Sentence Comprehension

Question 1

Introduction - Sentence Comprehension

Answer 1

• What happens after we recognise a word
  Parsing - Process of working out the syntactic structure of a sentence/syntactic analysis. First step is to determine the syntactic category to which each word belongs to. These then combine to form phrases.
• Ambiguous sentences often cause processing difficulty.

Globally Ambiguous = Sentence doesn’t tell you which possibility is correct. Has at least two possibilities.

Temporarily Ambiguous = When finished reading you know which sentence is correct. Has one correct interpretation. Often use knowledge about the world to work.

Syntactic Ambiguity = A sentence may be interpreted in more than one way due to an ambiguous sentence structure

Question 2

Explain what Garden-path sentences are

Answer 2

Sentences that lead you up to a dead end when you adopt one analysis, but it turns out to be wrong later when you go to re-analyse.

Contain a relative clause = Modifies the main noun

Question 3

Explain the garden-path model/theory

Answer 3

Theory that attempts to explain parsing of ambiguous or unambiguous sentences.
First applying the Minimal Attachment principle, and if this is unsuccessful, applying the principle of Late Closure

Question 4

Explain the principle of minimal attachment within garden-path theory

Answer 4

• First strategy of the model, using syntactic information
• Principle of if grammatically permissible, attaching an ambiguous phrase into the tree structure using the fewest number of additional nodes.
• Processor adopts the analysis that requires the simplest structure interpretation
• E.g. ‘the criminal shot the cop with pistol’. Have no issue with this as can use minimal attachment to solve global ambiguity.
• ‘The criminal shot the dog with the collar’ Principle states that this results in an implausible analysis as a a collar cannot be used to shoot. Have to reanalyse.

Question 5

7. Describe an experiment that shows evidence for the garden-path theory.

Answer 5

Traxler, Pickering and Clifton (1998)

• Eye-movement
• 1. ‘The steak with the sauce that was tough didn’t win a prize’
• 2. The steak with the sauce that was runny didn’t win a prize.
• Results showed the highest reading time for ‘that was tough/runny’ in 1.
• Supports late closure

Question 6

Explain the principle of late closure in the garden-path model

Answer 6

• Second strategy of GP.
• Used if minimal attachment does not make a prediction regarding resolution.
• The principle that if two structures involve the same number of nodes, you attach the incoming phrase to the phrase currently being processed (low in the tree).

E.g. ‘The man realised we left yesterday’
- Have to use late closure as minimal attachment won’t solve.

Recency Preference = An ambiguous phrase is attached to the most recent clause

Question 7

Rayner, Carlson and Fraizrer

Answer 7

1. ‘The spy saw the cop with the binoculars, but the cop didn’t see him’
2. ‘The spy saw the cop with a revolver, but the cop didn’t see him’

• When get to 2 have to re-analyse
• People took longer to read 2. Supports minimal attachment.

Question 8

Explain the assumptions of the garden-path model

Answer 8

• When we process sentences we first use syntactic info
• A lot of the time sentence structure is unambiguous which results in the correct structural analysis
• When structure is ambiguous, use syntactic info first.

Question 9

Explain the properties of the garden-path model

Answer 9

Modular - Initial syntactic processing is not influenced by non-syntactic info.

Serial Model - Only a single analysis happens at once

Reanalysis - Difficulty happens when the initial analysis is inconsistent with the info used later on.

Question 10

Describe constraint based models

Answer 10

• A group of models that claim that syntactic processing is not modular. Non-structural information is used immediately and at the same time as structural information
• Processor uses multiple sources of information (constraints) to assign a structure and activate alternative syntactic structures in parallel.

Question 11

Explain the properties of constraint-based models

Answer 11

Interactive - All sources of info used immediately and interact

Parallel - All analyses of an ambiguous structure are activated in parallel. The stronger the support for an analysis, the higher its activation.

Competition - Difficulty occurs when two or more analyses are equally activated by the different sources of information

Question 12

Main Clause Analysis

Answer 12

When the ambiguity is part of a main clause in the sentence

E.g. ‘The defendant examined the suspect

Question 13

Reduced Relative Analysis

Answer 13

When the ambiguity is part of a lesser relative clause in the sentence

E.g. ‘The defendant examined by the lawyer was unreliable’

Question 14

Trueswell, Tanehaus and Garnsey (1994)

Answer 14

• Eye-movement study with 4 conditions animacy (animate/inanimate) vs ambiguity (ambiguous vs unambiguous).
• Reading longer and harder for animate and ambiguous sentences, for which main clause analysis was plausible, than for animate but unambiguous sentences
• Plausibility did not affect ambiguity resolution for inanimate sentences
• Consistent with the Constraint-Based Models, plausibility information seemed to affect the structural preference

Question 15

Clifton et al. (2003)

Answer 15

• Argue that Trueswell et al. (1994) didn’t have enough experimental materials and subjects so used more in their study which used the same sentences.
• Inanimate and ambiguous sentences took longer and were harder to read than inanimate but unambiguous sentences
• Because main clause analysis was implausible in the former
• Consistent with the Garden-Path Model