pith. sign in

arxiv: 2408.05086 · v3 · submitted 2024-08-09 · 💻 cs.CL · cs.AI

A systematic framework for generating novel experimental hypotheses from language models

Kanishka Misra , Najoung Kim This is my paper

Pith reviewed 2026-05-23 21:51 UTC · model grok-4.3

classification 💻 cs.CL cs.AI
keywords hypothesis generationlanguage modelschild language acquisitiondative verbsgeneralizationsimulated experiments
0
0 comments X

The pith

Language models can simulate nonexistent child experiments to generate new hypotheses about how kids generalize verbs.

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

The paper presents a framework that runs language models as stand-ins for children to forecast what would happen in experiments no one has conducted. When applied to dative-verb learning, the simulation produces the prediction that matching argument order to discourse prominence in training sentences changes how readily children extend new verbs to unseen structures. The authors also lay out specific child experiments that could confirm or refute the prediction. A reader would care if the method turns out to let researchers generate fresh, testable ideas about human cognition directly from model runs instead of only from existing data.

Core claim

The authors claim that their framework, when instantiated on dative verb acquisition, produces the novel hypothesis that alignment between argument ordering and discourse prominence features of exposure contexts modulates how children generalize new verbs to unobserved structures, and they supply concrete experimental designs for testing this claim with children.

What carries the argument

A systematic framework that treats language models as simulated learners to predict outcomes of future behavioral experiments.

If this is right

  • The match between argument ordering and discourse prominence in exposure sentences modulates children's cross-structural generalization of dative verbs.
  • A set of lab experiments with children can be run to test the generated hypotheses.
  • The same simulation approach can be applied to other open questions in language acquisition.

Where Pith is reading between the lines

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

  • If the simulations prove reliable, researchers could generate candidate hypotheses faster by querying models before committing to child studies.
  • The method might surface cases where model predictions diverge from actual child data, highlighting specific limits of current language models as cognitive simulators.
  • Similar simulation pipelines could be tried in non-language domains of cognitive development where behavioral experiments are costly.

Load-bearing premise

Language models can accurately simulate how children would respond in language-learning experiments.

What would settle it

Running the proposed experiments with children and finding that alignment between argument ordering and discourse prominence does not affect generalization rates.

Figures

Figures reproduced from arXiv: 2408.05086 by Kanishka Misra, Najoung Kim.

Figure 1
Figure 1. Figure 1: Overview of our methodology for investigating an LM learner’s cross-dative generalization be [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Average ∆ values computed using our LM learners on NABA (N=12) and NANA (N=14) verbs from AO-CHILDES (Huebner and Willits, 2021). Error bars indicate 95% CIs. Across both datives, the average ∆ is significantly greater for NABA verbs than it is for NANA verbs (p < .01 for both). observed in the PP construction (i.e., LMs assign higher probability in general to PP constructions than DO constructions). Due t… view at source ↗
Figure 3
Figure 3. Figure 3: Asymmetric cross-dative generalization in our LM learners. Average log probability per token of [PITH_FULL_IMAGE:figures/full_fig_p018_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Average generalization set log probabilities per token for DO generalization instances for DO [PITH_FULL_IMAGE:figures/full_fig_p020_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Average log probabilities per token assigned to the generalization set across theme animacy con [PITH_FULL_IMAGE:figures/full_fig_p021_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: A visual depiction of the three-way interaction effects between pronominality, animacy, and def [PITH_FULL_IMAGE:figures/full_fig_p025_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: A visual depiction of the three-way interaction effects between pronominality, animacy, and def [PITH_FULL_IMAGE:figures/full_fig_p026_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Average Verbhood ∆s and Accuracies across different adaptation dative types. Note that there is no upper/lower-bound for Verbhood ∆, since they are differences in log probabilities, and can theoretically be infinite in either direction. difference measure with positive values signifying greater verbhood, an LM that has made the right category￾based inference should show Verbhood ∆ values that are substanti… view at source ↗
read the original abstract

Neural language models (LMs) have been shown to capture complex linguistic patterns, yet their utility in understanding human language and more broadly, human cognition, remains debated. While existing work in this area often evaluates human-machine alignment, few studies attempt to translate findings from this enterprise into novel insights about humans. To this end, we propose a systematic framework for hypothesis generation that uses LMs to simulate outcomes of experiments that do not yet exist in the literature. We instantiate this framework in the context of a specific research question in child language development: dative verb acquisition and cross-structural generalization. Through this instantiation, we derive novel, untested hypotheses: the alignment between argument ordering and discourse prominence features of exposure contexts modulates how children generalize new verbs to unobserved structures. Additionally, we also design a set of experiments that can test these hypotheses in the lab with children. This work contributes both a domain-general framework for systematic hypothesis generation via simulated learners and domain-specific, lab-testable hypotheses for child language acquisition research.

