Statistical differentiation of tracheoesophageal speech produced under four prosthetic/occlusion speaking conditions

Twelve male and 12 female total laryngectomy patients who received the tracheoesophageal puncture (TEP) as a means of vocal rehabilitation served as subjects for this investigation. Recordings were made of these subjects' speech produced with four prosthetic/occlusion conditions: (1) duckbill prosthesis with tracheostoma valve; (2) duckbill prosthesis with digital occlusion of the tracheostoma; (3) low pressure prosthesis with tracheostoma valve; and (4) low pressure prosthesis with digital occlusion. Speech tasks consisted of three trials of maximum phonation time on /a/ and reading of a 98-word standard passage. Acoustic analysis of the recorded speech samples included a total of 34 frequency, intensity, temporal, and noise measures. Eight acoustic measures (words per minute, harmonics-to-noise ratio, percent jitter, intensity range during vowel phonation, percent periodic phonation, mean intensity during reading, directional jitter, and directional jitter, and directional shimmer) were chosen as dependent variables for a repeated measures MANOVA. The overall repeated measures MANOVA, a set of complex contrasts, and paired t tests revealed that TEP speech produced with the low pressure prosthesis was significantly different from that produced with the duckbill prosthesis on a weighted linear combination of the eight acoustic variables. Tracheoesophageal voice produced with a low pressure prosthesis had greater amounts of periodic phonation than tracheoesophageal voice produced with a duckbill prosthesis. The use of a tracheostoma valve did not have a significant impact on the subset of acoustic measures used in the repeated measures MANOVA.