I will review some linguistic arguments for a Kaplan-style picture on which semantic values are determined with respect to a context parameter. I will point out that, if we accept these arguments, then the facts lead us to a position on which certain constituents have a semantics that implicate this parameter ... even though they do not contain pronounced elements that we recognize as indexical. Specifically, their semantic values depend on the world coordinate of the context parameter. I will suggest that these effects come from silent world indexicals sitting in argument positions projected by predicates, but maybe you have better ideas.