Authorship norms don’t just vary in what order names appear in — they vary in how many names appear at all, and why. A philosophy article and a particle-physics paper can both be legitimate, rigorously produced scholarship while looking almost nothing alike on the byline: one author versus, in extreme cases, thousands. That gap is not a difference in rigor. It is a difference in what each field treats “a contribution worth an authorship credit” as meaning in the first place, rooted in how each field actually produces knowledge.
This page is about why single authorship remains the humanities’ default and large team authorship became STEM’s default, and what that means for how authorship criteria get applied in each. For how author order is decided once a paper does have multiple authors, see How Authorship Order Is Decided: Conventions Across Disciplines; for the extreme end of the STEM pattern, thousand-plus-author collaborations, see Hyperauthorship: Definition and How Research Assessment Handles It. This page covers the more common, everyday version of the divide: a solo-authored humanities article against a five-to-fifteen-author STEM paper, not the outlier cases at either end.
The humanities default: single authorship as the norm
Bibliometric research on authorship patterns consistently finds that single-authored output remains the norm in the humanities and, to a lesser extent, in interpretive social sciences — co-authorship is common in some subfields but nowhere close to universal the way it has become in most of the natural sciences. Studies specifically looking at digital humanities, a subfield with an explicit collaborative, tools-and-infrastructure-building reputation, have found single authorship still predominates in its core journals despite that reputation.
The reasons are structural, not a matter of humanities scholars being less collaborative by temperament:
- The unit of scholarly output is an individual argument or interpretation. A literary analysis, a historical interpretation, or a philosophical argument is generally understood as the product of one sustained, coherent authorial voice reasoning through evidence and prior scholarship. Where two people jointly reasoned to the same interpretive conclusion, that is unusual enough to be worth stating explicitly, rather than the default assumption a multi-name byline invites in a lab-science context.
- The scholarly monograph is still the primary unit of career evaluation in most humanities fields, particularly in tenure review at research universities. A monograph is, by convention, single-authored; a field where the terminal, career-defining scholarly product is expected to be one person’s sustained argument naturally produces single-authored norms upstream in its article output too.
- Data acquisition is rarely the bottleneck that drives team formation. Much of STEM’s shift toward larger teams (below) is a direct response to needing many specialized hands and instruments to produce a single result. A historian working from archival sources or a philosopher working from a body of published argument typically does not need a comparable division of labor to produce a publishable analysis.
The STEM default: distributed, multi-author teams
The opposite pressure has pushed most STEM fields toward multi-author collaboration as the default, and toward steadily larger teams over time. Fields built around large shared instruments or infrastructure — high-energy physics, astronomy, large-scale genomics — show the most extreme version of this: authorship is granted collaboration-wide under a group’s own membership rules rather than assessed contribution-by-contribution, because no single paper’s list of specific tasks maps cleanly onto who gets named. Other STEM fields such as chemistry and biology sit between that extreme and the humanities’ single-author norm: co-authorship is the clear majority pattern, typically reflecting a genuine division of labor across data collection, analysis, and writing, without reaching the scale of a physics mega-collaboration.
The structural reasons run parallel to the humanities’ reasons for the opposite pattern:
- Producing a single result routinely requires distinct specialized skills — running an experiment, operating an instrument, performing statistical analysis, writing software, supervising the project — that are not interchangeable and are not typically held by one person. The CRediT taxonomy’s 14 contributor-role categories (Conceptualization, Data Curation, Formal Analysis, Investigation, Methodology, Software, and so on — see the CRediT overview) exist largely because this kind of divided labor is now the STEM norm, not the exception.
- Funding and infrastructure increasingly flow to teams, not individuals. Large instruments, shared datasets, and multi-site studies are typically only fundable and operable as collaborative efforts, which structurally requires distributing both the labor and the credit across a team.
- The unit of scholarly output is a discrete, reproducible result, not a sustained individual interpretive argument — which makes it comparatively natural to decompose that result into the separable tasks a multi-author team divides among its members.
Why authorship criteria fit unevenly across the two models
The ICMJE authorship criteria (substantial contribution to conception/design/data/analysis; drafting or critically revising the manuscript; final approval; and accountability for the work) were written for biomedicine, a field near the distributed-team end of this divide. Applied to a solo-authored humanities article, all four criteria are trivially satisfied by the one person who did the entire thing — the criteria add little that a byline of one didn’t already communicate. Applied to a large STEM team, the same four criteria do real, necessary work: they are the test that separates a genuine co-author from a technician, funder, or supervisor who contributed meaningfully but not in a way that clears all four thresholds, and who is credited instead through an acknowledgment (see gift authorship and ghost authorship for what goes wrong when that line is misapplied in either direction).
