This report is a summary of the Task Force discussion papers as well as of Task Force members' reactions to the papers.

The CEAL Task Force (TF) to review a possible change from the Wade-Giles (WG) to the Pinyin (PY) romanization system bases its recommendations upon the following parameters:

The TF is concerned with the problems of a WG/PY conversion for Chinese-language bibliographical records in which the major focus has been on describing materials written in Chinese characters. Ramifications for records written in other languages are mentioned, but not fully discussed in this report.

The TF bases its recommendations upon today's technology as well as upon information available at the time of this writing. While aware of possible future innovations, the TF has sought not to be unduly influenced by anticipated technological developments.

The TF agreed from the outset on the following:

Users will continue to need to access materials written in Chinese script by means of a romanization system.

The TF supports the Library of Congress (LC)'s plan to change to PY romanization in the near future. The TF emphasizes, however, that the change should be carried out only after a careful look at the impact of such a change on present national and local databases, on future additions to information about individual library collections, and on user access to the information.

The primary goal in the cataloging of Chinese is to faithfully transcribe the source for the user. Romanization of the original Chinese has been used in American libraries for two purposes: (1) to alphabetize the Chinese characters for filing and retrieving; and (2) to represent or to substitute the Chinese characters phonetically. Although romanization should not and cannot be used as a substitute for Chinese characters, it is an auxiliary tool needed for searching and retrieval, for filing guides, and for clarifying pronunciation.

Software has been developed that allows the transcription of Chinese script into romanized equivalents, including both the WG and PY systems. In addition, software now also allows entries input in the WG scheme to be changed to its PY counterpart, and vice versa. For example, in OCLC WorldCat, a user can input a search in PY and retrieve a record that has no PY in the original record. In essence, whatever the format of the original data, Chinese script, PY, or WG, the information can be changed to and viewed in other formats at the discretion of the user without altering the original data. It is also important to note that Chinese computing is now affordable from the smallest laptops to mainframes, and therefore multilingual computing in libraries is now possible and will soon become commonly affordable.

TF members have varying opinions about the degree to which technology will solve the problems of PY conversion. But in light of the TF's recognition of a continuing need for romanized access to Chinese materials, and in recognition of LC's plan to change to a PY scheme, the TF recommends:

1. That LC follow the guidelines for PY romanization set forth by the Chinese National Commission on Language as the model for devising their PY scheme.
2. That LC and other CEAL libraries strive to make a total retrospective change to PY if they choose to follow LC's plan.
3. That libraries explore new technologies which allow multiple input/output methods for access to Chinese materials without relying solely on one scheme.

Unresolved issue: Some libraries suggest adding parallel PY and WG access points during a transition period until PY changes are complete. Other libraries feel this is too input intensive. The TF is unable to make a recommendation on this question.

Word division basically refers to the way syllables are aggregated. For machine processing, one of three approaches is possible:

1. Word division can be ignored, with each syllable treated instead as a separate entity.
2. Word division can be achieved by adding an "aggregator" between twosyllables or by eliminating a space character between them.
3. Word division can be achieved by indirect aggregation such as adjacency searching.

1. That LC not ignore Word Division.
2. That LC use Chinese PY romanization rules as its model for determining word division.
3. The TF is unable to recommend either option 2 or 3 above as a better solution, but instead recommends that LC, in cooperation with the CEAL community, explore both possibilities for achieving word division before making a final decision.

The TF raises the issue without making a recommendation.

Split files have already been created and maintained while card catalogs are retrospectively being converted to machine readable files. Ideally the change from WG to PY will all be done on Day One, but if that goal becomes impossible, then another split file might be created until a global conversion of all records from WG to PY is completed.

The TF recommends:

1. That single libraries not maintain split OPAC files, but instead make all records PY accessible before Day One
2. That retrospective conversion of catalog cards to machine-readable form be done in PY romanization after Day One

LC will need to review and update all authority files affected by the change to PY.

The TF recommends:

1. That LC explore a way globally to do a WG/PY flip if both versions already exist in an authority record.
2. That LC preserve the "commonly known" form for names that fit that category.

LC call numbers for materials written in Chinese have been assigned according to general classification schedules. Up to now these call numbers have been based upon WG romanization. Future call numbers will be based upon PY romanization. Since almost all CEAL libraries follow LC's classification schedules, future changes in LC policy will affect all Chinese collections in North America.

The TF recommends:

That LC give clear guidelines on how it will deal with both old call numbers that follow WG romanization and the implementation of new call numbers that follow PY romanization

Unresolved issue for classification schemes:

If LC continues to assign classification numbers that reflect WG romanization, then these numbers will not reflect current practice, and everyone will be burdened with many "see references" that will become less and less clear in future years. If, on the other hand, LC adapts new numbers based upon PY romanization, then we will end up with varying editions and literary authors having two classification numbers since massive recataloging and remarking of currently cataloged materials is impossible for most libraries.

LCSH, based upon "source found", already has a mixture of WG, PY, and common-English-usage forms.

The TF recommends:

1. That LC review subject headings to resolve WG/PY conflicts, favoring the new PY form.
2. That LC give clearer guidelines for resolving WG/PY/English usage conflicts based upon the "source found" principle.

Users already have experienced difficulty in accessing bibliographic records since most younger users are unfamiliar with WG romanization. The change to PY access should make most user searches easier and therefore faster.

The TF recommends:

That LC develop on-line user education materials quickly in order to keep users informed of WG/PY changes.

Cost is an important consideration, especially since millions of WG records will be online in the national databases and in single library OPACs by the time LC begins the switch to PY. There is a strong possibility that two major costs will occur: the cost of the WG/PY switch both in the national databases and in local databases, as well as the cost of maintaining split files if that issue is unresolved at the time of the change.

The TF recommends:

1. That LC and other CEAL libraries cooperatively work with the national utilities to find a way to develop a cost effective global conversion project.
2. That LC and other CEAL libraries strive to avoid duplication of efforts when making the change to PY.
3. That LC and other CEAL libraries explore cooperative ways to reduce the cost of loading nationally converted records into local databases.
4. That LC and other CEAL libraries make a special effort to estimate the cost of a front-end program that can search both PY and WG entries, so that libraries will have alternative choices when the cost of conversion becomes clear.

CEAL TASK FORCE ON PINYIN CONVERSION

Karl Lo (Chair), University of California, San Diego
Timothy Connor, Harvard University
John DeFrancis, University of Hawaii
Martin Heijdra, Princeton University
Karl Kahler, Univesity of Pennsylvania
Tai-loi Ma, University of Chicago
Amy Tsiang, University of California, Los Angeles
Peter Zhou, University of Pittsburgh