pith. machine review for the scientific record. sign in

arxiv: 2605.08889 · v1 · submitted 2026-05-09 · 💻 cs.LG · cs.CL· cs.DL

Recognition: no theorem link

Machine Learning Research Has Outpaced Its Communication Norms and NeurIPS Should Act

Ajay Mandyam Rangarajan , Jeyashree Krishnan
Authors on Pith no claims yet

Pith reviewed 2026-05-12 01:26 UTC · model grok-4.3

classification 💻 cs.LG cs.CLcs.DL
keywords readability metricsNeurIPSmachine learning papersacronym densityresearch communicationcitation impactwriting standardsarXiv analysis
0
0 comments X

The pith

NeurIPS papers have become harder to read over decades as acronym use surged, and the conference should adopt explicit writing standards to reverse the trend.

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

The paper examines millions of papers across arXiv, NeurIPS, and PubMed and finds that NeurIPS abstracts score worse on every classical readability metric since 1987, with Flesch scores dropping and sensational language rising. Acronym density in titles has increased sharply, and most acronyms see little reuse compared to broader science. Readable NeurIPS papers receive more citations, while volume has grown fiftyfold, prompting the authors to argue that NeurIPS must act to keep communication accessible for human readers. They outline seven concrete standards to pilot in 2027, including acronym budgets and plain-language summaries. A sympathetic reader cares because declining readability risks fragmenting knowledge in a field that depends on cumulative progress.

Core claim

Machine learning research has grown exponentially while its communication norms have stayed the same or worsened, as shown by NeurIPS abstracts becoming harder to read on classical metrics, rising acronym density from 0.33 to 3.21 per 100 words, and increased sensational language. About 89 percent of acronyms are used fewer than ten times, and more readable papers tend to attract more citations. With volume up roughly fifty times since 1987, the authors conclude that NeurIPS should pilot seven standards starting in 2027 to prioritize human readers over surface statistics or LLM judgments.

What carries the argument

Longitudinal analysis of classical readability scores, acronym density, and citation correlations across 24,772 NeurIPS papers from 1987 to 2024, which documents the divergence from science-wide baselines and justifies enforceable standards.

Load-bearing premise

Classical readability metrics and the observed citation link accurately reflect human comprehension, and that imposing the proposed standards will improve actual understanding and impact rather than just changing surface statistics.

What would settle it

A controlled experiment measuring human readers' comprehension, retention, and ability to build on ideas from papers written under the proposed standards versus current norms, which would falsify the claim if scores improve but real understanding does not.

Figures

Figures reproduced from arXiv: 2605.08889 by Ajay Mandyam Rangarajan, Jeyashree Krishnan.

