pith. sign in

arxiv: 2605.15365 · v1 · pith:CIUNC33Hnew · submitted 2026-05-14 · 💻 cs.CL

Greedy or not, here I come: Language production under vocabulary constraints in humans and resource-rational models

Thomas Hikaru Clark , Sihan Chen , Laura Nicolae This is my paper

Pith reviewed 2026-05-19 15:29 UTC · model grok-4.3

classification 💻 cs.CL
keywords language productionvocabulary constraintsgreedy samplingresource-rational modelshuman communicationpsycholinguisticssequential sampling
0
0 comments X p. Extension
pith:CIUNC33H Add to your LaTeX paper What is a Pith Number? 
\usepackage{pith}
\pithnumber{CIUNC33H}

Prints a linked pith:CIUNC33H badge after your title and writes the identifier into PDF metadata. Compiles on arXiv with no extra files. Learn more

The pith

Humans produce language more like greedy step-by-step choices than global planning when vocabulary is restricted.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper tests how people answer questions when limited to a small set of frequent words, as few as the top 250. It compares their output to two model strategies: one that selects each word based on immediate local fit and another that searches for the best overall message. Human behavior tracks the local choice strategy more closely overall, yet higher-skilled speakers show more revision and backtracking. Both strategies naturally produce a reliance on semantically light words in the hardest cases. The work ties these patterns to resource-rational accounts of how people manage cognitive limits during speech.

Core claim

When vocabulary is constrained to the most frequent 250 words, human responses to questions align more closely with greedy sampling than with globally optimal sampling, as approximated through Sequential Monte Carlo inference on large language models, except that more skilled participants exhibit greater backtracking and revision.

What carries the argument

Sequential Monte Carlo inference on large language models used to approximate both greedy word-by-word selection and global search for the best message under vocabulary limits.

If this is right

  • Resource-rational models can explain everyday language production when speakers face lexicon limits.
  • Greater skill correlates with more backtracking, suggesting expertise enables non-greedy planning.
  • Reliance on light words emerges automatically from either sampling strategy.
  • The same modeling approach may apply to communication under other constraints such as time pressure or aphasia.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Default reliance on local decisions may be a general feature of how minds handle complex sequential tasks.
  • Training or practice could shift behavior toward more revision even for less skilled speakers.
  • The findings point toward testable predictions for second-language learners facing similar vocabulary bottlenecks.

Load-bearing premise

Sequential Monte Carlo sampling from language models gives a faithful stand-in for what truly greedy or truly optimal human production would look like under the same word limits.

What would settle it

A direct comparison showing that human word sequences under the 250-word limit match the globally optimal model outputs more closely than the greedy ones, or that speaker skill does not predict rates of revision.

Figures

Figures reproduced from arXiv: 2605.15365 by Laura Nicolae, Sihan Chen, Thomas Hikaru Clark.

Figure 1
Figure 1. Figure 1: Example keystroke sequence under a 250-word [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Mean weighted score (automated evaluation on a [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Bump plots for both models and humans showing [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Mean valid word deletions per response, by vo [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
read the original abstract

Communicating using only a limited vocabulary is a common but challenging cognitive phenomenon, requiring an ideal communicator to plan carefully to optimize for intelligibility while circumventing a constrained lexicon. In this work, we investigate how humans respond to a broad array of questions under variable vocabulary limitations, consisting of only 250 highly frequent words at the most restrictive. We provide theoretically motivated comparisons to greedy and globally optimal sampling algorithms using Sequential Monte Carlo inference with large language models. Humans generally resemble greedy sampling more than globally optimal sampling, though more skilled humans are more likely to backtrack and revise -- a non-greedy behavior. An observed human pattern of leaning on semantically light words in high-constraint settings falls out of both greedy and globally optimal sampling. We discuss the results and their broader implications for resource-rational cognition, psycholinguistics, L2 communication, and language impairments.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper investigates human language production when restricted to limited vocabularies (most restrictively, the 250 most frequent words). It compares human responses to questions under these constraints against greedy sampling and globally optimal sampling, the latter approximated via Sequential Monte Carlo inference on large language models. The central claim is that humans generally align more closely with greedy sampling, although more skilled humans exhibit non-greedy behaviors such as backtracking and revision; the increased reliance on semantically light words under high constraint emerges from both sampling regimes.

