Legacy: ASY: /banana/ example
The /banana/ example
This example illustrates how simple articulatory changes can have complex acoustic results.
You can listen to four different synthesized tokens: /banana/, /bandana/, /badnana/, or /baddata/, by clicking on the audio files, below. You can listen to all four by clicking on the token labelled "sequence" on the left.
In the synthesis process, the only thing that was varied to produce these tokens was the timing of when the velum (i.e., the nasal port) was opened. Opening and closing the velum determines whether or not a particular consonant is nasalized. For example, if the velum is closed during the production of a voiced alveolar (where the tongue tip is raised to touch the roof of the mouth behind the teeth), the consonant /d/ will be produced. If the velum is open during this same articulation, the resulting consonant will be /n/.
The figure, above, illustrates how the only thing changing during the production of these tokens was the timing of velum opening and closing. Everything else remained constant. At the top of the figure are small pictures of the midsagittal representation of the tract at different points in the articulation. These shapes are the same for all four of the tokens. The plots below these shapes show velum opening, which varies for the different tokens. For example, the velum is closed at the beginning of /banana/, opens rapidly, and stays open for the rest of the articulation. Contrast this with /baddata/, at the bottom, where the velum remains fairly closed throughout the entire articulation.
Another way to look at this is provided in an additional figure, below, that illustrates what remains constant across the different tokens, and what changes. At the top of this figure, the general pattern of the four different tokens is shown -- this remains constant for all of the tokens. Below these are four panels that represent the movement of several articulators (the jaw, the lips, and the tongue), which also remain constant for all of the tokens. The bottom panel, however, shows the varying patterns of velum movement for the different tokens. It is these patterns of movement that result in the different tokens being produced.
Simple articulatory changes can have complex acoustic (and phonetic) results.
You can hear this by listening to the tokens indicated above, or by observing the changes in acoustic detail in the figure, below, that shows pseudo-spectrograms (“formant tracks”) of the four tokens.
| ASY DEMO | VOWELS | VOCAL TRACT | DYNAMIC SYNTHESIS | INFORMATION |