Some Theoretical Implications of Cross-Modal Research in Speech Perception