Exam 2 Flashcards

win (94 cards)

1
Q

Problem with Meaning

A

Why do we initially pick one
interpretation (often the wrong one)
rather than another?

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Syntactic Ambiguity

A

the different meanings reflect different dependencies within the sentence “she said he tickled her yesterday”

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Frequency effects (Certain structures are more frequently used than
others) : Garden Path Sentences
Why do we do this?

A

Picking the wrong interpretation of the sentence and having to backtracking, we do this without taking into account meaning

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Ambiguity Resolution: structural accounts

A

Picking the most frequent structure when reading, always, without regard to the meaning of the actual sentence

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Ambiguity Resolution: Incremental Interaction

A

Basing meaning off the context being input - as each word is input, so is its interpretation with respect to context

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Referential context: Avoiding Garden Path…Using

A

the sentence structure in order to determine the meaning - then looking at the towel with the apple on it

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Language in visual contexts:
“Put the apple on the towel in the box.”

A

Looking at the apple, then the towel, then the apple again, then the box -

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Language in visual contexts:
“Put the apple that’s on the towel in the box.”

A

only looking at the apple on the towel, then putting it in the box - with the sentence being far less ambiguous than the one before, there is a faster response to that sentence than the other one

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Incremental Interpretation..par

A

As each word within a sentence is inputted, we process it, then building a partial interpretation of the sentence

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Prediction in Sentence Comprehension

A

The man and the girl: “will ride” - based on the pictures you can predict
“The boy will eat the cake”: based on the pictures”

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Sentence processing is ________: interpret each new word, and the sentence thus far, to
the. __________

A

Incremental; Fullest Extent

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Fullest Extent

A

Predicting what will come next but only based what has already been read

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Writing Systems

A

Logograms: Chinese, Japanese
Syllabic: Japanese Kana
Alphabetical: English, Hebrew, Arabic

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Convert text to meaning based on the….

A

Direct connections between words and meaning

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Converting alphabetic language and infrequent words in English

A

Word to sound, sound to meaning

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Dual-Route model of reading can be used to

A

.. to assess the reading profiles of people with developmental dyslexia

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

Direct/Surface Route is used for

A

Irregular and Frequent words

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

Phonological Route

A

mapping letters onto corresponding sounds for Infrequent and Non-Words

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

Phonological Dyslexia

A

prefer to use the direct route; not good with non-words, new words, tests on phonological awareness like speech minus “S” is peach with s; rhyming games

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

Surface Dyslexia

A

not good with irregular words, fine with phonological awareness tasks, rare dyslexia <20%.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

Fixations: eye stays still

A

200-250msec determined by, length, freq, Syntactic Cat, predicta, cognitive load.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

Saccades: when the eyes moves

A

25-30msec, length is language specific
English: 8 characters
Hebrew: 5.5 characters
Japanese: 3.5 characters
Chinese: 2 characters

