Page 27 - Fall2013

Figure 3. Wheatstone’s speaking machine. (Dudley and Tarnoczy, 1959)
models with head shapes \\\[1\\\] often give a rudimentary answer to that question. However, static bent models which were original- ly developed were limited to the formation of only the vowels /a/ and /i/. The work reported here describes successful efforts to find a more nearly appropriate set of vocal-tract configurations for static bent models for classroom demonstrations.
The dynamic straight models are useful for simple demonstrations, but they do not directly simulate the move- ments of tongue as mentioned above. Therefore, we have developed dynamic bent models, such as the gel-type tongue model \\\[5, 6\\\] and the head-shaped model with the sliding tongue \\\[12\\\]. With these models, relatively realistic tongue movements can be simulated. The dynamic bent models are useful when one demonstrates a tongue movement between the formation of the vowels /a/ and /i/. With the dynamics bent models , learners can see that the downward / backward tongue movement is needed for formation of vowel /a/, and that the upward / forward tongue movement is needed for the formation of vowel /i/. The gel-type tongue model \\\[5, 6\\\] has many advantages, including the flexibility of the tongue, enabling one to produce many different vowels. One disad- vantage of the gel-type tongue model is that it is difficult to manipulate, so that there is considerable challenge in repro- ducing the same configuration repeatedly. Because of this difficulty, this model is mainly used in classroom demonstra- tions when the author demonstrates the vowel production to learners in a class or workshop, but the individual learners are never asked to manipulate this model.
The head-shaped model with the sliding tongue \\\[10\\\] has the advantages of both the S3T (sliding three-tube) and the gel-type tongue models. The sliding tongue model has a lim- ited number of degrees of freedom, so that it is simpler for one to produce the target vowel. In addition, the vocal tract is bent in the middle at a right angle, so that one can move the tongue more realistically. The degrees of freedom for this model are as follows: (1) the 1st degree of freedom is the
diagonal movement of the tongue; (2) the 2-nd degree of freedom is the protrusion of the tongue dorsum; (3) the 3-rd degree of freedom is lip rounding. This model is able to pro- duce the vowel sequence between /a/ and /i/ relatively easily; however, the other vowels were difficult to produce in a sequential manner. Consequently, the mechanical bent-type models were redesigned, so that a single bent-type model with sliding blocks would cover all of the vowels, /i/, /e/, /a/, /o/, and /u/.
In the study reported in the present article, the CT (con- nected-tube) model, which was originally designed with cylindrical tubes, was redesigned using square tubes. Then, two bent-type models with sliding blocks were designed: one of these was for the formation of front vowels and the other for back vowels. A final design resulted in a single bent-type model with sliding blocks that produces all five vowels.
Bent-type models with sliding blocks
To find an appropriate set of the vocal-tract configura- tions for static bent models for classroom demonstration, two bent-type models with blocks were designed The redesign of the CT (connected-tube) models with square tubes resulted in bent-type models with a rectangular cross- section. The basic dimension of the cross-section was 45 mm x 20 mm for the neutral vowel, schwa. The length of the oral and pharyngeal cavities was 90 mm and 70 mm, respective- ly. There was a narrow constriction at the larynx, the length of which was 20 mm. The dimension of its cross-section was 9 mm x 9 mm. Two bent-type models with sliding blocks: one was for front vowels (Model A) and the other was for back vowels (Model B). In parts a to e of Fig. 4, the left panel shows the three-dimensional representations of the vocal- tract shape (the numbers are the lengths of sections in mm along the vocal- tract length) and the right panel is a picture of the actual model (the front plate was removed in the pho- tographing of the model).
26 Acoustics Today, October 2013