Causal Categories in Cognition and Language

1
Causal Categories in Cognition and Language

• Phillip Wolff
• Emory University
• Atlanta, GA

2
The phenomenon

• Causal chain Linguistic expression
• Sara caused the door to open.
  (periphrastic)
• Sara opened the door. (lexical)
• Sara caused the door to open.
• Sara opened the door.

3
More examples (Dowty, 1979)

• 1) a. The low air pressure caused the water to
  boil.
• b. The low air pressure boiled the water.
• 2) a. A change in molecular structure caused the
• window to break.
• b. A change in molecular structure broke the
  window.

4
Whats going on?
(e.g., Brennenstuhl Wachowicz, 1976 Comrie,
1985 Croft, 1991 Cruse, 1972 Dowty, 1979
Frawley, 1992 Gawron, 1985 Kozinsky Polinsky,
1993 Kemmer Verhagen, 1994 Levin Rappaport
Hovav 1994 McCawley, 1978 Pinker, 1989
Shibatani, 1976 Smith, 1970 Wierzbicka, 1988)

• Sara caused the door to open. (periphrastic)
• Sara opened the door. (lexical)
• Sara caused the door to open.
• Sara opened the door.

Direct causation
Indirect causation
5
Why should we care?
Davidson, 1969/2001 Wolff Gentner, 1996
Wolff, 2003 Croft, 1991, DeLancey, 1983, 1984,
1991 Déchaine, 1997 Frawley, 1992 Goldberg,
1995 Haiman, 1983 Kiparsky, 1997 Rappaport
Hovav Levin, 1997 Shibatani, 1976

• Sara caused the door to open. (periphrastic)
• Sara opened the door. (lexical)
• Sara caused the door to open.
• Sara opened the door.

Single event construal available
No single event construal available
6
The standard proposal

• Direct causation Event construals
• Causal expressions

7
Definitions of direct causation

• Temporal contiguity (Fodor, 1970 Smith, 1970
  Goldberg, 1995)
• Physical contact (Ammon,1980 Nedyalkov
  Silnitsky, 1973 Shibatani,
• 1976 Wierzbicka, 1975)
• Control (Brennenstuhl Wachowicz, 1976 Smith,
  1970)
• Efficiency (Gawron, 1985)
• Intentionality (DeLancey, 1983 Cary, Hilton,
  Keil, Morris, Spelke, Talmy, 1995
• Schlesinger, 1989 see also Kiparsky,
  1997 Talmy, 1976, 1988)
• Mediacy (Comrie, 1985 Cruse, 1972 Rappaport
  Hovav Levin, 1999 Kemmer
• Verhagen, 1994 Verhagen Kemmer, 1997)
• Conventionality (Shibatani, 1973)
• Stereotypicality (McCawley, 1978)
• Prototypicality (Lakoff Johnson, 1980)

8
Direct causation in terms of mediacy

• The distinction between direct and indirect
  causatives is concerned with the mediacy of the
  relationship between cause and effect. (Comrie,
  1985, p. 165)
• Indirect causation can be defined as a situation
  that is conceptualized in such a way that it is
  recognized that some other force besides the
  initiator is the most immediate source of energy
  in the effected event. (Verhagen Kemmer, 1997,
  p. 67)
• the primitive requirement for direct causation
  is that there be no intervening eventbetween the
  causing subevent and the result subevent
  (Rappaport Hovav Levin, 1999, p. 33)
• It appears that in discussing covert causatives
  we must understand direct to mean that no agent
  intervenes in the chain of causation between the
  causer (represented by the subject of the verb)
  and the sufferer of the effect (represented by
  the object) (Cruse, 1972, p. 524)

9
No-intervening-cause criterion (Wolff, 2003)
10
Testing the no-intervening-cause criterion

• Predictions of the standard proposal
• P(lexical unmediated) gt P(lexical mediated)
• P(1 event unmediated) gt P(1 event mediated)

mediated
unmediated
11
Experiment 1 Mediated vs. unmediated causal
chains

• Participants. 16 undergraduates
• Materials. 16 animations of causal chains
  involving 3 marbles
• Procedure.
• 1. Choose description
• 2. Count events

12
Procedure (continued)

• Choose a sentence
• a. The green marble moved the yellow marble.
• b. The green marble made the yellow marble move.
• c. neither of the above
• Count events
• How many events occurred between the green
• and yellow marbles?

