Likelog is a fuzzy–logic language with a Prolog-like syntax augmented with similarity equations (modeling the fuzzy component of the language) and with an operational semantics based on resolution where the classical syntactic unification algorithm of pure logic programming has been replaced by a similarity-based unification method. On the other hand, Curry is a functional–logic language with a Haskell-like syntax and an operational principle based on (needed) narrowing, that is, a combination of syntactic unification and rewriting. In this paper we propose a new similarity-based unification method which empowers the original one used in Likelog by also taking into account functional features like laziness. This calculus is specially well suited for being embedded into the kernel of the original needed narrowing strategy of Curry in a very natural way, also verifying nice formal properties such as termination, crispness (i.e., it computes at least the same elements of the crisp case) and fuzzyness (i.e., similarities collected in a given program are exploited as much as possible). Our final goal is to achieve the complete integration of the (also integrated) declarative paradigms of functional– logic and fuzzy–logic programming, in order to obtain a richer and much more expressive programming scheme where mathematical functions cohabit with fuzzy logic features.