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

Perceptual Span: information extracted

A

English: 15 characters to the right of the fixation

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
24
Q

Perceptual Span Equation

A

saccadic length x 2 = span

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
25
Rayner (1975)
Within ~7 characters of the fixation point: subjects notice the change More than ~12 characters from their fixation points: subjects don’t notice anything The more similar the original misspelling to the final word, the shorter the fixation durations on the final word
26
Why care about eye movement?
useful experimental too, diagnosing schizophrenia, Alzheimer type dementia, Parkinson's
27
What drives eye movements?
physical properties of the stimulus: - the number of letters in the word Linguistic properties of the stimulus: - -- Frequency, syntactic category, predictability
28
Semantic Knowledge: World Knowledge
Paris is in France
29
Semantic Knowledge: Object Knowledge
Cups are used for drinking - the purpose of the object
30
Semantic Knowledge: Language knowledge
Symbolism referring to the cup
31
Hints about how semantic memory is organized?
Categorical (taxonomic) priming: cat-dog; stool - table) Thematic (event-based) priming: dog - bone; professor - classroom Perceptual priming: frisbee - pizza Motoric priming: key- screwdriver
32
Semantic Memory relating to parts of Memory system: Episodic memory
Things that have happened
32
Semantic Memory relating to parts of Memory system: Perceptual Information
your perception within an item, how it smells, feels, tastes, sounds it makes
32
Semantic Memory relating to parts of Memory system: Procedural Knowledge
performance of a task or activity, knowledge of how to do that
32
How is semantic memory acquired?
through experience with that item
32
What is the dominant theory about the organization of semantic memory?
Sensorimotor-based (perceptual Symbols); Overlap between the items like baseball and orange - orange triggers more taste and smell because it has that
32
Overlap between thinking and doing
Brain regions support physical action, also represent the meaning of actions. So damage here will have an effect on the individuals understanding of action words
32
People with Parkinson's Disease have degeneration of cells...
that are involved in planning actions; more difficulty understanding words involving actions compared to words that don't
32
Semantic Relationships: Semantic Dementia (ATL) (3)
- Categorical relationships - Gradual atrophy of anterior temporal lobe (L > R) - Knowledge about object properties is lost, however specific knowledge is lost before general
32
Semantic Relationships: Thematic or event based relationships (AG)
problems here cause problems with: ambulance - hospital dog - bone professor - classroom
32
Semantic Relationships: Mapping word to meaning (PMTG)
- Damage causes problems with mapping sounds onto meaning
32
What computational models do, and what are its parts?
Establish the brain might compute meaning; parts include models of theory about how or what the brain computes, others are their just to make it work
32
What is meaning?
Knowledge of the contexts in which the word can occur words used in similar contexts have similar meanings
33
Latent Semantic Analysis
- distance correlates with e.g. priming and other measures of “semantic overlap” - the graph is a semantic space
34
Elman (Co-Occurence) Simple Neural network: Task?
predict what word will come next based on the input
35
SRN: Adjusting Internal Weights (what does the model learn)
Verbs come after nouns words like eat are followed by consumable items "Categories Emerge"
36
The Nature of Prediction: frequency of occurrence
choosing the word that is most frequent - also correlates more to the context
37
The Nature of Prediction: Accessing words...frequency
based on how a word is being spelt, understanding the context will assist in predicting what word is it before it is completed
38
Emergent Representations: Syntactic Knowledge
what we call verbs come after what we call nouns SRN learned about the contexts in which words could occur
39
Emergent Representations: Semantic Knowledge
Understanding the context and choosing words for example that are consumable after the word eat
40
Neural Networks: How they represent
Patterns of Activation Spreading Activation
41
Neural networks: How they learn
Adjusting the connection strength to get closer or further from the word you want versus the word you don't want We adjust the connection strengths so that “what we get” is closer to “what we want”
42
Giving SRN memory: structures
Input Layer, Hidden Layer, Copy of hidden layer, Recurrent Layer
43
Giving SRN memory: steps
1. Input first word, activation spreads to hidden layer, hidden unit spreads activation to output unit but also sends copy to recurrent unit (connection change the activation to reflect what may come next) 2. Next word, hidden unit gets the word and the copy from the recurrent copy, sends to output unit and copy to recurrent unit,
44
Modification of connection strengths
Error Driven Learning Backpropagation: when we want to get closer to what we want to output to the hidden layer - becoming more correct (not biologically plausible)
45
“Deep” neural networks
Input, hidden, output layers
46
Learning in deep neural networks
We keep the new weight to one side We change all the weights in one go at the end, rather than changing them as we go along
