Search results

All of us in the library community are familiar with the impact of the MARC (machine‐readable cataloging) record on library operations. Whether we specialize in administrative, systems, public, or technical services, we recognize the MARC record as a standard data storage format that has made it easier to produce, store, and retrieve data. Few disagree that widespread adoption of the MARC record has enabled libraries to realize the full benefits of automation.

Libraries must actively support humanities text files, but we must remember that to focus exclusively on texts tied to specific systems is to put ourselves in opposition to the needs of the researchers we intend to serve. A working model of the sort of system and resource provision that is appropriate is described. The system, one put in place at the University of Michigan, is the result of several years of discussions and investigation. While by no means the only model upon which to base such a service, it incorporates several features that are essential to the support of these materials: standardized, generalized data; the reliance on standards for the delivery of information; and remote use. Sidebars discuss ARTFL, a textual database; the Oxford Text Archive; InteLex; the Open Text Corporation; the Text Encoding Initiative (TEI); the machine‐readable version of the Oxford English Dictionary, 2d edition; and the Center for Electronic Texts in the Humanities.

Working with documents in electronic format is inherently different from dealing with materials in print; nor can all electronic formats be considered equivalent. Processing and presenting SGML is not the same as processing and presenting materials in other markup or word processing formats. To maximize flexibility and extensibility, SGML is highly modular, which complicates implementation. Its emphasis on content structure rather than appearance enhances searchability but makes consistent and precise display difficult. Mechanisms used to maximize platform and software independence (e.g., entities, link protocols), though effective, can be used incorrectly or in ways difficult to implement on some systems or using certain software. Difficult questions remain for libraries planning to implement SGML.