Lecture 8: Discourse Processing Flashcards

(20 cards)

1
Q

What is discourse?

A

Discourse:

  • ‘linguistic unit composed of several sentences’ (Harley, 2008)
  • can be relevant to both written and spoken language
  • can be relevant to both comprehension and production
Other related terms:
Text: restricted to written language
Narrative: story
Dialogue: conversation between 2 or more people
Conversation:
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Construction-integration Model

A

Kintsch (1988): two-stage model of discourse comprehension

CONSTRUCTION STAGE
Surface Model 
    -Word meanings
    -Syntactic parsing
Text-base
    -Propositions
    -Inferences

INTEGRATION STAGE
Situation Model
-ideas/events the text is about
-Network of inter-related propositions is integrated into a coherent structure.
-Any contradiction or incorrect inferences are resolved.

3 different types of mental representations

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Propositions

-2 examples/experiments

A

Organisation of propositions:
Kintsch (1994):
Proposition: predicate [agent, recipient, theme]

“The customer wrote the company a complaint.”
Proposition: write [customer, company, complaint]

“The geese crossed the horizon as the wind shuffled the clouds.”
Proposition 1: crossed [geese, horizon]
Proposition 2: shuffled [wind, clouds]

‘Psychological Reality’ of Propositions?:
Ratcliff & McKoon (1978): sentence/word recall
-“The geese crossed the horizon as the wind shuffled the clouds.”
-Task: Subjects read the sentences, did some distractor task, and then later were asked to write down what they remembered about the sentences.
-In each trial, a word from either the same or different clause (=proposition) was given as a cue.
-Retrieval rate: cue from same proposition > cue from different proposition (with distance controlled):
e.g., retrieval rate of ‘horizon’: cue - ‘geese’ > ‘wind’
-Propositions are encoded as a package.
-Elements in the same proposition are organised closely in memory.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Situation/Mental Model

A

Situation Model (van Dijk & Kintsch, 1983):

  • also called a mental model (Johnson-Laird, 1983).
  • internal representation of the external world
  • Comprehenders construct a model as they go along to represent what they hear and read.
  • Models represent different types of information about situations described in discourse (Zwaan & Radvansky, 1998):
    • space
    • time
    • causation
    • motivation of characters in discourse
  • Models are dynamically updated as discourse is processed.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Context and Situation Model

How does context help to form a situation model?

A

Bransford & Johnson (1973):
Comprehension and memory of abstract stories:
If the balloons popped, the sound wouldn’t be able to carry since everything would be too far away from the correct floor. A closed window would also prevent the sound from carrying, since most buildings tend to be well insulated. Since the whole operation depends on a steady flow of electricity, a break in the middle of the wire would also cause problems. Of course, the fellow could shout, but the human voices are not loud enough to carry that far. An additional problem is that a string could break on the instrument. Then there could be no accompaniment to the message. It is clear that the best situation would involve less distance. Then there would be fewer potential problems. With face to face contact, the least number of things could go wrong.

Bransford & Johnson (1973):
3 conditions: 
No context (picture)
Context (picture) before text
Context (picture) after text

Comprehensibility rating & numbers of ideas recalled:
Context before > Context after = No context.

  • Context helps readers to form a ‘mental model’ to understand and memorise texts.
  • However content should be available at the time of comprehension.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Space in Situation Model

How spatial information is represented in the situation model?

A

Bransford, Barclay, & Franks (1972): sentence recognition

(1) Three turtles rested on a log and a fish swam beneath them.
(2) Three turtles rested on a log and a fish swam beneath it.
(3) Three turtles rested beside a log and a fish swam beneath them.
(4) Three turtles rested beside a log and a fish swam beneath it.
(picture for 1 and 2 the same)

Task: read one of the four sentences, and later decide which of two sentences they had read.
Recognition rate: (1) & (2)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Verbatim vs Situation Model

Is the surface form of the sentence (verbatim) represented in comprehenders’ memory?

