Daniel, just a thought - maybe not such a good one, but Chris Darwin and I showed (QJEP 1991, 43A, 410-421) fairly robust phase vowel discriminations on an /I/-/E/ continuum by manipulating the phase of a single harmonic near F1. Now the acoustic manipulation here is faint, but the perceptual one (one vowel changes to another) is not. Interesting question though Denis McKeown, Leeds Uk