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 / 1 minor

Summary. The paper proposes a systematic framework that uses language models to simulate the outcomes of experiments that have not yet been run, with the goal of generating novel, testable hypotheses about human cognition. It instantiates the framework in the domain of child dative verb acquisition and cross-structural generalization, derives the hypothesis that alignment between argument ordering and discourse prominence in exposure contexts modulates generalization to unobserved structures, and outlines a set of corresponding child experiments.

Significance. If the framework can be shown to produce hypotheses that are both novel and grounded in faithful simulation of known human patterns, it would provide a domain-general method for accelerating hypothesis generation in cognitive science and language acquisition research. The concrete experimental designs offered are a practical contribution that could be directly implemented.

major comments (2)
  1. [Instantiation section] Instantiation section (framework application to dative acquisition): The central claim that the LM-derived hypotheses are valid outputs of the framework rather than model artifacts requires demonstrating that the LM reproduces established patterns from existing child dative acquisition studies (e.g., verb-class effects or dative alternation preferences). No such validation, comparison to published child data, or error analysis is reported, leaving the mapping from LM outputs to human generalization unsupported.
  2. [Hypothesis derivation step] Hypothesis derivation step: The abstract and described instantiation state that novel hypotheses were obtained via simulation, yet no model outputs, simulation parameters, or quantitative results from the LM runs are supplied. This absence makes it impossible to evaluate whether the reported hypothesis about argument ordering and discourse prominence follows from the simulation or from other sources.
minor comments (1)
  1. The abstract refers to 'a set of experiments that can test these hypotheses' but provides no details on design, stimuli, or predicted outcomes; moving a brief outline to the main text would improve clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments, which highlight important aspects for strengthening the presentation of our framework. We respond to each major comment below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Instantiation section] Instantiation section (framework application to dative acquisition): The central claim that the LM-derived hypotheses are valid outputs of the framework rather than model artifacts requires demonstrating that the LM reproduces established patterns from existing child dative acquisition studies (e.g., verb-class effects or dative alternation preferences). No such validation, comparison to published child data, or error analysis is reported, leaving the mapping from LM outputs to human generalization unsupported.

    Authors: We agree with the referee that demonstrating the LM's fidelity to known human patterns is essential to support the claim that the derived hypotheses are valid outputs of the framework. The current version of the manuscript focuses on the novel hypotheses and experimental designs but does not include this validation step. In the revision, we will add a validation subsection that applies the framework to existing child dative acquisition studies, comparing LM outputs to published data on verb-class effects and dative alternation preferences, along with quantitative metrics and error analysis. revision: yes

  2. Referee: [Hypothesis derivation step] Hypothesis derivation step: The abstract and described instantiation state that novel hypotheses were obtained via simulation, yet no model outputs, simulation parameters, or quantitative results from the LM runs are supplied. This absence makes it impossible to evaluate whether the reported hypothesis about argument ordering and discourse prominence follows from the simulation or from other sources.

    Authors: We acknowledge that the manuscript does not provide the specific LM outputs, simulation parameters, or quantitative results, which limits the ability to trace the hypothesis derivation. This was an oversight in the presentation. We will revise by including a detailed description of the simulation setup, example model outputs, and the step-by-step derivation process in a new section or appendix, ensuring transparency in how the hypothesis about argument ordering and discourse prominence was obtained from the simulations. revision: yes

Circularity Check

0 steps flagged

No circularity: forward simulation framework with no reduction to inputs by construction

full rationale

The paper proposes a domain-general framework that uses LMs to simulate outcomes of non-existent experiments in order to generate novel hypotheses about child dative generalization. The abstract and described instantiation contain no equations, fitted parameters, or self-citations that reduce the derived hypotheses to the LM training data or prior results by construction. The central output (alignment between argument ordering and discourse prominence modulating generalization) is presented as an emergent prediction from the simulation rather than a renaming or refit of known patterns. No uniqueness theorems, ansatzes smuggled via citation, or self-definitional loops are invoked. The derivation chain remains self-contained as a methodological proposal.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The framework depends on one central untested premise about the fidelity of LM simulation to child behavior; no free parameters or new entities are introduced in the abstract.

