Computers have, in the last twenty years, changed from experts' machines to tools used by anyone for a variety of purposes. This shift has brought the need for translation, or localisation, of the software and documentation. The use of computer programs is studied in the field of human-computer interaction, and manufacturers make usability studies for their programs before shipping them, but research with a linguistic point of view has started only recently. The data in the usability studies is most often collected from think-aloud protocols. This paper gives some suggestions for alternative methods for collecting data in the aim of studying the use of the original and localised versions of word-processor software and its documentation. The methods include the usage of log files to provide quantitative data and pair work to provide qualitative data.