You are performing a multifactorial analysis of the semantic and stylistic differences between the future markers will, be going to and gonna. You expect going to to be more frequently used in informal contexts than will, and gonna to be even more informal. Which modelling approach will be the most appropriate for testing the hypothesis?