Figure 1
Figure 1. Figure 1: End to end workflow. Three data sources feed a canonical Pydantic-validated schema. Five per-paper metric families are computed on titles and abstracts separately, then aggregated per venue-year and per arXiv primary category. The outputs feed the empirical evidence (Sections 2 through 6) and the proposed policy standards (Section 7). Pipeline details appear in Appendix A. 2 [PITH_FULL_IMAGE:figures/full_… view at source ↗
Figure 2
Figure 2. Figure 2: shows Flesch Reading Ease on NeurIPS, arXiv ML, arXiv non-ML, and PubMed from 1987 to 2025. NeurIPS falls from approximately 24 in 1987 to approximately 13 in 2024, a sustained decline across 37 years. The arXiv ML aggregate follows a similar downward trajectory. The arXiv non-ML baseline declines more slowly, confirming that the accelerated decline is specific to ML rather than a universal property of all… view at source ↗
Figure 3
Figure 3. Figure 3: shows total sensational language count per 100 abstract words for NeurIPS, arXiv ML, arXiv non-ML, and PubMed. The NeurIPS total rate rises by approximately 50 percent between 2015 and 2024 (from 1.10 to 1.69 per 100 words). The arXiv non-ML baseline rises more slowly over the same period. The divergence between ML and non-ML venues accelerates after 2022, consistent with the hypothesis that instruction-tu… view at source ↗
Figure 4
Figure 4. Figure 4: shows the standardized scores for all six models under each of the three prompt variants. All six models agree across all three prompts: LLM-judged readability is approximately flat or gently improving over four decades, with a notable upward shift after 2022. This result stands in direct contrast to every classical readability metric ( [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Acronym density in titles (left) and abstracts (right), 1987 to 2025. Lower values indicate fewer acronyms per word (↓ better). Dashed black horizontal line marks the Barnett and Doubleday science-wide peak of 2.40 per 100 words (2019). NeurIPS and arXiv ML diverge from the non-ML control around 2017 on titles and around 2020 on abstracts. 5 Citations and the Fragmentation Paradox Citation count is the fie… view at source ↗
Figure 6
Figure 6. Figure 6: Mean bibliography length per NeurIPS paper, 1987–2024. Reference list length grew from roughly 11 entries per paper in 1987 to approximately 61 entries in 2022, a fivefold to sixfold increase over 37 years. Mean bibliography length declined slightly in 2023 and 2024. The cause of this reversal is not yet established. Vertical dashed line marks the late-2022 availability of instruction￾tuned writing assista… view at source ↗
Figure 7
Figure 7. Figure 7: Papers per year, 1987 to 2024. arXiv ML submissions grew by roughly three orders of magnitude between 1992 and 2024. NeurIPS grew by a factor of roughly 50 over the same window, from fewer than 100 accepted papers in 1987 to over 4,500 in 2024. PubMed and arXiv non-ML show sustained but slower growth over the same period. The volume increase is the structural backdrop against which the readability and acro… view at source ↗
Figure 8
Figure 8. Figure 8: Classical readability metrics, part A: Flesch family and grade-level formulas, 1987– 2025. Arrows indicate the direction of easier reading (↑ = higher is better; ↓ = lower is better). Every grade-level metric rises on NeurIPS and arXiv ML after 2015 and accelerates after 2020; Flesch Reading Ease drops correspondingly. Vertical dashed lines mark the late-2022 availability of instruction-tuned writing assis… view at source ↗
Figure 9
Figure 9. Figure 9: Classical readability metrics, part B: Linsear Write, LIX, RIX, FORCAST, Powers– Sumner–Kearl, sentence length, and syllables per word, 1987–2025. Arrows indicate the direction of easier reading (↓ = lower is better). Vertical dashed lines mark the late-2022 availability of instruction-tuned writing assistants. Part A with the Flesch family and the first eight grade-level formulas appears in [PITH_FULL_IM… view at source ↗
Figure 10
Figure 10. Figure 10: Hohmann writing style metrics on NeurIPS, arXiv ML, arXiv non-ML, and PubMed, 1987–2025. PubMed data are available only for the four metrics computed by lightweight regex and word-level rules (sentence length, signposting, hedging, and active narration). The remaining metrics depend on full spaCy dependency parsing and were not run on the 24.5 million paper PubMed corpus because of compute cost; the corre… view at source ↗
Figure 11
Figure 11. Figure 11: All 10 sensational language categories per 100 abstract words, 1987–2025. Novelty and scale rise sharply on NeurIPS and arXiv ML after 2015 (roughly 2.2 times and 3.0 times their 2015 levels by 2024); the total rate and rigor rise more modestly. Vertical dashed lines mark the late-2022 availability of instruction-tuned writing assistants. 20 [PITH_FULL_IMAGE:figures/full_fig_p020_11.png] view at source ↗
read the original abstract