Significance. If the SMC approximations are shown to be faithful, the work would strengthen resource-rational accounts of language production by demonstrating how humans trade off planning depth against computational cost under lexical constraints, with direct implications for psycholinguistics, L2 communication, and models of language impairment. The parameter-free character of the model predictions and the use of SMC to operationalize global optimality constitute methodological strengths that could support falsifiable predictions.

major comments (2)
  1. [§4 (SMC implementation and global-optimality approximation)] §4 (SMC implementation and global-optimality approximation): no diagnostic is reported that compares log-probability, intelligibility, or downstream task performance of the SMC-derived 'globally optimal' trajectories against the greedy samples under the 250-word mask. Without evidence that the particle filter recovers meaningfully higher-utility sequences, the claimed distinction between the two model classes is not established and the human-greedy alignment result becomes difficult to interpret.
  2. [Results section (human-model alignment)] Results section (human-model alignment): the directional resemblance between human data and greedy sampling is asserted without reported sample sizes for participants, statistical tests, alignment metrics (e.g., correlation or divergence measures), or details on how model-generated sequences were matched to human responses, preventing quantitative evaluation of the central claim.
minor comments (2)
  1. [Abstract and §2] Abstract and §2: the exact vocabulary sizes and constraint levels used in each condition should be stated explicitly rather than summarized as 'variable vocabulary limitations.'
  2. [Discussion] Discussion: the implications for L2 communication and language impairments are noted but would benefit from concrete links to one or two key prior studies in those literatures.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments, which identify key areas where additional evidence and quantitative detail would strengthen the manuscript. We address each major comment below and outline planned revisions.

read point-by-point responses

  1. Referee: [§4 (SMC implementation and global-optimality approximation)] §4 (SMC implementation and global-optimality approximation): no diagnostic is reported that compares log-probability, intelligibility, or downstream task performance of the SMC-derived 'globally optimal' trajectories against the greedy samples under the 250-word mask. Without evidence that the particle filter recovers meaningfully higher-utility sequences, the claimed distinction between the two model classes is not established and the human-greedy alignment result becomes difficult to interpret.

    Authors: We agree that direct diagnostics would better substantiate the SMC approximation to global optimality. The manuscript motivates SMC as a standard method for approximating exhaustive search over sequences under the vocabulary mask and reports the specific particle count and resampling procedure used, but it does not include explicit utility comparisons. In the revision we will add to §4 a comparison of mean log-probabilities (under the base LLM) for SMC trajectories versus greedy samples in the 250-word condition, along with a brief note on sequence intelligibility via human ratings of a small subset. This will provide concrete evidence that the particle filter recovers higher-utility sequences. revision: yes

  2. Referee: [Results section (human-model alignment)] Results section (human-model alignment): the directional resemblance between human data and greedy sampling is asserted without reported sample sizes for participants, statistical tests, alignment metrics (e.g., correlation or divergence measures), or details on how model-generated sequences were matched to human responses, preventing quantitative evaluation of the central claim.

