Using Register-Diversified Corpora for General Language Studies

Style Non-abstract modals (prediction, necessity, possibility) suasive verbs conditional subordination agentless passives by passives passive dependent clauses professional letters editorials broadcasts (press reviews) technical prose (other academic prose) (official documents) conversations, fiction personal letters public speeches public conversations broadcasts Table 8 F scores and correlations for dimension score differences across 23 spoken and written registers (df = 22,459). Dimension F value Probability r2 1 111.9 p < .001 84.3% 2 32.3 p K .001 60.8% 3 31.9 p < .001 60.5% 4 4.2 p K .001 16.9% 5 28.8 p < .001 58.0% the results of an Analysis of Variance showing that the registers are significant discr iminators for each dimension, and the r 2 values show their s t rength (r 2 is a direct measure of the percentage of var ia t ion in the d imens ion score that can be predic ted on the basis of the register distinctions). Four of the five d imens ions have r 2 values over 50%, wi th only Dimens ion 4 hav ing a relatively small r 2 value of 16.9%. These