“…Speech synthesis by rule. At the present time, speech synthesis programs typically concatenate segments whose durations are determined by dictionary look-up, with adjustments in duration based upon immediate segment context (see, e.g., Flanagan, Coker, Rabiner, Schafer, & Umeda, 1970;Mattingly, 1971; see also Denes, 1970;Mattingly, 1966;Rabiner, 1969;Rabiner, Levitt, & Rosenberg, 1969;Teranishi & Umeda, 1968). In the present theory, the temporal patterning of utterances constructed this way can be expected to deviate from those of natural speech.…”