RLG is converting all bibliographic records in the RLIN database which are coded Chinese (CHI) in the 008/35-37 field, beginning with clusters containing Library of Congress records having library identifiers DCLP or DCLC. OCLC has converted Chinese serial records, and is now converting bibliographic records coded Chinese (CHI) in the 008/35-37 field, beginning with the most recent records; then proceeding to convert other records in WorldCat.
The conversion program either converts what it finds in a subfield, or analyzes the subfield to determine if it should proceed to convert the syllables therein. All variable fields will be subject to conversion, following Library of Congress specifications, with the following exceptions: 210, 222, 246$i, 520, 546, 561$a, 60030$a (family names), and 653.
Go to Table of ContentsChinese romanization will be consistent on the vast majority of bibliographic records after they have been converted. However, because data dictionaries and conversion sequences were not applied in the same manner to all subfields, certain syllable strings will convert differently in different subfields. For example, if mixed text procedures have been applied to subfield A but not to subfield B, a certain string of syllables in subfield A may not convert, while the same string in subfield B would. Because a data dictionary is applied to a subject heading but not a descriptive string, the subject heading may convert correctly but the same term appearing in a title may not. Most of the bibliographic records containing such inconsistencies will be marked for review. Some other problems may require special searching after conversion.
We have tried to explain these inconsistencies in the following section of the pinyin home page. If you encounter an inconsistency in conversion that you don't understand, or have difficulty in constructing searches to find all instances of a certain kind of conversion, please contact the Library of Congress or your users group for assistance.
A. Marker
A local MARC field 987 has been added to all bibliographic records being converted to pinyin by machine program at RLG and OCLC. After October 1, 2000, libraries should add the 987 field to all records that are coded Chinese in the 008/35-37 language field, or any other bib record containing romanized Chinese characters. The 987 field shows that a record has been created in pinyin. That record will not then be converted by the utilities' conversion programs. Thus, adding field 987 to a record eliminates the potential for possible erroneous conversion.
For example, the marker would be added to bibliographic records under these circumstances:
a fully romanized Chinese recorda romanized name appears in a statement of responsibility, with a corresponding romanized added entry
a romanized name appears in a statement of responsibility, but the corresponding added entry is not given in romanized form
a note includes romanized Chinese text, but there are no romanized Chinese access points on the record
Updated instructions, March 20, 2002
Continue adding a pinyin marker to all records that are coded Chinese in the language code.
Insofar as possible, the Library of Congress will add the pinyin marker to new non-Chinese records on which romanized Chinese is present, as well as to non-Chinese records that are converted to pinyin, until completion of the conversion project. We have found that this application of the marker makes it possible for the Library to eliminate converted non-Chinese records from being processed again.
OCLC users are being urged to add a pinyin marker to all non-Chinese records on which romanized Chinese is present.
RLIN users are not being asked to add a pinyin marker to non-Chinese records by RLG.
RLG and OCLC have substantially completed the machine conversion of bibliographic records in the RLG Union Catalog and WorldCat. Nevertheless, they urge the continued use of the 987 field until such time as there are no more Wade-Giles records being loaded into the two databases, or being created as a result of retrospective conversion projects.
B. Headings:
Headings in access points should convert in the same manner as authority records:
see Authority Records, 2. Extent of Conversion on Individual Records, for
a description of how headings on authority records converted.
The forms of heading for several place names have changed since the data
dictionaries in the conversion program were written. A list of these changed
headings will also be posted in the near future, giving the heading as it
will appear on the converted record, along with the more recent, correct forms.
The list can be used to manually update those subject headings on bibliographic
records so that they correspond with the subject authority records.
EXAMPLE:
- Headings on bib records were converted to:
- Shenzhen jing ji Tequ (Shenzhen, Guangdong Sheng, China)
- But these headings now need to changed to the current form :
- Shenzhen Jingji Tequ (Shenzhen, Guangdong Sheng, China)
C. Pre-AACR2 Headings:
Subfields in headings on bibliographic records were converted if they appeared to the conversion program to be in Wade-Giles romanization. Therefore, if a subfield in a pre-AACR2 heading was used on a bibliographic record that is converted by the machine program, it probably was converted to pinyin.
