This weekend I was listening to Terri Gross's interview with Steven Sondheim, and was struck by a very amazing claim that the musical master made: harmonies can resolve, but melodies cannot. He really said that. Gross was asking about the ending of a song in Sweeny Todd, and asked how Sondheim had picked the final note to sound so unresolved. Sondheim replied that the note couldn't sound unresolved by itself, it was because the harmony was unresolved that the note stuck out. This should be of immense surprise not only to the countless theory text authors who have distinguished between Perfect Authentic Cadences and Imperfect Authentic Cadences, or to every theorist who is an advocate of voice-leading, but also to Carole Krumhansl and her followers who showed through probe-tone experiments that melodies do indeed create expected resolutions. I think Sondheim is relying too much on his piano background, and not enough on the things he learned from Milton Babbitt.
Earlier in the interview he told Gross that "discordant" meant "incorrect" and was not synonymous with "dissonant." Not according to my dictionaries and thesaurus! I got the feeling that Sondheim was really making these statements to act as a high priest of music, creating barriers to the sacred mysteries of melody and harmony through insistence on idiosyncratic jargon. He was putting the layperson in her place, making clear that only a trained adept could comprehend how "Send In the Clowns" was structured.