small subset of the marks used by printers. Since ISO/IEC 10646 is intended to encode plain text, and the particular font used in a signature mark may be significant, the full and correct representation of signatures marks cannot be achieved without the use of a higher level protocol.
The Unicode Consortium and NCITS/L2 note that the bullets and the reversed pilcrow sign are general-purpose punctuation marks.
05F5 HEBREW ACCENT TSERE
through
05FC HEBREW ACCENT ASTERISK
These proposed characters are from Table 2: Special vowel points, accents, and marks of ISO
8957:1996. The contents of Table 2 are historic cantillation marks. There are no known implementations of this table.
The decision to equate certain Hebrew cantillation marks with diacritical marks (for example, 41 HEBREW ACCENT TSERE was mapped to U+0308 COMBINING DIAERESIS)
is questionable
WG1 also mapped 5A HEBREW ACCENT RAFE to U+05BF HEBREW POINT RAFE, even though the two characters had different names and images. Furthermore, 4C HEBREW POINT RAFE in Table 1 is mapped to U+05BF HEBREW POINT RAFE, so that the net effect is to unify 4C in the Basic Hebrew alphabet and 5A in the Special vowel points, accents, and marks.
NCITS/L2 and the Unicode Consortium recommend that the addition of characters from Table 2 of ISO 8957:1996 be deferred. This will allow experts on cantillation marks and historic pointing and scholars from Israel and from other countries to clarify the nature of specific marks, and to determine the correct mappings for certain TC 46 characters. This will allow a complete proposal for any missing characters to be developed.
Background information on library implementations of Hebraic script
Except for the name of one character, Table 1: Basic Hebrew alphabet corresponds to the USMARC Hebrew character set published by the Library of Congress. It is used by US libraries with significant collections of Hebraica.
This character set was developed by the Research Libraries Group and subsequently adopted by the Library of Congress for USMARC. To determine the repertoire of characters, RLG consulted librarians responsible for Hebraica collections. The librarians unanimously rejected the inclusion of cantillation marks.
A European implementation of Table 1 is under development for the Consortium of European Research Libraries.
In Israel, library data is encoded using an SII character set similar to ISO 8859-8, rather than ISO 8957. Nevertheless, the Standards Institution of Israel did participate in the development of ISO 8957.
ISO 5426-2:1996, Information and documentation – Extension of the Latin alphabet coded character set for bibliographic information interchange – Part 2: Latin characters used in minor European languages and obsolete typography.
ISO 6862:1996, Information and documentation – Mathematical coded character set for bibliographic information interchange.
ISO 10754:1996, Information and documentation – Extension of the Cyrillic alphabet coded character set for non-Slavic languages for bibliographic information interchange
ALA/LC Romanization Tables: Transliteration Schemes for Non-Roman Scripts. Tables compiled and edited by Randall K. Barry. Library of Congress, 1997.
Descriptive Cataloging of Rare Books, 2nd. ed., Library of Congress, 1991.
R. S. Gilyarevsky & V. S. Grivnin, Languages Identification Guide. Nauka, 1970.
Appendix: Characters from ISO 10754:1996
Disapprove addition for the reason shown in bold face:
04C5 CYRILLIC CAPITAL LETTER CHECHEN KA
04C6 CYRILLIC SMALL LETTER CHECHEN KA
Already encoded. The uppercase Chechen ka is U+041A + U+030A and the lowercase Chechen ka is U+043A + U+030A.
Editorial: The images at 04C5 and 04C6 in N 1744 are incorrect; the appendage is a circle, not a curve.
04C9 CYRILLIC CAPITAL LETTER CHUVASH NG
04CA CYRILLIC SMALL LETTER CHUVASH NG
Already encoded. The Chuvash ng is a glyphic variant of either of these Cyrillic letters: nje or en with descender. (ALA/LC Romanization Tables, p. 118)
04CD CYRILLIC CAPITAL LETTER KOMI NG
04CE CYRILLIC SMALL LETTER KOMI NG
Already encoded. The Komi ng is a glyphic variant of the Cyrillic letter nje. (ISO 10754:1996, Table A.1, also ALA/LC Romanization Tables, p. 123)
Annex B of ISO 10754:1996 states: In some cases, early efforts to "cyrillicize" the writing of certain languages made use of some letters from other alphabets." The following four characters exemplify this.
04EC CYRILLIC CAPITAL LETTER SELKUP OE
04ED CYRILLIC SMALL LETTER SELKUP OE
Already encoded.
04F6 CYRILLIC CAPITAL LETTER AISOR EL
04F7 CYRILLIC SMALL LETTER AISOR EL
Already encoded.
04FA CYRILLIC CAPITAL LETTER KURDISH QA
04FB CYRILLIC SMALL LETTER KURDISH QA
Already encoded.
