Reading...
Play button
Play button
Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
67 Cards in this Set
- Front
- Back
- 3rd side (hint)
|
Schemata
|
A framework that helps in interpreting
information by organizing the relations among concepts, situations, events, and/or actions in memory. |
|
|
|
Hierarchical Semantic Networks
|
Semantic memory viewed as represented by a
network (or net) of interconnected concepts. Some concepts are based on category membership and others define the properties of a concept. So there are category nodes and property nodes. |
|
|
|
isa arcs
|
Some connections in the net are called isa arcs
and they allow for the inheritance of properties. isa arcs are found between category nodes (in this example, they are noted by the thick white lines). |
|
|
|
Spreading Activation
|
Analogous to neuronal activation It is an automatic rather than a controlled process.
This early model of semantic net has a hierarchical organization. Thus, the canary node is closer to the bird node than it is to the animal node. If spreading activation is how semantic nets really work, we would expect differences in reaction time (RT) to different lexical tasks. |
|
|
|
semantic priming
|
Semantic priming using a lexical decision
task: I show you a series of strings of letters and you tell me whether the letters form a word or not. Some strings spell words (“batch”) and others don’t (“tahbc”). If I precede a test stimulus item (“robin”) with a related word (“bird”), your response time (RT) is faster than it is if preceded by an unrelated word. n This suggests that by seeing ‘bird’ activation is spread to ‘robin’ so that the concept of robin is activated before you see the word ‘robin.’ |
|
|
|
Hierarchical nets: The problems
|
Typicality effect: RTs faster when verifying more typical instances as members of a category
Relatedness effect: RTs slower when member is similar to category but answer is false |
|
|
|
The Typicality Effect
|
T/F lexical task: “A robin is a bird” is associated with faster RT than is “An ostrich is a bird”
The hierarchical model predicts that people use only hierarchical knowledge and that typicality should not matter. |
|
|
|
The Relatedness Effect
|
T/F lexical task: “A bat is a bird” takes longer to
respond “False” than “A bat is a plant” RTs are slowed due to the many similarities between bats and birds Again, the hierarchical model predicts that people use only hierarchical knowledge and that similarities should not matter. |
|
|
|
Hierarchical nets:
Fixing the Problems! |
So it seems as if more ‘typical’ and highly ‘related’ instances of a category are more strongly connected with category membership.
Associations are more appropriately conceived in terms of strengths of associations as opposed to distance in a hierarchical model |
|
|
|
Frequency effect
|
high frequency words (e.g., cat) are
recognized faster than low-frequency words (e.g., cab). |
|
|
|
Zipf’s Law:
|
High frequency words tend to be short words
The highest frequency word occurs approximately twice as often as does the second most frequent word, which occurs approximately twice as often as the third most frequent word and so on. Hence, it has been estimated that about 135 account for about half of the words spoken in various language copra. |
|
|
|
Bock & Levelt’s (1994)
Spreading Activation Model |
this model incorporates conceptual aspects of word knowledge during the ‘conceptual level’ of processing.
A second level of processing is at the ‘lemma level’ which refers to the syntactic aspects of word knowledge. A third level of processing is at the ‘lexeme’ level which is designed to capture a word’s phonological properties. Improves upon previous models by incorporating knowledge of the syntactic and phonological aspects of word knowledge. It also can help explain TOT phenomenon |
conceptual->lemma: syntax->lexeme: phonology
|
|
|
Forster’s
Autonomous Search Model |
Library analogy, a word (book) is found in only one place in the mental lexicon (library) but can
be located by using various resources: Access files: Orthographic - words are accessed via visual features (for visual word identification) Phonological - sound (for spoken word identification) Syntactic/semantic - meaning and grammatical class |
|
|
|
Logogen model (Morton)
|
Each word has its own Logogen or entry in the lexicon.
