Zipfian frequency distributions facilitate word segmentation in context

Kurumada, Chigusa; Meylan, Stephan C.; Frank, Michael C.

doi:10.1016/j.cognition.2013.02.002

Cited by 54 publications

(79 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In other words, the type of statistic that learners use to determine categorization is not the absolute frequency with which words occur in each context, but rather the conditional probabilities that words occur in these contexts, given their overall frequency. This is consistent with learners’ use of conditional rather than absolute frequency statistics in other experiments as well (see, for example, Aslin, Saffran & Newport, 1998; Kurumada et al, 2013). …”

Section: Discussionsupporting

confidence: 89%

“…Our results also suggest the type of statistics that learners utilize as they acquire grammatical categories. As has been shown in studies of word segmentation (Aslin et al, 1998; Kurumada et al, 2013), statistical learning does not appear to depend primarily on simple frequency statistics (such as lexical frequency or bigram frequency), but rather utilizes more complex calculations (such as conditional probabilities or Bayesian statistics) that involve the expected frequency with which element combinations should occur, given their individual element frequencies. While it might seem unlikely that infants and young children could be capable of these complex calculations, our studies to date with young learners (Aslin et al, 1998; Schuler et al, in preparation) support the notion that statistical learning involves such computations at many levels of analysis.…”

Section: Discussionmentioning

confidence: 99%

“…For example, learners are better at acquiring categories (Valian & Coulson, 1998) and learning both adjacent (Kurumada, Meylan, & Frank, 2013) and non-adjacent dependencies (Gomez, 2002) from distributional cues if the structures they are learning contain some high frequency elements. Researchers suggest that high frequency elements may facilitate learning because they provide an additional distributional cue to the learner.…”

Section: Introductionmentioning

confidence: 99%

“…Since words and contexts in real languages do indeed vary dramatically in frequency, studying the effects of such frequency variations is important for understanding how statistical learning works in such circumstances (see also Kurumada, Meylan, & Frank (2013), who used Zipfian frequency variations in statistical learning of word segmentation and found that such variations can be advantageous). By introducing large lexical frequency variations in the input, we will not only explore how this impacts category formation and generalization; we will also test whether previous distributional learning results scale-up to more naturalistic input, and precisely how computations in natural language acquisition are adjusted for frequency variations.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

The Effect of Zipfian Frequency Variations on Category Formation in Adult Artificial Language Learning

Schuler

Reeder

Newport

et al. 2017

Language Learning and Development

View full text Add to dashboard Cite

Successful language acquisition hinges on organizing individual words into grammatical categories and learning the relationships between them, but the method by which children accomplish this task has been debated in the literature. One proposal is that learners use the shared distributional contexts in which words appear as a cue to their underlying category structure. Indeed, recent research using artificial languages has demonstrated that learners can acquire grammatical categories from this type of distributional information. However, artificial languages are typically composed of a small number of equally frequent words, while words in natural languages vary widely in frequency, complicating the distributional information needed to determine categorization. In a series of three experiments we demonstrate that distributional learning is preserved in an artificial language composed of words that vary in frequency as they do in natural language, along a Zipfian distribution. Rather than depending on the absolute frequency of words and their contexts, the conditional probabilities that words will occur in certain contexts (given their base frequency) is a better basis for assigning words to categories; and this appears to be the type of statistic that human learners utilize.

show abstract

Section: Discussionsupporting

confidence: 89%

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

The Effect of Zipfian Frequency Variations on Category Formation in Adult Artificial Language Learning

Schuler

Reeder

Newport

et al. 2017

Language Learning and Development

View full text Add to dashboard Cite

show abstract

“…For instance, in work on segmentation, Kurumada, Meylan, and Frank (2013) found that presenting participants with Zipf-distributed input facilitated segmentation and Hay, Pelucchi, Estes, and Saffran (2011) found that presenting participants with natural language (Italian) containing peaked transitional probabilities facilitated subsequent word learning, bolstering previous findings that had used artificial languages. For instance, in work on segmentation, Kurumada, Meylan, and Frank (2013) found that presenting participants with Zipf-distributed input facilitated segmentation and Hay, Pelucchi, Estes, and Saffran (2011) found that presenting participants with natural language (Italian) containing peaked transitional probabilities facilitated subsequent word learning, bolstering previous findings that had used artificial languages.…”

Section: A Puzzle: the Mnpq Languagementioning

confidence: 68%

Semantic Coherence Facilitates Distributional Learning

2016

Self Cite

View full text Add to dashboard Cite

Computational models have shown that purely statistical knowledge about words' linguistic contexts is sufficient to learn many properties of words, including syntactic and semantic category. For example, models can infer that "postman" and "mailman" are semantically similar because they have quantitatively similar patterns of association with other words (e.g., they both tend to occur with words like "deliver," "truck," "package"). In contrast to these computational results, artificial language learning experiments suggest that distributional statistics alone do not facilitate learning of linguistic categories. However, experiments in this paradigm expose participants to entirely novel words, whereas real language learners encounter input that contains some known words that are semantically organized. In three experiments, we show that (a) the presence of familiar semantic reference points facilitates distributional learning and (b) this effect crucially depends both on the presence of known words and the adherence of these known words to some semantic organization.

show abstract

Cognitive and linguistic biases in morphology learning

Finley

2018

WIRES Cognitive Science

View full text Add to dashboard Cite

Morphology is the study of the relationship between form and meaning. The study of morphology involves understanding the rules and processes that underlie word formation, including the use and productivity of affixes, and the systems that create novel word forms. The present review explores these processes by examining the cognitive components that contribute to typological regularities among morphological systems across the world's language. The review will focus on research in morpheme segmentation, the suffixing preference, acquisition of morphophonology, and acquiring morphological categories and inflectional paradigms. The review will highlight research in a range of areas of linguistics, from child language acquisition, to computational modeling, to adult language learning experiments. In order to best understand the cognitive biases that shape morphological learning, a broad, interdisciplinary approach must be taken. This article is categorized under: Linguistics > Linguistic Theory Linguistics > Language Acquisition Psychology > Language.

show abstract

Zipfian frequency distributions facilitate word segmentation in context

Cited by 54 publications

References 45 publications

The Effect of Zipfian Frequency Variations on Category Formation in Adult Artificial Language Learning

The Effect of Zipfian Frequency Variations on Category Formation in Adult Artificial Language Learning

Semantic Coherence Facilitates Distributional Learning

Cognitive and linguistic biases in morphology learning

Contact Info

Product

Resources

About