Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 38 of the 38 references it cites were found in a registry or an index.

Colors in Context: A Pragmatic Neural Model for Grounded Language Understanding

Open access
Monroe, Will, Hawkins, Robert X.D., Goodman, Noah D., Potts, Christopher · 2017 · Transactions of the Association for Computational Linguistics

Abstract

We present a model of pragmatic referring expression interpretation in a grounded communication task (identifying colors from descriptions) that draws upon predictions from two recurrent neural network classifiers, a speaker and a listener, unified by a recursive pragmatic reasoning framework. Experiments show that this combined pragmatic model interprets color descriptions more accurately than the classifiers from which it is built, and that much of this improvement results from combining the speaker and listener perspectives. We observe that pragmatic reasoning helps primarily in the hardest cases: when the model must distinguish very similar colors, or when few utterances adequately express the target color. Our findings make use of a newly-collected corpus of human utterances in color reference games, which exhibit a variety of pragmatic behaviors. We also show that the embedded speaker model reproduces many of these pragmatic behaviors.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1162/tacl_a_00064

Cites

38 of 38 references verified0 citation errors · 0 not checkable

Statistics

Figures

Table 1: Examples of color reference in context, taken from our corpus. The target color is boxed. The speaker's description is shaped not only by this target, but also by the other context colors and their relationships.
Table 1: Examples of color reference in context, taken from our corpus. The target color is boxed. The speaker's description is shaped not only by this target, but also by the other context colors and their relationships.CC-BY · From the PDF
Figure 1 Example trial in corpus collection task, from speaker's perspective. The target color (boxed) was presented among two distractors on a neutral background.
Figure 1 Example trial in corpus collection task, from speaker's perspective. The target color (boxed) was presented among two distractors on a neutral background.CC-BY · From the PDF
Image 3
CC-BY · From the PDF
(d) The pragmatic listener l 2 uses Bayes' rule to infer the target using the speaker's utterance as evidence.
(d) The pragmatic listener l 2 uses Bayes' rule to infer the target using the speaker's utterance as evidence.CC-BY · From the PDF
(a) The L 0 agent processes tokens u i of a color description u sequentially. The final representation is transformed into a Gaussian distribution in color space, which is used to score the context colors c 1 . . . c 3 .
(a) The L 0 agent processes tokens u i of a color description u sequentially. The final representation is transformed into a Gaussian distribution in color space, which is used to score the context colors c 1 . . . c 3 .CC-BY · From the PDF
(b) The S 0 agent processes the target color c t in context and produces tokens u i of a color description sequentially. Each step in production is conditioned by the context representation h and the previous word produced.
(b) The S 0 agent processes the target color c t in context and produces tokens u i of a color description sequentially. Each step in production is conditioned by the context representation h and the previous word produced.CC-BY · From the PDF
Figure 4 Human and model reference game performance (top) and fraction of examples improved and declined from L 0 to L e (bottom) on the dev set, by condition.
Figure 4 Human and model reference game performance (top) and fraction of examples improved and declined from L 0 to L e (bottom) on the dev set, by condition.CC-BY · From the PDF
Figure 5 Conditional probabilities (%) of all agents for two dev set examples. The target color is boxed, and the human utterances ( blue , drab green not the bluer one ) are bolded . Boxed cells for alternative utterances indicate the intended target; largest probabilities are in bold . S 0 probabilities ( italics ) are normalized across all utterances. Sample sizes are reduced to save space; here, m = 2 and n = 1 (see Section 4.3).
Figure 5 Conditional probabilities (%) of all agents for two dev set examples. The target color is boxed, and the human utterances ( blue , drab green not the bluer one ) are bolded . Boxed cells for alternative utterances indicate the intended target; largest probabilities are in bold . S 0 probabilities ( italics ) are normalized across all utterances. Sample sizes are reduced to save space; here, m = 2 and n = 1 (see Section 4.3).CC-BY · From the PDF
Figure 6 L 0 's log marginal probability density, marginalizing over V (value) in HSV space, of color conditioned on the utterance drab green not the bluer one . White regions have higher probability. Labeled colors are the three colors from the right column of Figure 5.
Figure 6 L 0 's log marginal probability density, marginalizing over V (value) in HSV space, of color conditioned on the utterance drab green not the bluer one . White regions have higher probability. Labeled colors are the three colors from the right column of Figure 5.CC-BY · From the PDF

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 38 references

Cited by

See all 97

Metadata

Authors
Monroe, W., Hawkins, R. X. D., Goodman, N. D., & Potts, C.
Affiliations
Stanford University (US)
Corresponding author
  • Monroe, Will
  • Hawkins, Robert X.D.
  • Goodman, Noah D.
  • Potts, Christopher
Venue
Transactions of the Association for Computational Linguistics
Published
2017-12
Volume
5
Pages
325-338
ISSN
2307-387X (ISSN-L)
OpenAlex
W2604397416
Cited by
94
References
38
Citations by year
2026: 3 · 2025: 3 · 2024: 3 · 2023: 13 · 2022: 7 · 2021: 23 · 2020: 16 · 2019: 14 · 2018: 9 · 2017: 3
FWCI
14.96
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • Transactions of the Association for Computational Linguistics (publishedVersion, cc-by)
  • arXiv (Cornell University) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
  • arXiv (Cornell University) (cc-by)
Funders
  • National Science Foundation ((DGE-114747, 114747, 1456077)
  • Alfred P. Sloan Foundation
  • Defense Advanced Research Projects Agency (FA8750, FA8750‐14‐2‐0009)
Topics
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Categorization, perception, and language
  • Social Sciences › Arts and Humanities › Language and Linguistics › Syntax, Semantics, Linguistic Variation
  • Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
Keywords
referring expressions, pragmatic reasoning, color description, recurrent neural networks

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Colors in Context: A Pragmatic Neural Model for Grounded Language Understanding. https://app.cochat.ai/papers/10.1162/tacl_a_00064
The paper
Monroe, W., Hawkins, R. X. D., Goodman, N. D., & Potts, C. (2017). Colors in Context: A Pragmatic Neural Model for Grounded Language Understanding. Transactions of the Association for Computational Linguistics. https://doi.org/10.1162/tacl_a_00064

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.