First, we briefly advocate a need of natural language based methods in data mining, notably whew domain experts have a limited knowledge of modern tools of information technology. We present some approaches to linguistic summarization of sets of (numeric and/or textual) data, and show that a fuzzy logic based approach by Yager (1982), notably in its extended and implementable version of Kacprzyk and Yager (2001), and Kacprzyk, Yager and Zadrozny (2000), offers a simplicity and intuitive appeal, in particular in its new setting by Kacprzyk and Zadrozny (2005) based on Zadeh's computing with words and protoforms.