Speech rate normalization is a perceptual process in which identical durations are more likely to be perceived as "long" rather than "short" when embedded in fast speech compared to slow speech. In this study, we examine how rate normalization interacts with the age and gender of both the listener and the speaker when the target length contrast is undergoing a merger. Specifically, we focus on lenis and aspirated stops in Daejeon Korean, where a Voice Onset Time (VOT, a duration cue) merger is occurring while the F0 of the following vowel is becoming the primary cue. This change is being led by younger and female speakers. We explore how the age and gender of both the speaker and the listener modulate rate normalization. The study found that both VOT and F0 cues were robustly used across all speaker and listener conditions to distinguish aspirated from lenis stops. However, the relative strength of these cues depended on the age and gender of both the speaker and the listener. Regarding rate normalization, despite the robustness of the VOT cue across all contexts, evidence for rate normalization was very weak and did not align with VOT cue use. The rate effect was only significant for the older male speaker, who is expected to retain the most conservative speech pattern, where VOT differences are most useful for distinguishing stops. These findings suggest that rate normalization is not an automatic process that affects all duration contrasts equally. Instead, listeners selectively apply rate normalization based on their prior knowledge of whether the target duration dimension is useful in a given context. Future studies will explore how listeners' rate normalization is conditioned by their perception of the speaker's age at an individual level.