axioms (1)
  • domain assumption Language models can simulate outcomes of child language experiments with sufficient accuracy to generate valid novel hypotheses about human generalization.
    This premise is required for the simulation step to produce usable hypotheses but receives no supporting evidence or validation in the abstract.

pith-pipeline@v0.9.0 · 5698 in / 1350 out tokens · 38035 ms · 2026-05-23T21:51:50.941696+00:00 · methodology

discussion (0)

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

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Collocational bootstrapping: A hypothesis about the learning of subject-verb agreement in humans and neural networks

    cs.CL 2026-05 unverdicted novelty 6.0

    Collocational bootstrapping via co-occurrence regularities enables neural networks to learn subject-verb agreement robustly when input variability matches child-directed speech, indicating it as a viable acquisition strategy.

  2. Filling in the Mechanisms: How do LMs Learn Filler-Gap Dependencies under Developmental Constraints?

    cs.CL 2026-04 unverdicted novelty 6.0

    LMs develop shared yet item-sensitive filler-gap mechanisms with limited data but require substantially more data than humans to match generalizations.

Reference graph

Works this paper leans on

123 extracted references · 123 canonical work pages · cited by 2 Pith papers · 3 internal anchors

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

    work page

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

    work page

  3. [3]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

    work page

  4. [4]

    , year 1999

    author Aissen, J. , year 1999 . title Markedness and Subject Choice in Optimality Theory . journal Natural Language & Linguistic Theory volume 17 , pages 673--711

    work page 1999

  5. [5]

    , year 2003

    author Aissen, J. , year 2003 . title Differential Object Marking: Iconicity vs. Economy . journal Natural Language & Linguistic Theory volume 21 , pages 435--483

    work page 2003

  6. [6]

    , year 2020

    author Ambridge, B. , year 2020 . title Abstractions made of exemplars or ‘You’re all right, and I’ve changed my mind’: Response to commentators . journal First Language volume 40 , pages 640--659

    work page 2020

  7. [7]

    , author Bidgood, A

    author Ambridge, B. , author Bidgood, A. , author Twomey, K.E. , author Pine, J.M. , author Rowland, C.F. , author Freudenthal, D. , year 2015 . title Preemption versus entrenchment: Towards a construction-general solution to the problem of the retreat from verb argument structure overgeneralization . journal PloS one volume 10 , pages e0123723

    work page 2015

  8. [8]

    , author Pine, J.M

    author Ambridge, B. , author Pine, J.M. , author Rowland, C.F. , author Chang, F. , year 2012 . title The roles of verb semantics, entrenchment, and morphophonology in the retreat from dative argument-structure overgeneralization errors . journal Language volume 88 , pages 45--81

    work page 2012

  9. [9]

    , author Pine, J.M

    author Ambridge, B. , author Pine, J.M. , author Rowland, C.F. , author Young, C.R. , year 2008 . title The effect of verb semantic class and verb frequency (entrenchment) on children’s and adults’ graded judgements of argument-structure overgeneralization errors . journal Cognition volume 106 , pages 87--129

    work page 2008

  10. [10]

    , author Lao, S.Y.C

    author Arnold, J.E. , author Lao, S.Y.C. , year 2008 . title Put in last position something previously unmentioned: Word order effects on referential expectancy and reference comprehension . journal Language and Cognitive Processes volume 23 , pages 282--295

    work page 2008

  11. [11]

    , author Losongco, A

    author Arnold, J.E. , author Losongco, A. , author Wasow, T. , author Ginstrom, R. , year 2000 . title Heaviness vs. newness: The effects of structural complexity and discourse status on constituent ordering . journal Language volume 76 , pages 28--55

    work page 2000

  12. [12]

    , year 1976

    author Aronoff, M. , year 1976 . title Word formation in generative grammar . journal Linguistic Inquiry Monographs , pages 1--134

    work page 1976

  13. [13]

    , year 2017

    author Arunachalam, S. , year 2017 . title Preschoolers' Acquisition of Novel Verbs in the Double Object Dative . journal Cognitive science volume 41 , pages 831--854

    work page 2017

  14. [14]

    , year 1979

    author Baker, C.L. , year 1979 . title Syntactic theory and the projection problem . journal Linguistic Inquiry volume 10 , pages 533--581

    work page 1979

  15. [15]

