Conceptual engineering has played a prominent role in recent work in philosophy, and some have even argued that much traditional philosophical work should be understood or rethought in those terms. Here I argue that we need a notion of linguistic function to do work in conceptual engineering. For we need to understand how the relevant parts of language work, and what functions they have served, before we make forward-looking decisions about whether to retain, remove, repair, or revise them. But despite the usefulness of a notion of linguistic function, many philosophers have expressed skepticism against the idea that we can identify the functions of concepts or terms. Here I also respond these strains of skepticism and suggest how to begin to develop a notion of linguistic system function that can avoid these forms of skepticism, building on work in systemic functional linguistics, complemented by a 'concern-based' approach to linguistic functions. Better understanding how parts and aspects of our language function, I will argue, can enable us to make far better-informed decisions in conceptual engineering and can be especially useful in assessing and resolving many old philosophical problems.