Because markup languages are capable of mapping every detail of a text, they are especially useful for tagging dictionaries, which are among the most densely-structured of all texts. However, DTDs which have been developed for encoding a range of dictionaries must have a wide scope, with a consequent loss of descriptive power. Even a DTD which has been tailored for a specific dictionary must still allow a great level of flexibility to capture the variations in its structure.

This paper describes how greater precision can be attained if the dictionary editors themselves tag as they compose. Author-tagging not only enables the markup to be designed according to a specific lexicographic approach, but also has a number of significant advantages for the editors: by applying constraints to the writing process, it facilitates the development of a consistent style, and it constitutes a lexicographic tool, by allowing the editors to search the XML documents during composition, and to include annotations and other revision material.

The discussion is illustrated with examples taken from a DTD and its associated XSL stylesheets which have been developed for a Greek-English lexicon currently being written at Cambridge University.