Skip to Content

Department of Linguistics


Read this page in conjunction with the
EMA workshop video

Electromagnetometry (EMA) and the Production of Speech

Jonathan Harrington
October, 2001

Electromagnetometry: a technique for recording the movement of speech articulators.

The horizontal (X) and vertical (Y) positions of multiple sensors attached to the vocal organs can be recorded as a function of time.

General Principles of operation: transmitter coils

Three transmitter coils that are attached to a head mount produce an alternating magnetic field.

The magnetic fields induce an alternating current in sensors that are attached to the articulators (tongue, lips, jaw, etc).

The size of the induced current is (inversely) proportional to the X (horizontal) and Y (vertical) distance of each sensor from the transmitter coils.

Figure 1: Position of transmitter coils.

Figure 2: The new Carstens system: see

The sensors

Typical dimensions are 1.5 x 1.5 x 1.0 mm.

There are up to 10 sensors and these are attached using a dental adhesive to various articulatory landmarks.

Figure 3: Equipment needed for sensor placement.
Image source: UCLA Phonetics Laboratory

Transducer placement

Tongue and lips

Figure 4: Placement of a tongue and lip sensor.
Image source: UCLA Phonetics Laboratory

Typical transducer placement

Figure 5: Some typical positions for sensor placement.
Image source: UCLA Phonetics Laboratory
  1. a. tongue dorsum: approximately 4 cm from tip of tongue, depending on speaker.
  2. b. midpoint of tongue body
  3. c. tongue tip: approximately 1 cm from tip of tongue
  4. d. base of low front incisors: to measure jaw movement
  5. e. lower lip
  6. f. upper lip
  7. r. bridge of the nose (as a reference point)
  8. r. base of upper front incisors (as a reference point)

Digitisation and data rotation

The multichannel articulatory signals are synchronised with the acoustic speech signal and digitised directly to a SUN workstation disk.

Continuous digitisation is possible -- up to a maximum of about 30-40 minutes, after which the tongue sensors may loosen.

Typical digitisation rates:

500 Hz for the articulatory signals and 20000 Hz for the acoustic speech signal

After digitisation, the signals are often rotated so that the X-axis is parallel to the occlusal plane (a line extending from a bite plane held between the teeth; the Y-axis is then defined to be perpendicular to this).

Data rotation

Figure 6: A bite plate is used to determine the occlusal plane ("bite plane"). This permits data rotation so that x and y movements are standardised across subjects. That is differences in bite plane angle are factored out.


Mostly in basic speech research e.g.

  • analyses of coarticulation
  • articulatory speech synthesis
  • synthesising facial movements

Very rarely used:

  • in clinical speech pathology and treatment
  • in developmental studies (child language acquistion)

Assessing the system


  • The only safe and painless way to obtain quantifiable movement data on articulatory movement (alternatives: electromyography, X-ray microbeam)


  • the equipment is very expensive ($100,000+) and very complicated to use
  • there are many sources of potential error: e.g. sensors are incorrectly placed/become unstuck/break
  • the size/shape of the vocal organs influence trajectory movement