F0

In the methodology, there are three types of measurements (normalized F0, onset F0, and first 100 ms F0) were adopted. Tones have different durations. Tone 3 has the longest duration and the Tone 4 is the shortest one (Xu, 1997). The normalized measurement can equalize the different tone durations. F0 was extracted by every 10%. The second one measurement is onset F0. If aspiration has the rising or lowing effect, onset F0 will be affected most. The third one is first 100 ms F0. F0 was extracted every 10 ms and there were 10 points recorded. There are three types of

58

measurement adopted in the present study; however, only the first 100 ms was discussed. Although normalized F0 reduces the duration differences among tones, the perturbation effect of aspiration decrease by the normalization. On the other side, the onset F0 has the most obvious perturbation effect but how to define the onset of F0 raise other issue. Hombert et al. (1979) indicated the voiceless stops may still affect F0 at least 100 ms after vowel onset. That is why first 100 ms was adopted in this paper. The discussion was illustrated as below.

Aspiration

There are various parameters controlling F0 which are affected by aspirated and unaspirated distinction. These parameters could explain why aspiration distinctions have different effects on the F0 of the following vowels. The fact is that the glottis is widely open upon the release of an aspirated stop, as opposite to a more closed position for an unaspirated stop (Hombert, 1978). The more the glottis opened, the more airflow passed. According to the flow rate calculation, it is often estimated by determining the velocity at which flow is through a given cross-section area. The flow equation is: Q= VA (where Q= volumetric flow, V=flow velocity, and A=

cross-sectional area).

On the other hand, Müller (1987) indicated that an increase in subglottal pressure leads to an increase in frequency. Chomsky and Halle (1968) also suggested that aspirated stops are produced with heightened subglottal pressure in contrast to unaspirated stops which would have normal subglottal pressure. Previous studies also illustrated that sound required a high rate of air flow; fricative and aspirated stops tend to be produced with a separated opening gestures of glottis (Löfqvist et al., 1984;

Löfqvist and Yohsioka, 1980; Yoshioka, 1981, 1982). Subtelny et al. (1969) indicated

59

that higher airflow rate after the aspirated versus the unaspirated series. According to the Bernoulli Effect, when the speed of flow through a fluid increases, the pressure decreases. Higher airflow rate passes through the glottis, the vocal folds vibration increases due to the decrease of pressure. Then higher airflow raises F0. Based on Bernoulli Effect, it might be suspected that F0 is higher after aspirated stops.

In addition to glottis opening, the stiffness of vocal folds is distinguished in aspiration.

Increasing the stiffness of the vocal folds makes the coupling between upper and

The average F0 is the lowest for vowels following lax stops, higher following tense stops, and highest following aspirated stops (Kim et al., 2002). Further investigation indicated that Korean aspirated and tense stops correspond to the aspirated and unaspirated stops in Mandarin (Reetz and Jongman, 2009). What is more, Hirose (1974) examined the actions of intrinsic laryngeal muscles in production of Korean stop consonants by electromyography (EMG). The results indicated that the aspirated stop appeared to be characterized by the suppression of all the adductor muscles (vocalis, cricothyroid muscle, interarytenoid muscle, lateral cricoarytenoid, etc.) of the larynx immediately preceding the articulatory release. This suppression was always followed by a steep increase in activity which seemed to correspond to the rapid closure of the glottis after stops release. In the tense stops, the pattern of vocalis and lateral cricoarytenoid activity were most characterized. Both muscle, vocalis

60

particularly, showed a marked increase in activity before the stop release, which presumably results in an increase of inner tension of the vocal folds and constriction of the glottis during or immediately after the articulatory closure.

Results in the first 100 ms F0 data indicated that the aspiration at the initial position has the rising effect on F0 and the results conform to our hypothesis that F0 is higher after aspirated stops.

Aspiration and Rate

The main effect of speaking rate was significant. F0 was higher at fast speaking rate in Tone 1 and Tone 2 which is supported by Dromey & Ramig (1998). However, the main effect of speaking rate was also significantly higher at slow speaking rate in Tone 3 and Tone 4. It is quite strange to see the contradictory results. If there exists contradictory conditions, it should be that Tone 1 and Tone 4 have the same result and Tone 2 and Tone 3 for their onset tonal contour range. From the present results, it still shows that different rates indeed have influence upon F0 which might help to clarify the conflicting findings of previous researches. The different effect of aspiration of previous studies may come from the non-well controlled speaking rate.

In addition, according to flow equation, the fast speaking rate has a higher airflow.

Under the same cross-area, the higher airflow increases higher airflow rate. Thus the fast airflow rate is assumed to trigger a higher F0 (Ladfoged, 1967). Moreover, Verneuil (1996) found that the increasing airflow corresponds with the increasing of F0. This finding supports the results that F0 is higher at fast speaking rate. Besides, the result could be supposed that as more muscular effort while producing fast speaking rate is expended to increase the stiffness of vocal folds. As mentioned above,

61

stiffening of the vocal folds affects glottal vibration. Halle and Stevens (1971) indicated that when the vocal folds are slackened, there is a decrease in the glottal vibration frequency.