    Authors: We accept that the Results section would be improved by explicit quantitative reporting. The current text describes qualitative alignment in word-choice patterns and revision rates but omits sample sizes, formal tests, and matching procedures. We will revise the Results section to report the total number of participants and trials, describe the sequence-matching protocol (pairing by question and constraint level), and include alignment metrics such as type-overlap proportions and a distributional divergence measure between human and model outputs. Where sample sizes permit, we will also add simple statistical comparisons of key behavioral measures. revision: yes

Circularity Check

0 steps flagged

Independent LLM-based SMC simulations; no reduction of predictions to human-data fits

full rationale

The paper compares human responses under vocabulary constraints to greedy and globally optimal sampling implemented via Sequential Monte Carlo on large language models. These model procedures operate on pretrained LLMs with a fixed 250-word mask and do not fit parameters or define optimality from the human behavioral data itself. No self-definitional loops, fitted-input predictions, or load-bearing self-citations appear in the derivation chain. The observed human resemblance to greedy sampling is therefore evaluated against an external computational benchmark rather than being forced by construction from the same data.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Based on abstract only; no explicit free parameters, invented entities, or additional axioms beyond standard assumptions in cognitive modeling are stated.

axioms (1)
  • domain assumption Humans approximate resource-rational agents when producing language under constraints
    The study frames results in terms of resource-rational cognition without deriving this from first principles.

pith-pipeline@v0.9.0 · 5683 in / 1211 out tokens · 44744 ms · 2026-05-19T15:29:14.660106+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

86 extracted references · 86 canonical work pages · 2 internal anchors

  1. [1]

    , editor =

    Lombrozo, Tania and Wilkenfeld, Daniel A. , editor =. Mechanistic versus Functional Understanding , booktitle =

    work page

  2. [2]

    Adolphs, Svenja and Schmitt, Norbert , year = 2003, journal =. Lexical

    work page 2003

  3. [3]

    , year = 2011, journal =

    Arnon, Inbal and Clark, Eve V. , year = 2011, journal =. Why

    work page 2011

  4. [4]

    Beeke, Suzanne , year = 2012, pages =. 13. 13

    work page 2012

  5. [5]

    Language

    Bleakley, Hoyt and Chin, Aimee , year = 2004, journal =. Language

    work page 2004

  6. [6]

    Handbook of Psycholinguistics , author =

    Language Production:. Handbook of Psycholinguistics , author =

    work page

  7. [7]

    Cognition , volume =

    Perceptual Adaptation to Non-Native Speech , author =. Cognition , volume =. doi:10.1016/j.cognition.2007.04.005 , pmcid =

    work page doi:10.1016/j.cognition.2007.04.005 2007

  8. [8]

    Brysbaert, Marc and Stevens, Micha. How. Frontiers in Psychology , volume =. doi:10.3389/fpsyg.2016.01116 , pmcid =

    work page doi:10.3389/fpsyg.2016.01116 2016

  9. [9]

    Calderon, Nitay and Reichart, Roi and Dror, Rotem , year = 2025, number =. The. doi:10.48550/arXiv.2501.10970 , archiveprefix =. 2501.10970 , primaryclass =

    work page doi:10.48550/arxiv.2501.10970 2025

  10. [10]

    Carnie, Andrew , year = 2013, publisher =. Syntax:

    work page 2013

  11. [11]

    Cooperative Explanation as Rational Communication , booktitle =

    Chandra, Kartik and Chen, Tony and Li, Tzu-Mao and. Cooperative Explanation as Rational Communication , booktitle =

    work page

  12. [12]

    Journal of Language Evolution , volume =

    Discrete and Systematic Communication in a Continuous Signal-Meaning Space , author =. Journal of Language Evolution , volume =

    work page

  13. [13]

    Aspects of the

    Chomsky, Noam , year = 2014, publisher =. Aspects of the

    work page 2014

  14. [14]

    Chomsky, Noam , year = 2014, publisher =. The

    work page 2014

  15. [15]

    Behavioral and Brain Sciences , volume =

    Rules and Representations , author =. Behavioral and Brain Sciences , volume =

    work page

  16. [16]

    IRE Transactions on Information Theory , volume =

    Three Models for the Description of Language , author =. IRE Transactions on Information Theory , volume =

    work page

  17. [17]

    Clark, Thomas Hikaru and Vigly, Jacob Hoover and Gibson, Edward and Levy, Roger , year = 2025, journal =. A

    work page 2025

  18. [18]

