An English summary of all articles for the course Language acquisition of the Master Multilingualism and Language Acquisition at the University of Utrecht must be learned for the exam.
Human simulations of vocabulary learning – Gillette, Gleitman, Gleitman & Lederer (1999)
Investigated Generalization: Noun learning is superior to verb learning in early child language
learning. Dominant explanation: verbs are abstract and cognitively complex, so children must
develop mentally first.
This study investigates an alternative hypothesis it is the information requirements of verb
learning, not the conceptual requirements, that determine the acquisition order. Verb learning
requires access to structural features of the language and can’t take place until a scaffolding of noun
knowledge enables the acquisition of clause-level syntax.
Another hypothesis that is investigated vocabulary acquisition takes place via an incremental
constraint-satisfaction procedure that bootstraps itself into successively more sophisticated linguistic
representations which, in turn, enable new kinds of vocabulary learning.
How children learn words: they pair the word to the world. Hardest question: how to make good on a
description of extralinguistic circumstances that will support the word-to-world pairing procedure.
Principle of Charity young learners will correctly interpret the world in view and the adult’s speech
intent just because child and adult are creatures of the same sort conceptually and motivationally.
However, adults and children differ in the way they speak.
Gentner (1978, 1981) proposed a conceptual explanation: nouns describe objects, while verbs label
the relations among those objects. Nouns are therefore simpler and more readily learnable. In this
view, the object representations that support noun learning have to be in place before verb learning
could begin.
Explanation that this paper gives not all words are learnable from a single kind of input evidence.
The information becomes available after solving prior parts of the language learning task. If so, verb
learning would be delayed until the requisite linguistic representations are constructed. Several
studies have shown a close correlation between increase in the verb vocabulary and first signs of
syntactic knowledge at the end of the second year.
- Goal 1: to understand the kinds of words that are most efficiently acquired via observation of
the ongoing scene.
- Goal 2: to examine the internal properties that a successful cross-situational learning
procedure must have.
- Procedure Subjects were shown silent video of mothers playing with their child. The task
was to identify the “mystery” noun/verb. The word was represented by a beep. For each
subject, the procedure was repeated for 16 different words, half nouns and half verbs.
- Subjects 84 undergraduates, 50 men and 34 women.
- Instructions they were told that the investigators were interested in how well they could
identify a word by observing the contexts in which it was uttered. Each target was identified
to them in advance as a noun or verb.
Findings Experiment 1:
- The percentage of correct choices was much higher for nouns than for verbs.
, - 1/3 of the verbs were never correctly identified by any subject, whereas every noun target
was identified by at least one subject.
- This is a suggestion that extralinguistic observation is sufficient for the learning of first nouns,
but that it may be too weak for efficient verb learning taken by itself, regardless of the age or
mentality of the learner.
- Performance for both lexical classes improves across trials, as theory of cross-situational
learning would predict.
- Tiger constancy = the greater tendency for tigers, in preference to any other object, to be in
view (salient) when “tiger” is uttered. This operates well for nouns in this experiment.
- The learner is hypothesized to remember the past and present context for the occurrence of
the novel word, and to parse out those properties of the scenarios that are applicable to
them all. Sometimes, subjects make the mistake of increasing the generality of their chosen
word, selecting one that covers just about everything.
Experiment 2: Imageability and contextual learning
- There may be constraints on the learner’s information base, with physicalistic concepts the
only ones that can be readily matched up with the scenes that accompany input utterances.
- Procedure Subjects rated the maternal words for imageability.
- Materials Three lists of words. List 1 was all words in alphabetical order, with ‘to’ and ‘a’
preceding the verbs and nouns. List 2 were nouns, and list 3 were verbs. Each word was
followed by a labeled scale, from “not at all imageable” to “extremely imageable”.
- Instructions The task was to rate each word for imageability on a 7-point scale.
Findings Experiment 2:
- Imageability is almost perfectly correlated with the noun/verb distinction, with all the nouns
except “thing” being rated as more imageable than any verb.
- There is a statistically reliable correlation between imageability and identifiability for verbs.
For nouns, there is a trend but it is not significant.
- The results suggest that it is not a difference between nouns and verbs that accounts for the
ease with which words are identified by inspecting the environments for their use, but an
account that only observables can be efficiently acquired by observation operating alone.
This paper hypothesizes an incremental learning machinery with at least these properties: (1) it
acquires a small stock of nouns by word-to-world pairing; and then (2) uses that stock of nouns as a
scaffold for constructing representations of the linguistic input that will support a more efficient
learning procedure.
Experiment 3: Cues for verb learning
- Question is whether adult subjects do better when provided with alternative/additional cues.
