VII. Verb Learning

Question 1

Why might it be easier for children to learn nouns than verbs?

Answer 1

1. many nouns refer to physical object concepts, whereas verbs refer to event or action concepts
   —indeterminacy of reference: the referents of nouns are typically more stable, whereas the referents for verbs are typically dynamic events
2. nouns can occur independently, whereas verbs express relations and therefore must involve nouns

Question 2

conceptual complexity hypothesis

Answer 2

1. word learning is dependent on conceptual development
2. action/event concepts are more complex than object concepts: action/event concepts involve movement
3. object concepts must develop before action/event concepts: action/event concepts involve object concepts

Question 3

linguistic experience hypothesis

Answer 3

1. verbs require local dependencies, whereas nouns can stand alone: verbs require a subject and/or object
2. infants must build syntactic structure before they can understand how verbs relate to other words: syntax is important in understanding how verbs function (“The cat eats food” is not the same as “The food eats cat”, but without knowledge of syntax (VP, NP, DO) a child might not be able to understand the relationship between cat and food.

Question 4

How might computational models help researchers in language acquisition?

Answer 4

using computational models to simulate human learning
- advantages: helps researchers investigate hypotheses without using actual infants
—ethicality: it is unethical to deprive children of language or limit/change their conceptual development for the purposes of research
—control: researchers can control for the amount of linguistic input that computational models receive

Question 5

human simulation paradigm

Answer 5

using human subjects to simulate learning
- similar to using computational models in that researchers use learning one kind of subject (adults or computers) to make inferences about the possibilities of learning for other kinds of subject (infants)
- used in Gillette et al. (1999) to investigate the linguistic experience hypothesis, in which subjects were undergraduates at UPenn

Question 6

Gillette et al. (1999): goal, methods, findings, conclusions

Answer 6

goal: to investigate the linguistic experience hypothesis; what input (linguistic or visual) is necessary for adults to learn the meaning of a word? can a visual scene give away the meaning of a word? when is a visual scene informative or noninformative?
methods: human simulation paradigm in which subjects were UPenn undergrads
exp1: undergrads shown silent videos of mothers interacting with children (limiting linguistic evidence) and when the target word (the word that participants are supposed to guess the meaning of) was used in the video, a beep was played
exp2: undergrads rated frequent maternal words for “imageability,” or the ability to form a mental picture for those words
exp3: to test verb-learning, undergrads were placed into 6 conditions with different kinds of evidence: silent videos (no lign evidence), co-occurring nouns (no video), co-occurring nouns and video, jabberwocky syntax (syntactic information, no visual information), full sentence excluding verb (no video), scene + full sentence excluding verb
findings:
exp1: undergrads were better overall at guessing the meaning of nouns than verbs, but still struggled with nouns that were not present in the scene, or perpetually present (e.g. mommy, daddy, shoes)
exp2: nouns predictably had higher overall imageability, however verbs such as hammer, throw, and fell had high imageability
exp3: undergrads’ overall accuracy at guessing the meaning of verbs increased with increased linguistic info available; highest accuracy when undergrads are given visual and true syntactic information
conclusions:
linguistic evidence is a highly influential factor in being able to correctly identify the meanings of words
—support for the linguistic experience hypothesis

Question 7

syntactic bootstrapping theory

Answer 7

the theory of word learning in infants may use the syntactic environment of a word to learn something about that word’s meaning
- a refinement of the linguistic experience hypothesis developed by Lila Gleitman
- words w/ similar meanings occur in similar kinds of sentences: there could be an underlying link between the meanings of sentences and their syntactic structures

Question 8

Naigles (1990); goal, methods, findings, conclusions

Answer 8

goal: can 2 yr old children use syntactic bootstrapping to learn the meaning of a novel verb?
methods: using a preferential-looking paradigm, researchers showed children a visual stimuli in which two actions simultaneously occurred. kids were also given auditory stimuli which either labelled the visual stimuli with a verb in a transitive or intransitive frame (“The duck is gorping the bunny” vs. “The duck and the bunny are gorping.”). kids then saw two separate videos, in which one action was now occurring on one screen, the other action on another screen. researchers then prompted the children with “where’s gorping?” and measured children’s looking times towards the two screens
findings: children overall spent longer looking at the matching screen during the test phase (transitive frame-causative verb) (intransitive frame-non causative verb) than the non-matching screen, and shifted their attention towards the matching screen during the test phase
conclusions: 2-year old children can use syntactic bootstrapping methods to learn the meanings of novel verbs
—support for syntactic bootstrapping theory of word learning