FunGramKB is a multipurpose lexico-conceptual knowledge base for natural language processing systems, and more particularly, for natural language understanding. The linguistic layer of this knowledge-engineering project is grounded in compatible aspects of two linguistic accounts, namely, Role and Reference Grammar (RRG) and the Lexical Constructional Model (LCM). RRG, although originally a lexicalist approach, has recently incorporated constructional configurations into its descriptive and explanatory apparatus. The LCM has sought to understand from its inception the factors that constrain lexical-constructional integration. Within this theoretical context, this paper discusses the format of lexical entries, highly inspired in RRG proposals, and of constructional schemata, which are organized according to the descriptive levels supplied by the LCM. Both lexical and constructional structure is represented by means of Attribute Value Matrices (AVMs). Thus, the lexical and grammatical levels of FunGramKB are the focus of attention here. Additionally, the need for a conceptualist approach to meaning construction is highlighted throughout the discussion.