13
Exp. 1 Results
14
Role of intention (DeLancey, 1983 Carey,
Hilton, Keil, Morris, Spelke, Talmy, 1995
Schlesinger, 1989 see also Brennenstuhl
Wachowicz, 1976 Kiparsky, 1997 Talmy, 1976,
1988)

• The assassin killed the ambassador with poison.
• The bowler toppled the pin.
• The woman extinguished the flame.

15
Experiment 2 Describing mediated chains

• Participants. 48 undergraduates
• Materials. 12 pairs of mediated causal chains

16
Procedure Predictions

• Sentence choices
• a. The man collapsed the house of cards.
• b. The man caused the house of cards to
  collapse.
• c. neither of the above
• Event judgments
• Yes or No This animation shows a single
  event.

17
Exp. 2 Results
18
Summary

• Evidence for the standard proposal

19
Direct or indirect?

• The woman spread out the handkerchief.
• The woman caused the handkerchief to spread out.

20
Direct or indirect?

• The truck tipped over the bookcase.
• The truck caused the bookcase to tip over.

21
Definitions of direct

• Having no intervening persons, conditions, or
  agencies
• Proceeding without interruption in a straight
  course or line

22
Directness in terms of direction

• Direct causation the causal influence and the
  result are roughly in the same direction
• Indirect causation the causal influence and the
  result are in different directions

23
Force dynamics (Talmy, 1988 see also
Jackendoff, 1991 Kemmer Verhagen, 1994
Pinker, 1989)

• Elements of a causal interaction

Patient
Result
Affector
(Antagonist)
(Agonist)
Sunlight caused the gases to react.
24
Vector model (Jackendoff, 1991 Wolff Song,
2002, 2003)

• The blast caused the boat to heel.
• Vitamin B enables the body to digest food.
• Heavy overnight rain prevented the tar from
  bonding.

25
Support for the model (Wolff, Song Driscoll,
2002 Wolff Song, 2003)

• Shared features
• CAUSE PREVENT 1
• CAUSE ENABLE 1
• PREVENT ENABLE 1

Stress .09 R2 0.97
26
Designations

• A Force exerted on patient by the affector
• P Force vector associated with the patient
• (i.e., the patients tendency)
• O S of all Other forces acting on patient
• R Resultant force acting on the patient
  (APO)
• E Position vector

27
Dimensions of the vector model
CAUSE
ENABLE
PREVENT
28
Testing the vector model
a. The fans caused the boat to hit the cone. b.
The fans helped the boat to hit the cone. c. The
fans prevented the boat from hitting the
cone. d. None of the above.
29
Experiment 3 1D interactions

• Participants 18 undergraduates
• Materials Eight animations generated from 3D
  Studio Max and the Havok Reactor physics engine
• 1-4 A gt P
• 5-8 A lt P

30
Predictions
Cause
Help
31
Predictions
Prevent
No verb
32
Results E3
33
Experiment 4 2D interactions

• Participants 18 undergraduates
• Materials Ten animations, A P

34
Predictions

• Cause Prevent
• Help No verb

35
Results E4
36
Configurations associated with CAUSE
E
E
E
E
37
Configurations and direct causation
Direct causation
Indirect causation
A-R angle
0?
90?
E
E
45?
135?
E
E
38
Experiment 5 Direct vs. Indirect

• Participants 16 undergraduates
• Materials Eight animations
• Procedure
• Choose descriptions
• Make event judgments

39
Procedure (continued)

• Description task
• a. The fans pushed / blew the boat into the cone.
• b. The fans caused / made the boat move into the
  cone.
• c. None of the above
• Event task
• Can this animation be viewed as a single event?

40
Description predictions
41
Event predictions
42
Description results
43
Event results
44
Experiment 6 Complex Scenes

• Participants 15 undergraduates
• Materials 12 animations
• 6 Direct A-R angle lt 90?
• 6 Indirect A-R angle gt 90?

45
Procedure Predictions

• Procedure
• Description task
• Direct Lexical causative
• Indirect Periphrastic causative
• Event task
• Direct 1 Event
• Indirect 1 Event

46
Results
47
Conclusions

• Causal expressions depend on the direction of the
  affector vis-à-vis the resultant
• Mediacy is not the whole story
• The results support the vector model
• Implications for the standard proposal

48
What next?

• Relation to other categories of events / verbs
• P manner verbs (walk, crawl)
• P E path verbs (approach, leave)
• A P two-argument activities (push, pull)
• A, P, E cause verbs (break, make)
• E preps (above, below) (Regier Carlson,
  2001 Zwarts Winter, 2000)