Sentence processing Flashcards
Processing sentence
Recognise individual words
Assign syntactic structure - put words together in meaningful groups (+ group groups) to determine literal meaning
Interpretation - determine intended meaning of sentence in context
Assigning syntactic structure - tree diagram
Words in sentence are structured into phrases and clauses
The way words are structures into phrases can be depicted by tree diagrams - each sentence broken down into smaller constituents (nodes) which are connected via branches.
There is often many different ways in which words can be grouped together - syntactic ambiguity = when each combination gives different meaning to a sentence
Types of syntactic ambiguity
Global ambiguity = remain ambiguous between 2 of more interpretations even when get to end of sentence
Temporary ambiguity = start off as being ambiguous but able to resolve them in favour of one interpretation by time end of sentence
What is the process of providing syntactic structure?
Parsing
Serial model - Garden Path Theory
Only one syntactic structure is initially considered
Sentence meaning not involved in selection of structure
The simplest structure is chosen following
- minimal attachment - initially build sentence with simplest structure - fewer nodes
- late closure - same number of nodes - reader associates incoming material with more recent material
If the simplest structure is incorrect, sentence meaning can influence re-analysis
Parallel model - constraint satisfaction
All relevant sources of information are immediately available to parser
Initial interpretation of sentence depends on multiple sources of information - constraints:
- context
- plausability
- general world knowledge
- verb bias
Constraint satisfaction = competing sentence structures are activated simultaneously - the syntactic structure receiving most support from all constraints is most activated and chosen
Hybrid model considering both serial and parallel?
Unrestricted Race model
What is non-literal language? Give examples.
An expression is non-literal when intended meaning cannot be derived by direct composition of literal meanings of words as guided by grammer.
Examples - metaphor, idiom, irony
Theory of figurative language processing - serial model = Standard pragmatic theory
Function of sentence is to communicate opposite of what is saying
Processing = literal meaning assessed first, mismatch with context is detected, the utterance reanalysed as ironic
Results in processing cost for ironic language compared to the same utterance intended meaning
Theory of figurative language processing - Graded Salience Hypothesis
For highly familiar ironies, the Ironic meaning is accessed straight away
Unfamiliar ironies - literal meaning assessed first and reanalysis must occur - processing cost for unfamiliar ironies
Theory of figurative language processing - Direct Access View
Literal and ironic language are essentially treated in the same way
Processing - ironic meaning can be assessed without accessing the literal meaning first.
No additional processing cost for ironic language as treated in same way as literal language
Context and world knowledge - Hagoort (2004)
Event related brain potentials (ERP) - measure electrical activity generated by brain while performing task e.g., reading sent when encountering a word that doesn’t fit with the sentence
Produces N400 - large spike in electrical activity by brain 400ms after seeing the word with issue
