This study investigates the effect of native dialect on the use of prosodic cues to English stress by Standard Mandarin (SM) listeners, Taiwanese Mandarin (TM) listeners, and English listeners. Both SM and TM use fundamental frequency (F0) to realize lexical tones, but only SM uses duration together with F0 to realize lexically contrastive full-full vs. full-reduced stress patterns. Native English listeners and second language learners of English who spoke SM or TM as native language and were at similar proficiencies in English completed a sequence-recall task. English disyllabic non-words that differed in stress placement were resynthesized to contain only F0 cues, only duration cues, or converging F0 and duration cues. The results showed that SM-speaking learners used duration more than TM-speaking learners to recall English non-words. Native dialect is suggested to be considered in second language speech processing models.