Although fuzzy logic has been successfully applied in many different problem domains, fuzzy solutions typically structure the problem task as one of static categorisation. This may not be appropriate for dynamic systems, especially where higher-order information is unavailable. The paper proposes an approach to reasoning about dynamic systems drawing upon the first-order approach to temporal logic. A brief overview of the classical theory is provided and extended to a fuzzy theory modelling the inherent vagueness of linguistic terms such as "recently". The notion of a temporal qualification is developed as a domain independent method for augmenting base domains and term set of a linguistic variable. Such extended variables are termed temporal linguistic variables. The technique is then applied to a simple dynamic system highlighting its advantage over traditional approaches. Since temporal linguistic variables are a special case of linguistic variable they are immediately applicable in fuzzy knowledge based system and machine learning contexts.