Aphasic subjects' comprehension of synthetic and natural speech

This study investigated the ability of aphasic patients with mild auditory comprehension problems to respond to synthetic speech produced by an inexpensive speech synthesizer attached to a personal computer. Subjects were given four practice sessions with synthetic speech; testing of synthetic speech comprehension was performed during Sessions 1 and 4. During testing, aphasic subjects' comprehension of synthetic speech was compared with their comprehension of natural speech on four tasks: (a) picture identification, (b) following commands, (c) yes/no questions, and (d) paragraph comprehension with yes/no questions. Aphasic subjects comprehended natural speech better than synthetic speech in Session 1 but not in Session 4. Their synthetic speech scores improved between Sessions 1 and 4. There was also a significant difference among scores on the four tasks for both sessions. The means for picture identification were highest, followed by yes/no questions, commands, and finally paragraph comprehension for both sessions. Although performance by some subjects on some tasks was accurate enough to indicate that an inexpensive speech synthesizer could be a useful tool for working with mild aphasic patients, considerable caution in selecting both tasks and patients is warranted.