- Procedure Subjects were asked to identify mystery verbs, like in Experiment 1. They were
asked to offer their answer after receiving the full set of cues for that word.
- Condition 1: Cross-situational observation 2 judges selected the videotaped stimuli. They
were asked to inspect 2 min of silent video and to pick out a segment that in their view
“accounted for why the mother said [verb]”. Subjects would then watch this while hearing a
beep to guess a word.
- Condition 2: Noun co-occurrence Subjects saw a written list of the nouns that occurred
with the verb in the sentences used in the videotape. This condition models a situation in
which the learner, in possession of knowledge of many nouns but not of language-specific
, syntax, notices that some of these known nouns occur frequently with a particular novel
verb, providing a basis for inferring its meaning.
- Condition 3: Observation and nominal context (Conditions 1 and 2 combined) Subjects
saw the videoclip with the written list of noun for each sentence.
- Condition 4: Syntactic frames The mother’s sentences for the various mystery verbs
appeared on a written list in partial nonsense form. All the nouns and verbs were converted
to nonsense but the structural information was left intact. This condition models a situation
in which a learning device extracts the syntactic frame privileges of novel verbs.
- Condition 5: Selectional and syntactic information (Conditions 2 and 4 combined)
Combined the noun and structural information by presenting the nouns of Condition 4 in
their original locations in the mothers’ utterance.
- Condition 6: Full information (syntactic bootstrapping) Subjects saw the video along with
the sentences, with only the verbs as nonsense. This models the real input situation.
Results and discussion Experiment 3:
- Condition 1 Percentage correct in the video condition was even lower than in the
equivalent situation of Experiment 1. This is consistent with the idea that very young children
learn a new verb best if it is introduced when the event is impeding rather than when it is
already ongoing. The outcome was a low success rate for verb identification if the observer’s
only resource is to pair the passing scene with a single word.
- Condition 2 Identifiability scores were higher than for Condition 1, but not significant.
- Condition 3 Subjects achieved a respectable level of identification of new verbs.
- Condition 4 Subjects identified 51.7% of the verbs, which is an improvement. Subjects
were able to use syntax to make inferences about the verb meanings even though they were
artificially disbarred from observing the contexts of use and the co-occurring nouns. The 12
verbs that were never identified in Condition 1 are the 12 that are the most easily identified
in the present condition. The difference is between verbs whose content is mental versus
those that encode physical, observable action.
- Condition 5 The subjects identified the targets 75% of the time, a 24% improvement.
- Condition 6 The subjects approached perfect identification at 90.4%.
Discussion:
- The primary finding was that the environment isn’t so simply and generally informative.
- The second finding was that the extralinguistic contexts were more informative for some
kinds of words than for others. The greatest identification difficulties were for verbs that
refer to mental states/acts.
- The main predictor of identifiability was the words’ concreteness of imageability.
- The discontinuity hypothesis holds that children organize the world in terms of mental
representations fundamentally different from the way adults do. If this is true, there would
be no reason to except adult performance in Experiment 1 to reproduce child performance.
- The continuity approach assumes that children are conceptually equipped to entertain the
concepts encoded by most of the words adults say to them. Within this hypothesis, it is still
possible to account for why children are at first restricted to noun learning, because different
kinds of words require different kinds of information to identify.
- We want to be able to explain why in the earliest child vocabularies verbs, while rare, are not
absent. This is why the results of Experiment 2 are useful. It is not a noun versus verb
distinction, but it’s about imageability. This also explains why the first verbs learned are
concrete instead of abstract ones.
The benefits of buying summaries with Stuvia:
Guaranteed quality through customer reviews
Stuvia customers have reviewed more than 700,000 summaries. This how you know that you are buying the best documents.
Quick and easy check-out
You can quickly pay through credit card or Stuvia-credit for the summaries. There is no membership needed.
Focus on what matters
Your fellow students write the study notes themselves, which is why the documents are always reliable and up-to-date. This ensures you quickly get to the core!
Frequently asked questions
What do I get when I buy this document?
You get a PDF, available immediately after your purchase. The purchased document is accessible anytime, anywhere and indefinitely through your profile.
Satisfaction guarantee: how does it work?
Our satisfaction guarantee ensures that you always find a study document that suits you well. You fill out a form, and our customer service team takes care of the rest.
Who am I buying these notes from?
Stuvia is a marketplace, so you are not buying this document from us, but from seller jessievanhoof. Stuvia facilitates payment to the seller.
Will I be stuck with a subscription?
No, you only buy these notes for $5.89. You're not tied to anything after your purchase.
