This work introduces the exploitation of some language resources, namely word association norms, for building lexical search engines. We used the Edinburgh Associative Thesaurus and the University of South Florida Free Association Norms for the construction of knowledge graphs that will let us execute algorithms over the nodes and edges in order to do a lexical search. The aim of the search is to perform an inverse dictionary search that, given the description of a concept as a query in natural language, will retrieve a target word. We evaluated two graph approaches, namely Betweenness Centrality and PageRank, using a corpus of human-definitions. The results are compared with the BM25 text-retrieval algorithm and also with an online reverse dictionary– OneLook Reverse Dictionary. The experiments show that our lexical search method is competitive with the IR models in our case study, even with a slight outperformance. This demonstrates that an inverse dictionary is possible to build with these kind of resources, no matter the language of the Word Association Norm.