Cross-situational learning in a Zipfian environment

Research output: Contribution to journalArticleScientificpeer-review

Abstract

Both adults and children have shown impressive cross-situational word learning in which they leverage the statistics of word usage across many different scenes in order to isolate specific word meanings (e.g., Yu & Smith, 2007). However, relatively little is known about how this learning scales to real language. Some theoretical analyses suggest that when words follow a Zipfian distribution, as they do in natural language, it should be more difficult to learn a lexicon because of the many low-frequency words that are only observed a few times (Blythe, Smith, & Smith, 2010; Vogt, 2012). Although this effect can be mitigated somewhat by assuming mutual exclusivity (Reisenauer, Smith, & Blythe, 2013), no mathematical analyses suggest that learning in a Zipfian environment should be easier. In this work, we show the opposite of the predicted effect using cross-situational learning experiments with adults: when the distribution of words and meanings is Zipfian, learning is not impaired and is usually improved. Over a series of experiments, we provide evidence that this is because Zipfian distributions help people to disambiguate the meanings of the other words in the situation.
Original languageEnglish
Pages (from-to)11-22
Number of pages12
JournalCognition
Volume189
DOIs
Publication statusPublished - 1 Aug 2019

Fingerprint

learning
Language
experiment
language
statistics
Situational
evidence
Experiment
Mutual Exclusivity
Natural Language
Lexicon
Word Learning
Word Meaning
Word Usage
Statistics
Word Frequency

Keywords

  • Cross-situational statistical learning
  • Language learning
  • Zipfian distributions

Cite this

@article{b688d87f92e94967a581492cc02fea54,
title = "Cross-situational learning in a Zipfian environment",
abstract = "Both adults and children have shown impressive cross-situational word learning in which they leverage the statistics of word usage across many different scenes in order to isolate specific word meanings (e.g., Yu & Smith, 2007). However, relatively little is known about how this learning scales to real language. Some theoretical analyses suggest that when words follow a Zipfian distribution, as they do in natural language, it should be more difficult to learn a lexicon because of the many low-frequency words that are only observed a few times (Blythe, Smith, & Smith, 2010; Vogt, 2012). Although this effect can be mitigated somewhat by assuming mutual exclusivity (Reisenauer, Smith, & Blythe, 2013), no mathematical analyses suggest that learning in a Zipfian environment should be easier. In this work, we show the opposite of the predicted effect using cross-situational learning experiments with adults: when the distribution of words and meanings is Zipfian, learning is not impaired and is usually improved. Over a series of experiments, we provide evidence that this is because Zipfian distributions help people to disambiguate the meanings of the other words in the situation.",
keywords = "Cross-situational statistical learning, Language learning, Zipfian distributions",
author = "Hendrickson, {Andrew T.} and Amy Perfors",
year = "2019",
month = "8",
day = "1",
doi = "10.1016/j.cognition.2019.03.005",
language = "English",
volume = "189",
pages = "11--22",
journal = "Cognition",
issn = "0010-0277",
publisher = "Elsevier Science BV",

}

Cross-situational learning in a Zipfian environment. / Hendrickson, Andrew T.; Perfors, Amy.

In: Cognition, Vol. 189, 01.08.2019, p. 11-22.

Research output: Contribution to journalArticleScientificpeer-review

TY - JOUR

T1 - Cross-situational learning in a Zipfian environment

AU - Hendrickson, Andrew T.

AU - Perfors, Amy

PY - 2019/8/1

Y1 - 2019/8/1

N2 - Both adults and children have shown impressive cross-situational word learning in which they leverage the statistics of word usage across many different scenes in order to isolate specific word meanings (e.g., Yu & Smith, 2007). However, relatively little is known about how this learning scales to real language. Some theoretical analyses suggest that when words follow a Zipfian distribution, as they do in natural language, it should be more difficult to learn a lexicon because of the many low-frequency words that are only observed a few times (Blythe, Smith, & Smith, 2010; Vogt, 2012). Although this effect can be mitigated somewhat by assuming mutual exclusivity (Reisenauer, Smith, & Blythe, 2013), no mathematical analyses suggest that learning in a Zipfian environment should be easier. In this work, we show the opposite of the predicted effect using cross-situational learning experiments with adults: when the distribution of words and meanings is Zipfian, learning is not impaired and is usually improved. Over a series of experiments, we provide evidence that this is because Zipfian distributions help people to disambiguate the meanings of the other words in the situation.

AB - Both adults and children have shown impressive cross-situational word learning in which they leverage the statistics of word usage across many different scenes in order to isolate specific word meanings (e.g., Yu & Smith, 2007). However, relatively little is known about how this learning scales to real language. Some theoretical analyses suggest that when words follow a Zipfian distribution, as they do in natural language, it should be more difficult to learn a lexicon because of the many low-frequency words that are only observed a few times (Blythe, Smith, & Smith, 2010; Vogt, 2012). Although this effect can be mitigated somewhat by assuming mutual exclusivity (Reisenauer, Smith, & Blythe, 2013), no mathematical analyses suggest that learning in a Zipfian environment should be easier. In this work, we show the opposite of the predicted effect using cross-situational learning experiments with adults: when the distribution of words and meanings is Zipfian, learning is not impaired and is usually improved. Over a series of experiments, we provide evidence that this is because Zipfian distributions help people to disambiguate the meanings of the other words in the situation.

KW - Cross-situational statistical learning

KW - Language learning

KW - Zipfian distributions

U2 - 10.1016/j.cognition.2019.03.005

DO - 10.1016/j.cognition.2019.03.005

M3 - Article

VL - 189

SP - 11

EP - 22

JO - Cognition

JF - Cognition

SN - 0010-0277

ER -