Machine learning research has grown exponentially while its communication norms have not. We argue NeurIPS should adopt explicit, measurable writing standards. We analyze 2.8 million arXiv papers (1991-2025), 24,772 NeurIPS papers (1987-2024), and 24.5 million PubMed papers (1990-2025), applying classical readability scores, the Hohmann writing style suite (including sensational language), acronym density and reuse, an LLM as judge readability protocol, and citations from OpenAlex and Semantic Scholar. Four patterns emerge. First, NeurIPS abstracts score harder to read on every classical readability metric: Flesch Reading Ease falls from about 24 in 1987 to 13 in 2024, and sensational language rises by about 50 percent in NeurIPS abstracts between 2015 and 2024. Second, acronym density in NeurIPS titles has grown from 0.33 per 100 words in 1987 to 3.21 in 2024, and about 89 percent of NeurIPS acronyms are used fewer than ten times, ten points above the science-wide baseline. Third, more readable NeurIPS papers tend to receive more citations, suggesting readability and impact are correlated and that less readable papers risk remaining fragmented. LLM as judge scores rate NeurIPS abstracts as roughly stable from 1987 to 2022, with early signs of improvement thereafter, a pattern that disagrees with every classical readability metric and raises a design question for enforcement: is the target reader a human or an LLM? Lastly, NeurIPS volume has grown roughly 50-fold between 1987 and 2024. Assuming the goal is to optimise for human readers, we propose seven standards NeurIPS could pilot at NeurIPS 2027: an acronym budget with a venue-approved term list, a human readability threshold, stricter citation standards, standalone visual elements, a plain language summary, a pre-registered acronym glossary, and open source audit tooling.

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

3 major / 2 minor

Summary. The paper analyzes trends in 2.8M arXiv papers, 24k NeurIPS papers, and 24.5M PubMed papers using classical readability metrics (e.g., Flesch Reading Ease declining from ~24 in 1987 to 13 in 2024 for NeurIPS abstracts), acronym density (rising from 0.33 to 3.21 per 100 words in NeurIPS titles), sensationalism, LLM-as-judge protocols, and citation data. It reports that NeurIPS communication has become harder to read and more acronym-heavy than baselines, that more readable papers receive more citations, that LLM scores conflict with classical metrics, and that NeurIPS volume has grown 50-fold; it proposes seven pilot standards for NeurIPS 2027 including acronym budgets, readability thresholds, plain-language summaries, and audit tooling.

Significance. If the observed trends are robust and the metrics validly proxy human comprehension, the work provides a large-scale empirical basis for concerns about accessibility in ML research and offers concrete, measurable policy proposals that could influence conference norms. The scale of the multi-corpus analysis (millions of papers, multiple metric families, cross-domain comparisons) and the explicit contrast between classical and LLM-based scores are strengths that could stimulate further empirical work on scientific communication.

major comments (3)
  1. [citation correlation analysis] The citation-readability correlation (reported in the abstract and the third pattern) is presented without error bars, without details on sampling or pre-processing of the 24,772 NeurIPS papers, and without controls for confounders such as subfield, author prestige, paper length, or novelty; this leaves open whether the positive link is causal or confounded and weakens the claim that less readable papers risk remaining fragmented.
  2. [comparison of metric families] Classical readability metrics (Flesch, etc.) and acronym counts are applied directly to domain-dense ML abstracts containing equations and novel jargon, yet the paper notes that its own LLM-as-judge protocol shows stability or slight improvement (disagreeing with every classical metric) without human-subject validation (expert ratings or comprehension quizzes) to establish that the metric declines track actual difficulty for ML readers.
  3. [proposed standards] The proposal of seven standards (acronym budget, human readability threshold, etc.) rests on the assumption that enforcing them will improve human understanding or impact, but the manuscript provides no pilot data, prior evidence from other fields, or discussion of potential unintended effects (e.g., on technical precision) to support that the surface-statistic changes will translate to better comprehension.
minor comments (2)
  1. [methods] The abstract and text would benefit from explicit statements of the exact sampling criteria and preprocessing steps used for the NeurIPS and arXiv corpora to allow replication.
  2. [results figures] Figure or table captions for the trend plots should include the number of papers per year and any confidence intervals to improve clarity.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive and detailed feedback. Their comments help clarify how to strengthen the statistical presentation, metric interpretation, and framing of the policy proposals. We respond to each major comment below and indicate the revisions that will be made to the manuscript.

read point-by-point responses

  1. Referee: The citation-readability correlation (reported in the abstract and the third pattern) is presented without error bars, without details on sampling or pre-processing of the 24,772 NeurIPS papers, and without controls for confounders such as subfield, author prestige, paper length, or novelty; this leaves open whether the positive link is causal or confounded and weakens the claim that less readable papers risk remaining fragmented.

