While most empirical studies have investigated the improvement of learners' L2 spoken proficiency via speaking-related interventions, the present study examined the same topic through a different modality: listening. Ninety-five first-year tertiary-level students of English as a second language (ESL) in Hong Kong participated in this three-month experiment. They were separated randomly into three groups: (A) narrow listening with latitude for learners to choose their own materials; (B) narrow listening without such latitude; and (C) a control group without narrow listening. All participants attended a compulsory course on English phonology. Narrow listening was set as homework for the two experimental groups A and B. The overall results indicated that groups A and B improved significantly more than group C in pronunciation and fluency in reading aloud, performing a dialogue, and free production. There were also differences between group A and B in these tasks. The participants were generally highly positive about the narrow listening set as homework. This article ends with a discussion of the findings from theoretical and pedagogical perspectives.