One of the most important features of fuzzy set theory is its potential for the modeling of natural language expressions. Most works done on this topic focus on some parts of natural language, mostly those that correspond to the socalled "evaluating linguistic expressions". We build constraints for the mathematical substitutes of these expressions to mark characteristic limits on an ordered scale. We discuss a case of creating the total list of terms of the linguistic variable "symptom presence in diagnosis" replaced by fuzzy sets all emerging from one presence description "seldom". To accomplish this task we insert a particular parameter in the membership functions of the sets.