[HOLD] 📘 Glossary update - 2026-11-12 (replaces #1092)

RI
4h ago
+3773-2981

[HOLD] Don't merge before #1111. Like #1092, this restores the italic p in related terms, which needs #1111's rendering on main.

Glossary regeneration from the Glossary sheet (12 November). It replaces #1092: it carries all of #1092's changes, plus the Arabic fixes below, which the Arabic glossary team asked for after reviewing the Doors preview. All were made in the sheet; every changed cell is logged with its old value.

Arabic fixes (new in this run)

  • Related terms inside definitions removed (258 definitions). Nearly every Arabic definition ended with its own "المصطلحات ذات الصلة: …" list, so the related terms appeared twice. They now appear only in the Related terms section. Lists for alternative definitions went too. The alternative definitions themselves stay.
  • Duplicate "التعريف:" label removed from the start of 19 definitions, which the page already labels "تعريف". Leftover bold markers and label fragments went with them. Confirmation bias no longer opens with its English definition.
  • Titles, as the team worded them:
    • Meta-analysis is التحليل التلوي. Before, it shared التحليل البعدي with Post Hoc, so the two pages had the same address and Meta-analysis was missing.
    • شارات العلوم المفتوحة; علم الوجود (الذكاء الاصطناعي); تحيز النشر (مشكلة درج الملفات); نادي رايوت للعلوم.
    • The stray "نشر" is gone from Research Contribution Metric.
    • SIPS gets its missing space, and OER Commons its corrected shadda.
  • Leading English acronyms removed from 25 Arabic titles, for example APC and NHST. The English name still follows in brackets.
  • Credits: "Ruwayshid" is now "Ruwayshid Alruwaili" in 17 credit cells. External Validity, Ideological bias, Induction, PARKing and Universal design for learning had their translators and reviewers merged in one cell and "NA" as reviewer. They now have separate translator and reviewer lists, as the team gave them. "Dr. Nazik Alnour" and "MOhammed Mohsen" are written as everywhere else.
  • Meta-analysis wording: Arabic definitions now use التحليل التلوي (and its plural and adjective forms) for meta-analysis wherever it means meta-analysis. This covers Meta-analysis, Psychometric meta-analysis (title too), Abstract bias, Evidence synthesis, Salami slicing and Systematic review. التحليل البعدي stays only for Post Hoc, and البعدي stays where it means posterior (Bayesian parameter estimation). Meta-science and meta-synthesis keep their wording.
  • Related terms joined by commas (closes #1124): DOI, Open Scholarship Knowledge Base and Post Hoc listed several related terms as one. They are now separate in every language, so ORCID, COS, OERs, A priori and Ad hoc link where they have pages.
  • URLs in definitions (all languages): sources are cited instead of written out, and listed under References. This covers Academic Impact, Consortium authorship, Open Data, OER Commons, Open Source software and Preregistration Pledge. Pointers to tools and sites stay as links with short visible text: orcid.org/register, protocols.io, sortee.org, credit.niso.org, OpenNeuro (Chinese) and the zcurve package by its DOI. The DOI translations drop the DOI Handbook quotation that the English no longer has.
  • BibTeX Doc: group authors are braced, so they no longer render as personal names ("Definition, T. O.", "Knowledge, F. O.", "Council, E. a. S. R."). ESRC now points to UKRI's current "Defining impact" page; the old address redirects to ESRC's homepage. The DOI Handbook is credited to the DOI Foundation, and its URL is shown. BMJ and JMIR block automated requests, so I checked those two pages through Wayback snapshots (April 2026 and September 2023).
  • Old addresses: 31 Arabic titles changed (30 at first, plus Psychometric meta-analysis), so their page addresses changed. _renamed_terms.json lists every old address as an alias, and each was live on forrt.org before this run. The German aliases from #1092 (ba618d2) are carried in the same commit.

New in this run (all made in the sheet or the BibTeX Doc) (from #1092)

  • Contribution (EN, DE, AR, ZH): a pull request and its review are now described as two steps. Contributors propose changes in a pull request, and other project members review them before they are merged. The Larivière citation now has its year: (Larivière et al., 2016, p. 430). The year matches the Lariviere2016 BibTeX entry and Crossref (10.1177/0306312716650046).
  • WEIRD (EN, DE, AR): research is now "conducted with relatively homogeneous samples from WEIRD societies, and by researchers from those societies", not "by" samples. The Chinese text never had this problem and is unchanged.
  • WEIRD related terms: Bias, Populations and Sampling bias are removed. None of them has a glossary page, and no existing term clearly matches. The Arabic definition's inline list of related terms drops the same three.
  • Italic p in related terms: *p*-hacking, *p*-value, *p*-value fishing and Reverse *p*-hacking are back in ten Related_terms cells. Doors renders and links them (#1085). Until Doors is live, forrt.org shows the asterisks and these related terms are not linked.
  • Research integrity (all languages): now cites the 2017 ALLEA code (new ALLEA2017 entry in the BibTeX Doc: Revised Edition, Berlin, ISBN 978-3-00-055767-5). The definition quotes wording from the 2017 code that the 2023 revision changed, e.g. "consider negative results to be as valid as positive findings" became "can be as relevant as". Authorship still cites the 2023 edition.

Carried over from #1084 (from #1092)

  • The cherry-picked commit ba618d2 restores the German alias addresses for Rechenmodell and statistisches Modell (_renamed_terms.json plus the regenerated aliases). It applied without conflicts. Both files now match #1084.
  • Everything else from #1084 was regenerated from the fixed sources and is unchanged:
    • Arabic Gift (or Guest) Authorship and Git;
    • "et al." for truncated author lists;
    • the Onie (2020) pages;
    • reference order;
    • Authorship casing.

Added on 11 November (review fixes, made in the sheet or the BibTeX Doc) (from #1092)

  • Direct replication (EN): the example's two clauses are now separate sentences.
  • FORRT (TR, ZH): the embedded English name is now "Framework for Open and Reproducible Research Training".
  • DOI: "DataCite" in the related terms (EN, and TR through the English column) and in the DE, AR and ZH definitions.
  • Full author lists from Crossref for Etz et al. (2018), Bahlai et al. (2019) and Hardwicke et al. (2020). The other truncated entries are tracked in #1090.
  • References: Ikeda et al. (2019) has its DOI (10.24602/sjpr.62.3_281), so PARKing links the article again. The DataCite Metadata Schema titles use DataCite's spelling. Closed access cites only its archived snapshot, without the retrieval date that predated it.
  • Picked up from sheet edits made after the 9 November run: Reference cells no longer repeat as bare links (for example www.zenodo.org) sources whose APA reference already gives that link. Syed (2019) now has its DOI, and DataCite Schema and Closed access have APA references.

Checks

  • Outside content/glossary/arabic/, the glossary content matches #1092's branch except for lastmod dates.
  • Arabic: 262 terms, one more than before (Meta-analysis). No slugs collide.
  • After this run, no Arabic definition holds a related-terms label, a leading definition label, an unbalanced **, an "NA" credit or a bare "Ruwayshid".
  • I built the Doors branch (#1121, on #1118) with this content and checked the Arabic DOI and External Validity pages and the English External Validity page.

🤖 Generated with Claude Code