    Authors: We agree that the correlation analysis requires greater statistical transparency. In the revised manuscript we will add error bars to the citation-readability plots, provide complete details on the sampling frame (all 24,772 NeurIPS papers 1987–2024 with citation data from OpenAlex) and pre-processing steps (abstract extraction, cleaning, and tokenization), and include multivariate regressions controlling for paper length, year, and subfield. Author prestige and novelty are difficult to operationalize at this scale, so we will explicitly discuss them as unmeasured confounders and rephrase the fragmentation claim as an observational association that motivates further investigation rather than a causal assertion. revision: yes

  2. Referee: Classical readability metrics (Flesch, etc.) and acronym counts are applied directly to domain-dense ML abstracts containing equations and novel jargon, yet the paper notes that its own LLM-as-judge protocol shows stability or slight improvement (disagreeing with every classical metric) without human-subject validation (expert ratings or comprehension quizzes) to establish that the metric declines track actual difficulty for ML readers.

    Authors: We deliberately juxtapose classical metrics with acronym density and the LLM protocol precisely because classical formulas were not designed for equation-laden technical text; the disagreement is presented as a substantive finding that raises the human-versus-LLM reader question. We will add an expanded limitations subsection that acknowledges the absence of direct human comprehension studies in this work, cites prior domain-specific readability research, and clarifies that classical metrics are used as established proxies while the LLM results are exploratory. This will make the metric comparison more balanced without overstating what the data demonstrate. revision: partial

  3. Referee: The proposal of seven standards (acronym budget, human readability threshold, etc.) rests on the assumption that enforcing them will improve human understanding or impact, but the manuscript provides no pilot data, prior evidence from other fields, or discussion of potential unintended effects (e.g., on technical precision) to support that the surface-statistic changes will translate to better comprehension.

    Authors: The seven standards are offered as concrete, measurable pilots for NeurIPS 2027 rather than as interventions already proven to raise comprehension. In revision we will insert a dedicated subsection that (a) references analogous plain-language and terminology-control efforts in biomedicine and physics, (b) explicitly discusses possible unintended consequences such as reduced technical precision or displacement of necessary jargon, and (c) frames the standards as testable hypotheses whose effects on reader understanding and citation impact should be evaluated during the pilot. This will present the proposals as an empirical invitation rather than an unexamined prescription. revision: yes

Circularity Check

0 steps flagged

No circularity: purely observational trends from external data

full rationale

The paper performs large-scale empirical analysis of readability metrics, acronym counts, and citation correlations across arXiv, NeurIPS, and PubMed corpora. No derived quantity is defined in terms of itself, no fitted parameter is relabeled as a prediction, and no load-bearing claim reduces to a self-citation or ansatz. All central patterns (Flesch decline, acronym growth, readability-citation link) are computed directly from the cited external datasets and standard formulas; the LLM-as-judge comparison is an additional measurement, not a definitional loop. The proposal of standards follows from the observations without circular justification.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The analysis rests on the assumption that classical readability formulas and LLM judgments are valid proxies for human understanding of technical abstracts; no free parameters are fitted to the target claim, and no new entities are postulated.

axioms (2)
  • domain assumption Classical readability scores (Flesch, etc.) and the Hohmann sensational-language suite measure actual human comprehension difficulty in scientific abstracts.
    Invoked when interpreting the drop from 24 to 13 and the 50% rise in sensational language as evidence of declining readability.
  • domain assumption The citation-readability correlation reflects a causal effect of readability on impact rather than confounding factors such as topic popularity or author reputation.
    Used to argue that less readable papers risk remaining fragmented.

pith-pipeline@v0.9.0 · 5676 in / 1464 out tokens · 32566 ms · 2026-05-12T01:26:34.500689+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

42 extracted references · 42 canonical work pages

  1. [1]

    eLife , volume =

    The growth of acronyms in the scientific literature , author =. eLife , volume =. 2020 , doi =

    work page 2020

  2. [2]