D. Subject headings:
D1. Personal Names (600 fields), Meetings (611 fields)
Subject headings for personal names and meeting names converted in the same manner as they did in other access points. OCLC flagged 611$c names containing "Taiwan" for manual review, and "Taiwan" is noted in the 987$f subfield.
D2. Corporate headings (610 fields), Uniform titles (630 fields), Topical Subject Headings (650 fields), Geographics (651 fields)
Topical subject headings (650 fields) are converted by data dictionary only: if a topical subject heading in a 650 field exactly matches an entry in the data dictionary, the conversion is made; if there is not an exact match, no conversion takes place. Examples:
650 -0 $a Ju-I (Scepters) == 650 0 $a Ru yi (Scepters)
650 -0 $a Feng-shui == 650 -0 $a Feng shui
In fields 610 (corporate bodies) and 651 (geographics), the machine program first checks data dictionaries: if there is an exact match, a conversion is made; if there is not an exact match, then the machine program tries other routines to convert a string of syllables.
EXAMPLES:
The forms of heading for a number of subject headings have changed since the data dictionaries in the conversion program were written. Also, several subject headings were not converted. Lists of these changed and unconverted headings, as well as lists of all Chinese subject headings that converted in the years 1999 and 2000, and the Chinese subject headings that were changed to descriptive headings, are posted on the pinyin home page. These lists should be used to manually correct subject headings on bibliographic records so that they correspond with headings on subject authority records. [updated November 5, 2001]
- 651 -0 $a Wei River Valley (Kansu Province and Shensi Province, China) == 651 -0 $a Wei River Valley (Gansu Sheng and Shaanxi Sheng, China)
- [exact subject match]
- 610 20 $a Yüan tung fang chih Group == Yuan dong fang zhi Group
- [exact subject match]
- 610 20 $a Yen-ch'ing ta hsüeh == Yanjing da xue
- [no exact subject match; standard conversion]
- 651 -0 $a Chen-ching shih (China) == 651 -0 $a Zhenjiang (Jiangsu Sheng, China)
- [no exact subject match, but there is an exact match in the data dictionary of conventional place names]
- 651 -0 $a Fang-shan hsien (China) == 651 -0 $a Fangshan Xian (China)
- [no exact subject match; Xian is capitalized (generic place name); then standard conversion]
D3. Headings for Regionsx30 00 $a Dunhuang manuscripts
- x30 00 $a Tun-huang manuscripts
- -- did not convert; it should convert to:
651 -0 $a Muruin Sum Reservoir (China)
- 651 -0 $a Mo-li miao Reservoir (China)
- -- has been converted to:
- 651 -0 $a Moli Miao Reservoir (China)
- -- the heading now needs to be changed to the current form:
Subject headings for regions that included certain multi-syllable generic terms were converted by data dictionary (see below, section E3). Most converted correctly when they appeared in subject headings. It appears at present that fewer than 100 Chinese bib records need to be corrected. At least that many non-Chinese records will also have to be located and corrected.
Conversion failed to occur for several different reasons. Because of an error in the specifications (which has since been corrected), the heading Sinkiang Uigur Autonomous Region usually did not correctly convert to Xinjiang Uygur Zizhiqu in 651$a or 650$z subfields. Some headings for regions did not convert because of typographical errors; others because they included terms for former (non-Wade-Giles) conventional names.
EXAMPLES:A few regions that should have converted as 'conventional place names' may also not have converted.
651 -0 $a Canton Region (China)… [change manually to Guangzhou Region (China)]
651 -0 $a Taiyuan Shi Region (China) [change manually to Taiyuan Region (Shanxi Sheng, China)]
650 -0 … $z Xingyi Shi Region (Guizhou Sheng) [change manually to Xingyi Region (Guizhou Sheng)]
EXAMPLE:Reviewing and cleaning up headings for regions also provides us with the opportunity to correct some existing errors in word order.
