In language comprehension, mental representations of temporal relations between described situations are often construed by inference. While the basis for these inferences remains unclear, growing evidence suggests that abstract predicate properties – such as dynamicity and causal structure – play a crucial role in temporal event construal. Across two self-paced reading experiments, replicating and extending Gennari (2004), we find that temporal proximity inferences are shaped by these properties, but only for stative predicates that generally encode non-dynamic situations without causal structure: Participants consistently expected states to overlap in time, in both complement (Experiment 1) and relative clause constructions (Experiment 2). These findings indicate that temporal proximity inferences arise as a general feature of (non)-dynamicity, supporting models of language comprehension that prioritize abstract event structural properties in shaping temporal inferences.