    , year 2020

    author Baroni, M. , year 2020 . title Linguistic generalization and compositionality in modern artificial neural networks . journal Philosophical Transactions of the Royal Society B volume 375 , pages 20190307

    work page 2020

  16. [16]

    , M \"a chler, M

    author Bates, D. , author M \"a chler, M. , author Bolker, B. , author Walker, S. , year 2015 . title Fitting Linear Mixed-Effects Models Using lme4 . journal Journal of Statistical Software volume 67 , pages 1--48 . :10.18637/jss.v067.i01

    work page doi:10.18637/jss.v067.i01 2015

  17. [17]

    , year 2011

    author Beavers, J. , year 2011 . title An Aspectual Analysis of Ditransitive Verbs of Caused Possession in English . journal Journal of Semantics volume 28 , pages 1--54

    work page 2011

  18. [18]

    , year 1909

    author Behaghel, O. , year 1909 . title Beziehungen zwischen umfang und reihenfolge von satzgliedern . journal Indogermanische Forschungen volume 25 , pages 110

    work page 1909

  19. [19]

    , author Albrecht, J.E

    author Birch, S.L. , author Albrecht, J.E. , author Myers, J.L. , year 2000 . title Syntactic focusing structures influence discourse processing . journal Discourse Processes volume 30 , pages 285--304

    work page 2000

  20. [20]

    , author Garnsey, S.M

    author Birch, S.L. , author Garnsey, S.M. , year 1995 . title The effect of focus on memory for words in sentences . journal Journal of Memory and Language volume 34 , pages 232--267

    work page 1995

  21. [21]

    , year 1986

    author Bock, J.K. , year 1986 . title Syntactic persistence in language production . journal Cognitive Psychology volume 18 , pages 355--387

    work page 1986

  22. [22]

    , author Goldberg, A.E

    author Boyd, J.K. , author Goldberg, A.E. , year 2011 . title Learning what not to say: The role of statistical preemption and categorization in a-adjective production . journal Language volume 87 , pages 55--83

    work page 2011

  23. [23]

    , author Cueni, A

    author Bresnan, J. , author Cueni, A. , author Nikitina, T. , author Baayen, R.H. , year 2007 . title Predicting the dative alternation , in: booktitle Cognitive foundations of interpretation . publisher KNAW , pp. pages 69--94

    work page 2007

  24. [24]

    , author Nikitina, T

    author Bresnan, J. , author Nikitina, T. , year 2009 . title The Gradience of the Dative Alternation . journal Reality exploration and discovery: Pattern interaction in language and life , pages 161--184

    work page 2009

  25. [25]

    , author Tomasello, M

    author Brooks, P.J. , author Tomasello, M. , year 1999 . title How children constrain their argument structure constructions . journal Language volume 75 , pages 720--738

    work page 1999

  26. [26]

    , author Embley Emonds, J

    author Citko, B. , author Embley Emonds, J. , author Whitney, R. , year 2017 . title Double Object Constructions . journal The Wiley Blackwell Companion to Syntax, Second Edition , pages 1--46

    work page 2017

  27. [27]

    , author Clark, E.V

    author Clark, H.H. , author Clark, E.V. , year 1977 . title Psychology and language: an introduction to psycholinguistics . publisher Harcourt Brace Jovanovich New York

    work page 1977

  28. [28]

    , year 1995

    author Collins, P. , year 1995 . title The indirect object construction in english: an informational approach . journal Linguistics volume 33 , pages 35--49

    work page 1995

  29. [29]

    , year 2019

    author Conwell, E. , year 2019 . title The effects of the pronoun me on dative comprehension . journal Journal of Child Language volume 46 , pages 1127--1141

    work page 2019

  30. [30]

    , author Demuth, K

    author Conwell, E. , author Demuth, K. , year 2007 . title Early syntactic productivity: Evidence from dative shift . journal Cognition volume 103 , pages 163--179

    work page 2007

  31. [31]

    , author O’Donnell, T.J

    author Conwell, E. , author O’Donnell, T.J. , author Snedeker, J. , year 2011 . title Frozen chunks and generalized representations: The case of the english dative alternation , in: booktitle Proceedings of the 35th Boston University conference on language development , organization Citeseer . pp. pages 132--144

    work page 2011

  32. [32]

