ABSTRACT
The domain of modality is structurally diverse and may be described in multiple ways. Another way of classification concerns possible functions of modal words, such as possibility, ability, permission, necessity, obligation, or probability. These categories usually come from traditional grammars and textbooks. For each modal word, a sample of 250 independent observations was extracted and annotated. Multiple correspondence analysis (MCA) is a post-hoc exploratory technique for visualizing clusters of a dataset containing more than two variables. This method was designed as an analogue to principal component analysis (PCA) for handling discrete and categorical data. MCA plotting is no more than an exploratory technique: two points may appear close to each other on a 2D plot, but they may occur far apart if we change our perspective and look at them from a different dimension. The polytomous logistic regression is a further development of the logit regression approach adopted for predicting categorical variables that have more than two levels.
