Scientists Are Scrubbing Their Own Vocabulary to Survive Federal Funding Pressure

A quietly extraordinary thing is happening inside American biomedical science, and it is happening without a single official censor ever being hired. In the wake of the mass termination of U.S. federal research grants—terminations that a federal district court in Massachusetts described as arbitrary and capricious—medical and public health scientists have begun editing their own scientific vocabularies to match the edited language of their grants. That this has occurred on federal grant records is well documented. There is now mounting evidence that the same pressures are propagating into the manuscripts that scientists submit for peer-reviewed publication, a development with profound implications for what the scientific record will look like a generation from now.

The clearest window into this phenomenon comes from an editor who has seen it firsthand. Alexander C. Tsai of Harvard Medical School and Massachusetts General Hospital, writing in the journal SSM – Mental Health, describes handling a manuscript during review when the author disclosed that the grant supporting the study had been renegotiated. The revised award language was understood to extend its restrictions to publications acknowledging support from the grant. What that meant in practice was anyone’s guess. The author received conflicting advice about how broadly the restriction should be interpreted, including whether it reached the manuscript text itself. Faced with this ambiguity, the author adopted a conservative interpretation and surgically excised the offending terminology from prominent parts of the manuscript, while leaving in place other language that might plausibly have raised similar concerns but had not been specifically flagged.

Tsai argues that his experience is unlikely to be unique, given the scale at which grant award documents have been revised under duress. The proximate cause is not mysterious. Biomedical research funding in the United States has been aggressively politicized. Grants have been terminated mid-cycle for proposing to study topics the administration disfavors. Other grants, spared outright termination, have had their noncompeting continuations slow-walked. The National Institutes of Health has reminded grantees that the force-revised language represents substantive changes in the scope of their awards, and that those changes constitute new terms and conditions with which recipients are expected to comply. Importantly, the NIH notice did not explicitly prohibit particular words from appearing in publications. Instead, it places grantees in the unenviable position of determining how award-specific changes in scope apply to the research activities and scientific products arising from the award.

These executive branch actions unfold within a broader context of attacks on science and academic freedom, compounded by concurrent state-level measures. State legislatures in Indiana, Ohio, and North Dakota, among others, have introduced or passed bills altering post-tenure review procedures, expanding grounds for disciplining or dismissing tenured faculty, restricting institutional diversity, equity, and inclusion programs and expenditures, and imposing politically prescribed requirements on faculty evaluation and teaching. The provisions differ substantially by state, but their conjoint effect has been to deploy a tightly woven dragnet of political oversight over public higher education while weakening the institutional buffers that traditionally protect academic autonomy. Many American scientists now work in environments where tenure protections are eroded, DEI activities are restricted, and the climate for academic freedom is measurably colder.

The result, Tsai writes, is that science now stands in the meat locker of anticipatory compliance. Because the NIH compliance notice does not specifically instruct grantees to eliminate specified terms, but the consequences of noncompliance could be catastrophic for a laboratory, institutional counsel, research management offices, and investigators have strong incentives to adopt conservative interpretations of what the revised terms require. The sanitization takes familiar forms. Terms that have been standard scientific vocabulary for decades—gender-affirming, health equity, justice, racial segregation, sexual and gender minorities, structural racism—closely correspond to terms appearing on screening lists that NIH staff reportedly supplied to Senate staffers and investigative reporters. In manuscripts written under award-specific restrictions, such terms may be replaced with oblique constructions, with older behavioral categories the field has outgrown, or with awkward periphrases whose only virtue is that they do not contain a disfavored word.

The sophistication of this censorship, Tsai argues, is that it requires no censors at all, because scientists can be induced to do the work themselves—in exchange for the continued possibility of maintaining their effort and employing their postdocs. He is careful not to fault individual scientists for making these calculations, acknowledging he would likely do the same. But this is a collective action problem. If manuscript sanitization becomes common, each individually rational act of compliance reveals a degree of elasticity that the administration can then extrapolate to an entire field, and an initially bounded list of keywords that trigger review is unlikely to stop expanding once it has been shown to work. He likens the dynamic to episodes in which major American law firms, targeted by executive orders threatening financial consequences, entered into accommodation agreements that failed to insulate them from further executive pressure. The arrangement, he writes, is elegant: it converts the coercive effects of state action into private editorial decisions, distributes the moral injury across thousands of individual scientists each making a defensible compromise, and leaves executive power free to deny with curious vigor that any censorship is occurring at all.

