Dimension scores

The interpretation of a factor requires the computing of dimension scores (factor scores). By means of these scores the respective registers can be distributed along the dimension and thus characterised more fully (cf. Biber and Gray, Chapter 21). The scores can be computed by summing the frequencies of salient features of the texts, that is, summing the frequencies of features with positive loadings and then subtracting the frequencies of negative ones (cf. Biber, Conrad and Reppen 1998: 279). Beforehand, the raw frequency counts need to be standardised in order to ensure equal weight of each feature. The average count values 0.0 and the standard deviation 1.0. The score +1.0 thus represents one standard deviation above the mean score of a feature (cf. Biber, Conrad and Reppen 1998: 280). Therefore, these scores illustrate whether a feature is common or rare in a certain text compared to the average count of that feature in all texts. Finally, the mean dimension score for each register can be computed, taking account of the single scores of every text representing the respective register (cf. Biber and Gray, Chapter 21).

Created with the Personal Edition of HelpNDoc: Easily create iPhone documentation