Information fusion and hesitant information fusion represent an important part of decision making processes. This paper focuses on hesitant expressions and the way to take them into account in the computations, using weights served by a simple but efficient process. In a previous paper we have proposed to use an operator called the symbolic weighted median to express hesitant linguistic assessments such as "I hesitate between this and that but I tend to lean toward that alternative". Now we go further in explaining in detail how to transform such expressions into our hesitant operators. Inspired by language science research, several hesitant linguistic expressions are discussed, including linguistic modifiers and qualifiers, then they are transformed into weight vectors before being aggregated to complete information fusion.