A Ta-ka-ti-ki Afternoon

18 April 2013 by Kerstin Hoppenhaus, posted in Linguistics

When you say a (as in aardvark), the back of your tongue moves towards your palate.

When you say ka, it pushes even further.

When you say e (as in eel), the tip of your tongue moves towards your palate.

And with ti, it moves further.

It’s true. I’ve seen it.

Yesterday, Sven Grawunder and Leonardo Lancia tested a new setup for recording ultrasound images for automated image analysis. They are phoneticians and are preparing a study on the relative movements of tongue and larynx in specific speech situations. They wouldn’t tell me more about it at this point, not because of excessive secrecy, but because they want to use me as a subject in the study and subjects are not supposed to know too much about a study’s goals in advance, because they might, consciously or not, react to the information and skew the outcome.

But this was only a technical test, to get all the machinery to play nicely with each other, so I was explained everything.

Observing the physical production of language is not an easy thing. The tongue is one of the main articulators, and somewhere in the medium range on the „resistence to observation-scale". Lips are easier, the glottis down the throat much harder.

The tongue is in your head, so you need something to look inside. Ultrasound is good for this (or good enough, anyway).

The tongue moves fast, so you need a good resolution (in this case: 60 frames per second).

While you speak, your head moves. Automatic image analysis can’t tell one movement from the other, so you have to track the head movement and correct for it in the analysis. That’s what the two sticks with blue spots are for. And the video camera. And the lights.

The movements of the vocal folds in the larynx are measured with a electroglottograph, two electrodes on your throat.

Synchronized with all that, a microphone records the sounds that actually comes out of your mouth. In this case: Ta, Ka, Ti, Ki.

The vowels are selected to be well separated (a and i/e), the consonants to strengthen or weaken the tongue movement, depending on the combination with the vowels. The aim is to have samples of well known tongue movements, so they can test whether the analysing software is producing useful results.

When your mouth is too dry, you can’t see the palate.
When you move your head, the transducer slips.
When you put it back, make sure the markers are in the right place.
The ultrasound wants a different contact gel from the electroglottograph.
The glasses tend to blur.
And so on.

Ultimately, they will need to refine and simplify this setup enough to take it to the field. Because most languages are not spoken near a lab. And those are the ones they are after.




2 Responses to “A Ta-ka-ti-ki Afternoon”

  1. Leonardo Lancia Reply | Permalink

    Hi Kerstin,
    Let me be clearer than 2 days ago on the difference between /a/ and /i/.
    In /a/ the highest point of the tongue is in back in the mouth, while in /i/ the highest point is in the front part of the mouth (or as my colleagues would say) of the oral cavity.

  2. Kerstin Hoppenhaus Reply | Permalink

    Thanks for clarifying, Leonardo.
    I am sure you explained it perfectly the first time and then it somehow got lost on its way through my brain. Sorry for that.
    Very curious about the next steps, though. Keep me posted.

Leave a Reply

six − 5 =