    , year 2009

    author Coppock, E. , year 2009 . title The logical and empirical foundations of Baker's paradox . Ph.D. thesis. Stanford University

    work page 2009

  33. [33]

    , author Grimm, S

    author De Marneffe, M.C. , author Grimm, S. , author Arnon, I. , author Kirby, S. , author Bresnan, J. , year 2012 . title A statistical model of the grammatical choices in child production of dative sentences . journal Language and cognitive processes volume 27 , pages 25--61

    work page 2012

  34. [34]

    , author Manning, C.D

    author De Marneffe, M.C. , author Manning, C.D. , author Nivre, J. , author Zeman, D. , year 2021 . title Universal dependencies . journal Computational linguistics volume 47 , pages 255--308

    work page 2021

  35. [35]

    , author Chang, M.W

    author Devlin, J. , author Chang, M.W. , author Lee, K. , author Toutanova, K. , year 2019 . title BERT : Pre-training of deep bidirectional T ransformers for language understanding , in: booktitle NAACL 2019 , pp. pages 4171--4186

    work page 2019

  36. [36]

    , year 2018

    author Dupoux, E. , year 2018 . title Cognitive science in the era of artificial intelligence: A roadmap for reverse-engineering the infant language-learner . journal Cognition volume 173 , pages 43--59

    work page 2018

  37. [37]

    , year 1990

    author Elman, J.L. , year 1990 . title Finding structure in time . journal Cognitive science volume 14 , pages 179--211

    work page 1990

  38. [38]

    , author Pylyshyn, Z.W

    author Fodor, J.A. , author Pylyshyn, Z.W. , year 1988 . title Connectionism and cognitive architecture: A critical analysis . journal Cognition volume 28 , pages 3--71

    work page 1988

  39. [39]

    , author McElree, B

    author Foraker, S. , author McElree, B. , year 2007 . title The role of prominence in pronoun resolution: Active versus passive representations . journal Journal of Memory and Language volume 56 , pages 357--383

    work page 2007

  40. [40]

    , year 2023

    author Frank, M.C. , year 2023 . title Bridging the data gap between children and large language models . journal Trends in Cognitive Sciences volume 27 , pages 990--992

    work page 2023

  41. [41]

    , year 1995

    author Goldberg, A.E. , year 1995 . title Constructions: A construction grammar approach to argument structure . publisher University of Chicago Press

    work page 1995

  42. [42]

    , year 2011

    author Goldberg, A.E. , year 2011 . title Corpus evidence of the viability of statistical preemption . journal Cognitive Linguistics volume 22 , pages 131--153

    work page 2011

  43. [43]

    , year 2016

    author Goldberg, A.E. , year 2016 . title Partial productivity of linguistic constructions: Dynamic categorization and statistical preemption . journal Language and cognition volume 8 , pages 369--390

    work page 2016

  44. [44]

    , author Zada, Z

    author Goldstein, A. , author Zada, Z. , author Buchnik, E. , author Schain, M. , author Price, A. , author Aubrey, B. , author Nastase, S.A. , author Feder, A. , author Emanuel, D. , author Cohen, A. , et al., year 2022 . title Shared computational principles for language processing in humans and deep language models . journal Nature Neuroscience volume ...

    work page 2022

  45. [45]

    , author Bicknell, K

    author Goodkind, A. , author Bicknell, K. , year 2018 . title Predictive power of word surprisal for reading times is a linear function of language model quality , in: editor Sayeed, A. , editor Jacobs, C. , editor Linzen, T. , editor van Schijndel, M. (Eds.), booktitle Proceedings of the 8th Workshop on Cognitive Modeling and Computational Linguistics ( ...

    work page doi:10.18653/v1/w18-0102 2018

  46. [46]

    , author Pinker, S

    author Gropen, J. , author Pinker, S. , author Hollander, M. , author Goldberg, R. , author Wilson, R. , year 1989 . title The learnability and acquisition of the dative alternation in english . journal Language volume 65 , pages 203--257

    work page 1989

  47. [47]

    , author Martin, A.E

    author Guest, O. , author Martin, A.E. , year 2023 . title On logical inference over brains, behaviour, and artificial neural networks . journal Computational Brain & Behavior volume 6 , pages 213--227

    work page 2023

  48. [48]

    , year 1988

    author Gundel, J.K. , year 1988 . title Universals of topic-comment structure . journal Studies in syntactic typology volume 17 , pages 209--239

    work page 1988

  49. [49]

