The article reports the results of a study on the perception of reduced forms by non-native users of English. It tests three hypotheses: (i) reduced forms with context are recognized more accurately and faster than reduced forms without context; (ii) gradient reduction is perceived less robustly than the categorical one; and (iii) subjects with musical background perceive reduced forms better than those without. An E-Prime study on 102 Polish learners of English was implemented, comparing participants' accuracy and reaction times with a control group of 14 native speakers. The study was corpus-based and used 287 reduced forms from a corpus of Lancashire. The results indicate that (i) lexical context and phone density significantly affect perception, (ii) the category of reduction process (gradient or categorical) is irrelevant, and (iii) musical background only partially impacts non-native perception.