A

Fletcher & Chrysler (1990): sentence recognition
(story about george and purchasing art treasures that cost a lot, wife was angry bought necklace, statue most expensive, carpet cost the least. ect)

Task: read the text, and later decide whether a test sentence appeared in the text.

Test sentence (all false):

(1) surface: George says his wife was angry when she found out that the necklace cost more than the rug. (was actually carpet)
(2) text-base: George says his wife was angry when she found out that the painting cost more than the carpet. (was actually necklace)
(3) situation model: George says his wife was angry when she found out that the necklace cost more than the vase. (carpet not vase)

Task: Did you see this sentence?
Recognition rate (correct rejection): (1)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Summary of Discourse

A
  • Construction-Integration model (Kintsch, 1988): 2-stage model of discourse processing
  • Situation / Mental model: part of CI model (also as a separate model)
  • Representations of discourse are often organised with a proposition as a unit.
  • The construction of a situation model can be helped with context when presented at the time of processing.
  • The situation model contains representation of spatial relationships between objects in the discourse.
  • Experiments suggest there are different types of discourse representation: surface model; text-base; situation model.
  • The surface form of discourse might not necessarily is kept maintained as it is. The situation model, on the other hand, tends to be maintained in memory well.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Moses Illusion

A

How many animals of each type did Moses take on the Ark?”
It wasn’t Moses, but Noah, who took animals on the Ark!

Many readers don’t notice the anomaly, and give an answer. (e.g., Erickson & Mattson, 1981).

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Mose Illusion in Discourse

-experiment

A

story: (There was a tourist flight travelling from vienna to barcelona. On the last leg of the journey, it developed engine trouble. Over the Pyrenees, the pilot started to lose control. The plane eventually crashed right on the border. Wreckage was equally strewn in France and Spain. The authorities were trying to decide where to bury the survivors, What is the solution to the problem?

Many people don’t notice the anomaly, and give an answer = Moses illusion replicated in discourse (Barton & Sanford, 1993)

  • Situation-specific real-world knowledge is automatically activated during comprehension, and readers focus on specific parts of the knowledge. (Sanford & Garrod, 1981)
  • Thus, the situation model could override semantic information in individual words in the sentences.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

How to reduce the illusion?

A

Is there any way to eliminate / reduce the illusion?
-Perhaps guiding comprehenders’ focus to the critical words (‘Moses’) would help?

-“Moses took two of each kind of animal on the Ark.”

  • “It was Moses who took two of each kind of animal on the Ark.” (it-cleft) (Bredart & Modolo, 1988)
  • “There was a guy called Moses who took two of each kind of animal on the Ark.” (there-insertion)
  • “MOSES take two of each kind of animal on the Ark.” (capitalisation) (Bredart & Docquier, 1989)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

In what other ways, can we attract attention to certain parts of the sentence(s)?
-2 experiments

A
  • Comprehenders take longer to process a given word if it appears as the first word in the sentence (Aaronson & Scarborough, 1976).
  • And they remember the first-mentioned object better than subsequent ones - ‘Advantage of First Mention’ (Gernsbacher & Hargreaves, 1988; Gernsbacher, Hargreaves, & Beenman, 1989):

(1) Tina beat Lisa in the tennis match.
(2) Tina and Lisa beat Susan in the tennis match.
(3) Tina was beaten by Lisa in the tennis match.

Does the content in the most recent substructure have advantage?

Caplan (1972): word recognition

(1) Now that artists are working fewer hours, oil prints are rare.
(2) Now that artists are working in oil, prints are rare.

Task: Did this word (‘oil’) appear in the sentence?
RT: (1)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Summary of Attention

A
  • Misguiding of attention can lead comprehenders to not noticing semantic anomalies in discourse.
  • Languages have their own ways to attract focus on certain parts of sentence / discourse.
  • Comprehenders tend to remember first-mentioned objects well.
  • Also, objects in the most recent unit (e.g., proposition) tend to be remembered well.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What is an Inference?

-inference in discourse processing example

A

Inference:

  • the derivation of additional knowledge from facts already known.
  • This might involve going beyond what is actually presented in the linguistic input.

