The literal variables you mentioned should be statistical nominal variables, belonging to a kind of classified data (including ordered variables). For nominal variables, chi-square test is widely used to determine whether the distribution difference of observation units among attributes is statistically significant. However, there are several other nonparametric test methods for nominal variables, such as rank sum test, correlation analysis of contingency table data, Ridit analysis and so on. And we need to choose different test methods according to different purposes.