Klaske van Sluis

74 4.3. Results tistical significance is corrected for false discovery rate [10]. To estimate the importance of the factors studied for the acoustic realizations of the alveolar plosives, linear mixed effect models were created [11, 12], The model analyses the relationship between prevoicing, burst duration, and vowel duration on the one hand and the fixed effects time (pre- / post-treatment) and phoneme iden- tity (/t/ /d/) on the other hand, including the interaction term. The by-subject and by-word intercepts and random slopes of the effects of time and phoneme identity were used as random effects. P values were estimated by approximat- ing the Student t-test statistics of the coefficients by Z-test statistics, following Barr et al. [11]. An R Markdown script and the original /t/ /d/ data have been made available on http://www.fon.hum.uva.nl/rob/ICPhS19/. 4.3 Results In Table 4.1 the acoustic characteristics of the sustained vowel /a/ recorded from the participants before and after surgery are presented. In three partic- ipants after treatment no median F0 could be determined. After treatment acoustic features are changed. Median F0 values of tracheo-esophageal speech have a mean of 61Hz, coming from a pre-operative value of mean 138Hz. The mean percentage voiced frames drops from 93% to 55%. Harmonics-to-noise ratio (HNR) and maximum voicing duration (MVD) decreases. A statistically significant difference between pre- and post-treatment values was found for F0-median, %V, HNR and MVD (p ≤ .05, WMPSR). Table 4.1: Range, mean, standard deviation and statistical testing of the acous- tic values pre- and post-operative of sustained vowel /a/ n =17. Abbreviations: F0: fundamental frequency, %V: percentage voiced, HNR: harmonics-to-noise ratio, MVD: maximum voicing duration. * p ≤ .05 tested with WMPSR test ˆ ( n =14) Acoustic Range Mean SD p -value Parameter F 0 -median Pre 79-277 138ˆ 56 (Hz) Post 17-132 61 40 .002* F 0 -SD Pre 1-27 8 10 (Hz) Post 1-42 11 13 .391 %V Pre 15-100 93 21 Post 0-100 55 38 .003* HNR Pre 4-20 15 5 (dB) Post 0-8 3 4 < .001* MVD Pre .3-2 1.9 .4 (sec) Post 0-2 1.2 .8 .013* Table 4.2 shows the results with acoustic features of the consonant /t/