This article reassesses W.'s view on Biber's multi-dimensional analysis of register variation, and deals with tagging procedures, quantitative vs. qualitative research methods and functional interpretation of statistical co-occurence.