Pith. Sign in

RECORD

Neural Networks and Quantifier Conservativity: Does Data Distribution Affect Learnability?

PaperPDF from arXiv

data files

Paper breakdownThe authors' claims, conditions and measurements, as JSON Signed receiptProof this review has not been altered
X Bluesky LinkedIn Reddit Hacker News
Vishwali Mhasawade, Ildikó Emese Szabó, Melanie Tosik, Sheng-Fu Wang

Computation and Language · Submitted 15 Sep 2018

abstract compiled · 2026-09-23 · no language-model review

The authors' main claim We demonstrate that the aquisitional issues with non-conservative quantifiers can not be explained by the distribution of natural language data, which favors conservative quantifiers.

how it is stated · asserted 1.00 read the record →

Comment (0) Read on arXiv ↗
Cite ▾
arXiv ID1809.05733 (version 1)
Pith IDVSOUU5IP, Pith's permanent record of this paper and its review
CitationIldik\'o Emese Szab\'o, Melanie Tosik, Sheng-Fu Wang, Vishwali Mhasawade. "Neural Networks and Quantifier Conservativity: Does Data Distribution Affect Learnability?." arXiv:1809.05733 (2018). https://pith.science/paper/VSOUU5IP
Claim paper

The compiled record

The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.

Open the compiled record

Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-09-23: 1 request, 5,485 input tokens.

What the abstract itself states, as probabilities: claims first 0.04 · states limitation 0.23 · human participants 0.21 · controlled comparison 0.55 · randomized assignment 0.03 · states quantitative result 0.27 · compares with prior results 0.77 · mentions code or data availability 0.03.

The abstract, sentence by sentence

arXiv listing abstract v1. Results and the main claim are underlined; .

All known natural language determiners are conservative. Psycholinguistic experiments indicate that children exhibit a corresponding learnability bias when faced with the task of learning new determiners. However, recent work indicates that this bias towards conservativity is not observed during the training stage of artificial neural networks. In this work, we investigate whether the learnability bias exhibited by children is in part due to the distribution of quantifiers in natural language. We share results of five experiments, contrasted by the distribution of conservative vs. non-conservative determiners in the training data. We demonstrate that the aquisitional issues with non-conservative quantifiers can not be explained by the distribution of natural language data, which favors conservative quantifiers. This finding indicates that the bias in language acquisition data might be innate or representational.

background contribution method result interpretation main claim

What the classifier reports for each sentence. The number is the probability it gave the label.
#sentencerolethe authors' own workstates the main conclusionstrength
s0 All known natural language determiners are conservative. background 1.00 0.04 0.02 asserted 1.00
s1 Psycholinguistic experiments indicate that children exhibit a corresponding learnability bias when faced with the task of learning new determiners. background 1.00 0.04 0.02 hedged possibility 0.93
s2 However, recent work indicates that this bias towards conservativity is not observed during the training stage of artificial neural networks. background 0.75 0.07 0.07 hedged possibility 0.75
s3 In this work, we investigate whether the learnability bias exhibited by children is in part due to the distribution of quantifiers in natural language. contribution 0.73 0.28 0.04 hedged possibility 0.82
s4 We share results of five experiments, contrasted by the distribution of conservative vs. non-conservative determiners in the training data. method 0.67 0.91 0.03 asserted 0.85
s5 We demonstrate that the aquisitional issues with non-conservative quantifiers can not be explained by the distribution of natural language data, which favors conservative quantifiers. result 1.00 0.93 0.80 asserted 1.00
s6 This finding indicates that the bias in language acquisition data might be innate or representational. interpretation 1.00 0.69 0.67 hedged possibility 0.81

What kind of paper the classifier reports this is: empirical study 0.91 · evidence from observation or experiment 0.86.

What was examined, and what was not

sources read
  • arXiv listing abstract v1
not examined
  • full text
  • figures and tables
  • references and citing sentences
  • relations to other papers
levels run
  • P abstract
not run
  • S sections
  • T body sentences
  • C citations
  • R cross-paper relations

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. Every number in the measurements table is the authors', checked by code. Record as JSON

The paper
Original abstract

All known natural language determiners are conservative. Psycholinguistic experiments indicate that children exhibit a corresponding learnability bias when faced with the task of learning new determiners. However, recent work indicates that this bias towards conservativity is not observed during the training stage of artificial neural networks. In this work, we investigate whether the learnability bias exhibited by children is in part due to the distribution of quantifiers in natural language. We share results of five experiments, contrasted by the distribution of conservative vs. non-conservative determiners in the training data. We demonstrate that the aquisitional issues with non-conservative quantifiers can not be explained by the distribution of natural language data, which favors conservative quantifiers. This finding indicates that the bias in language acquisition data might be innate or representational.

Human review

How Pith checks papers → · Full audit of this paper →

Discussion

Know something about this paper? Leave a comment, or highlight any passage above to comment on it.