A domain-specific language (DSL) can have graphical, textual or hybrid syntaxes. Certain domain concepts are better suited to be represented graphically, whereas a textual representation is often more appropriate for modeling the behavior, complex expressions, and fine details of a domain. As such, the best of both worlds of graphical and textual modeling can be delivered by a DSL that has a hybrid (part-graphical and part-textual) syntax. The engineering of hybrid graphical-textual DSLs and their supporting workbenches is a non-trivial endeavor, as a substantial amount of hand-written code is required. Existing workbenches for hybrid graphical-textual DSLs pose several limitations, e.g., they do not enforce the consistency between the graphical and textual parts of the model and do not provide uniform error reporting. This work will propose a methodology and provide tooling for systematic engineering that aims to minimize the accidental complexity involved in designing and developing hybrid graphical-textual DSLs and their supporting workbenches.