Page 18 - Fall 2011
P. 18

 You say tomato, I say tomato The types of changes to which I refer in this paper involve sometimes subtle, sometimes not-so-subtle, changes in the pronunciation of particular sounds. It is important to familiarize ourselves with the ways in which linguists and speech scientists talk about speech sounds. Speech production boils down to a manipulation of the airstream. For example, say the word tomato; do this with your hand in front of your mouth and you will have a tactile impression of this airstream manipulation, in addition to the auditory one. Figure 1 presents a spectrogram and waveform from two speakers’ pronunciations of tomato. My production is on the left, and a male’s production of this word is on the right. We return to some key differences in female and male productions later. Tomato, like all words, is made up of a series of speech sounds. To produce the word tomato, your tongue tip moves up and makes contact in the region behind your front teeth, sometimes making contact with your teeth themselves, to make the /t/. If you put your hand in front of your mouth, you will feel a rather strong puff of air as you release the /t/; this is called aspiration. From the /t/, your mouth changes its configuration seamlessly as it moves towards a more neutral configuration for the initial vowel, which is a short- ened, indistinct schwa-like vowel. Then, to produce an /m/, you close your lips and open the passageway to your nasal cavity, allowing the air to flow through your nasal cavity and sinuses on the way to the open atmosphere. Following the /m/, the vowel you produce will vary considerably depend- ing on the variety of English you speak. Most speakers from North America will produce the vowel sound which also occurs in bake and cake: /e͡􏰀/. If you speak a variety of British English, you will likely produce this sound as an /ɑ/, which is more similar, although not identical, to the vowel that North American English speakers use in the word father. Next we come to a sound that is like /t/, but which is pro- duced much more quickly and with a different movement trajectory in natural speech; this is called a flap: /ɾ/. Finally, there is an /o/ sound, which involves rounding your lips a little bit. Note that this last vowel sound is produced and sounds quite different from the first vowel in the word, despite the fact they are both spelled with an “o.” It is rather amusing how long it takes to describe how to produce a word, compared to how long it takes to simply say the word. Our mouths do some impressive articulatory gymnastics at incredible speed in speech production, and we do not even give it much, if any, thought. These articulatory movements modify the airstream, making constrictions in the oral cavity to varying degrees. In making a /t/, there is a complete constriction to the point where air is trapped inside the oral cavity. To produce a vowel sound, the oral cavity is left relatively unconstricted, but the oral cavity and tongue are shaped in particular ways. These constrictions and configurations result in particular acoustic consequences when the air column in your vocal tract is excited. You can move your articulators to your heart’s content, but without the excitation of the air column in your vocal tract, no sound will be emitted. A crash course on the acoustics of speech production The acoustics of the speech signal are determined by two main factors: the sound source and the filter through which that sound passes. The role of the filter is to modify the spectral shape of what was produced by the sound source. In the production of voiced sounds with a relatively open vocal tract—sounds like vowels, /r/, and /l/—the SPEECH COMMUNICATION 101  Fig. 1. A waveform (top) and spectrogram (bottom) of the word “tomato” produced by a female speaker (left) and a male speaker (right) of American English. Imitation in Speech 17 

   16   17   18   19   20