Frame
This file states the problem in one sentence and lists what the fields that study papers call its parts.
The problem in one sentence
How should a research article be ordered so that a reader can find, judge and reuse its claims, and does the standard order do that?
What the field calls it
| Name | Field | What it covers |
|---|---|---|
| IMRaD (also IMRAD, IMRD) | scientific writing, publication science | Introduction, Methods, Results, and Discussion as the sections of an article |
| Research article (RA) genre | genre analysis, English for Specific Purposes (ESP) | the article as a recurring social form with expected parts |
| Move and step | genre analysis | a stretch of text that does one rhetorical job, and its sub-jobs |
| CARS (Create a Research Space) | genre analysis | the three moves of an introduction (Swales 1990) |
| Hourglass model | genre analysis, writing teaching | general to specific in the introduction, specific to general in the discussion (Hill 1982) |
| Metadiscourse | applied linguistics | text about the text: signposts, hedges, boosters, self-mention (Hyland 2004) |
| Structured abstract (also "more informative abstract") | medical editing, information science | an abstract with fixed labelled parts (Ad Hoc Working Group 1987) |
| Graphical abstract, visual abstract | publishing | one image that summarises the article |
| Highlights | publishing | three to five short bullet points with the main findings (Elsevier highlights, example of practice only) |
| Reporting guideline | research methodology | a checklist of what a report of one study design must contain (EQUATOR) |
| Completeness of reporting | meta-research | the share of checklist items a report actually contains |
| Scholarly reading patterns, information-seeking behaviour | library and information science | how many articles researchers read, for how long, and which parts |
| Strategic reading | information science | working across many articles at once, by fragments (Renear 2009) |
| Registered Report (RR), results-free review | meta-science | peer review of question and method before the results exist |
| HARKing | meta-science | presenting a post hoc hypothesis as if it were a priori (Kerr 1998) |
| Reproducible research, executable paper, literate programming | computational science | the article shipped with the code and data that produce it |
| Nanopublication, modular publishing | semantic publishing | the article split into machine-readable claims, or into one unit per research stage |
| Readability formula (Flesch Reading Ease, FRE; New Dale-Chall, NDC) | psycholinguistics, bibliometrics | a score from sentence length, word length or word familiarity |
| Document class, article type | typesetting, publishing (JATS) | the template a publisher supplies, and the label a journal gives a manuscript |
| Paper checklist, reproducibility checklist, Responsible NLP Research checklist | machine learning (ML) and natural language processing (NLP) venues | a form the author fills at submission: yes, no or not applicable, per reporting item (NeurIPS, ACL) |
| Artifact evaluation (AE), artifact badge | software engineering, systems, programming languages | a review of the code and data behind a paper, and a pictogram printed on the article if it passes (ACM Artifact Review and Badging) |
| Model card, datasheet for datasets, system card | ML, responsible AI | a short fixed-section document that travels with a model or a dataset |
| STAR Methods, key resources table, RRID | life-science publishing (Cell Press) | a methods format with a table of reagents and software, each with a persistent identifier |
| Plain language summary (PLS), lay summary, digest, significance statement, author summary | health and science publishing | a second, short text for readers outside the field, beside the abstract (Cochrane, eLife, PNAS, PLOS) |
| Data availability statement (DAS), data sharing statement (DSS), code availability statement | publishing policy | a sentence or section saying where the data and code are, or why they are not shared |
| Technical report, model report, release blog post | industry AI labs | a long report or a web page issued by a company in place of, or beside, a peer-reviewed paper |
| Generative AI policy, AI disclosure, LLM-modified content, excess vocabulary | publishing ethics, meta-research | what a venue allows and asks authors to declare about language-model use, and how much use has been estimated |
| Multivocal literature review (MLR), grey literature | evidence synthesis, software engineering (Garousi 2019) | a review that adds sources outside journals (rules, reports, pages) to the journal literature, each scored for quality |
| Reporting-guideline adherence interventions, journalology | meta-research | what is done to make authors follow a guideline, and the science of publishing (Blanco 2019, Galipeau 2015) |
What practitioners call it
Publishers and venues use their own words; the grey search used these as strings (protocol.md 5).
| Practitioners' name | Where | Field's name above |
|---|---|---|
| instructions for authors, guide for authors, author guidelines | publishers, journals | article type, document class |
| call for papers, author guide, paper checklist | conferences | reproducibility checklist |
| author summary, digest, lay summary, PLSP | PLOS, eLife, Taylor & Francis | plain language summary |
| key resources table | Cell Press | STAR Methods |
| LLM policy, AI use policy | ICLR, ICML, NeurIPS, ACL | generative AI policy |
| TOP guidelines, MDAR checklist | Center for Open Science, journals | reporting guideline |
Method
This is a lighter multivocal literature review after Garousi 2019, written for software engineering and applied here to publication science.
- The formal pool was built by a narrative search with snowballing, not by screening database counts.
- The grey search was run with stated strings and a stopping rule.
- Grey sources are scored on 20 criteria in three tiers: tier 1 books, standards and reports; tier 2 organisers' pages, presentations and abstracts; tier 3 blogs and posts.
- Steps and pool: protocol, sources.
Search terms, three ways
| Academic name | Acronym | Author |
|---|---|---|
| introduction methods results discussion | IMRaD, IMRAD, IMRD | Sollaci, Day |
| create a research space, move analysis | CARS | Swales, Hyland |
| structured abstract, more informative abstract | none | Hartley, Haynes |
| reporting guideline, completeness of reporting | CONSORT, PRISMA, EQUATOR | Moher, Altman |
| scholarly article reading patterns | none | Tenopir, King |
| registered reports, results-free review | RR | Chambers, Nosek |
| executable paper, computational notebook | none | Donoho, Pimentel |
| readability of scientific abstracts | FRE, NDC | Plavén-Sigray, Hartley |
| reproducibility checklist, machine learning | none | Pineau, Dodge |
| artifact evaluation, research artifact badge | AE, AEC | Winter, Heumüller, Timperley |
| model documentation, dataset documentation | none | Mitchell, Gebru |
| plain language summary, lay summary | PLS | Buljan, Jelicic Kadic |
| data availability statement, data sharing statement | DAS, DSS | Gabelica, Hardwicke, Hamilton |
| foundation model transparency | FMTI | Bommasani |
| LLM-modified content, excess vocabulary | LLM | Liang, Kobak |
| generative AI policy for authors | GAI | Ganjavi |
Out of scope
- Mathematics, theoretical computer science and the humanities depart from IMRaD; a companion review in this repository covers mathematical articles with proofs.
- Peer review as a process, citation practice and authorship rules are left out, except where a venue's rule about language-model use or about a reporting checklist changes what an author writes.