Lecture 2- Visual and auditory blocks of language Flashcards
describe the structure of language
- phonology (sounds )- move form sounds to phonemes to words as sounds isnt enough
- orthography (written representation of phonology- eg letters and characters)- moving from images to graphemes
what is an allaphone
same phonemes vary dependant on :
where the person is from
how they say it
coarticulation (filling in missing sounds)
whats the difference between a sound and a phoneme ?
=- pjhonemes distinguished by voice onset time (time from starting to produce words up until they hit the vocal chords and start vibrating)
whatr is meant by categorical perception of phonemes
you either hear B or P - no inbetween
who creatred the Logogen model
Morton 1964
what are logogens
word detectors- which try and collect info for one word
these are linked to visual and auditory info
describe the logogen model
logogen activated by sensory input (Visual or auditory info which is presented)
analysed and evidence provided
logogen model gives cues to cognitive system (bottom up processing)- cog system gives semantic info back to the logogen system
when the threshold of logogen ‘wins’ and all info about words becomes available - enters buffer and provides responses
responses could be- eg definition of ice , or knowing cream follows the word ice
how does the logogen model relate to cognitiv systems
logogen model gives cues to cognitive system (bottom up processing)- cog system gives semantic info back to the logogen system
what is a key assumption of the logogen model
Firing threshold depends on how often youve seen the logogen before - when threshold is reached logogen wins
give some strengths of the logogen model
Begins to explain how different inputs are integrated
Can explain word frequency effects by assuming different thresholds (e.g. the threshold for “student” is lower than that for “steward”)
Allows context influence
Not an exhaustive or particularly realistic model by any means, but ideas (e.g. the threshold for word identification) are present in almost every modern model of language comprehension
give some weaknesses of the logogen model
Activation can only rise, not fall
Can’t explain how nonwords can be processed
what are the 3 levels of the cohort model
stage one : accessing stage
2: selection stage
3: integration stage
who created the cohort model and when
Marslen- wilson (1980)
whats the difference between the logogen model and the cohort model
Refinement of logogen
uses time course of speechpercep- start identifyingf words as soon as the first phoneme is heard
can correct potential errors and use this to our advantage
when are words recognised according to the cohort model
at recognition point; word identification occurs only when theres no other words left matching input - at uniqueness point
name some strengths of the Cohort model
Predicts (correctly) that the first phoneme should be the most important
The uniqueness point is indeed very important in speech perception
Allows some context effects: Uniqueness point is earlier for some words when context is known
name some weaknesses of the Cohort model
Predicts (incorrectly) that you can’t identify a word if the first phoneme is missing
What about nonwords?
where is brocas area
Left frontal lobe- close to motor cortex ( toungue and mouth)
where is wernickes area
superior part of left temporal lobe- although diffferent parts of lobe now may play a part
close to auditory cortex
whats the difference between wernickes and Brocas aphasia
brocas= issues with producing speech- but understand it -> slow halting speech, poor articulation, speech omits prepositions, conjunctions and other grammatical filler words
Wernickes = speak perfectly and effortlessly- but phonemes often in wrong order -> reduced comprehensions of spoken and written language
what is meant by anomia in wernickes aphasia
Non meaningful speech and difficulty finding the right word
what are the levels of the connectionist model of speech production
semantic level ( activating meaning of word )
lexical level (activating word)
phonological level (activates phonemes)
what processing is used in the connectionist model
algorithmic model
top down and bottom up- to produce language
what is meant by inhibitroy connections between levels in the connectionist model
- within levels have inhibitory connections- different nodes inhibit each other - eg higher activation and meaning of dog = less of cat and cod
nesacary to increase contrast and have a winner
link the connectionist model of speech production to wernickes aphasia
in wa - spread out activation between inhibitory connection- so one word activates the meaning of another- also due to working in both ways
When connections between levels are weaker, phonological substitutions and nonwords are produced more frequently (e.g. “cod” or “coa”)
But how do patients get from one word to a semantically related one (e.g. “cat” instead of “dog”)?
Unclear, but when connection within a level are damaged more, that seems to lead to more such errors.
