Breakdowns in writing intentions when simultaneously deploying SGML-marked texts in hard copy and electronic copy

Many writers of technical documentation must consider two different presentation media, namely traditional printed books and electronic forms. This appears to be a long-term situation, not a transitional phase: for some reading tasks, hard copy will be preferred, but for others, electronic copy will be preferred. In some settings, it is thus necessary to prepare material that is of high quality in both media, often with the constraint that a single source file be used. The problem is to specify the structure of a text so that whether it is printed or deployed electronically, neither version contains textual problems caused by its dual role. Several examples are presented to show how a writer's structuring intentions can be effective in hard copy but not in electronic copy. The difficulty of preserving structuring intentions in both media stems from declarative markup languages that are rhetorically impoverished. While standard markup languages can be used to specify what text elements comprise a text, they cannot be used to specify the intended roles of the text elements. To preserve structuring intentions, it is proposed that a rhetorical markup language is needed. Two potential advantages of such a language are improved media-transferabitity and improved visibility of text structure.