The ability of listeners to identify specific words in known and unknown languages using only acoustic information was examined. Each of ten listeners (representing eight different languages) listened to eight passages of fluent dialog, including one in his primary language and one in English (a secondary language for eight listeners). The average number of accurate identifications was slightly more than 50% with nearly 30% of responses being false positive. Word-spotting ability varied widely and appeared to be language dependent, it also appeared to be subject dependent, but the language background of a subject did not affect his ability. Phonetic transcriptions of passages containing false alarm responses revealed that, in unknown languages, false positive responses often appeared to be elicited by utterances that were phonetically similar to the target word. In known languages, semantic similarity was the eliciting factor. Although the average subject's word-spotting ability was better in a known language than an unknown language, the most efficient word-spotters performed better in some unknown languages than the native speakers of that language.