Class Notes (808,703)
United States (313,202)
PSYCH 150 (8)
pearl (8)
Lecture

# Review Qs 4 -Word Meaning.docx

5 Pages
68 Views

School
University of California - Irvine
Department
Cognitive Sciences
Course
PSYCH 150
Professor
pearl
Semester
Fall

Description
Psych156A/ Ling150 Spring 2012 Review Questions: Word Meaning (1) Terms/concepts to know: mapping problem, fast mapping, cross-situational learning, Bayesian inference, posterior probability, likelihood, prior probability, positive examples, whole object constraint, taxonomic constraint, subordinate, basic, superordinate, lexical contrast Mapping problem: An infinite number of hypotheses about word meaning are possible given the input the child has. That is, the input underspecifies the word’s meaning. Fast mapping: Children begin by making an initial fast mapping between a new word they hear and its likely meaning. They guess, and then modify the guess as more input comes in. Cross-situational learning: New approach: infants accrue statistical evidence across multiple trials that are individually ambiguous but can be disambiguated when the information from the trials is aggregated. Bayesian inference: In Bayesian inference, the belief in a particular hypothesis (H) (or the probability of that hypothesis), given the data observed (D) can be calculated the following way: P(H | D) = P(D | H) * P(H) P(D) Posterior probability: P(H | D) Likelihood: P(D | H) Prior probability: P(H) Positive examples: Whole object constraint: Taxonomic constraint: Subordinate: Least-general ex. dalmation Basic: ex. dog Superordinate: Most general ex. animal Lexical contrast a. Mapping problem – even if something is explicitly labeled in the input, how does the child know what specifically that word refers to, an infinite number of hypotheses about word meaning are possible given the input the child has, that is the input underspecifies the words meaning b. Fast mapping – children begin by making an initial fast mapping between a new word they hear and its likely meaning, they guess, and then modify the guess as more input comes in c. Cross-situational learning – infants accrue statistical evidence across multiple trials that are individually ambiguous but can be disambiguated when the information from the trials are aggregated d. Baysesian inference – the belief in a particular hypothesis (H) or the probability of that hypothesis given that the data observed (D) can be calculated e. Posterior probability – how strongly you believe in a hypothesis that ou’ve seen data, combo of likelihood of prior belief that your hypothesis is true f. Likelihood – chance of seeing data, give that the hypothesis is true g. Prior probability – previous knowledge of probability of hypothesis H h. Positive examples – children usually only see examples of what something is, rather than being explicitly told what it isn’t example “I love my dax” i. Whole object constraint – first guess is that a label refers to a whole object, rather than part of the object EX: dog as opposed to dog parts j. Taxonomic constraint – first guess children make is about an unknown label is that it applies to the taxanomic class EX: dog instead of all running things k. Subordinate – least general item EX: dalmatian l. Basic – EX: dog m. Supererdinate – most general item EX: animal n. Lexical contrast – all words seems to differ, not impossible to find exact synonyms, most of the time subtle differences if (2) While fast mapping may sound like a good strategy in theory, why is it unlikely to be easy to carry out in real world situations? (Hint: Think about how many potential referents there are in a real world situation.) “…not all opportunities for word learning are as uncluttered as the experimental settings in which fast-mapping has been demonstrated. In everyday contexts, there are typically many words, many potential referents, limited cues as to which words go with which referents, and rapid attentional shifts among the many entities in the scene.” - Smith & Yu (2008) Fast mapping is unlikely to be carried out in real world situations because not all opportunities for world learning are as uncluttered as the experimental settings in which fast mapping has been demonstrated. In everyday contexts, there are typically many words, many potential referents, limited cues as to which words go with which referents and rapid attentional shifts among the many entities in the scene. Scenes have many more things to get confused with, record could refer to a million diff things, don’t remember where mom was looking when said this is a record, don’t know names for many diff things (3) What evidence is there that infants can do cross-situational learning in experimental scenarios? Infants accrue statistical evidence across multiple trials that are individually ambiguous but can be disambiguated when the information from the trials is aggregated. The evidence that supports infants can do cross-situational learning is through the work by Yu & Smith who found that infants preferentially look at target over distracter, and 14 mo olds looked longer than 12-mo olds. This means that they were able to tabulate distributional information across situations, therefore they can do cross-situational learning (4) As with fast mapping, why might cross-situational learning be more difficult in realistic scenarios, as compared with experimental scenarios where it has been shown to be present in infants? Is there any reason to believe that realistic scenarios (which have more potential refere
More Less

Related notes for PSYCH 150

OR

Don't have an account?

Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Join to view

OR

By registering, I agree to the Terms and Privacy Policies
Just a few more details

So we can recommend you notes for your school.