650 -0 … $z Sinkiang Uighur Autonomous Region [change manually to Xinjiang Uygur Zizhiqu]
EXAMPLES:E. Descriptive fields:
650 -0 … $z Tangshan (Hebei Sheng) Region [change manually to Tangshan Region (Hebei Sheng)]
650 -0 … $z Luoyang (Henan Sheng) Region [change manually to Luoyang Region (Henan Sheng)]
651 -0 $a Luoyang Shi Region (China) [change manually to Luoyang Region (Henan Sheng, China)]
E1. 2-syllable place names.
The conversion programs almost always identified 2-syllable place names in corporate headings. Because testing revealed that 2-syllable place names could not reliably be identified and converted in descriptive subfields, conversion sequence G2 (Two-syllable place names, in which the second syllable is a generic term) was applied to headings for place names, but not to descriptive subfields (such as the 245$a, 260$b, 500$a and 740$a subfields).
To compensate for this necessary shortcoming, records in which these sequences fired have been marked for review in the 987$f subfield. At least in these records, the descriptive strings in which these names are most likely to occur can be scrutinized and corrected if necessary. LC will correct these errors on a priority basis. When reviewing one of these records,
1) check to make sure that the access point converted correctly, andE2. Single syllable generic terms for jurisdictions
2) check descriptive fields (245, 260, 440, etc.): if the 2-syllable place name appears there, correct the romanization of the name
EXAMPLE 1:
245 00 $a Lixian zhi… [change manually to Li Xian zhi…]
245 00 $a
651 -0 $a Li Xian (Sichuan sheng, China)… [change manually to Li Xian (Sichuan Sheng, China)…]
651 -4 $a(Sichuan sheng, China)… [change manually to
987 $a PINYIN $b… $c… $d r $e… $f see descriptive cataloging for 2-syllable place name
EXAMPLE 2:
245 00 $a Huaxian zhi… [change manually to Hua Xian zhi…]
245 00 $a…
651 -0 $a Hua Xian (Henan Sheng, China)
651 -4 $a(Henan Sheng, China)
987 $a PINYIN $b… $c… $d r $e… $f see descriptive cataloging for 2-syllable place name
Single syllable generic terms for jurisdictions have not been capitalized in descriptive strings because the conversion program could not identify them reliably. Conversion sequence G1 (Single syllable generic terms for jurisdictions used in proper names) has been applied to headings for place names, but not to descriptive subfields. Records with these terms have not been marked for review in the 987 field. LC will correct these sorts of capitalization problems on an as-encountered basis.
EXAMPLE 1 from a single record:E3. Multi-syllable generic terms may have converted incorrectly
245 00 $a Xinmi shi zhi, 1986-1995 ... [change manually to Xinmi Shi zhi, 1986-1995...]
245 00 $a, 1986-1995 ...
651 -0 $a Xinmi Shi (China)
651 -4 $a(China)
987 $a PINYIN $b… $c… $d r $e… $f G1
EXAMPLE 2:
245 00 $a Chengdu shi zhi… [change manually to Chengdu Shi zhi…]
245 00 $a…
710 2- $a Chengdu Shi di fang zhi bian zuan wei yuan hui.
710 2- $a..
987 $a PINYIN $b… $c… $d r $e… $f G1
E3A. How these terms should have converted
Conversion specifications called for the identification and conversion of 10 multi-syllable generic terms for jurisdictions (see list in specification G3). Wade-Giles romanization separated these syllables, both in descriptive text and access points. The conversion programs were to identify these 10 terms when they were used as part of place names, connect the previously unconnected syllables, and capitalize the resulting term.
E3B. 3 categories of errors, with examples
These terms converted correctly in most instances. However, several kinds of errors did occur in the conversion of these multi-syllable terms on LC's Chinese bibliographic records by RLG. These errors were not marked for review in the 987 field.
