摘要：A decade of research on errors as contained in learner corpora, usually known as
Computer-aided Error Analysis (CEA), has proved not just that increasing attention
is being paid to the field (Leech 1997: 15; Dagneaux et al. 1998: 163 et
passim; Granger 2003b: 542; Tono 2003: 804–805), but also shown what elements
are basic for this purpose: a learner corpus and an error tagset for annotation
of errors in the corpus (cf. Granger 1999; Tono 2000).