|
|
|
|
Meter analogy
|
once word activated a meter registers some degree of activation, when word is no longer activated the meter does not decline immediately (but slowly), so if the word is present again, the meter continues to rise until a logogen reaches a threshold of recognition. Individual threshold love
|
|
|
|
Logogen Model
|
This is an activation model, not a search
model. Logogens compete for activation (i.e., high frequency words have lower activation thresholds) Information is fed into the system by orthographic (visual), auditory (phonological), or semantic (the cognitive system) inputs. trengths: n Can account for the frequency effect. n Weaknesses: n What constitutes the perceptual unit that maps acoustic or visual information onto the logogen? Since the model operates on words as units, it is not clear how sublexical units (syllables, morphemes) and nonwords are processed. Does not specify how auditory and visual information are integrated. |
|
|
|
Connectionist Model
Seidenberg & McClelland hidden units-> orthographic units-> hidden units-> phonological units |
Accounts for lots effects we see in word recognition studies.
e.g. frequency effect, neighborhood effects, regularity effects, semantic priming, TOT. It does so without the use of rules and with no mechanism that ‘looks up’ words in a mental lexicon. In fact, there is no lexicon |
|
|
|
Grammar/Syntax
|
rules for building
sentences |
|
|
|
Morphology
|
rule for building words
|
|
|
|
Morphology
|
The study of the structure of words (often used to refer to the rules for building words)
|
|
|
|
Derivational morphemes:
|
Bound morphemes that change the meaning of a word. These usually create new stems!
learn-> learnable; |
|
|
|
Inflectional morphemes
|
Bound morphemes that change an ‘aspect’ of the word (e.g., tense, number, possession to name only a few).
walk-> walked |
|
|
|
stress test
|
compounds generally have stress on the first component, phrases on the second
a dark ROOM vs. a DARK room |
|
|
|
Prescriptive rules
|
what your grammar teachers tell you- It is how one ‘ought’ to talk.
|
|
|
|
Descriptive rules
|
describe how people actually talk. These rules never need to be mentioned in style manuals.
|
|
|
|
The word watchers
|
train their binoculars on the
especially capricious, eccentric, and poorly documented words and idioms that get cited from time to time. |
|
|
|
The Jeremiahs
|
express bitter laments and righteous
prophecies of doom. |
|
|
|
The entertainers
|
show off collections of palindromes,
puns, anagrams, rebuses, malapropisms, Goldwynisms, eponyms, and bloopers. Seems all in good run. |
|
|
|
The Modular View (Chomsky):
Rules all the way down |
Rule:N à Nstem + Ninflection
wug s N Nstem Ninflection |
|
|
|
The Connectionist View
(Seidenberg & McClelland) Memory and Associations all the way up! |
hidden-ortho-hidden-phonological
We don’t need rules to explain the patterns. n If a child hears the regular forms pat à patted and skate à skated; then she can generalize from walk à walked on the basis of associative learning alone! |
|
|
|
The Theory of Words and Rules
A simple principle |
If a word can provide its
own past tense from memory (PDP), the rule is blocked; elsewhere, the rule (by default) applies (modularity). |
|
|
|
irregulars
|
Irregulars are the most common in
English and most all other languages. n Explanation is simple: Irregulars must be memorized repeatedly, generation after generation, to survive in a language. n If an irregular slips in popularity, a generation of children will not hear it enough to remember its past tense. |
|
|
|
Words are syntactic
atoms. |
Regulars are formed by
rules Irregulars are learned through memory |
|
|
|
Lexical ambiguity
|
A sentence contains a word with more than one meaning.
|
|
|
|
Syntactic ambiguity
|
Where a sentence may have more than one interpretation because more than one phrase tree structure can be associated with it.
|
|
|
|
Polysemy
|
one word has more than one meaning
|
|
|
|
Meaning dominance
|
refers to the relative frequency of each meaning of an ambiguous word
|
|
|
|
Strength of context
|
how strongly the context suggests a particular word.
|
|
|
|
Syntactic ambiguity
|
A sentence has more than one interpretation because it can be associated with more than one tree.
|
|
|
|
Garden path sentence
|
The early part of the sentence makes the reader or listener go “down the garden path” which initially leads to a wrong interpretation
|
|
|
|
Modular view:
|
Parse, then interpret it semantically (syntax then semantics)
Backtrack & fix if we can’t finish the parse or if it makes little semantic sense In this view, syntax is the default ‘sentence decoder’. Syntactic parsing recruited during sentence processing to inform comprehension and semantics only recruited if something goes wrong |
|
|
|
Interactive (PDP) view
|
Simultaneous activation at the lexical, syntactic, and semantic levels.
