Reading scholarly papers in unfamiliar fields

A scholarly paper is a highly conventionalized document. Once you know the conventions, you can read a paper in a field you know nothing about and extract its central claim, its evidence, its limitations, and its place in the literature — even if half the technical vocabulary is opaque. This is the C2 skill that matters most for adult intellectual life in English: the ability to read across disciplines, to assess work outside your training, and to distinguish what the paper actually shows from what its abstract claims.

The mistake that even strong C1 readers make is to read a paper linearly, page by page, dictionary in hand, trying to understand every sentence before moving on. This reads at one-quarter speed and misses the architecture. The professional move is the opposite: read the paper non-linearly, by section, with a specific question for each section. Within twenty minutes you should know the paper’s claim, its method’s plausibility, its results’ robustness, and its standing in the field.

This lesson teaches that non-linear reading. It walks through IMRaD (Introduction, Methods, Results, and Discussion) — the empirical-paper convention — and the parallel conventions of humanistic and theoretical papers, which do not follow IMRaD but have their own discoverable structures. It teaches you to distinguish results from interpretation, to read the limitations section like a senior reviewer, and to recognize the rhetorical signals of academic prose — hedging, intensification, citation moves, the gap-claim-contribution opening.

IMRaD — the empirical paper convention

Most empirical papers in the natural and social sciences follow IMRaD: Introduction, Methods, Results, and Discussion. Many add an Abstract at the start and a Limitations subsection inside Discussion. Some add Conclusions at the end.

The order in which sections appear is not the order in which you should read them. The professional reading order is:

1. Abstract — for the gist.
2. Discussion (sometimes called Conclusions) — for what the authors claim and acknowledge.
3. Figures and tables — for the actual results.
4. Introduction — for the field context, only if you need it.
5. Methods — for the procedure, only if you intend to assess the method’s validity.
6. Results text — for results that the figures did not show clearly.

This order lets you decide within ten minutes whether the paper is worth a full read.

The abstract — what every word is doing

A scientific abstract is 150-300 words and has a stable structure:

• Background sentence (one or two): Although X has been widely studied, Y remains poorly understood.
• Aim/objective: We sought to determine whether…
• Methods compressed: Using a cohort of 1,247 patients with diagnosed Z, we measured…
• Results compressed: We found that the intervention was associated with a 32 percent reduction in…
• Conclusion: These findings suggest that…

Read the abstract twice. The first time, locate the four parts. The second time, watch for hedging vocabulary: suggest, may, appears to, is associated with, is consistent with. Strong hedges mean modest claims. Weak hedges mean strong claims. The absence of hedging in a results sentence is sometimes a red flag.

Read this in the style of a contemporary biomedical abstract:

Background. Long COVID, the persistent multi-system illness following SARS-CoV-2 infection, affects an estimated 10 to 30 percent of acutely infected individuals, yet the underlying pathophysiology remains poorly understood and disease-modifying treatments are lacking. Objective. To determine whether persistent viral RNA in gut tissue is associated with symptom severity in patients with long COVID at six months post-acute infection. Methods. In this prospective cohort study, we enrolled 174 adults with confirmed SARS-CoV-2 infection and persistent symptoms at six months. Participants underwent intestinal biopsy and symptom assessment using the validated PASC Symptom Scale. We assayed biopsy specimens for SARS-CoV-2 RNA using droplet digital PCR. Results. Detectable intestinal viral RNA was present in 41 of 174 participants (23.6 percent). Mean PASC Symptom Scale score was 27.4 (SD 9.1) in the RNA-positive group and 18.9 (SD 8.3) in the RNA-negative group (P less than 0.001). After adjustment for age, sex, and baseline severity, intestinal viral RNA remained associated with symptom severity (adjusted beta 7.8, 95% CI 4.9 to 10.7). Conclusions. Persistent intestinal viral RNA is associated with greater long-COVID symptom severity at six months. These findings support the hypothesis that viral persistence contributes to the pathophysiology of long COVID and may identify a treatable subgroup.