    Scientometrics , volume =

    The evolution of scientific writing: an analysis of 20 million abstracts over 70 years in health and medical science , author =. Scientometrics , volume =. 2025 , doi =

    work page 2025

  3. [3]

    Biological Conservation , volume =

    Comparing the writing styles of highly and rarely cited papers in conservation biology , author =. Biological Conservation , volume =. 2025 , doi =

    work page 2025

  4. [5]

    PLOS ONE , volume =

    Narrative Style Influences Citation Frequency in Climate Change Science , author =. PLOS ONE , volume =. 2016 , doi =

    work page 2016

  5. [6]

    JAMA Network Open , volume =

    Trends in the Use of Promotional Language (Hype) in Abstracts of Successful National Institutes of Health Grant Applications, 1985--2020 , author =. JAMA Network Open , volume =. 2022 , doi =

    work page 1985

  6. [7]

    Journal of Educational Psychology , volume =

    Effects of text structure on use of cognitive capacity during reading , author =. Journal of Educational Psychology , volume =. 1982 , doi =

    work page 1982

  7. [8]

    2011 , doi =

    Scientific Writing = Thinking in Words , author =. 2011 , doi =

    work page 2011

  8. [9]

    The Chicago Guide to Communicating Science , author =

    work page

  9. [10]

    The Sense of Style: The Thinking Person's Guide to Writing in the 21st Century , author =

    work page

  10. [11]

    Stylish Academic Writing , author =

    work page

  11. [12]

    How can we boost the impact of publications?

    Ryba, Ren and Doubleday, Zo. How can we boost the impact of publications?. Proceedings of the National Academy of Sciences , volume =. 2019 , doi =

    work page 2019

  12. [13]

    Journal of Marketing , volume =

    The Quest for Citations: Drivers of Article Impact , author =. Journal of Marketing , volume =. 2007 , doi =

    work page 2007

  13. [14]

    eLife , volume =

    The readability of scientific texts is decreasing over time , author =. eLife , volume =. 2017 , doi =

    work page 2017

  14. [15]

    Proceedings of the Royal Society B , volume =

    Specialized terminology reduces the number of citations of scientific papers , author =. Proceedings of the Royal Society B , volume =. 2021 , doi =

    work page 2021

  15. [16]

    FACETS , volume =

    Knowledge in the dark: scientific challenges and ways forward , author =. FACETS , volume =. 2019 , doi =

    work page 2019

  16. [17]

    , year =

    Barnett, Adrian G. , year =

    work page

  17. [18]

    Learning and Instruction , volume =

    Cognitive load theory, learning difficulty, and instructional design , author =. Learning and Instruction , volume =. 1994 , doi =

    work page 1994

  18. [19]

    Public Understanding of Science , volume =

    Jargon as a barrier to effective science communication: evidence from metacognition , author =. Public Understanding of Science , volume =. 2019 , doi =

    work page 2019

  19. [20]

    Journal of Language and Social Psychology , volume =

    The effects of jargon on processing fluency, self-perceptions, and scientific engagement , author =. Journal of Language and Social Psychology , volume =. 2020 , doi =

    work page 2020

  20. [21]

    2013 , doi =

    Berlin, Leonard , journal =. 2013 , doi =

    work page 2013

  21. [22]

    Scientometrics , volume =

    Readability and citations in information science: evidence from abstracts and articles of four journals (2003--2012) , author =. Scientometrics , volume =. 2016 , doi =

    work page 2003

  22. [24]

    The growth of acronyms in the scientific literature

    Adrian Barnett and Zoe Doubleday. The growth of acronyms in the scientific literature. eLife, 9: 0 e60080, 2020. doi:10.7554/eLife.60080. URL https://doi.org/10.7554/eLife.60080. Science wide acronym baseline: 2.4 acronyms per 100 title words (2019); roughly 30 percent of acronyms used once; 1.1M unique acronyms across 24M titles and 18M abstracts (1950 to 2019)

    work page doi:10.7554/elife.60080 2020

  23. [25]

    TAC : AOITROMJA ? (the acronym conundrum: advancing or impeding the readability of medical journal articles?)

