THE MACHINE-READABLE TEXT: MARKUP AND USE
The sections of the Workshop that dealt with machine-readable text tended to be more concerned with access and use than with preservation, at least in the narrow technical sense. Michael SPERBERG-McQUEEN made a forceful presentation on the Text Encoding Initiative's (TEI) implementation of the Standard Generalized Markup Language (SGML). His ideas were echoed by Susan HOCKEY, Elli MYLONAS, and Stuart WEIBEL. While the presentations made by the TEI advocates contained no practicum, their discussion focused on the value of the finished product, what the European Community calls reusability, but what may also be termed durability. They argued that marking up—that is, coding—a text in a well-conceived way will permit it to be moved from one computer environment to another, as well as to be used by various users. Two kinds of markup were distinguished: 1) procedural markup, which describes the features of a text (e.g., dots on a page), and 2) descriptive markup, which describes the structure or elements of a document (e.g., chapters, paragraphs, and front matter).
The TEI proponents emphasized the importance of texts to scholarship. They explained how heavily coded (and thus analyzed and annotated) texts can underlie research, play a role in scholarly communication, and facilitate classroom teaching. SPERBERG-McQUEEN reminded listeners that a written or printed item (e.g., a particular edition of a book) is merely a representation of the abstraction we call a text. To concern ourselves with faithfully reproducing a printed instance of the text, SPERBERG-McQUEEN argued, is to concern ourselves with the representation of a representation ("images as simulacra for the text"). The TEI proponents' interest in images tends to focus on corollary materials for use in teaching, for example, photographs of the Acropolis to accompany a Greek text.
By the end of the Workshop, SPERBERG-McQUEEN confessed to having been converted to a limited extent to the view that electronic images constitute a promising alternative to microfilming; indeed, an alternative probably superior to microfilming. But he was not convinced that electronic images constitute a serious attempt to represent text in electronic form. HOCKEY and MYLONAS also conceded that their experience at the Pierce Symposium the previous week at Georgetown University and the present conference at the Library of Congress had compelled them to reevaluate their perspective on the usefulness of text as images. Attendees could see that the text and image advocates were in constructive tension, so to say.
Three nonTEI presentations described approaches to preparing machine-readable text that are less rigorous and thus less expensive. In the case of the Papers of George Washington, Dorothy TWOHIG explained that the digital version will provide a not-quite-perfect rendering of the transcribed text—some 135,000 documents, available for research during the decades while the perfect or print version is completed. Members of the American Memory team and the staff of NAL's Text Digitization Program (see below) also outlined a middle ground concerning searchable texts. In the case of American Memory, contractors produce texts with about 99-percent accuracy that serve as "browse" or "reference" versions of written or printed originals. End users who need faithful copies or perfect renditions must refer to accompanying sets of digital facsimile images or consult copies of the originals in a nearby library or archive. American Memory staff argued that the high cost of producing 100-percent accurate copies would prevent LC from offering access to large parts of its collections.
THE MACHINE-READABLE TEXT: METHODS OF CONVERSION
Although the Workshop did not include a systematic examination of the methods for converting texts from paper (or from facsimile images) into machine-readable form, nevertheless, various speakers touched upon this matter. For example, WEIBEL reported that OCLC has experimented with a merging of multiple optical character recognition systems that will reduce errors from an unacceptable rate of 5 characters out of every l,000 to an unacceptable rate of 2 characters out of every l,000.
Pamela ANDRE presented an overview of NAL's Text Digitization Program and Judith ZIDAR discussed the technical details. ZIDAR explained how NAL purchased hardware and software capable of performing optical character recognition (OCR) and text conversion and used its own staff to convert texts. The process, ZIDAR said, required extensive editing and project staff found themselves considering alternatives, including rekeying and/or creating abstracts or summaries of texts. NAL reckoned costs at $7 per page. By way of contrast, Ricky ERWAY explained that American Memory had decided from the start to contract out conversion to external service bureaus. The criteria used to select these contractors were cost and quality of results, as opposed to methods of conversion. ERWAY noted that historical documents or books often do not lend themselves to OCR. Bound materials represent a special problem. In her experience, quality control—inspecting incoming materials, counting errors in samples—posed the most time-consuming aspect of contracting out conversion. ERWAY reckoned American Memory's costs at $4 per page, but cautioned that fewer cost-elements had been included than in NAL's figure.
OPTIONS FOR DISSEMINATION
The topic of dissemination proper emerged at various points during the Workshop. At the session devoted to national and international computer networks, LYNCH, Howard BESSER, Ronald LARSEN, and Edwin BROWNRIGG highlighted the virtues of Internet today and of the network that will evolve from Internet. Listeners could discern in these narratives a vision of an information democracy in which millions of citizens freely find and use what they need. LYNCH noted that a lack of standards inhibits disseminating multimedia on the network, a topic also discussed by BESSER. LARSEN addressed the issues of network scalability and modularity and commented upon the difficulty of anticipating the effects of growth in orders of magnitude. BROWNRIGG talked about the ability of packet radio to provide certain links in a network without the need for wiring. However, the presenters also called attention to the shortcomings and incongruities of present-day computer networks. For example: 1) Network use is growing dramatically, but much network traffic consists of personal communication (E-mail). 2) Large bodies of information are available, but a user's ability to search across their entirety is limited. 3) There are significant resources for science and technology, but few network sources provide content in the humanities. 4) Machine-readable texts are commonplace, but the capability of the system to deal with images (let alone other media formats) lags behind. A glimpse of a multimedia future for networks, however, was provided by Maria LEBRON in her overview of the Online Journal of Current Clinical Trials (OJCCT), and the process of scholarly publishing on-line.