    , editor =

    Clark, Thomas Hikaru and Vigly, Jacob Hoover and Gibson, Edward and Levy, Roger P. , editor =. Resource-. Proceedings of the 2025

    work page 2025

  19. [19]

    Nature Human Behaviour , volume =

    How Laypeople Evaluate Scientific Explanations Containing Jargon , author =. Nature Human Behaviour , volume =. doi:10.1038/s41562-025-02227-0 , copyright =

    work page doi:10.1038/s41562-025-02227-0

  20. [20]

    and Graf, Caroline and Kreiss, Elisa and Goodman, Noah D

    Degen, Judith and Hawkins, Robert D. and Graf, Caroline and Kreiss, Elisa and Goodman, Noah D. , year = 2020, journal =. When Redundancy Is Useful:. doi:10.1037/rev0000186 , pmid =

    work page doi:10.1037/rev0000186 2020

  21. [21]

    Communication

    D. Communication. Language Learning , volume =. doi:10.1111/0023-8333.51997005 , copyright =

    work page doi:10.1111/0023-8333.51997005

  22. [22]

    Agrammatic Output in Non-Fluent, Including

    Fedorenko, Evelina and Ryskin, Rachel and Gibson, Edward , year = 2022, journal =. Agrammatic Output in Non-Fluent, Including

    work page 2022

  23. [23]

    Nature , volume =

    Language Is Primarily a Tool for Communication Rather than Thought , author =. Nature , volume =. doi:10.1038/s41586-024-07522-w , copyright =

    work page doi:10.1038/s41586-024-07522-w

  24. [24]

    Good-Enough Representations in Language Comprehension , author =

    work page

  25. [25]

    , year = 2010, journal =

    Florian Jaeger, T. , year = 2010, journal =. Redundancy and Reduction:. doi:10.1016/j.cogpsych.2010.02.002 , pmcid =

    work page doi:10.1016/j.cogpsych.2010.02.002 2010

  26. [26]

    Proceedings of the National Academy of Sciences , volume =

    Information-Theoretic Principles in Incremental Language Production , author =. Proceedings of the National Academy of Sciences , volume =

    work page

  27. [27]

    Amortized

    Gershman, Samuel and Goodman, Noah , year = 2014, journal =. Amortized

    work page 2014

  28. [28]

    and Dautriche, Isabelle and Mahowald, Kyle and Bergen, Leon and Levy, Roger , year = 2019, journal =

    Gibson, Edward and Futrell, Richard and Piantadosi, Steven P. and Dautriche, Isabelle and Mahowald, Kyle and Bergen, Leon and Levy, Roger , year = 2019, journal =. How. doi:10.1016/j.tics.2019.02.003 , pmid =

    work page doi:10.1016/j.tics.2019.02.003 2019

  29. [29]

    Goulden, Robin and Nation, Paul and Read, John , year = 1990, journal =. How

    work page 1990

  30. [30]

    and Lieder, Falk and Goodman, Noah D

    Griffiths, Thomas L. and Lieder, Falk and Goodman, Noah D. , year = 2015, journal =. Rational

    work page 2015

  31. [31]

    Grogger, Jeffrey and Steinmayr, Andreas and Winter, Joachim , year = 2020, series =. The. doi:10.3386/w26719 , archiveprefix =. 26719 , publisher =

    work page doi:10.3386/w26719 2020

  32. [32]

    Gu, Jiawei and Jiang, Xuhui and Shi, Zhichao and Tan, Hexiang and Zhai, Xuehao and Xu, Chengjin and Li, Wei and Shen, Yinghan and Ma, Shengjie and Liu, Honghao and Wang, Saizhuo and Zhang, Kun and Wang, Yuanzhuo and Gao, Wen and Ni, Lionel and Guo, Jian , year = 2025, number =. A. doi:10.48550/arXiv.2411.15594 , archiveprefix =. 2411.15594 , primaryclass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2411.15594 2025

