Yes IF we accept some restrictions regarding the fixed size of the input, and we consider the outputs to be part of the domain of discourse appended at the end of the input (or, for single-bit outputs, use it as an indicator of valid inputs).
There is a mapping betweeen types of grammar and the type of automaton that can recognise them:
Type Grammar Accepted Automaton
Type 0 Unrestricted grammar Turing Machine
Type 1 Context-sensitive grammar Linear-bounded automaton
Type 2 Context-free grammar Pushdown automaton
Type 3 Regular grammar Finite state automaton
One one hand, we can always build a finite state automaton that accepts the same set of valid inputs, arriving at the same final state / output, as any given combinational circuit.
Inversely, the set of valid inputs for a combinational circuit does define a language which is produced by a grammar, and a regular (type 3) grammar is guaranteed to be enough to generate that language (albeit it may not be the best approach to model a given combinational circuit, and the complexity and semantics of the resulting grammar may be no better than a huge input/output table).
Let's think, for example, of a circuit that validates a parity bit for a given byte, with a 9-bit input: you could say that such combinational circuit implements a grammar, and the produced language is the set of bytes with correct parity bits.
Again, in many applications of combinational logic, however, thinking in terms of formal languages theory is not the most useful approach (since their input is limited, we often think in terms of truth tables and boolean logic).
This answer provides further insights: What is the connection between combinatorial circuits and finite state automata?
