In this paper, I explore the implications for ontology of what Jared Warren has dubbed a top-down approach to metasemantics, explaining semantic features of subsentential expressions in terms of semantic features of whole sentences. Among authors who have adopted such an approach to metasemantics are Gottlob Frege, Michael Dummett, Crispin Wright, Donald Davidson, W.V.O. Quine, Agustín Rayo, Oystein Linnebo, and Warren himself, and many of these authors have also taken this top-down approach to have ontological implications. I discuss what a top-down approach amounts to, the extent to which such an approach is plausible, and what its implications for ontology are, if any. Often a top-down approach is assumed to have broadly deflationary implications, and to provide a separate path to defending the view that abstract objects exist.