What to notice:

• Hedging vocabulary in the conclusion. Is associated with, support the hypothesis, may identify. The authors are careful not to claim causation. Association is weaker than cause; support is weaker than prove; may identify is weaker than identifies.
• The structure of the result. A prevalence (23.6 percent), a between-group difference (27.4 vs 18.9), an adjusted effect with confidence interval (beta 7.8, 95% CI 4.9 to 10.7). Three layers of evidence, each more controlled than the last.
• The P value and the adjusted analysis. P less than 0.001 signals statistical significance; the adjusted analysis controls for confounders. A reader who can parse these markers can assess the result without the full paper.
• The translational reach. May identify a treatable subgroup. The paper is positioning itself for clinical relevance, which is the field’s reward structure.

Reading the discussion section — the rhetorical machinery

The discussion section of a scientific paper has a near-standardized rhetorical sequence:

1. Restate the principal finding in one sentence.
2. Place the finding in the context of prior literature (this is consistent with X, extends Y, contradicts Z).
3. Propose mechanisms that might explain the finding.
4. Acknowledge limitations.
5. Suggest future research.
6. State implications (clinical, theoretical, policy).

Sections four and five are where the paper’s honesty lives. Read them slowly. A paper with a thin or evasive limitations section is often hiding something. A paper with a thoughtful, specific limitations section is usually the more reliable one.

Read this in the style of a discussion section:

In this prospective cohort of 174 adults with long COVID at six months post-acute infection, we found that persistent intestinal viral RNA was present in nearly a quarter of participants and was associated with substantially greater symptom severity. The magnitude of the association — an adjusted increase of 7.8 points on the PASC Symptom Scale — corresponds to a clinically meaningful difference and exceeds the threshold reported in prior validation studies.

Our findings are consistent with the proposal that viral persistence in gut-associated lymphoid tissue may contribute to long-COVID pathology, a hypothesis advanced by Su and colleagues (2023) and supported by autopsy and biopsy studies in smaller cohorts. They extend prior work by demonstrating a quantitative dose-response relationship between detectable RNA and symptom severity within a well-characterized post-acute population.

Several limitations of this study should be acknowledged. First, the cross-sectional design at six months precludes inference about whether viral persistence is a cause, consequence, or biomarker of severe long COVID. Longitudinal data on viral clearance and symptom trajectory will be required to establish temporality. Second, our cohort was drawn from a single academic medical center and was predominantly white and college-educated; generalizability to other populations is uncertain. Third, the PASC Symptom Scale, while validated, is patient-reported and subject to recall and reporting biases. Fourth, we did not assess viral RNA in tissues other than intestine, and we cannot rule out that persistence at other sites accounts for the observed association. Fifth, droplet digital PCR detects RNA but does not establish whether the RNA represents replication-competent virus.

What to notice:

• The limitations sequence: design, generalizability, measurement, completeness, mechanism. A serious limitations section names problems by category. A thin section says our study has limitations and stops.
• The methodological vocabulary. Cross-sectional, longitudinal, temporality, generalizability, recall and reporting biases, replication-competent. At C2 you should recognize these terms even outside your field. They are the cross-disciplinary vocabulary of empirical research.
• Hedging vocabulary throughout. May contribute, is consistent with, extends prior work. The discussion is making claims, but the claims are pitched carefully against the evidence.
• The literature placement. Su and colleagues (2023). The discussion locates the paper in the conversation; reading the citation graph is part of reading the paper.

Reading methods — when you need to and when you do not

For most casual reading, you do not need to read the methods section in detail. For papers whose conclusions you intend to rely on, you do. The minimum methodological literacy at C2:

• Study design: randomized controlled trial, prospective cohort, retrospective cohort, case-control, cross-sectional, ecological, case series, case report. The strength of evidence varies by design.
• Sample size and power: was the study large enough to detect a real effect?
• Outcome measure: how is the dependent variable measured, and is the measure validated?
• Statistical approach: which test, what confounders adjusted for, was the analysis pre-registered?
• Confounders considered: what alternative explanations did the authors address, and which did they not?

A reader who can ask these five questions of any empirical paper has crossed the threshold of methodological literacy.

Humanistic papers — the alternative architecture

Humanistic papers — in literature, history, philosophy, cultural studies — do not follow IMRaD. They follow a different but equally discoverable structure:

1. An opening case or anecdote that signals the paper’s terrain.
2. A statement of the gap or question in the literature.
3. A claim about what the paper will argue.
4. A sequence of close readings or historical analyses that build the argument.
5. A turn or counterexample that complicates the central claim.
6. A conclusion that restates the contribution in more general terms.