1. Several strings of syllables which were not listed among the terms given in the conversion specifications were joined together erroneously. As far as we are aware today, these are the terms which have been created incorrectly, along with the number of LC Chinese records on which they occurred:
The individual syllables in these terms should be separated, not joined. The LC bib records that had these errors have been corrected.
diquan 78 diqueh 5 dujiaqu 1 minzu 65 tequan 7 xingzhengquan 0 zizhiquan 1 zhuanquan 4
EXAMPLES:2. In certain instances, strings of syllables given in the specifications were joined in parts of the bibliographic record where they should have been left separate. We have just begun to identify the scope of the problem. We have searched for all of the terms given in the specification. The following chart shows the 10 multi-syllable terms, the number of LC Chinese records on which these converted terms appear, and, insofar as we can estimate at this time, the approximate number that will have to be corrected.
245 00 $a Bi qiu ni Zhuanquan ji [change manually to Bi qiu ni zhuan quan ji]
245 00 $a
240 10 $a Zuo zhuan
240 10 $a
245 10 $a Zuo Zhuanquan yi [change manually to Zuo zhuan quan yi]
245 10 $a
245 10 $a Qian yi shi Diquan shi [change manually to Qian yi shi de quan shi]
245 10 $a
3. Following the specifications, the conversion programs occasionally converted a multi-syllable term in the wrong context. In other words, even though the terms were converted accurately according to the specifications, correction is still needed.
TERM HITS NEED CORRECTION Diqu 1670 ca. 1100 Dujiaqu 1 0 Tequ 88 ca. 55 Xingzhengqu 75 ca. 40 Zhuanqu 11 2 ziran 33 0 Zizhiqi 11 0 Zizhiqu 11 0 Zizhiqi 1300 0 Zizhixian 253 0 Zizhizhou 356 0
EXAMPLES:
245 00 $a Guan yu fa hui Diqu you shi di yan jiu [change manually to Guan yu fa hui di qu you shi de yan jiu]
245 00 $a
245 00 … $c Hua dong Diqu da xue… [change manually to Hua dong di qu da xue…]
245 00 … $c…
245 00 $a Zhongguo li dai Xingzhengqu hua [change manually to Zhongguo li dai xing zheng qu hua
245 00 $a
EXAMPLE:E4.
245 10 $a Zhongguo Xingzhengqu hua gai lun [change manually to Zhongguo xing zheng qu hua gai lun]
245 10 $a
Zu, the generic term for PeopleThe term
zu ("people") is normally separated from the "name of racial, linguistic, or
tribal (grouping) of mankind", as stated in the pinyin romanization guidelines
("Separation of Character Romanizations", section 1; example given in section
2c). The exception to this rule occurs when the name of the people, along with
the term zu, are included as part of a place name (section 2b). The exception
is made to conform with BGN romanization practice.The pinyin conversion specifications make this distinction in conversion sequence G3 (Multi-syllabic generic terms for jurisdictions used as part of a proper name). Consider the following examples:
EXAMPLE 1:E5.
245 00 $a Maowen Qiangzu Zizhixian Heihuxiang she hui diao cha bao gao
245 00 $a
440 -0 $a Qiang zu diao cha cai liao
440 -0 $a
The term zu is connected to the name of the people, Qiang, in the 245$a subfield by conversion sequence G3 because this term is part of a place name (Maowen Qiang People's Autonomous County). However, it is separated in the series statement because the title reads "Qiang People research material". Here, "Qiang People" is not part of a place name. [The conversion specifications did not call for 'xiang' to be connected to the preceding place name; this change will have to be made manually.]
EXAMPLE 2:
245 00 $a Yunnan Sheng Dehong Daizu Jingpozu Zizhizhou she hui gai kuang. $p Jingpo zu diao cha cai liao
245 00 $a. $p
The term zu is connected to the name of the people, Jingpo, in the 245$a subfield because it is part of a place name (Yunnan Province Dehong Dai People and Jingpo People's Autonomous District). The term zu is separated in the 245$p subfield because it is not part of a place name ("Jingpo People research material").
de/di.
See http://lcweb.loc.gov/catdir/pinyin/di2.htmlE6. A data dictionary may have been applied in one subfield but not another
Data dictionaries were used in certain situations by the conversion program to replace one string of syllables with another. This made it possible to accomplish non-standard conversions. For example, topical subject headings (field 650$a) were compared with terms in a data dictionary; when a match occurred, the Wade-Giles form on the bibliographic record was replaced with the pinyin form from the data dictionary. If an exact match did not occur, no conversion was carried out. Data dictionaries were also used to convert conventional place names appearing in access points such as the 110$a and 651$a subfields. The basic conversion of syllables was accomplished through use of a data dictionary.