04FC CYRILLIC CAPITAL LETTER KURDISH WE
04FD CYRILLIC SMALL LETTER KURDISH WE
Already encoded.
0508 CYRILLIC CAPITAL LETTER YAKUT I WITH STROKE
0509 CYRILLIC SMALL LETTER YAKUT I WITH STROKE
Already encoded. The uppercase Yakut I with stroke is U+0406 + U+0335 and the lowercase Yakut I with stroke is U+0456 + U+0335.
050A CYRILLIC CAPITAL LETTER JE WITH STROKE
050B CYRILLIC SMALL LETTER JE WITH STROKE
Already encoded. The uppercase Je with stroke is U+0408 + U+0335 and the lowercase Je with stroke is U+0458 + U+0335.
050C CYRILLIC CAPITAL LETTER KOMI ELJ
050D CYRILLIC SMALL LETTER KOMI ELJ
Already encoded. The Komi elj is a glyphic variant of the Cyrillic letter lje. . (ISO 10754:1996, Table A.1, also ALA/LC Romanization Tables, p. 123)
050E CYRILLIC CAPITAL LETTER EL WITH MIDDLE HOOK
050F CYRILLIC SMALL LETTER EL WITH MIDDLE HOOK
Already encoded. The el with middle hook is a glyphic variant of the Cyrillic letter lje. (ALA/LC Romanization Tables, p. 118)
0510 CYRILLIC CAPITAL LETTER MORDVIN EL KA
0511 CYRILLIC SMALL LETTER MORDVIN EL KA
ALA/LC: Mordvin-Moksha dialect (1923) (=Lkh/lkh with dot on h)
Typographical digraph. "The letter combinations
k[, h[ are characteristic of the Mordvin-Moksha language." (Gilyarevsky & Grivnin, p. 42). The discrete letters are already encoded.Editorial comment: These characters should be named CYRILLIC CAPITAL LIGATURE EL HA and CYRILLIC SMALL LIGATURE EL HA respectively, based on (a) the quoted information, and (b) the ALA/LC romanizations for the case forms. In the basic Cyrillic letters at the beginning of the ALA/LC romanization table for non-Slavic languages (p. 114),
K/k is romanized as L/l, {/[ is romanized as Kh/kh, which agrees with the romanizations for the case forms of the digraph in the table for Mordvin-Moksha Dialect (1923) (p. 127), i.e., Lkh/lkh (with dot on h).0512 CYRILLIC CAPITAL LETTER EN WITH MIDDLE HOOK
0513 CYRILLIC SMALL LETTER EN WITH MIDDLE HOOK
Already encoded. The letter is listed for Chuvash and Khanty (Vahk) in Table A.1 of ISO 10754:1996, but is not included in ALA/LC romanization tables for these alphabets. Treat as a glyphic variant of the Cyrillic letter nje.
0514 CYRILLIC CAPITAL LETTER MORDVIN ER KA
0515 CYRILLIC SMALL LETTER MORDVIN ER KA
Typographical digraph. "The letter combinations
k[, h[ are characteristic of the Mordvin-Moksha language." (Gilyarevsky & Grivnin, p. 42). The discrete letters are already encoded.Editorial comment: These characters should be named CYRILLIC CAPITAL LIGATURE ER HA and CYRILLIC SMALL LIGATURE ER HA respectively, based on (a) the quoted information, and (b) the ALA/LC romanizations for the case forms. In the basic Cyrillic letters at the beginning of the ALA/LC romanization table for non-Slavic languages (p. 114),
H/h is romanized as R/r, {/[ is romanized as Kh/kh, which agrees with the romanizations for the case forms of the digraph in the table for Mordvin-Moksha Dialect (1923) (p. 127), i.e., Rkh/rkh (with dot on h).
The following proposed characters require further study:
04FE CYRILLIC CAPITAL LETTER YA IE
04FF CYRILLIC SMALL LETTER YA IE
0500 CYRILLIC CAPITAL LETTER KOMI DE
0501 CYRILLIC SMALL LETTER KOMI DE
0502 CYRILLIC CAPITAL LETTER KOMI DJE
0503 CYRILLIC SMALL LETTER KOMI DJE
0504 CYRILLIC CAPITAL LETTER KOMI DZE
0505 CYRILLIC SMALL LETTER KOMI DZE
0506 CYRILLIC CAPITAL LETTER KOMI ZJE
0507 CYRILLIC SMALL LETTER KOMI ZJE
0516 CYRILLIC CAPITAL LETTER KOMI ESJ
0517 CYRILLIC SMALL LETTER KOMI ESJ
0518 CYRILLIC CAPITAL LETTER KOMI TJE
0519 CYRILLIC SMALL LETTER KOMI TJE