  33. [33]

    Lexical Teddy Bears and Advanced Learners: A Study into the Ways

    Hasselgren, Angela , year = 1994, journal =. Lexical Teddy Bears and Advanced Learners: A Study into the Ways

    work page 1994

  34. [34]

    Charles , year = 2010, journal =

    Healey, E. Charles , year = 2010, journal =. What the. doi:10.1055/s-0030-1265756 , copyright =

    work page doi:10.1055/s-0030-1265756 2010

  35. [35]

    Semantics in

    Heim, Irene and Kratzer, Angelika , year = 1998, publisher =. Semantics in

    work page 1998

  36. [36]

    Cognitive Psychology , volume =

    Sources of. Cognitive Psychology , volume =. doi:10.1006/cogp.1998.0713 , copyright =

    work page doi:10.1006/cogp.1998.0713 1998

  37. [37]

    Honnibal, Matthew and Montani, Ines and Van Landeghem, Sofie and Boyd, Adriane , year = 2020, doi =

    work page 2020

  38. [38]

    Journal of speech, language, and hearing research : JSLHR , volume =

    The Relationship between Listener Comprehension and Intelligibility Scores for Speakers with Dysarthria , author =. Journal of speech, language, and hearing research : JSLHR , volume =. doi:10.1044/1092-4388(2008/040) , pmcid =

    work page doi:10.1044/1092-4388(2008/040 2008

  39. [39]

    and Black, S

    Kagan, A. and Black, S. E. and Duchan, F. J. and. Training Volunteers as Conversation Partners Using ". Journal of speech, language, and hearing research: JSLHR , volume =. doi:10.1044/1092-4388(2001/051) , pmid =

    work page doi:10.1044/1092-4388(2001/051 2001

  40. [40]

    Kahneman, Daniel , year = 2003, journal =. Maps of

    work page 2003

  41. [41]

    Prospect

    Kahneman, Daniel and Tversky, Amos , year = 1979, journal =. Prospect. 1914185 , eprinttype =

    work page 1979

  42. [42]

    Word Knowledge in the Crowd:

    Keuleers, Emmanuel and Stevens, Micha. Word Knowledge in the Crowd:. Quarterly Journal of Experimental Psychology , volume =. doi:10.1080/17470218.2015.1022560 , pmid =

    work page doi:10.1080/17470218.2015.1022560 2015

  43. [43]

    Klein, Wolfgang and Perdue, Clive , year = 1997, journal =. The

    work page 1997

  44. [44]

    and Jaeger, T

    Kleinschmidt, Dave F. and Jaeger, T. Florian , year = 2015, journal =. Robust Speech Perception: Recognize the Familiar, Generalize to the Similar, and Adapt to the Novel , shorttitle =. doi:10.1037/a0038695 , pmcid =

    work page doi:10.1037/a0038695 2015

  45. [45]

    Kong, Anthony Pak-Hin and Wong, Cherie Wan-Yin , year = 2018, journal =. An. doi:10.1044/2018_AJSLP-18-0015 , pmcid =

    work page doi:10.1044/2018_ajslp-18-0015 2018

  46. [46]

    and Weinheimer, Sidney , year = 1964, journal =

    Krauss, Robert M. and Weinheimer, Sidney , year = 1964, journal =. Changes in Reference Phrases as a Function of Frequency of Usage in Social Interaction: A Preliminary Study , shorttitle =

    work page 1964

  47. [47]

    Florian , year = 2015, journal =

    Kurumada, Chigusa and Jaeger, T. Florian , year = 2015, journal =. Communicative Efficiency in Language Production:

    work page 2015

  48. [48]

    Behavioral and Brain Sciences , volume =

    Building Machines That Learn and Think like People , author =. Behavioral and Brain Sciences , volume =

    work page

  49. [49]

    The Development of

    Laufer, Batia , year = 1991, journal =. The Development of

    work page 1991

  50. [50]

    Is Bilingualism Associated with Enhanced Executive Functioning in Adults?