History suggests how such pressures play out. Three recent episodes offer instructive precedents. The first is the near-total freeze on federal gun violence research at the Centers for Disease Control and Prevention after the National Rifle Association campaigned against the agency’s injury center following a landmark 1993 study linking household firearms to homicide risk. The 1996 Dickey Amendment did not explicitly ban gun violence research, but its language was ambiguous enough to achieve its intended effects. As researchers Arthur Kellermann and Frederick Rivara observed, no federal employee was willing to risk a career or agency funding to find out what was permitted. Publication on gun violence subsequently declined by 64 percent relative to the growth of the indexed literature between 1998 and 2012, and the number of scientists focused on the topic plateaued, trends that did not reverse until the amendment’s meaning was clarified years later. The second episode concerns the Agency for Health Care Policy and Research, which came under organized attack after issuing a clinical guideline favoring conservative treatment of acute back problems; it survived only through a political compromise that cut its budget and ended the guideline program, and its successor agency, AHRQ, now faces renewed existential threats, mass staff reductions exceeding two-thirds of early-2025 levels, website removals of peer-reviewed publications, and a collapse in new grantmaking.

The third precedent may be the closest analogy, because it targeted vocabulary itself. In 2003, then-Representative Patrick Toomey introduced an amendment to rescind funding for five NIH grants studying sexual behavior; it failed by two votes, but NIH Director Elias Zerhouni was subsequently compelled to justify roughly 200 grants on a circulating hit list, many concerning HIV prevention and sexual behavior. No grants were ultimately terminated, yet the chilling effect persisted. When researcher Joan Kempner surveyed affected scientists three years later, half had removed red flag words—gay, lesbian, homophobia, sex workers, needle exchange—from the titles and abstracts of subsequent applications, and a quarter had abandoned entire research topics. NIH-funded research on lesbian, gay, bisexual, and transgender health declined by more than 50 percent in the two years after the failed amendment, and sexual and gender minority health research remains below one percent of the NIH portfolio today. Crucially, those earlier attacks traveled through Congressional processes; the present ones proceed through executive and administrative action, a distinction that gives additional weight to concerns about recovery.

That concern is grounded in a specific asymmetry, one that the history of Soviet biology under Trofim Lysenko illustrates. After Lysenko’s fall, formal genetics research revived with surprising speed, but the accumulated institutional damage did not: no Soviet genetics textbook was published between 1948 and 1963, a generation of biology instructors had been trained in Lysenkoist doctrine, and administrators who rose under his patronage remained in power into the 1970s. Restrictions can be imposed quickly and removed quickly, but rebuilding lost personnel, training pipelines, institutional trust, and scientific trajectories takes considerably longer. Tsai is careful not to over-extrapolate—Soviet scientists were fired, imprisoned, and executed, circumstances plainly different from those in the contemporary United States—but the lesson about institutional inertia stands.

Why does vocabulary matter so much? Because scientific vocabulary is the medium through which the field conveys findings about the pathways by which large-scale forces produce disease. Concepts like institutionalized racism, minority stress, racial segregation, and structural stigma are central to mature literatures on how social conditions shape health. Consider a study measuring gender-based violence—a construct that specifies the gendered relations and structures of power in which violence occurs—and relabeling it in publications as generic interpersonal violence. Readers must reconstruct a critical feature of what was actually studied, and the populations most affected lose the precision of the finding. Tsai canvasses the coping strategies available to authors, from awkward periphrases to coded substitutions—his satirical suggestion of replacing structural racism with pink polka dotted bananas exposes the absurdity of purely lexical enforcement, which would not even defeat automated semantic screening—and portfolio segregation, in which investigators with non-federal funding allocate their most vocabulary-vulnerable work to genuinely non-federal support. None of these strategies will preserve scientific communication against a determined state adversary, but they make the pressures visible rather than subsidizing them invisibly. The deeper remedy, he argues, is collective: editorial statements, professional society position papers, and coordinated commentary that turn isolated, private accommodations into a documented collective act performed under duress. Scientists, he insists, deserve to have their compliance recorded accurately as compliance rather than as consent.

Subject of Research: Anticipatory self-censorship among U.S. biomedical scientists under federal funding and political pressure on scientific vocabulary

Article Title: Inside the meat locker: anticipatory self-censorship and pressure on the vocabulary of American science

Article References: Tsai, A. C. (2026). Inside the meat locker: anticipatory self-censorship and pressure on the vocabulary of American science. SSM – Mental Health, 10, Article 100696. https://doi.org/10.1016/j.ssmmh.2026.100696

Image Credits: AI Generated

DOI: 10.1016/j.ssmmh.2026.100696

Keywords: self-censorship, NIH, scientific censorship, academic freedom, federal grant terminations, chilling effect, research funding, public health, structural racism, AHRQ, Dickey Amendment, Lysenkoism