Speech production

1
Speech production
2
People

• Dr Antje S. Meyer
• Dr Jane Morgan
• Dr Femke F. van der Meulen
• Fieras Lethaus
• Changming Wu

3
Research strategy

• In most experiments conducted by the members
  of the group, participants are asked to name or
  describe objects or events they see. Their speech
  is recorded on tape so that researchers can
  analyse what they said, when they hesitated or
  made speech errors. In addition, researchers are
  often interested in the speakers speech onset
  latencies and speech rates because they reflect
  on the difficulty of the speech planning
  processes. Finally, it is sometimes useful to
  monitor the speakers eye movements because they
  show which parts of a picture attract the
  speakers attention and in which order and for
  how long they are processed. This information can
  be linked to the record of the speech output in
  order to study how speakers co-ordinate speech
  planning and speech output in time (see Figure
  2).

4

• Figure 1 Participant wearing the headband of
  the SMI Hi-Speed 2D eye tracker.

5
3
.
.
2
.
1
.
.
.
5
.
6
7
4
The scooter the bed the bell and the book
Speech
6
Facilities

• The Centre houses test rooms dedicated to
  speech production research. The laboratory
  computers and associated software (NESU) can be
  used to displays images for very short (16 ms) or
  longer time periods (several seconds) and to
  measure the speakers speech onset latencies.
  Picture onset and offset times can be
  pre-programmed or determined on-line depending on
  properties of the participants speech.
  Audio-equipment is available to present auditory
  stimuli (tones or speech) and to record the
  speakers speech for later analyses. In one room,
  there is an SMI HiSpeed-2D eyetracker (see Fig.
  1). The eye movements can be recorded for
  off-line analyses, and they can be used on-line
  to trigger gaze-dependent display changes.

7
Basic findings

• When speakers name or describe sets of
  objects they havent seen before, their eye
  movements are tightly co-ordinated with their
  speech.output. They usually look at each object
  just before naming it (see Figure 2) and their
  gaze remains on the object until they are about
  to say the last word of the phrase referring to
  the object. When speakers use a single syllable
  to name an object (dog), they will usually
  fixate the object for a much shorter time (about
  500 ms) than when they use a longer expression
  (the little brown dog, 1300 ms). These results
  suggest that, contrary to widely held believes,
  speakers produce descriptions in a very
  sequential fashion, completing all planning for
  one object and its name before starting to
  process another object.

8
Repeated reference

• . When speakers refer again to objects they
  have named before, they usually look at the
  objects again. For instance, when speakers
  produce utterances such as The dog next to the
  tree is brown, they usually look at the dog
  twice, shortly before saying dog and shortly
  before saying brown. Sometimes, however,
  speakers refer to known objects without looking
  at them again. This is particularly likely to
  happen when the referring expression is short and
  high in frequency (e.g., a pronoun such as it).
  We assume that as default speakers plan an eye
  movement to each object they refer to. However,
  when an appropriate referring expression becomes
  available very rapidly (as may be the case for
  pronouns), the planned eye movement to the object
  may be cancelled and replaced by an eye movement
  to another object.

9
Generating sentences

• When speakers describe simple events and
  select syntactic structures (The man gives the
  boy a key or The man gives a key to the boy),
  their gaze patterns become more complex, but they
  are far from chaotic. We observe that speakers
  often first fixate the middle of the display,
  which may be a good vantage point to gain an
  overview of the scene and identify the type of
  action that is occurring. After that brief
  preview phase, speakers tend to fixate upon the
  entities they name in the order of mention. The
  temporal co-ordination between eye gaze and
  speech is similar to the co-ordination observed
  in multiple-object naming.

10
References

• Levelt, W.J.M., Roelofs, A. Meyer, A.S. (1999).
  A theory of lexical access in language
  production. Behavioral and Brain Sciences, 22,
  1-38.
• Meyer, A.S. Bock, K. (1999). Representations
  and processes in the production of pronouns Some
  perspectives from Dutch. Journal of Memory and
  Language, 41, 281-301.
• Meyer, A.S., Sleiderink, A.M. Levelt, W.J.M.
  (1998). Viewing and naming objects Eye movements
  during noun phrase production. Cognition, 66,
  B25-B33.
• Meyer, A.S. Meulen, F.F. van der (2000).
  Phonological priming of picture viewing and
  picture naming. Psychonomic Bulletin Review, 7,
  314-319.
• Meulen, F.F van der, Meyer, A.S. Levelt, W.J.M.
  (2001). Eye movements during the production of
  nouns and pronouns. Memory and Cognition, 29,
  512-521.