Machine Learning and Cognitive Science of Language Acquisition

A Bayesian approach to the Poverty of the stimulus

author: Amy Perfors, Computational Cognitive Science Group, Department of Brain and Cognitive Sciences, MIT

Description

Shown that given reasonable domain-general assumptions, an unbiased rational learner could realize that languages have a hierarchical structure based on typical child-directed input. Can use this paradigm to explore the role of recursive elements in a grammar: the “winning” grammar contains additional non-recursive counterparts for complex NPs; perhaps language, while fundamentally recursive, contains duplicate non-recursive elements that more precisely match the input?

You might be experiencing some problems with Your Video player.

*Slides*
0:00	A Bayesian Approach to the Poverty of the Stimulus
0:00	Innate - Learned
0:21	Explicit Structure - No explicit Structure
0:50	Language has hierarchical phrase structure
1:40	Why believe that language has hierarchical phrase structure?
2:44	Why believe that structure dependence is innate?
3:36	Why believe it’s notinnate? - part 1
4:15	Why believe it’s notinnate? - part 2
4:52	Why believe it’s notinnate? - part 3
5:05	Our argument - part 1
5:08	Our argument - part 2
6:11	Plan
6:57	The model: Data
8:11	Data
8:35	Data: variation - part 1
9:00	Data: variation - part 2
9:18	Data: amount available
10:03	Data: amount comprehended
10:56	The model
12:14	Grammar types
13:01	Specific hierarchical grammars: Hand-designd
14:22	Specific linear grammars: Hand-designed - part 1
14:51	Specific linear grammars: Hand-designed - part 2
15:22	Specific linear grammars: Hand-designed - part 3
16:07	Specific linear grammars: Hand-designed - part 4
16:22	Automated search
17:28	The model
17:32	Grammars - part 1
18:04	Grammars - part 2
18:17	Tradeoff: Complexity vs. Fit
18:48	Measuring complexity: prior
19:37	Measuring fit: likelihood
20:11	Plan
20:14	Results: data split by frequency levels (estimate of comprehension)
21:42	Results: data split by age (estimate of availability) - part 1
21:43	Results: data split by age (estimate of availability) - part 2
22:44	Generalization: How well does each grammar predict sentences it hasn’t seen? - part 1
22:56	Generalization: How well does each grammar predict sentences it hasn’t seen? - part 2
23:44	Take-home messages
25:41	Implications for innateness?

Lecture rating

People found this lecture:
Worth seeing
because it is:
Valuable and informative
Well presented
Easily understandable
Acceptably recorded
You need to login to cast your vote.

Report a problem or upload files

If you have found a problem with this lecture or would like to send us extra material, articles, exercises, etc., please use our ticket system to describe your request and upload the data.
Enter your e-mail into the 'Cc' field, and we will keep you updated with your request's status.