The abilities of second language speakers to discriminate the prosodic contrasts between idiomatic and literal meanings of ambiguous sentences were investigated using utterances previously shown to be reliably identified by acoustic cues. Four listener groups of varying proficiency, native speakers of American English, native speakers of non-American English, fluent nonnative speakers of English, and advanced students of English as a second language (ESL), judged whether single and paired, tape-recorded, literal and idiomatic utterances were spoken with intended idiomatic or literal meanings. Both native speaker groups performed significantly better than fluent nonnatives, while ESL students performed at chance. These results lend support to the hypothesis that abilities to discriminate subtle prosodic contrasts are learned later than other components of speech and language.