    Leonard Berlin. TAC : AOITROMJA ? (the acronym conundrum: advancing or impeding the readability of medical journal articles?). Radiology, 266 0 (2): 0 383--387, 2013. doi:10.1148/radiol.12121776. URL https://doi.org/10.1148/radiol.12121776

    work page doi:10.1148/radiol.12121776 2013

  24. [26]

    Britton, Shawn M

    Bruce K. Britton, Shawn M. Glynn, Bonnie J. F. Meyer, and M. J. Penland. Effects of text structure on use of cognitive capacity during reading. Journal of Educational Psychology, 74 0 (1): 0 51--61, 1982. doi:10.1037/0022-0663.74.1.51. URL https://doi.org/10.1037/0022-0663.74.1.51

    work page doi:10.1037/0022-0663.74.1.51 1982

  25. [27]

    Evaluating the evaluators: Are readability metrics good measures of readability? arXiv preprint arXiv:2508.19221, 2025

    Isabel Cachola, Daniel Khashabi, and Mark Dredze. Evaluating the evaluators: Are readability metrics good measures of readability? arXiv preprint arXiv:2508.19221, 2025. URL https://arxiv.org/abs/2508.19221

    work page arXiv 2025

  26. [28]

    Position: T he current AI conference model is unsustainable! diagnosing the crisis of centralized AI conference, 2025

    Nuo Chen, Moming Duan, Andre Huikai Lin, Qian Wang, Jiaying Wu, and Bingsheng He. Position: T he current AI conference model is unsustainable! diagnosing the crisis of centralized AI conference, 2025. URL https://arxiv.org/abs/2508.04586

    work page arXiv 2025

  27. [29]

    Kelly, and Terrie Klinger

    Ann Hillier, Ryan P. Kelly, and Terrie Klinger. Narrative style influences citation frequency in climate change science. PLOS ONE, 11 0 (12): 0 e0167983, 2016. doi:10.1371/journal.pone.0167983. URL https://doi.org/10.1371/journal.pone.0167983

    work page doi:10.1371/journal.pone.0167983 2016

  28. [30]

    Mollie Hawkes Hohmann and Sean D. Connell. Comparing the writing styles of highly and rarely cited papers in conservation biology. Biological Conservation, 307: 0 111125, 2025. doi:10.1016/j.biocon.2025.111125. URL https://doi.org/10.1016/j.biocon.2025.111125

    work page doi:10.1016/j.biocon.2025.111125 2025

  29. [31]

    Barnett, Neil King, and Sean D

    Mollie Hawkes Hohmann, Adrian G. Barnett, Neil King, and Sean D. Connell. The evolution of scientific writing: an analysis of 20 million abstracts over 70 years in health and medical science. Scientometrics, 130: 0 3349--3366, 2025. doi:10.1007/s11192-025-05353-8. URL https://doi.org/10.1007/s11192-025-05353-8. Source of the Hohmann writing style metric s...

    work page doi:10.1007/s11192-025-05353-8 2025

  30. [32]

    Jeschke, Sophie Lokatis, Isabelle Bartram, and Klement Tockner

    Jonathan M. Jeschke, Sophie Lokatis, Isabelle Bartram, and Klement Tockner. Knowledge in the dark: scientific challenges and ways forward. FACETS, 4 0 (1): 0 423--441, 2019. doi:10.1139/facets-2019-0007. URL https://doi.org/10.1139/facets-2019-0007

    work page doi:10.1139/facets-2019-0007 2019

  31. [33]

    Readability and citations in information science: evidence from abstracts and articles of four journals (2003--2012)

    Lei Lei and Su Yan. Readability and citations in information science: evidence from abstracts and articles of four journals (2003--2012). Scientometrics, 108 0 (3): 0 1155--1169, 2016. doi:10.1007/s11192-016-2036-9. URL https://doi.org/10.1007/s11192-016-2036-9

    work page doi:10.1007/s11192-016-2036-9 2003

  32. [34]