    Lehtonen, Minna and Soveri, Anna and Laine, Aini and J. Is Bilingualism Associated with Enhanced Executive Functioning in Adults?. Psychological Bulletin , volume =. doi:10.1037/bul0000142 , pmid =

    work page doi:10.1037/bul0000142

  51. [51]

    Revista da ABRALIN , pages =

    A Functionalist Typology of Redundancy , author =. Revista da ABRALIN , pages =. doi:10.25189/rabralin.v19i3.1722 , copyright =

    work page doi:10.25189/rabralin.v19i3.1722

  52. [52]

    Lew, Alexander K. and. Sequential. doi:10.48550/arXiv.2306.03081 , archiveprefix =. 2306.03081 , primaryclass =

    work page doi:10.48550/arxiv.2306.03081

  53. [53]

    , year = 2020, journal =

    Lieder, Falk and Griffiths, Thomas L. , year = 2020, journal =. Resource-Rational Analysis:

    work page 2020

  54. [54]

    Lipkin, Benjamin and LeBrun, Benjamin and Vigly, Jacob Hoover and Loula, Jo. Fast. doi:10.48550/arXiv.2504.05410 , archiveprefix =. 2504.05410 , primaryclass =

    work page doi:10.48550/arxiv.2504.05410

  55. [55]

    Cognition , volume =

    Functional Explanation and the Function of Explanation , author =. Cognition , volume =

    work page

  56. [56]

    Inferring Design: Evidence of a Preference for Teleological Explanations in Patients with

    Lombrozo, Tania and Kelemen, Deborah and Zaitchik, Deborah , year = 2007, journal =. Inferring Design: Evidence of a Preference for Teleological Explanations in Patients with. doi:10.1111/j.1467-9280.2007.02015.x , pmid =

    work page doi:10.1111/j.1467-9280.2007.02015.x 2007

  57. [57]

    Syntactic and semantic control of large language models via sequential

    Loula, Jo. Syntactic and. doi:10.48550/arXiv.2504.13139 , archiveprefix =. 2504.13139 , primaryclass =

    work page doi:10.48550/arxiv.2504.13139

  58. [58]

    Proceedings of Meetings on Acoustics , volume =

    Human Language Diversity and the Acoustic Adaptation Hypthesis , author =. Proceedings of Meetings on Acoustics , volume =

    work page

  59. [59]

    Cognition , volume =

    Grammatical Cues to Subjecthood Are Redundant in a Majority of Simple Clauses across Languages , author =. Cognition , volume =

    work page

  60. [60]

    Marr, David , year = 1982, publisher =. Vision:

    work page 1982

  61. [61]

    and Keil, Frank C

    McCarthy, Amanda M. and Keil, Frank C. , year = 2023, journal =. A Right Way to Explain?

    work page 2023

  62. [62]

    27184953 , eprinttype =

    Morley, Brendan Arkell , year = 2022, journal =. 27184953 , eprinttype =

    work page 2022

  63. [63]

    , year = 1985, journal =

    Na, Liu and Nation, I.S.P. , year = 1985, journal =. Factors

    work page 1985

  64. [64]

    The Behavioral and Brain Sciences , volume =

    An Integrated Theory of Language Production and Comprehension , author =. The Behavioral and Brain Sciences , volume =. doi:10.1017/S0140525X12001495 , pmid =

    work page doi:10.1017/s0140525x12001495

  65. [65]

    Behavioral and Brain Sciences , volume =

    Natural Language and Natural Selection , author =. Behavioral and Brain Sciences , volume =

    work page

  66. [66]

    Poulisse, Nanda , year = 2011, pages =. A. The

    work page 2011

  67. [67]

    Proceedings of the National Academy of Sciences of the United States of America , volume =

    A Syntax--Lexicon Trade-off in Language Production , author =. Proceedings of the National Academy of Sciences of the United States of America , volume =. doi:10.1073/pnas.2120203119 , pmcid =

    work page doi:10.1073/pnas.2120203119

  68. [68]

