Virtually all IE use systems use lexicons, and there is universal agreement that lexicons need to be adapted or tuned to new user domains. The disagreement is about what tuning implies and whether there is real IE benefit in terms of recall and precision. Those in the Information Retrieval tradition of information access are usually skeptical about the latter, since statistical measures tend to bring their own internal criterion of relevance and semantic adaptation. Researchers like Strzalkowski [53] and Krovetz [38] have consistently argued that lexical adaptation, taken as far as domain-based sense tagging, does improve IE. In this paper we intend to adapt and continue our work on lexical tuning to provide some evaluable measure of the effectiveness or otherwise of lexical tuning for IE. That term has meant a number of things: the notion (as far back as Wilks 1972 [56]) has meant adding a new sense to a lexicon on corpus evidence because the text could not be accommodated to an existing lexicon. In 1990 Pustejovsky used the term to mean adding a new subcategorization pattern to an existing sense entry from corpus evidence. In the IE tradition there have been a number ([48], [32]) of pioneering efforts to add new words and new subcategorization/preference patterns to a lexicon from a corpus as a prolegomenon to IE.