However, it was not practicable to use data dictionaries other than the standard syllable table in descriptive subfields. For that reason, the same term appearing in a heading and in a descriptive string may have converted differently. Sometimes the results are correct, as shown in example 1; in other cases, one or more descriptive subfields may need to be corrected (as in example 2).
EXAMPLE 1:E7. Ambiguous subfields
245 00 $a Taiyuan bao wei zhan...
245 00 $a...
651 -0 $a Taiyuan (Shanxi Sheng, China)...
651 -4 $a(Shanxi Sheng, China)
987 $a PINYIN $b… $c… $d c
EXAMPLE 2:
245 00 $a Dongzhai niao lei zi ran bao hu qu ke xue kao cha ji… [change manually to Dongzhai Niaolei Ziran Baohuqu ke xue kao cha ji...]
245 00 $a...
651 -0 $a Dongzhai Niaolei Ziran Baohuqu (China)
651 -4 $a(China)
987 $a PINYIN $b… $c… $d c
The specifications direct the conversion program to process the 245$a subfield as mixed text, but to simply convert what is found in the 260$b subfield. Because the Wade-Giles syllables pei and tou were identified as being "common" to both Wade-Giles and pinyin, but representing different pronunciations (see the mixed text portion of the conversion specifications), the subfield could represent either Wade-Giles or pinyin romanization. Therefore, to be on the safe side, the conversion program did not convert the subfield, but marked it for review. The same syllables in the 260$b subfield, however, were converted to bei dou.
EXAMPLE:E8. Guangzhouese
245 00 $a Pei tou. [change manually to Bei dou]
245 00 $a.
260 $a Xianggang : $b Bei dou chu ban wei yuan hui,
260 $a: $b
,
987 $a PINYIN $b… $c… $d r $e… $f 245$a
On Chinese bib records converted in RLIN, the word Cantonese was converted to Guangzhouese when it appeared in subject headings. This term has been manually corrected on all LC records.
EXAMPLES:F. Mixed text:
650 -0 $a Guangzhouese dialects – change manually to
650 -0 $a Cantonese dialects
650 -0 $a Cookery, Chinese $x Guangzhouese style – change manually to
650 -0 $a Cookery, Chinese $x Cantonese style
The conversion program analyzes the descriptive subfields which were most likely to contain a combination of Wade-Giles syllables along with non-Wade-Giles syllables. When mixed text is identified, the program does this:
STEP 1 tries to further break down the subfield into smaller subsections;
STEP 2 it analyzes the syllables in each of the subsections;
STEP 3 it converts any of those units that contained purely Wade-Giles text;
STEP 4 it then analyzes the resulting subsections again;
STEP 5 if the subsections consist of pure Wade-Giles and entirely non-Wade-Giles syllables, the program considers the subfield to have been converted;
STEP 6 if mixed text remains in any of the subsections, then the record is marked 'r' for review
EXAMPLES:
- 500 $a Fu lu: wu sa shih ch'i wen hsüeh ta shih chi: p. 284-289. == 500 $a Fu lu: wu sa shi qi wen xue da shi ji: p. 284-289.
- [Wade-Giles text was separated from non-Wade-Giles text and converted]
- 245$a Ta T'ang pi hua / $c Li Kuo-chen pien chuan ; [lin mo T'ang Ch'ang-tung] = Magnificent frescos from the great Tang dynasty / compiled by Li Guozhen ; [reproduction Tang Changdong]. == 245 $a Da Tang bi hua / $c Li Guozhen bian zhuan ; [lin mo Tang Changdong] = Magnificent frescos from the great Tang dynasty / compiled by Li Guozhen ; [reproduction Tang Changdong].
