Skip to Content

Department of Linguistics


Electromagnetic Articulography (EMA)

Important: If you have not yet either installed the phonetic font "Charis SIL" or tested this installation to determine if the phonetic characters installed properly then click here to go to the phonetic font help pages.

In this workshop the terms "electromagnetic articulography", "electromagnetometry" and "kinematics" will sometimes be used interchangeably. "Kinematics" is a generic term for the study of articulator movement whilst "electromagnetic articulography", "electromagnetometry" and "EMA" are alternative names for a particular procedure used for studying kinematics.

Click here to view some notes on electromagnetic articulography.


Click here to display a print formatted PDF version of the following figures.

Figure 1: Tongue-X, tongue-Y and speech waveform for two tokens of 'beeber' produced by a Australian English female talker.

Figure 2: Tongue-X, tongue-Y and speech waveform for two (more) tokens of 'beeber' produced by the same talker.

Figure 3: Tongue positions are marked as 'bi' (taken from a production of 'Hector beeber') or 'ba' (taken from a production of 'Hector barber'). The tongue positions were extracted in the schwa of 'Hector' (ellipses on the left) and the word-final schwa in 'beeber' or barber'.

Figure 4: Jaw height trajectory in the 'barb' syllable of 'barber' produced by a Australian English female talker.

Figure 5: Averaged jaw height and velocity trajectories of accented (solid -darker) and unaccented (dotted - lighter) productions of 'barber'. The trajectories were time aligned at the peak jaw displacement in the vowel before averaging.

Figure 6: Displacement of the jaw opening (left column) and closing (right column) gestures as a function of duration (top row) and peak velocity (bottom row) for accented ad unaccented tokens of 'barb' from 'barber' produced by an Australian English female talker.

Workshop Questions

Please Note: The workshop questions below are NOT part of the 2006 assessment.

Figures 1 and 2 show tongue data and the synchronised acoustic signal for four productions of 'Hector Beeber' (the left and right columns of page 1; the left and right columns of figure 2), extending from the beginning of the last schwa of 'Hector' to the offset of 'Beeber' (so what you have in each case is articulatory and acoustic data of [əbi:bə]).

1. For all four tokens in figures 1 and 2, mark the acoustic onset and the acoustic offset of the [i:] vowel as two vertical lines on the speech waveform. Also mark as a single vertical line on the speech waveforms the acoustic onset of the final schwa vowel.

2. Let's call the interval between the acoustic onset and offset of the vowel the acoustic section of the vowel. Look at the tongue-X and tongue-Y positions within the acoustic section of the vowel. How would you (briefly) describe the movement of the tongue over this interval?

3. On the tongue-X trajectory, mark the maximum point of tongue fronting (which occurs within the acoustic section of the vowel). Also mark the maximum point of tongue backing on the tongue-X trajectory which occurs in the final schwa. Estimate (to the nearest couple of mm) from the tongue-X displays the magnitude of the tongue backing gesture from the [i:] vowel to the schwa.

4. Draw a set of tongue X-Y axes similar to the ones shown in figure 3. Mark the position of the four [i:] vowels in this plane using data extracted at the maximum point of tongue fronting from the four tokens.

5. The display in the left column of figure 3 shows tongue X-Y positions that were extracted in the schwa of 'Hector' in either 'Hector Beeber' or 'Hector Barber'. The schwas from 'Hector Beeber' are marked with 'bi' in this plot, and the schwas from 'Hector Barber' are marked with 'ba'. The right column is of the same kind of data, but from the word-final schwa: that is, they are extracted from the schwa of 'Beeber' (and marked 'bi') and from the schwa of 'Barber' (and marked 'ba'). Look at the data in the left column (the schwas in 'Hector'). Although all the data points are from the same phonetic unit (schwas), their distribution is different according to phonetic context: that is, the schwas that precede 'Beeber' are to the left and slightly above those that precede 'Barber'. What could account for this difference?

6. Perseverative coarticulation is where a speech sound X influences another speech sound Y that occurs after X in time. By comparing the ellipse plots on the right with those on the left in figure 3, would you say that the effects of perseverative coarticulation are as marked as those of anticipatory coarticulation?

7. Figure 4 shows a jaw height trajectory in the production of 'barb' of 'Barber' (produced by an Australian English talker). Mark as vertical lines (a) the peak displacements associated with the initial and final consonants and (b) the peak displacement associated with the vowel. Label the opening gesture and closing gesture and estimate (in mm) the magnitude of the opening gesture and the magnitude of the closing gesture.

8. The solid and dotted line jaw height trajectories on figure 5 show data from 'Barber' produced in two different ways. The solid line trajectory is of accented 'Barber' (i.e., with sentence stress) as in a production of 'Hector Barber' (in response to the question: 'What's your name'?'). The dotted line trajectory is of unaccented 'Barber' as in a production of 'That's Dr. Barber. Not Mr Barber' (here, the nuclear accents are on 'Dr' and 'Mr'). What are the main differences between the jaw trajectories of the accented and unaccented productions in terms of (a) the magnitude of the opening gesture and (b) the peak velocities of the opening and closing gestures?

9. There is a relationship between velocity, distance, and time (assuming a constant speed):

velocity = distance/time
time = distance/velocity

This relationship can be applied to kinematic data. For example, if the jaw moves through a greater distance from one production to the next, either the time taken for the jaw to move through the greater distance must increase, or else, if the jaw covers a greater distance in the same amount of time, the velocity must increase.

The data in figure 6 shows the opening and closing jaw height gestures of accented and unaccented productions of 'Barber'. There are two plots: (a) displacement (i.e., magnitude of the opening/closing gestures) as a function of the duration of the gesture (top) and (b) as a function of the peak velocity of the gesture (bottom). Points with a higher displacement are usually accented.

Would you say that this talker produces bigger jaw displacements by (a) taking longer or (b) speeding up her gestures? Justify your answers with evidence from the plots.