Language sentence comprehension

Question 1

Give an example of globally and temporarily ambiguous sentence

Answer 1

G: the criminal shot the cop with the pistol (automatically choose one meaning)
T: The criminal shot the dog with the pistol/collar (can tell by the end of the sentence)

Question 2

Explain minimal attachment principle

Answer 2

Fewest number of additional nodes: ‘the criminal shot the cop with the pistol’ Ambiguity occurs when deciding to attach the pp ‘with the pistol’ to VP ‘shot’ or NP ‘cop’. In this case VP uses less nodes -> preferred option.

Question 3

Explain three main properties of the garden path model.

Answer 3

1. modular model: initial syntactic processing not influenced by non-syntactic information
2. serial model: only single analysis is adopted first time
3. Reanalysis model: when initial analysis is inconsistent with information that is used later.

Question 4

‘the lifeguard rescued the man with the motorboat- Which structure is easiest to process according to GPM

Answer 4

VP attachment because it’s got fewer nodes meaning the lifeguard owns the motor boat. Because this makes sense, there is no need for reanalysis according to theory

Question 5

How does the GPM explain that ‘the lifeguard rescued the man with the mustache’ is hard to process

Answer 5

This is because we initially use VP minimal attachment therefore we have to reanalysis and use NP late closure attachment

Question 6

Describe an experiment that shows evidences supporting GPM

Answer 6

Rayner (1983): eye-tracking study had 2 conditions
1. ‘the spy saw the cop with the binoculars but the cop didn’t see him.’-VP
2. ‘The spy saw the cop with a revolver but the cop didn’t see him.’-NP
Difficulty should arrive at ‘revolver’ therefore reading time should be longer. This was found supporting the GPM

Question 7

Explain syntactic reanalysis with an example

Answer 7

Difficulty occurs when the initial analysis is inconsistent with the information that is used later e.g. ‘the criminal shot the dog with the collar’. VP attachment doesn’t make sense therefore need to reanalysis and use late closure process which creates a NP attachment.

Question 8

What are the three main properties in Constraint-based models

Answer 8

1. Interactive models: all sources of information (e.g. context, meaning, frequency) are used immediately and interact with each other.
2. Parallel model: all possible analysis of an ambiguous structure are activated in parallel. The stronger the support for an analysis, the higher the activation.
3. Competition model: when two or more are equally activated by different sources of information.

Question 9

Describe an experiment that investigates whether semantic plausibility information is used during syntactic processing

Answer 9

Ferreira & Clifton, 1986:
GPM: the use of plausibility/semantic information is delayed relative to structural strategies.
CBM: comprehends use plausibility information immediately
1. ‘the defendant examined by the lawyer turned out to be unreliable’ (temporarily ambiguous; reduced relative)
2. ‘The defendant that was examined by the lawyer turned out to be unreliable (unambiguous).
Reading time for ‘by the lawyer’ longer in 1 than 2 consistent with GPM suggest that it is not used during initial syntactic processing but is during reanalysis

Question 10

Describe an experiment that investigates whether semantic plausibility information is used during syntactic processing

Answer 10

Ferreira & Clifton, 1986:
GPM: the use of plausibility/semantic information is delayed relative to structural strategies.
CBM: comprehends use plausibility information immediately
1. ‘the defendant examined by the lawyer turned out to be unreliable’ (temporarily ambiguous; reduced relative)
2. ‘The defendant that was examined by the lawyer turned out to be unreliable (unambiguous).
Reading time for ‘by the lawyer’ longer in 1 than 2 consistent with GPM suggest that syntactic processing is not initially used.

Question 11

How is the sentence ‘the evidence examined by the lawyer turned out to be unreliable’ processed?

Answer 11

CBM: comprehenders should immediately use plausibility information and adopt the reduced relative analysis.
GPM: Comprehenders should initially ignore plausibility and therefore initially adopt the main clause analysis and use plausibility later.

Question 12

Contrasting findings of Trueswell at al (1994) and Clifton et al (2003) regarding the use of plausible information

Answer 12

Trueswell: Used eye-tracking and found reading time for ‘by the lawyer’ was compatible with constraint-based models therefore plausibility information affects the structural preference,
Clifton: same study but used more materials & more subject and found reading times compatible with the GPM

Question 13

explain the reference theory of sentence production

Answer 13

Explains how context may affect syntactic ambiguity resolution. It predicts that people prefer the interpretation that contains the fewest unsatisfied presuppositions. I.e. ‘the criminal shot the cop with the pistol’, in the NP attachment ‘with the pistol’ indicates which cop we are referring too. However, there is no context that indicates that there is more than one cop therefore NP attachment is odd and the presupposition is not satisfied therefore people prefer the VP attachment.