47
What is meaning (neural activity)
Neural activity with a meaning will be different in everyone, however as long as the activity occurs in the same situations - it does not matter
48
How does the mind represent discourse? : Mental Models
When reading you create a mental model of what is occuring, so when it says John put on a sweater, he is wearing it in your mental model which is why you will respond quicker when prompted on if a sweater was mentioned
49
How does the mind represent discourse? : Mental Models (Birds)
Pictures that were congruent with what the sentence was saying, were faster as responding to the pictures
50
3 Types of Discourse
Surface Form, Propositional Form, Situation Model
51
Surface Form
same meaning, same words
52
Propositional
same meaning, different words
53
Situation (mental) Model
different meaning, but same idea.
54
Joining up words requires
Knowledge of Grammar
55
Joining up sentences requires
Knowledge of grammar...and other stuff Knowledge of how the world works knowing you wouldnt drink something hot on a hot day but instead cold Knowledge allowing for the creation of mental models
56
Dialog
Taking account of the (specific) speaker/listener
57
How do both speakers and listeners adjust their assumptions in specific communicative circumstance?
What's Common ground The reliability of the speaker Audience Design
58
What is needed in order to establish CG, take mental state into account, and practice AD
Theory of Mind
59
Common Ground: Adults
Adults & kids adjust their interpretations based on what is in common ground Adults are sensitive to what is in common ground: the information shared between those speaking to eachother (interlocutors)
60
Common Ground in Kids: 5-6 year olds
they are also sensitive to what the person they are speaking to knows however this is not all or nothing
61
Evaluating the reliability of speakers: Adults
Interpreting Ambiguous Sentences When told that the speaker has inpariment so they adjust the reliabilty of what the person is saying which affects their reponse time when told about a specific item or location
62
Reliability of Speakers (kids)
- based on the confidence of both experimentors, it affected the learning for kids on a specific item - if experimentor appeared knowledgable on the specific toy, the kids were also just as confident and vice versa - learning is influenced
63
Audience Design (speaker)
Adjusting your language to communication more effectively with your audience
64
Audience Design: Referential Uniqueness in kids
- Developmental in children - Percentages of unambiguous first descriptions was higher for those who were older
65
Theory of Mind
keeping track of what’s going on in other people’s minds, independently of their own
66
False Belief Task
Kids under 4 say "In the box" Kids over 4 say "in the basket" Shows that when they are able to understand that when sally leaves, she will check in the place she left her item, they engage of theory of mind
67
Theory of Mind: Identifying common ground
Understanding and being aware that other people can have knowledge that may differ from your knowledge
68
Theory of Mind: Evalutating reliability of speaker knowledge
have some awareness that their mental states can differ
69
Theory of Mind: engaging in audience design
In order to successfully adjust communication, you must understand what the person you are speaking to knows and that their knowledge differs from your own
70
Pragmatics
going beyond the literal meaning
71
Indirect Speech
Talking about one thing in order to express another
71
Pragmatic/Conversational Implicatures are:
The parts of the speaker’s meaning that go beyond what the speaker explicitly said Inferred by the listener based on listener’s expectation that speaker is cooperative conversational partner
72
Indirect Speech : Ability
Can you pass me that bag of candy?
73
Indirect Speech : Knowledge
Do you know what time it is?
74
Indirect Speech : Fact
I'd like to move the table away from the whiteboard
75
Inferences are made when
the literal meaning does not give the entire intended meaning
76
Pragmatic Inferences occur when the listener
infers things that the speaker has not overtly stated
77
implicatures are
unstated meanings
78
Grice’s Conversational Maxims : Quality (truth)
Listener assumes that what the speaker is saying is true and not saying anything they do not have evidence for When it is violated, the listener has to assumed why the speaker is doing so: Sarcasm
79
Grice’s Conversational Maxims : Relevance
Listener assumes that what was said or being said is relevant in the context of their comments - so when they say something that may not seem too relevant, they must infer why they are doing so - Butter example
80
Grice’s Conversational Maxims : Quantity
listener assumes that the speaker is providing as much information as possible, not more, in order to get their point across - when asked how all class are going - you say, one of them is good - you then assume that with that much information, it is enough to figure out that they do not enjoy their other classes
81
Grice’s Conversational Maxims : Manner
Listeners assume that speakers avoid obscurity, ambiguity, disorderly communication - so when they do it, you must infer why. Spelling out walk so the dog doesn't go wild -
82
Pragmatics & Gricean Maxims help explain:
Where divergences from the literal meaning come from Why, when meaning is ambiguous, we choose a particular interpretation