    Gaps in the

    Rissman, Lilia and Liu, Qiawen and Lupyan, Gary , year = 2023, journal =. Gaps in the

    work page 2023

  69. [69]

    A Mathematical Theory of Communication , author =

    work page

  70. [70]

    Rspeer/Wordfreq: V3.0 , shorttitle =

    Speer, Robyn , year = 2022, doi =. Rspeer/Wordfreq: V3.0 , shorttitle =

    work page 2022

  71. [71]

    Cognition , volume =

    Explanations in the Wild , author =. Cognition , volume =

    work page

  72. [72]

    and Delogu, Francesca and Crocker, Matthew W

    Tourtouri, Elli N. and Delogu, Francesca and Crocker, Matthew W. , year = 2021, journal =. Rational. doi:10.1111/cogs.13071 , copyright =

    work page doi:10.1111/cogs.13071 2021

  73. [73]

    Current biology: CB , volume =

    Evidence for Cognition without Grammar from Causal Reasoning and 'theory of Mind' in an Agrammatic Aphasic Patient , author =. Current biology: CB , volume =. doi:10.1016/s0960-9822(00)00538-8 , pmid =

    work page doi:10.1016/s0960-9822(00)00538-8

  74. [74]

    Wang, Hao and Shimizu, Hirofumi and Kawahara, Daisuke , year = 2024, number =. Kanbun-. doi:10.48550/arXiv.2305.12759 , archiveprefix =. 2305.12759 , primaryclass =

    work page doi:10.48550/arxiv.2305.12759 2024

  75. [75]

    and Kirkovski, Melissa and Lum, Jarrad A

    Ware, Anna T. and Kirkovski, Melissa and Lum, Jarrad A. G. , year = 2020, journal =. Meta-. doi:10.3389/fpsyg.2020.01458 , pmcid =

    work page doi:10.3389/fpsyg.2020.01458 2020

  76. [76]

    Wicha, Nicole Y. Y. and Moreno, Eva M. and Kutas, Marta , year = 2004, journal =. Anticipating. doi:10.1162/0898929041920487 , pmcid =

    work page doi:10.1162/0898929041920487 2004

  77. [77]

    and Mueller, Aaron and Warstadt, Alex and Choshen, Leshem and Zhuang, Chengxu and Williams, Adina and Cotterell, Ryan and Linzen, Tal , year = 2025, journal =

    Wilcox, Ethan Gotlieb and Hu, Michael Y. and Mueller, Aaron and Warstadt, Alex and Choshen, Leshem and Zhuang, Chengxu and Williams, Adina and Cotterell, Ryan and Linzen, Tal , year = 2025, journal =. Bigger Is Not Always Better:

    work page 2025

  78. [78]

    Bilingual Children Outperform Monolingual Children on Executive Function Tasks Far More Often than Chance:

    Yurtsever, Asl. Bilingual Children Outperform Monolingual Children on Executive Function Tasks Far More Often than Chance:. Developmental Review , volume =

    work page

  79. [79]

    and Chronis, Andrea M

    Zechmeister, Eugene B. and Chronis, Andrea M. and Cull, William L. and D'Anna, Catherine A. and Healy, Noreen A. , year = 1995, journal =. Growth of a

    work page 1995

  80. [80]

    Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena

    Zheng, Lianmin and Chiang, Wei-Lin and Sheng, Ying and Zhuang, Siyuan and Wu, Zhanghao and Zhuang, Yonghao and Lin, Zi and Li, Zhuohan and Li, Dacheng and Xing, Eric P. and Zhang, Hao and Gonzalez, Joseph E. and Stoica, Ion , year = 2023, number =. Judging. doi:10.48550/arXiv.2306.05685 , archiveprefix =. 2306.05685 , primaryclass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2306.05685 2023

Showing first 80 references.