Reverse linguistic stereotyping (RLS) is a process whereby a speaker’s perceived group membership triggers differential perception of aspects of their speech. RLS has been suggested to cause drops in intelligibility and recall, though why perception of a non-existent accent can negatively affect listening outcomes has not been sufficiently elucidated. The current study suggests an explanation may lie in differential levels of engagement among listeners. A sample of 430 Japanese university students listened to a short speech from either a speaker of Chinese Pronunciation of English or Received Pronunciation and rated them on aesthetics, comprehensibility, perceived intelligibility, engagement, and recall. Multiple linear regression suggested that only engagement served as a significant predictor to recall, though the other variables all had significant indirect effects when engagement was included as a mediating variable. These results indicated that listener engagement is a key variable which may help improve our understanding of accented speech perception.