    , year 1997

    author Hadley, R.F. , year 1997 . title Cognition, systematicity and nomic necessity . journal Mind & language volume 12 , pages 137--153

    work page 1997

  50. [50]

    , author Oaksford, M

    author Hahn, U. , author Oaksford, M. , year 2008 . title Inference from absence in language and thought . journal The probabilistic mind: Prospects for Bayesian cognitive science , pages 121--42

    work page 2008

  51. [51]

    , author Risley, T.R

    author Hart, B. , author Risley, T.R. , year 2003 . title The early catastrophe: The 30 million word gap by age 3 . journal American educator volume 27 , pages 4--9

    work page 2003

  52. [52]

    , author Yamakoshi, T

    author Hawkins, R. , author Yamakoshi, T. , author Griffiths, T. , author Goldberg, A. , year 2020 . title Investigating representations of verb bias in neural language models , in: editor Webber, B. , editor Cohn, T. , editor He, Y. , editor Liu, Y. (Eds.), booktitle Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (...

    work page doi:10.18653/v1/2020.emnlp-main.376 2020

  53. [53]

    , year 2021

    author Hewitt, J. , year 2021 . title Initializing new word embeddings for pretrained language models . https:/nlp.stanford.edu/ johnhew//vocab-expansion.html

    work page 2021

  54. [54]

    , author Schmidhuber, J

    author Hochreiter, S. , author Schmidhuber, J. , year 1997 . title Long short-term memory . journal Neural computation volume 9 , pages 1735--1780

    work page 1997

  55. [55]

    , author Weissenborn, J

    author H \"o hle, B. , author Weissenborn, J. , author Kiefer, D. , author Schulz, A. , author Schmitz, M. , year 2004 . title Functional Elements in Infants’ Speech Processing: The Role of Determiners in the Syntactic Categorization of Lexical Elements . journal Infancy volume 5 , pages 341--353

    work page 2004

  56. [56]

    , author Montani, I

    author Honnibal, M. , author Montani, I. , author Van Landeghem, S. , author Boyd, A. , year 2020 . title spaCy : Industrial-strength natural language processing in python . :10.5281/zenodo.1212303

    work page doi:10.5281/zenodo.1212303 2020

  57. [57]

    , author Sulem, E

    author Huebner, P.A. , author Sulem, E. , author Cynthia, F. , author Roth, D. , year 2021 . title B aby BERT a: Learning more grammar with small-scale child-directed language , in: editor Bisazza, A. , editor Abend, O. (Eds.), booktitle Proceedings of the 25th Conference on Computational Natural Language Learning , publisher Association for Computational...

    work page doi:10.18653/v1/2021.conll-1.49 2021

  58. [58]

    , author Willits, J.A

    author Huebner, P.A. , author Willits, J.A. , year 2018 . title Structured semantic knowledge can emerge automatically from predicting word sequences in child-directed speech . journal Frontiers in Psychology volume 9 , pages 133

    work page 2018

  59. [59]

    , author Willits, J.A

    author Huebner, P.A. , author Willits, J.A. , year 2021 . title Using lexical context to discover the noun category: Younger children have it easier , in: booktitle Psychology of learning and motivation . publisher Elsevier . volume volume 75 , pp. pages 279--331

    work page 2021

  60. [60]

    , year 1990

    author Jackendoff, R. , year 1990 . title On larson's treatment of the double object construction . journal Linguistic inquiry volume 21 , pages 427--456

    work page 1990

  61. [61]

    , author Levy, R

    author Jara-Ettinger, J. , author Levy, R. , author Sakel, J. , author Huanca, T. , author Gibson, E. , year 2022 . title The origins of the shape bias: Evidence from the tsimane’. journal Journal of Experimental Psychology: General volume 151 , pages 2437

    work page 2022

  62. [62]

    , author Zuidema, W

    author Jumelet, J. , author Zuidema, W. , author Sinclair, A. , year 2024 . title Do language models exhibit human-like structural priming effects? journal arXiv:2406.04847

    work page arXiv 2024

  63. [63]

    , author Choi, J

    author Kember, H. , author Choi, J. , author Yu, J. , author Cutler, A. , year 2021 . title The Processing of Linguistic Prominence . journal Language and Speech volume 64 , pages 413--436

    work page 2021

  64. [64]