Begins as bottom-up (stimulus driven) process but immediate activation of top-down processes Syntax is NOT the default sentence decoder. Sentence processing determined by simultaneous activation of lexical, syntactic and semantic components of a sentence |
lexical+syntax+semantics
|
|
|
Garden Path Theory Lyn Frazier, 1970s and 1980s
|
During comprehension, we commit to a particular syntactic parsing that is unaffected by semantics unless the comprehension of sentence falters. There are two basic parsing strategies
|
|
|
|
Minimal Attachment
|
Prefer the interpretation that is accompanied by the simplest structure.
simplest = fewest branchings (all you need to do it count the nodes or branchings points) |
|
|
|
Late Closure
|
Prefer to incorporate incoming material into the phrase or clause currently being processed. That is, we associate incoming material with the most recent material possible.
|
|
|
|
Integration of Visual and Linguistic Information in Spoken Language Comprehension Tanenhaus, Spivey-Knowlton, Eberhard, & Sedivy (1995)
|
Purpose: Provides a strong test of whether nonlinguistic information can influence the earliest moments of syntactic processing
|
|
|
|
Eye tracking
|
observe rapid mental processes that accompany spoken language
|
|
|
|
THE BIG PICTURE
|
Overall, the data from studies in both written and spoken contexts confirms that immediacy of processing is the general rule in our language system.
Whenever possible, we try to integrate each word into the developing mental representation of what is described in the sentence. This process very much depends on the bottom-up processing of each word. However, we use context to help guide our lexical, syntactic, and semantic processing. |
|
|
|
Blend
|
two units are combined grizzly + ghastly - grastly
|
|
|
|
Anticipation
|
a speech unit is activated too early Take my bike - bake my bike
|
|
|
|
Perseveration
|
a speech unit is activated too late
Pulled a tantrum - pulled a pantrum |
|
|
|
Substitution
|
a unit is changed into a different unit
The place opens - the places closes |
|
|
|
Misdeviation
|
the wrong unit is attached to a word Intervening node - intervenient node
|
|
|
|
Shift:
|
the speech unit moves to a different location
She decides too hit it - She decide to hits it |
|
|
|
Exchange:
|
two units swap positions
Katz & Fodor - Fats and Kodor |
|
|
|
Addition:
|
a unit is added
Carefully enough - clarefully enough |
|
|
|
Deletion:
|
a unit is deleted
Plastic - plattic |
|
|
|
Levels of Speech Production
|
message level- idea or proposition to be communicated
sytactic level- selection and organization of lexical items morphemic level- complex words are built out of stems phonemic level- sound structure of each word is built all lead to the articulation of speech |
|
|
|
speech production
|
build syntactic structure of sentence
build words from morphemes build sound structure from phonemes |
|
|
|
The Modular View
|
message level- not discussed
syntactic level- module: does only syntactic stuff morphemic level- module: does only morphemic stuff phonemic level- module: does only sound stuff |
|
|
|
PDP view
|
message level- not discussed
syntactic level- syntactic features are activated morphemic level- morphemic features are activated phonemic level- phonological features are activated |
|
|
|
syntactic category rule:
|
Slips that involve whole words are almost always from the same syntactic category
|
|
|
|
Stranding errors
|
the inflection (morpheme) gets stranded by its intended form.
|
|
|
|
Consonant-vowel rule:
|
consonants are exchanged with consonants and vowels are replaced with vowels
|
|
|
|
Covert self repair
|
monitoring at the planning stages
GSR (galvanic skin response) is an index of physiological activity (including emotional arousal) |
|
|
|
Overt self repair
|
monitoring what you’ve actually said
18% of speech errors are corrected within the troublesome word (Hand me the yel…uh…red one). 51% are corrected immediately after the word (hand me the yellow, uh, red one) 31% are delayed for one or more words (hand me the yellow one, uh, red one). |
|
|
|
Types of repairs (proposed by Levelt):
|
Phonological: They have a nithe ^ nice boat.
Morphological: Sot he man have ^ has got his hats back Lexical: If you must read ^ uh write the English word... Inappropriate with replacement: It turns out to be a film ^ a movie scene… Inappropriate with insertion: You see a policeman ^ an English policeman… |
|