- [the $c subfield was identified as having mixed text; the convoersion program them subdivided the subfield into smaller units; the first two units (up to the semicolon that follows Wade-Giles chuan) were converted; the units that followed could not be converted, so the record was marked; do not convert those portions, since they were transcribed from the item]
-
500 $a A collection of essays published in various journals written by Hu P'eng and Hu K'o. - [Wade-Giles text could not be separated from the non-Wade-Giles text, so the field was not converted and the record was marked; convert manually]
- 500 $a Reprint of the 1940 ed. published by Kuo hsüeh cheng li she, Shanghai.
- [Wade-Giles text could not be separated from the non-Wade-Giles text, so the field was not converted and the record was marked; convert manually]
- 245 $b A study of Chang-sha ware in Tang dynasty
- [Wade-Giles text could not be separated from the non-Wade-Giles text, so the field was not converted and the record was marked; review but do not convert because the text was not romanized but transcribed]
- 830 -0 $a Shan-tung sheng chih (Series)
- [Wade-Giles text could not be separated from the non-Wade-Giles text, so the field was not converted and the record was marked; convert manually]
- 246 14 Reminiscences of Mr. Wei Yung-ning
- [Wade-Giles text could not be separated from the non-Wade-Giles text, so the field was not converted and the record was marked; review but do not convert because the text was not romanized but transcribed]
EXAMPLES:In the example below, mixed text processing was applied to the 245$a and 580$a subfields. The 245$a subfield did not convert because the syllables were identified by the conversion program as being ambiguous (see section E below). The personal names and corporate body in the 580$a subfield were not converted because the program could not separate out a convertible string of syllables from those which are not convertible. However, Wade-Giles personal names in access points, such as the 700 fields, were always converted to pinyin (but see section 3, Excluded from Conversion, below).
- 245 10 … /$c Hsü Hüeh-lu pien chu.
- [typo Hüeh identified by conversion program as non-Wade-Giles syllable, so the field was not converted and the record was marked; convert manually]
- 246 30 $a Chung-kuo ching chi she hui fa chan chan lu¨eh
- [typo lu¨eh identified by conversion program as non-Wade-Giles syllable, so the field was not converted and the record was marked; convert manually]
EXAMPLE:G. Non-roman (880) fields: Certain qualifiers for headings in non-roman (880) fields may not convert to the same form found in their matching roman fields. The conversion program identifies conventional place names from a list (data dictionary) and converts them to a specific form. When the geographic qualifier for one of these conventional place names changes on other than a one-to-one basis, the qualifier in the parallel field may not have converted in the same manner.
245 00 $a Kuo feng pao.
580 $a Photoreprint of the periodical (3 no. a month; some numbers called also: Kouk fong po) published by Kuo feng pao kuan, Shanghai and edited by Ho Kuo-chen and Liang Ch'i-ch'ao.
700 1 $a He, Guozhen.
700 1 $a Liang, Qichao, $d 1873-1929.
987 $a PINYIN $b… $c… $d r $e… $f 245, 580
EXAMPLES:Go to Table of Contents651 -0 $a Hai-k'ou shih (China)
651 -0 $a(China) converted to
651 -0 $a Haikou (Hainan Sheng, China)
651 -4 $a
651 -0 $a An-yang shih (China)
651 -0 $a(China) converted to
651 -0 $a Anyang (Henan Sheng, China)
651 -0 $a
Headings for the personal names on the Exclusion List, numbering approximately 2400, were not converted to pinyin (see Authority Records, 3. Excluded from Conversion, for a description of the Exclusion List). The Exclusion List itself may be viewed in the conversion specifications. Certain pre-AACR2 forms of names of jurisdictions were also excluded from conversion.
Go to Table of Contents
If you use a converted pinyin heading on an unconverted bibliographic record,
you risk its being converted to another form by a bibliographic conversion
program. If you're not sure, look at the marker in the 008/07 field to see
if the heading has been converted (or is subject to manual review). It is
safe to use a converted heading on a bib record that uses a marker in the
987 field to indicate that it is in pinyin form. Conversion programs for bib
records will not further change any record with a 987 field marked "PINYIN"
in the $a subfield.
Go to Table of Contents
Library of
Congress
Library of Congress Help Desk
(11/07/01)