How does Aspiration interaction with Rate? The interaction between Aspiration and Rate is significant in Tone 1, Tone 2, and Tone 4. Sounds or words could be misheard easier at the fast speaking rate. It was found that in the present study that the differences between aspirated stops and unaspirated stops are larger in fast speaking rate than in slow speaking rate. F0 is getting higher after aspirated stops and lower after unaspirated stops. The reason why perturbation effect is stronger at fast speaking rate might be that subjects pronounce sentence more clearly with consciousness. The conscious aspiration emphasis caused by speaker is to increase the discrimination among words in the fast speaking rate.

Generally, the results showed that the different speaking rates have different effect on F0, although the results were not consistent. F0 is significantly higher at fast speaking rate in Tone 1 and Tone 2, and higher at slow speaking rate in Tone 3 and Tone 4.

Does the coarticulation make F0 different distribution at different tones? This raises another issue to investigation in future.

Aspiration and Gender

The main effect of Gender is significant. Unsurprisingly, F0 of females is higher than that of males. Some important morphological differences between the male and the female larynx have been described by many researchers (Kahane, 19198; Hirano, 1983; Tieze, 1989). Because of the vocal fold size and mass as well as vocal tract length, physiological differences between genders have different influence on F0. The

62

anterior-posterior dimension of the male cartilage is approximately 20% larger than that of females (Kahane, 19198). Titze (1989) reported the different larynx size between genders. The thickness of the male vocal fold is about 20% to 30% greater than that of the female as shown in Fig. 59 (a). The greater thickness creates a difficulty in elongation of the membranous portion of the vocal folds. Apart from the thickness of vocal folds, Fant (1976) also showed that the ratio of the total length of the female vocal tract to that of a male is about 0.87 or the membranous vocal fold lengths differ by 60% (Fig. 59 (b)) (Kahane, 1987). The longer vocal fold length triggers the lower F0. These differences in physiological parameters lead to different performance in acoustical performances between genders.

Figure 38. Male-female comparisons of dimensions of the larynx (a) Sagittal view of the thyroid cartilage and (b) horizontal section showing difference in membranous length (Adopted by Kahane,

19198)

Ajamani (1990) pointed out that the thyroid angle in the female larynx is significantly larger than that in males. It varies from 106∘ to 60∘ in males and 132∘ to 88∘ in females. Because the front parts of the vocal folds are attached at the thyroid cartilage, it could be hypothesized that wider the thyroid angle is, the more the vocal folds are strengthened. To put that differently, the female vocal folds can be strengthened more

63

widely than male vocal folds. As mentioned above, the wider glottis opened, the more airflow pass. Thus, the high airflow triggers the higher F0 (Ladefoged, 1967).

In addition the different physiological structure of vocal folds, the CT muscle may work differently between genders (Ximenes Filho, 2005). In mammals, the CT articulation has acquired an important function in voice modulation, permitting elongation of the vocal folds as a result of contraction of the CT muscle. The cricothyroid, which by drawing the front edges of the thyroid and cricoids cartilages closer together, lengthens and tenses the vocal cords. Ximenes Filho (2005) investigated the angulations of cricothyroid articulation in the cricoid ring, and distance between cricothyroid and cricoarytenoid articulations from16 cadavers (9 men, 7 women). He indicated that the CT articulation angle was narrower in men than in women. The major diameter of the thyroid cartilage was wider in men. These different intrinsic physiological structures might explain why female‘s F0 is higher than male‘s.

The interaction between Aspiration and Gender was significant across four tones. F0 difference between aspirated and unaspirated stops is significantly greater in female when compared to male participants. The raising effect seems greater in female speakers as well. The physiological difference between females and males could explain the different perturbation effect of aspiration. A good illustration of this is the thickness of vocal folds. The thicker the vocal folds are, the more effort will be exhausted to make the vibration. Simply stated, based on the same subglottal pressure velocity (supposed the velocity of subglottal pressure is the same when males and females pronounce aspirated stops), the thicker the vocal folds are, the harder they vibrate. Furthermore, the range of F0 value between different genders is different. F0

64

range of females is larger than that of males (Iseli et al., 2006). Thus the F0 differences of aspirated and unaspirated stops could be expanded more because of the larger female F0 range.

Compared with Xu and Xu (2003), the result, opposite to the result of Lai‘s (2004) and the results of present study, indicated that F0 was lower after the aspirated stops.

There are only 7 female participants in Xu and Xu‘s experiment. Based on the significant interaction between Aspiration and Gender in the present study, it can be speculated that there exists a possible gender influence in the results if researchers did balance the number of each gender. Interestingly, the female results between Xu and Xu and the present study are different either. F0 is higher after the unaspirated stop (Xu and Xu, 2003) which is opposite to that of present results. F0 is about 20-30 Hz higher after unaspirated stops in Xu and Xu (2003). Somehow the F0 differences between aspirated and unaspirated stop are too large. Furthermore, only alveolar stops were used as stimuli in Xu and Xu‘s experiment. There are 3 places of articulation and four vowels were adopted in the present study. It might cause the different results with these dissimilarities.