Question 14

What is a discourse presupposition according to referent theory

Answer 14

“the criminal shot the cop with the pistol”
NP attachment= presupposes multiple cops
VP attachment= presupposes one cop
Referent theory suggests we use minimal attachment unless context suggest that the presupposition is left unsatisfied or doesn’t make sense. E.g. ‘there was a cop with a pistol and a cop with a stick, the criminal shot the cop with the pistol’ NP attachment should be preferred (to satisfy presupposition) however GPM prefers the VP attachment initially through minimal attachment.

Question 15

Explain how the sentence ‘the cow injured the farmer with the axe’ is processed according to referent theory

Answer 15

VP attachment is initially preferred because there is no context to suggest that there is more than one farmer therefore presupposition is not satisfied by context to make the NP attachment initially.

Question 16

Why do the GPM and Referent theory make the same predictions when ‘the cow injured the farmer with the axe’ is processed in the absence of context

Answer 16

Both use minimal attachment in the absence of context (VP attachment) because NP attachment would suggest that there is a presupposition that isn’t met by VP which we cannot tell without context therefore minimal attachment is applied.

Question 17

Describe the experiment by Altmann and Steedman (1988)

Answer 17

2 ambiguous sentences:
1. ‘the burglar blew open the safe with the new lock and made away with the loot’
2. ‘the burglar blew open the safe with the dynamite and made away with the loot’
if the context before sentence 1. has 2 reference (two safe) then this takes less time than if there is only 1 reference because without the presupposition we use minimal attachment (VP)
If context for sentence 2. has 2 reference (2 safes) then ‘with the dynamite’ doesn’t help distinguish between safe therefore VP attachment initially made taking longer than if there was only 1 reference in the context because VP attachment is the correct interpretation of the sentence.

Question 18

How does constraint-based theories make use of the ideas of referential theory?

Answer 18

Constraint-based models suggest that context is one of many sources of information that play a role so does not always have as strong of an effect as claimed by referent theory. Although referent theory makes non-minimal attachment earlier, it is not always easier than minimal attachment analysis.

Question 19

describe an experiment that investigated how frequency information is used during sentence processing

Answer 19

Mitchell (1987)
1. ‘ After the child visited the doctor prescribed the injection.’
2. ‘After the child sneezed the doctor prescribed the injection.’
Frequency information uses transitively (with object) or intransitively (without). In the case of ‘visit’ transitive more likely and ‘sneeze’ intransitively.
Therefore if frequency information is only used, then sentence 1 would be proceed with the direct object adopted at ‘doctor’ and wrong at ‘prescribed’. In sentence 2 the subject analysis should be adopted at ‘doctor’ and correct at ‘prescribe’

Question 20

How have sentences such as ‘After the child sneezed the doctor prescribed a course of injections’ been used to test the garden-path model and constraint-based model

Answer 20

Mitchell (1987): If frequency info is used immediately (and nothing else) then subject analysis should adopt at doctor making sentence 1 longer to read than 2 but this was not found.
Direct object analysis was first adopted for both and cause reanalysis at sentence 2 than 1. Therefore frequency does not provide strong enough effect to prevent process of direct object. (inconsistent with CBM and assumes a weak constraint over others and is consistent with GPM.

Question 21

What is meant by serialism and parallelism in sentence processing?

Answer 21

Serialism suggest processer adopts only a single analysis at a time (GPM) whereas parallelism suggest multiple analysis in case of syntactic ambiguity (CBM)

Question 22

Describe a study that investigates whether competition occurs during syntactic ambiguity resolution. What model does it support?

Answer 22

Van Gompel (2001): 
1. the hunter killed only the poacher with the rifle not long after sunset (ambiguous)
2. the hunter killed only the  leopard with the rifle not long after sunset (VP) 
3. the hunter killed only the leopard with the scars not long after sunset (NP) 
RT: ambiguous VP=NP

Question 23

How is ‘the hunter killed only the poacher with the rifle not long after sunset’ processed according to a race model of syntactic ambiguity resolution?

Answer 23

The one that is supported by most information is constructed first because there is no strong preference in this case (both NP & VP). When initial analysis is implausible, process reanalysed on either VP or NP. When initial analysis is plausible, no reanalysis is necessary therefore faster (ambiguous makes sense if NP or VP attachment is applied.

Question 24

Describe evidence that syntactically ambiguous sentences are not always fully reanalysed,

Answer 24

Christianson et al 2001: ‘while Anna dressed the baby that was small and cute spit up on the bed’ found subjects that subjects answered yes to both questions ‘did the baby spit up?’ and ‘did Anna dress the baby?’ 60%
Ferreira (2003) suggested ‘good enough representation theory’: this was because people often don’t process syntactic structure of sentences fully but instead construct a ‘good enough’ incomplete representation. e.g. the dog was bitten by the man. Inconsistent with other theories because assume people construct syntactic representations