The solution Monty gave me some time back works great. He suggested subclassing match() in the parser and adding logic to not consume() those token types that you want to be (re)recognized by a different lexer. SusanJ