Copyediting vs developmental editing: the editing taxonomy

Last updated: 2026-05-17

Synthesis

Professional editing is not one activity but a gradient of activities, each with its own scope, license to modify, and characteristic failure modes. The standard division in the English-language editing literature runs from proofreading (typographical, mechanical) to copyediting (grammar, style, internal consistency) to line editing (sentence-level voice, flow, clarity) to substantive / developmental editing (structure, argument, organization). The boundaries are conventions, not laws; different houses and professional bodies (CSE, Editorial Freelancers Association, Council of Editors of Learned Journals) draw them slightly differently. But the underlying principle is shared: heavier-touch editing requires more explicit authorization and produces more potential for distortion of the author’s intent.

This gradient matters for scriptorium because skills sit at different points on it. terminology-normalization and journal-style conversion are copyediting. compression and active-voice transformations are line editing. argumentative-flow and any future restructuring skill are developmental editing. Mixing them — calling “clean up this paragraph” and getting back a structurally rearranged argument — is the operating-mode mismatch that makes single-prompt LLM editing untrustworthy. Scriptorium’s one-skill-one-responsibility discipline is the editorial-taxonomy discipline ported into agent design.

Evidence

Einsohn (2011, 2019), The Copyeditor’s Handbook (3rd ed., UC Press, ISBN 978-0520286726). The most widely used English-language reference on copyediting. Einsohn is explicit about scope: “Copyeditors are expected to query structural and organizational problems, but they are not expected to fix these problems. Reorganizing or restructuring a manuscript is called developmental editing.” And: “copyeditors are not rewriters, ghost writers, or substantive editors. Although copyeditors are expected to make simple revisions to smooth awkward passages, copyeditors do not have license to rewrite a text line by line.”¹

Einsohn divides copyediting into four “Cs”: clarity, coherency, consistency, and correctness. These map onto sentence- and paragraph-level interventions, not paragraph-restructuring or argument-redesign work. The handbook’s repeated injunction is that the copyeditor queries the author when something larger is wrong — they do not silently fix it.

Mossop (2020), Revising and Editing for Translators (4th ed., Routledge, ISBN 978-1138895164). Though framed for translation, Mossop’s parameter system applies cleanly to scientific editing. He defines twelve parameters in four groups: (1) meaning transfer (accuracy, completeness); (2) content (logic, facts); (3) language and style (smoothness, tailoring to readership, sub-language conventions, idiom, mechanics); (4) physical presentation (layout, typography, organization).²

The Mossop framing is useful for scriptorium because each parameter is a separable check. A skill can target accuracy without addressing idiom; a skill can target logic without addressing smoothness. This separability is what enables skills to compose: the editorial trace is the parameters touched, in order, by which skill.

The Council of Science Editors (CSE) Manual, 9th edition (2024, ISBN 978-0226683942). The CSE Manual is the field-standard reference for scientific publishing in the English-language biomedical and life-sciences literature. Originally the CBE Manual (Council of Biology Editors, 1965–2000). The 9th edition delivers comprehensive guidance on style, citation systems, and editorial practice for scientific manuscripts. CSE recognizes three citation systems (citation-sequence, name-year, citation-name), which scriptorium’s csl-and-bibliographic-standards tooling must accommodate.³

CSE’s scope is closer to copyediting and house-style enforcement than to substantive editing, reflecting the journal-production reality that most scientific copyediting happens after acceptance and after the developmental editing has already been done (typically by the authors themselves, via peer review).

Practitioner consensus on the gradient. The Editorial Freelancers Association, the Chartered Institute of Editing and Proofreading (UK), and most editorial training programs converge on roughly four levels:

Level	Scope	Author authorization needed
Proofreading	Typos, layout, mechanical errors	Minimal
Copyediting	Grammar, style, consistency, fact-spot-check	Standard
Line editing	Sentence-level voice, flow, clarity	Explicit
Developmental / substantive editing	Structure, argument, organization	Explicit + collaborative

These distinctions are not academic. Pricing in professional editing markets correlates closely with scope; reputational risk for the editor scales with scope (a copyeditor who restructured your manuscript is a problem; a developmental editor who only fixed commas is a different problem); and time required scales with scope.

How this informs scriptorium

The editing taxonomy maps cleanly onto scriptorium’s skill categories as defined in [[DESIGN]] §“Design philosophy”:

Editing level	DESIGN.md category	Example skills (current + planned)
Proofreading	Normalization	(none yet — typography is downstream of scriptorium)
Copyediting	Normalization	`terminology-normalization`, journal-style-conversion
Line editing	Transformation (narrow)	`compression`, redundancy-removal, active-voice
Developmental	Transformation (broad)	`argumentative-flow`, restructuring
(Cross-cutting)	Critique	`citation-audit`, `reviewer-simulation`
(Cross-cutting)	Validation	figure-text-alignment, statistics-consistency

Three concrete design consequences follow.

1. Skill descriptions must declare their editing level. A skill’s SKILL.md should make explicit whether it operates at the copyediting, line-editing, or developmental level. This sets author expectations and informs the auto-invocable vs explicit-only decision: copyediting skills can default to auto-invocable (low rewrite risk); developmental-level skills must be explicit-only (high rewrite risk).

2. The conservative-edit posture is the copyediting default, extended. Einsohn’s “copyeditors do not have license to rewrite a text line by line” is the same principle as DESIGN.md’s “preserving every citation, preserving every quantitative statement, minimizing rewrite surface area.” Scriptorium’s restraint is the editorial profession’s restraint, made enforceable in the MANUSCRIPT_STATE.yaml constraints.

3. Developmental edits must surface “remaining weaknesses” explicitly. DESIGN.md already calls for this: “Skills that need to make aggressive changes should be invoked explicitly … and emit a clear ‘Remaining Weaknesses’ section so the author sees what they didn’t fix.” This mirrors the developmental-editor’s query letter — the document that accompanies a substantive edit and explains what the editor changed, what they considered changing and didn’t, and what they think the author should still address.

For Mossop’s parameter system specifically: scriptorium can use it as a checklist vocabulary in critique outputs. When citation-audit flags a quotation mismatch, that is a Logic / Facts finding under Mossop. When argumentative-flow flags a paragraph that doesn’t follow from the previous one, that is a Smoothness finding. Vocabulary parity with established editing literature makes the output legible to anyone trained in the editorial tradition.

Open questions / weak evidence

The English-language editing taxonomy is not universal. In some scientific publishing traditions (e.g., German, Japanese), the developmental-vs-copyediting split is less sharp; “lektorat” or equivalent terms span multiple Anglophone categories.
Practitioner-vocabulary documents (EFA’s editorial levels, CIEP’s editorial fact sheets) are the dominant source for the four-level gradient. Academic empirical work on whether these levels are cognitively or practically distinct is thin.
The boundary between line editing and developmental editing is particularly soft: a heavy line edit and a light developmental edit produce similar artifacts. Scriptorium should not pretend the boundary is sharper than it is.

References

Einsohn A, Schwartz M. The Copyeditor’s Handbook: A Guide for Book Publishing and Corporate Communications. 4th ed. University of California Press; 2019. ISBN 978-0520286726. ↩
Mossop B. Revising and Editing for Translators. 4th ed. Routledge; 2020. ISBN 978-1138895164. See chapter 17, “The Revision Parameters.” ↩
Council of Science Editors. Scientific Style and Format: The CSE Manual for Authors, Editors, and Publishers. 9th ed. University of Chicago Press; 2024. ISBN 978-0226683942. https://www.csemanual.org/ ↩