    Scientific Writing = Thinking in Words

    David Lindsay. Scientific Writing = Thinking in Words. CSIRO Publishing, 2011. doi:10.1071/9780643101579. URL https://doi.org/10.1071/9780643101579

    work page doi:10.1071/9780643101579 2011

  33. [35]

    Specialized terminology reduces the number of citations of scientific papers

    Alejandro Mart \'i nez and Stefano Mammola. Specialized terminology reduces the number of citations of scientific papers. Proceedings of the Royal Society B, 288 0 (1947): 0 20202581, 2021. doi:10.1098/rspb.2020.2581. URL https://doi.org/10.1098/rspb.2020.2581

    work page doi:10.1098/rspb.2020.2581 1947

  34. [36]

    Trends in the use of promotional language (hype) in abstracts of successful national institutes of health grant applications, 1985--2020

    Neil Millar, Bojan Batalo, and Brian Budgell. Trends in the use of promotional language (hype) in abstracts of successful national institutes of health grant applications, 1985--2020. JAMA Network Open, 5 0 (8): 0 e2228676, 2022. doi:10.1001/jamanetworkopen.2022.28676. URL https://doi.org/10.1001/jamanetworkopen.2022.28676

    work page doi:10.1001/jamanetworkopen.2022.28676 1985

  35. [37]

    Montgomery

    Scott L. Montgomery. The Chicago Guide to Communicating Science. University of Chicago Press, 2003

    work page 2003

  36. [38]

    The Sense of Style: The Thinking Person's Guide to Writing in the 21st Century

    Steven Pinker. The Sense of Style: The Thinking Person's Guide to Writing in the 21st Century. Viking, 2014

    work page 2014

  37. [39]

    The readability of scientific texts is decreasing over time

    Pontus Plav \'e n-Sigray, Granville James Matheson, Bj \"o rn Christian Schiffler, and William Hedley Thompson. The readability of scientific texts is decreasing over time. eLife, 6: 0 e27725, 2017. doi:10.7554/eLife.27725. URL https://doi.org/10.7554/eLife.27725

    work page doi:10.7554/elife.27725 2017

  38. [40]

    Doubleday, and Sean D

    Ren Ryba, Zo \"e A. Doubleday, and Sean D. Connell. How can we boost the impact of publications? T ry better writing. Proceedings of the National Academy of Sciences, 116 0 (2): 0 341--343, 2019. doi:10.1073/pnas.1819937116. URL https://doi.org/10.1073/pnas.1819937116

    work page doi:10.1073/pnas.1819937116 2019

  39. [41]

    Shulman, Graham N

    Hillary C. Shulman, Graham N. Dixon, Odalys M. Bullock, and Dolores Col \'o n Amill. The effects of jargon on processing fluency, self-perceptions, and scientific engagement. Journal of Language and Social Psychology, 39 0 (5--6): 0 579--597, 2020. doi:10.1177/0261927X20902177. URL https://doi.org/10.1177/0261927X20902177

    work page doi:10.1177/0261927x20902177 2020

  40. [42]

    Stefan Stremersch, Isabel Verniers, and Peter C. Verhoef. The quest for citations: Drivers of article impact. Journal of Marketing, 71 0 (3): 0 171--193, 2007. doi:10.1509/jmkg.71.3.171. URL https://doi.org/10.1509/jmkg.71.3.171

    work page doi:10.1509/jmkg.71.3.171 2007

  41. [43]

    Cognitive load theory, learning difficulty, and instructional design

    John Sweller. Cognitive load theory, learning difficulty, and instructional design. Learning and Instruction, 4 0 (4): 0 295--312, 1994. doi:10.1016/0959-4752(94)90003-5. URL https://doi.org/10.1016/0959-4752(94)90003-5

    work page doi:10.1016/0959-4752(94)90003-5 1994

  42. [44]

    Stylish Academic Writing

    Helen Sword. Stylish Academic Writing. Harvard University Press, 2012

    work page 2012