At present, the easiest (and frustratingly traditional) way out of this dilemma for the individual researcher is to ‘freeze’ at least core sections of one’s web data in paper-based or permanent electronic form (through screen shots, for example, or by saving html pages locally using the browser’s ‘save as’ facility). The ‘frozen’ version of the corpus, it must be said, is inevitably a distortion of the dynamic original (not least because its interactivity and substantial parts of its multi-modality have been lost), but this is preferable to having no record at all, and in the worst case seeing one’s corpus virtually disintegrate (excuse the pun) before the analysis has been completed.
This problem is also increasingly being addressed by several large-scale projects compiling web archives. The ‘Wayback Machine’ is one such facility, available at http://www.archive.org/. However, such general web archives are unlikely to meet the specific needs that a critical discourse analyst may have, with gaps in the archive perhaps affecting the very sites that are crucial to a particular project. Diachronic studies are, therefore, likely to be fraught with dif-
ficulty and would best be planned ex nunc, starting from the present moment.