The Correspondence Analysis on Generalised Aggregated Lexical Tables (CA-GALT) is an original textual statistical method to uncover the relationships between opinions, expressed as free-text answers, and respondent s characteristics. This method also identifies the specific links between each characteristic and certain words used in the answers. Promising results are obtained as shown by an application to real data (Enquesta de participació cultural a Catalunya 2014), substituting words by cultural activities for the first time. In that sense, we try to identify what social characteristics are most influential in the segmentation of cultural habits. We use the framework that gives us the most important set of theories on social stratification in cultural consumption, namely the homology argument, the individualisation argument and the omnivore-univore argument.