The causes of the (low) intelligibility of deaf speech were studied, following a so called "Speech Transformation Method". By manipulation of analysis parameters, errors of articulation, intonation and timing, that occurred in sentences spoken by deaf children, were artificially corrected one-by-one. Intelligibility tests showed that suprasegmental corrections caused only a small improvement (from 24% to 34% word-intelligibility), whereas segmental correction increased intelligibility scores to 74%. The results of this study formed one of the starting points for a research project aiming at the development of a visual speech training aid: a device that supplies visual information on the acoustic quality of deaf speech.