Previous work has shown that listeners rapidly update their interpretations of vague expressions such as quantifiers and expressions of uncertainty when they observe a speaker's usage of such terms. However, previous studies focused on instances involving two reasoning steps: inferring a world state from a visual scene and communicating the world state. Based on these experiments, it has been argued that listeners infer speaker-specific mappings between world states and vague expressions rather than listeners making inferences about how the speaker infers the world state from a visual scene. Here, we extend the work on semantic-pragmatic adaptation to a new class of expressions, namely vague temporal expressions, such as for a bit and for a while, and employ an experimental paradigm in which the inference from the visual scene to the world state is deterministic. We replicate previous findings in this setting, suggesting that adaptation indeed targets semantic representations.