    , author Linzen, T

    author Kim, N. , author Linzen, T. , year 2020 . title COGS : A compositional generalization challenge based on semantic interpretation , in: editor Webber, B. , editor Cohn, T. , editor He, Y. , editor Liu, Y. (Eds.), booktitle Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) , publisher Association for Compu...

    work page doi:10.18653/v1/2020.emnlp-main.731 2020

  65. [65]

    , author Smolensky, P

    author Kim, N. , author Smolensky, P. , year 2021 . title Testing for grammatical category abstraction in neural language models . journal Proceedings of the Society for Computation in Linguistics volume 4 , pages 467--470

    work page 2021

  66. [66]

    , author Smolensky, P

    author Kim, N. , author Smolensky, P. , year 2024 . title Structural generalization of modification in adult learners of an artificial language , in: booktitle Proceedings of the Annual Meeting of the Cognitive Science Society , pp. pages 856--863

    work page 2024

  67. [67]

    , year 1982

    author Kiparsky, P. , year 1982 . title Lexical phonology and morphology . journal Linguistics in the Morning Calm

    work page 1982

  68. [68]

    , author Payne, S

    author Kodner, J. , author Payne, S. , author Heinz, J. , year 2023 . title Why linguistics will thrive in the 21st century: A reply to Piantadosi (2023) . journal arXiv:2308.03228

    work page arXiv 2023

  69. [69]

    , author Hupkes, D

    author Lakretz, Y. , author Hupkes, D. , author Vergallito, A. , author Marelli, M. , author Baroni, M. , author Dehaene, S. , year 2021 . title Mechanisms for handling nested dependencies in neural-network language models and humans . journal Cognition volume 213 , pages 104699

    work page 2021

  70. [70]

    , year 2023

    author Lenth, R.V. , year 2023 . title emmeans: Estimated Marginal Means, aka Least-Squares Means . https://CRAN.R-project.org/package=emmeans. note r package version 1.9.0

    work page 2023

  71. [71]

    , year 1993

    author Levin, B. , year 1993 . title English verb classes and alternations: A preliminary investigation . publisher University of Chicago press

    work page 1993

  72. [72]

    RoBERTa: A Robustly Optimized BERT Pretraining Approach

    author Liu, Y. , author Ott, M. , author Goyal, N. , author Du, J. , author Joshi, M. , author Chen, D. , author Levy, O. , author Lewis, M. , author Zettlemoyer, L. , author Stoyanov, V. , year 2019 . title RoBERTa : A robustly optimized bert pretraining approach . journal arXiv:1907.11692

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  73. [73]

    , year 2000

    author MacWhinney, B. , year 2000 . title The CHILDES project: Tools for analyzing talk, Volume I: Transcription format and programs . publisher Psychology Press

    work page 2000

  74. [74]

    , year 1988

    author Massaro, D.W. , year 1988 . title Some criticisms of connectionist models of human performance . journal Journal of Memory and Language volume 27 , pages 213--234

    work page 1988

  75. [75]

    , year 1988

    author McClelland, J.L. , year 1988 . title Connectionist models and psychological evidence . journal Journal of Memory and Language volume 27 , pages 107--123

    work page 1988

  76. [76]

    , year 1991

    author McCloskey, M. , year 1991 . title Networks and Theories: The Place of Connectionism in Cognitive Science . journal Psychological science volume 2 , pages 387--395

    work page 1991

  77. [77]

    , author Russin, J

    author McGrath, S. , author Russin, J. , author Pavlick, E. , author Feiman, R. , year 2023 . title How can deep neural networks inform theory in psychological science? osf.io/preprints/psyarxiv/j5ckf, :10.31234/osf.io/j5ckf

    work page doi:10.31234/osf.io/j5ckf 2023

  78. [78]

    , year 2022

    author Misra, K. , year 2022 . title minicons: Enabling flexible behavioral and representational analyses of transformer language models . journal arXiv:2203.13112

    work page arXiv 2022

  79. [79]

    , author Kim, N

    author Misra, K. , author Kim, N. , year 2023 . title Abstraction via exemplars? A representational case study on lexical category inference in BERT , in: booktitle BUCLD 48: Proceedings of the 48th annual Boston University Conference on Language Development , address Boston, USA

    work page 2023

  80. [80]

    , author Mahowald, K

    author Misra, K. , author Mahowald, K. , year 2024 . title Language Models Learn Rare Phenomena from Less Rare Phenomena: The Case of the Missing AANNs . journal arXiv:2403.19827

    work page arXiv 2024

Showing first 80 references.