Abstract:
User surveys of printed dictionaries may be characterised as non-representative and non-realistic laboratory tests, often with retrospective questions based on memory. Log file analyses concerning the use of Internet dictionaries, on the other hand, are based on large numbers of users and look-ups. However, log file analyses have also been characterised by a juggling of numbers based on data calculations of limited direct relevance to practical and theoretical lexicography. This article proposes the development of lexicographically relevant log files for the use in log file analyses in order to give a true picture of how and why different dictionaries are employed for different purposes.
