On the role of computational, statistical, and interpretive techniques in multi-dimensional analyses of register variation: A reply to Watson

The present paper addresses several issues relating to the research goals and methodological techniques used in multi-dimensional analyses of register variation (e.g., Biber, 1988). The paper is prompted by the allegations of inadequacy and error published in Watson (1994). In my discussion, I point out that Watson fails to substantiate his claimed inadequacies with empirical evidence, and that previous investigations have shown that many of these criticisms are factually incorrect. In addition, I attempt to correct several fundamental misunderstandings of the multi-dimensional approach and the relationships among computational, statistical, and interpretive methodologies within the approach.