A study of the interpretation of prosodic features in backchannels (Swedish /a/ and /m/) produced by speech synthesis is presented. The study is part of work-in-progress towards endowing conversational spoken dialogue systems with the ability to produce and use backchannels and other feedback.