Sentence comprehension Flashcards
What are the levels of syntactic structure?
Grammar categories e.g. determiner, noun, verb
Rules e.g. noun-phrase and verb-phrase –> sentence
What is “syntax”?
The rules governing permissible word ordering (which we have an implicit understanding of)
Word meaning is not involved at this stage of processing
Why is syntax crucial to understanding meaning?
It encodes the information of who did what
Surface syntactic structures can differ while a deep syntactic structure coveys the same meaning
What is parsing and what are the 3 possibilities for how readers do this?
The process of computing syntactic structure:
1) Readers don’t compute syntactic structure
2) Readers wait until the end of the sentence before assigning syntactic structure (reliant on memory, wasteful of resources)
3) Make decisions about structure on word-by-word basis
What is the “stops making sense” task for analysis of parsing?
Sentence such as “the man stood by the window…” –> this is ambiguous as to whether it is describing the man or his action
We make an incremental decision here
When the word “waved” is added we go back to the ambiguous part of the sentence and fix the interpretative mistake
Suggests that we are making decisions on word by word basis
What is meant by the Garden Path Effect?
Sentences are very frequently structurally ambiguous and lead to reading difficulty - when a word seems incongruent with current syntactic structure
We are “led down the garden path” in analysis of the ambiguity and have to backtrack to fix the error when we obtain more information
How does “the horse raced past the stables” illustrates the garden path effect?
The word “stumbled” is added
There is ambiguity between that main clause and this relative clause which provides modifying information specifying which horse is under consideration
We have a preference for main clause analysis which is what leads to processing difficulty (structurally simpler and more frequently encountered)
What other preferences do we show which lead to processing ambiguities?
Preferences for adjective nouns e.g. “the old man” as opposed to noun verb analyses
Preferences for verb argument analyses e.g. “jogs a mile”
How do we process the sentence “the boy ate the broccoli naked”?
In this scenario, upon reaching the end of the sentence, we have to backtrack and use contextual knowledge to resolve the ambiguity
How can parsing and the Garden Path effect be investigated?
By comparing reading times for portions of ambiguous sentences and portions of unambiguous counterparts e.g. compare “the man stood by the window waved” and “the man who stood by the window waved”
This is a GRAMMATICALITY DECISION TASK - the sentence is viewed one word/phrase at a time and the participant judges grammatical acceptability and point at which the sentence stops making sense
What is a potential issue with grammaticality decision tasks?
Don’t reflect naturalistic behaviour - when we read we don’t normally make explicit grammatical decisions, we don’t normally read in chunks of words/phrases at a time, and we don’t have to read with the aim of decision making through button pressing
Why is self-paced reading a slightly better technique than decision tasks?
Sentences are still viewed one word at a time but the participant themselves indicates transition to next word and is not asked to indicate when the sentence stops making sense
Instead inferences are made based on how long it takes to indicate need for next word, no explicit decision making involved
Why are self-paced reading tasks criticised?
Self-paced judgements slow reading rate and the presentation of text is still not naturalistic
What is done in eye-tracking experiments?
Eye movement are recorded as participants read sentences
Length of fixations reflect difficult processing words, and this differs between ambiguities
May also see regressions (back tracking) when fixing interpretation errors
What are 2 advantages of the eye tracking technique?
Allows experimenter to examine real-time processes in reading without including a secondary task e.g. button pressing
Participant doesn’t have to be aware of exact experimental purpose –> more naturalistic behaviour