Ling 423/640G: Cognitive Linguistics
Ben
Bergen
Meeting 12: Mental
simulation II
October 2, 2008
Experimental
methods for simulation semantics
Experiments in this
area typically present participants with:
á
Some piece of
language that they have to process
o
A word
o
A sentence
á
A perceptual
or motor task, e.g.:
o
Categorizing
an object as a circle or square (when it appears in different places on the
screen)
o
Deciding
whether a depicted object had been mentioned in the preceding sentence
o
Pushing a
button, which is either close to their body or far from their body
The dependent measure is usually reaction
time (but need not necessarily be), e.g.
á
Time to press
a button indicating whether they saw a circle or square
á
Time to press
a button indicating that a sentence is meaningful
The effect they look for is either compatibility
or interference.
á
Compatibility: Faster responses when perceptual/motor
processes, sentence are compatible
o
After
processing a sentence like You gave Andy
the pizza, people are faster to press a button if it requires them to move
their hand in a direction compatible with giving someone pizza (away from their
body).
o
After
processing a sentence like The ranger saw the
eagle in the sky, people are faster to press a button indicating that they
read a sentence about an eagle when the eagle has the appropriate
(spread-winged) shape.
á
Interference: The response is faster when the perceptual or
motor processes are incompatible with the sentence
o
After
processing a sentence about an object in the upper part of the visual field,
categorizations of an object in the same location are slower.
á
Compatibility
effects appear when there is some time between the two tasks, while
interference seems to occur when they happen at the same time.
o
This makes
sense – you can't do two things with one part of your brain at the same
time. If processing language about visible objects uses your visual system,
then processing such language should interfere with – be incompatible
with – simultaneously using the same system to see.
o
But if
there's a time delay between the processes, it should be facilitatory
priming
Case
study – Zwaan et al. (2002)
This paper compares predictions of two
theories of language understanding.
á
"Amodal propositional representations" is the
conventional view. This is the idea that the meanings of words and sentences
are:
o
Amodal, meaning they
don't use specific modalities (like vision, motor control, etc.)
o
Propositional
representations, meaning that they are basically just rephrasings
of the meaning, in terms of some "language of the mind"
o
These are
views like the "features" or "truth-conditional semantics"
views we talked about at the beginning of the course.
á
"Perceptual
symbols" is the view aligned with the idea that meaning uses mental
simulation
o
Basically,
the idea is that words and sentences activate perceptual and motor systems
o
People
internally (re-)experience the sights, sounds, and
actions language describes
The two theories make different
predictions
á
Amodal propositional
representations doesn't say anything at all about whether or not people perform
mental imagery of perceptual properties of described objects
á
Perceptual
symbols theories make a clear prediction: when people process language about
objects, they should mentally represent their perceptual properties, like shape
and orientation
The two experiments reported in this
paper test for the prediction that people mentally simulate objects when they
hear them described, in particular whether they build mental images of what
those objects are implied to look like.
Participants read sentences, which
mentioned objects in one of two locations.
á
The ranger say the eagle in the sky
á
The ranger saw the eagle in the nest
Then they saw a picture depicting the
object (in this case, the eagle) in one of two shapes (in this case, either
with outstretched wings or with tucked-in wings)
Participants had to either decide as
quickly as possible whether the depicted object had been mentioned in the
preceding sentence (Exp 1) or just name the depicted object (Exp 2).
The predicted effect (the experimental
hypothesis) was that responses would be faster when the object was depicted in
the same shape implied by the sentence:
|
faster slower
|
|
|||||||||||
|
faster slower "eagle in
the nest" |
|
|||||||||||
The key independent variable, Condition,
was categorical and had two levels (Match or Mismatch). There were also two
other independent variables – Picture Version and List, neither of which
are directly relevant to the experimental hypothesis.
Both experiments showed a significant
main effect of Condition (using repeated-measures ANOVA)