Inference in discourse processing:

(1) The dentist drilled the tooth. The broom was tattered.
(2) The worker swept the floor. The broom was tattered.

(1) is incoherent, whereas (2) is coherent.
However, the coherency in (2) depends on inference that works out the relationship between ‘swept’ and ‘broom’ in (2).

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Bridging Inference

-experiment

A
Haviland & Clark (1974):
(1) repetition: 
We got some beer out of the trunk. 
The beer was warm.
(2) inference: 
We checked the picnic supplies. 
The beer was warm.

Reading Times for ‘The beer was warm’: (1)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Causal Bridging Inference

-2 different experiments

A

Keenan, Baillet, & Brown (1984); Myers, Shinjo & Duffy (1987)
(1) highly related:
Tony’s friend suddenly pushed him into a pond.
He walked home, soaking wet, to change his clothes.
(2) moderately related:
Tony met his friend near a pond in the park.
He walked home, soaking wet, to change his clothes.

Reading Times for ‘He walked… his clothes.’: (1)

17
Q

Elaborative Inference

A

Elaborative Inference: inference that requires extending what’s in the text to real-world knowledge.

Explicit: The tourist took the picture with the camera. The scene was more beautiful that he remembered.
Bridging Inference: The tourist took the picture of the church. The camera was the best he had ever owned.
Elaborative inference: The tourist took the picture of the church. The scene was more beautiful that he remembered.
Test sentence: The tourist used a camera.

Singer (1980):
Task: to judge whether the test sentence was true.
Judgement times:
Explicit = Bridging

18
Q

Open Role in Inference

  • example
  • experiment
A

To maintain the coherence in discourse, comprenders fill an ‘open role’ in bridging inference.

-Yuki drove to Edinburgh yesterday. The car kept overheating.
 “ ‘The car’ must be the one Yuki drove to Edinburgh.”
 The open role is filled with ‘the car’:
“Yuki drove [a car] to Edinburgh yesterday…”
Is the inference driven by real-world knowledge (a car is a typical vehicle to drive) or a lexical association (drive – car)?

Garrod & Terras (2000): reading (eye-tracking)
2x2 design:
Context:
(1) The teacher was busy writing a letter of complaint to a parent.
(2) The teacher was busy writing an exercise on the blackboard.
(Instrument – open role)

Target:
(3) However, she was disturbed by a loud scream from the back of the class and the pen dropped on the floor.
(4) However, she was disturbed by a loud scream from the back of the class and the chalk dropped on the floor.
(lexical association with ‘write’: pen > chalk)

Conditions: appropriate context: (1)(3); (2)(4)
inappropriate context: (1)(4); (2)(3)

First-pass reading times (‘early’ measure) at ‘dropped’:
‘pen’: (1)(3) = (2)(3) ‘chalk’: (2)(4) F6, F8->F11

First fixation duration:
F4
First-pass reading time:* 
F4+F5
Regression pass duration (Go past time):
F4+F5+F6+F7
Second-pass reading time:*
F7+F9+F10+F12
Total reading time: F4+F5+F7+F9+F10+F12
  • relevant to this expt
19
Q

What Model did Garrod and Terras (2000) propose based on their experiment?

A

Garrod & Terras (2000):

Two-stage inference model:
Inference processes involve two processes:

Stage 1 – “bonding”
Items suggested by lexical content are automatically activated and bonded with the verb (‘write’  ‘pen’).
Open roles (Instrument) are filled with those items.

Stage 2 – “resolution”
The link between the open role and the filler is checked against the discourse.

20
Q

Summary of Inference

A
  • Comprehenders often employ inferences in order to make sense of discourse.
  • There are different types of inference: bridging; causal bridging, elaborative.
  • Elaborative inference is often used to understand discourse, but it does not take place automatically (rather, at a later stage of processing).
  • In contrast, bridging inference can take place early.
  • Comprehenders fill open roles in discourse processing.
  • Open roles are filled using lexical associations between verb and open roles first, then real-world knowledge later.
  • Two-stage inference model: Bonding + Resolution