4 User Survey

To complement the study reported above, we though it would be useful also to survey the current user community by means of questionnaire. This was posted on the Lancaster website, and its presence was widely publicized during September/October 1997. Email was sent to a large number of relevant public bulletin boards and mailing lists, around the world urging interested research centres or individuals to make their views known by visiting the web site and completing a form.

The questionnaire stated that ``we are gathering feedback on what the users of corpora would like to see encoded within publically available corpora. We have devised a questionnaire which broadly outlines all of the features available in the TEI Lite markup scheme. We would like to know which of the features you would like to see being used in the encoding of corpus data. Please specify for each feature whether it is of absolute importance, or whether you would like it if possible. '' and contained a series of tables, of which users were invited to complete as many as they wished. (In practice most respondents actually filled in all of them). Section 12 below lists the results for each table. Despite the length of the questionnaire, we received a total of 26 responses during the month that the survey was carried out, from corpus building centres world wide.

In addition to questions about their preferred method of delivery, and encoding system, respondents were asked to state the extent to which they would prefer corpora to mark up each of a large number of text features. A total of 137 features were grouped into 40 `header features', 42 `primary features' and 55 `morphosyntactic features'. For each of these, respondents could indicate a preference on a four point scale, valued `essential', `if possible', `no opinion' and `don't want'. Detailed results are given in Table III below, and summarised in the next section.