Encoding, Interpretation and Theory

The object of the panel is to discuss the extent to which SGML (and the use of SGML by the Text Encoding Initiative) impose pre-set interpretations on texts that are to be encoded, and its unintended effects on the computer processing and internal representation of texts in textual databases. The issue arose in public on the Humanist Discussion in December 1995 when some discussants maintained that SGML's syntax imposes an interpretive framework on text that in some ways violates the way humanities scholars approach such texts. In addition to this criticism, others, including Darrell Raymond at the University of Waterloo, have charged that SGML and the TEI make poor use of computer technology in general, and database technologies and intelligent servers in particular. Because SGML is coming into widespread use in the publishing industry, and the TEI Guidelines are emerging as important in the library and scholarly communities, the time has come to discuss the strengths, weaknesses, interpretations and misunderstandings about SGML and the TEI Guidelines in particular, and encoding, interpretation and theory in general.

This panel will raise some issues that will be of central importance to scholars, whether or not they are involved in markup by making clear what encoding is, by discussing the theoretical foundations of formal and meta languages, and by exposing the theoretical underpinnings of the TEI with the intention of assessing its consequences, both good and not-so-good, for encoding projects.

The panel will help clear the air of misconceptions regarding markup, and provide them with a set of theoretical and operational principles that will help them when deciding on markup. For instance, there is widespread anecdotal evidence to suggest that many well-intentioned practitioners are conflating SGML with TEI, procedural and descriptive markup, and analytic from presentational. Such confusions lead people to believe that SGML and the TEI Guidelines can function in ways for which they were never intended, or to believe that some text types or textual features cannot be encoded using SGML or the TEI Guidelines.

The session will engender discussion on just what formal languages are, the relationship between meta-languages and the languages they define, and whether interpretive frameworks are embedded in meta-languages and if they carry over into defined languages. This is of central concern to the encoding community because it implies that the encoding to ols at their disposal force them to make interpretations, both conscious and implied, that they would not otherwise make. For instance, one can argue that SGML is no more than a formal language for defining a syntax of markup languages and that, in itself, it is not interpretative. One might object, however, by maintaining that a formal language or metalanguage defines rules of syntax, legality is not the issue in all textual objects, and thus SGML imposes a high-level interpretive framework on documents. Moreover, SGML invites us to organize information hierarchically and makes us particularily conscious of overlapping between such hierarchies, a concern not at all obvious to anyone working with a syntax which does not encourage hierarchical structuring, or in an environment (such as the encoding of some spoken texts) where hierarchy is not necessaily the issue. SGML, the argument continues, might be suitable for the Department of Defense or to publishers and has within it their concerns which are not shared by working scholars.

Finally, the session will address the theoretical models underpinning the TEI itself and its work. One line of argument here is that, rather than being constrained, the TEI-DTD, in trying to be all things to all people, is not constrained enough. To really work with TEI encoded texts, one must invest considerable time/money into building complex parsers that can read DTDs and then parse TEI instantiations. As such, TEI serves neither the individual scholar, who rarely has the technical capabilities to decode TEI texts nor the large data provider who needs to be able accurately and cheaply to parse large numbers of text.

The panelists will prepare position papers in advance of the meeting, which will be made available to participants on the World Wide Web in advance of the meeting. Our intention is that there will be an intelligent discussion about encoding in general, and not a tutorial about SGML or the TEI Guidelines.