CRediT statements show the same asymmetry. A 14-role contribution taxonomy is built to describe how distinct tasks were divided among several people; for a single-authored humanities article, there is no division to describe — the one author necessarily performed every role that applies, and a CRediT statement adds no information a reader couldn’t already infer. This is a genuine, structural reason CRediT has seen far more adoption in STEM and biomedical journals than in humanities journals, not an oversight in how the taxonomy was designed. It is worth noting explicitly that neither ICMJE nor CRediT excludes humanities work — both apply perfectly well to a multi-author humanities or social-science project, such as an edited-volume chapter with several contributors or a collaborative digital-humanities dataset. The asymmetry is about how much the underlying scholarly production model actually calls for that apparatus, not a rule that formally exempts one domain.
What each field uses to signal author-order meaning is also different for the same underlying reason. See How Authorship Order Is Decided: Conventions Across Disciplines for how STEM’s first-author/senior-author-last, alphabetical, and contribution-statement conventions work once a team exists — that entire question is close to moot in a single-authored humanities article, where “order” has nothing to determine.
Where the two models blend
The divide is a strong statistical tendency, not a hard rule, and several areas sit deliberately between the two defaults:
- Digital humanities and computational social science combine humanities questions with the kind of technical, tool-building labor (corpus construction, software development, data visualization) that drives multi-author collaboration elsewhere. Even here, though, bibliometric studies of core digital-humanities journals have found single authorship still predominates — the field’s collaborative reputation runs ahead of its actual publishing pattern.
- Edited volumes and multi-author humanities projects (co-authored monographs, jointly written book chapters, large digital-archive or critical-edition projects with named contributor teams) apply authorship criteria the same way STEM does once a genuine division of labor exists — the criteria track actual contribution structure, not disciplinary identity.
- Team science and interdisciplinary humanities-STEM collaborations — a historian working with a conservation scientist on the technical study of an artifact, or a linguist collaborating with a computer scientist on a natural-language-processing project — routinely mix contributors whose home disciplines carry opposite defaults. These are exactly the projects where an unstated assumption about “how authorship normally works here” causes friction, because each collaborator’s normal is different.
Practical implications for mixed and interdisciplinary teams
The practical risk is not that either norm is wrong within its own field — it is that a mixed team can silently apply the wrong one, or apply neither, because nobody named which was in effect. A few concrete failure modes worth watching for:
- A humanities scholar contributing interpretive framing to a STEM project may be substantial enough to clear ICMJE-style criteria, or may fall short of them and belong in acknowledgments instead — the field one trained in doesn’t answer that question; the actual contribution does.
- A technical specialist (statistician, data engineer, conservation scientist) contributing to a humanities project faces the reverse version of the same problem: humanities fields have less-codified authorship criteria than ICMJE provides, which makes it easier for a real contribution to go under-credited by default rather than through any deliberate decision.
- Career evaluation expectations differ sharply by field, which matters directly for early-career researchers on mixed teams: a solo-authored output typically carries more individual evaluative weight in humanities tenure review, while STEM tenure review is built to interpret a position within a multi-author byline (see Does Author Position Affect Tenure and Promotion Evaluation?). A researcher moving between these evaluation cultures, or being evaluated by a committee unfamiliar with the other field’s norm, benefits from stating the applicable convention explicitly rather than assuming the committee already knows it.
As with author order, the practical fix is the same: agree explicitly, in writing, on which authorship model and criteria apply to a given collaborative output, early enough that the agreement can still shape who does what — not after a draft exists and assumptions have already hardened into expectations.
Frequently asked questions
Is single authorship becoming less common in the humanities?
Bibliometric studies show co-authorship has been increasing gradually in some humanities and interpretive social-science subfields, but single authorship remains the clear majority pattern overall, including in subfields like digital humanities that have a more collaborative reputation than their actual publishing pattern reflects.
Does ICMJE apply to humanities scholarship at all?
ICMJE’s criteria are written for and formally scoped to biomedical journals, but the underlying logic (contribution, drafting/revision, final approval, accountability) applies fine to any multi-author collaborative output, including a humanities project with a genuine division of labor. It simply has less work to do on a solo-authored humanities article, where one person trivially satisfies all four criteria.
Should a solo-authored humanities article still include a CRediT statement?
Most journals that require CRediT statements are STEM and biomedical journals, and a statement adds little for a single author who necessarily performed every applicable role. Some interdisciplinary or multi-author humanities projects do use CRediT or a similar contributions statement where a real division of labor exists to document; check the specific journal’s policy.
How should an interdisciplinary team with both humanities and STEM members decide whose authorship norm applies?
There is no default answer, and that is exactly the risk: agree explicitly, in writing, early in the project on which criteria will determine authorship and how contribution will be documented, rather than letting each collaborator assume their home field’s convention governs. See Resolving Authorship Order Disputes for a structured process when disagreement arises anyway.
Is a large multi-author STEM team always evidence of real collaboration?
Not automatically — team size can reflect genuine distributed labor, but it can also reflect gift or guest authorship inflating the byline beyond who actually meets authorship criteria. See gift authorship and the COPE flowchart on ghost, guest, or gift authorship for how editors and institutions are meant to catch this regardless of field.