The vocabulary is different too. Humanistic papers deploy:

• Theoretical frameworks named by scholar: a Foucauldian reading, a Marxist analysis, a queer-theoretical frame, a postcolonial perspective.
• Critical moves: I argue, I suggest, I want to complicate, I take issue with, I extend, I revise.
• The disciplinary footnote: humanistic footnotes often contain substantive argument, not just citations.

Read this in the style of a humanities journal article opening:

In the closing pages of Toni Morrison’s Beloved, the narrator declares that “it was not a story to pass on” — a sentence that has occasioned a generation of scholarly disagreement about whether the novel ultimately commemorates or refuses commemoration. I want, in this essay, to take that disagreement as my starting point, but to displace its terms. The interpretive impasse, I will argue, is not between competing readings of the novel’s ending but between two distinct theories of historical memory that Beloved sets against each other and refuses to resolve.

Recent criticism, drawing on the work of trauma theorists such as Cathy Caruth and Dominick LaCapra, has tended to read Beloved as a novel about the impossibility of full witnessing — a text whose form enacts the structural belatedness of traumatic memory. This reading, however illuminating, has obscured what I take to be an equally important strand in the novel: a more communal, AAE-inflected theory of remembrance in which forgetting is not a failure but a deliberate act of ethical self-protection. The novel, I want to suggest, holds these two theories in productive tension and does not adjudicate between them. The ambiguity of the closing line is not a hermeneutic puzzle to be solved but a formal commitment to the unresolved.

What to notice:

• The opening textual hook. In the closing pages of Toni Morrison’s Beloved… The paper does not begin with method or with literature review; it begins with the text it will discuss.
• The gap-claim opening. Recent criticism… has obscured what I take to be… The two-step move — name the existing reading, identify what it misses — is the central rhetorical engine of humanistic scholarship.
• First-person assertion. I want to argue, I want to suggest, I take that disagreement as my starting point. Humanistic papers are openly first-person; empirical papers are openly third-person or first-person plural. Knowing the convention tells you what kind of paper you are reading.
• Theoretical citation. Cathy Caruth and Dominick LaCapra. The paper locates itself in a specific theoretical conversation by named scholar.

Hedging and intensification — reading the modal landscape

Academic writing operates on a controlled scale of certainty. The same finding can be stated with different strengths:

A claim’s actual strength is the weakest hedge in its sentence. Our results suggest that the intervention may be associated with improved outcomes in some patients. That sentence has three hedges (suggest, may, in some) and is making a very weak claim, even though it is dressed in confident-looking prose.

At C2 you read the hedges as carefully as the claims. A reviewer’s first move on a draft is to demand more hedging where the data does not support a strong claim, or to demand less hedging where the author is hiding behind language. Apply the same standard as a reader.

Citation moves — the conversation visible on the page

Citations are not decoration. They are how a paper takes a position in the field. The major citation moves:

• Citing as support. Smith (2019) found similar results in a larger cohort. The cited work agrees and strengthens.
• Citing as contrast. In contrast to Smith (2019), our findings suggest… The cited work disagrees.
• Citing to extend. Building on the framework proposed by Smith (2019)… The paper takes another’s work as a starting point.
• Citing to dismiss. Smith (2019), drawing on a small selected sample, claimed… The double markers — drawing on a small selected sample, claimed — signal disagreement without open polemic.
• Citing to box in. A growing literature (Smith 2019; Jones 2020; Park 2021) has documented… The pile of citations establishes consensus the new paper can rely on.

Reading citations as moves lets you map the paper’s position in its field within minutes.

Reading strategies

1. Read non-linearly. Abstract, then discussion, then figures, then introduction, then (if needed) methods. Linear reading wastes time.
2. Read the limitations section at the start of the discussion. A serious limitations section is the best signal that the paper is honestly reported.
3. Track the hedges. Mark every hedge in the abstract. The paper’s actual claim is the strongest claim that survives the strongest hedge.
4. Read figures and tables before the text describing them. Figures often carry the result more clearly than the prose; read them first and then read the text as commentary.
5. Use the citation graph. Note three to five papers cited multiple times — these are the field’s center. Follow up the most-cited names if you need depth.
6. Time-box the first read. Twenty minutes for a non-specialist read of an empirical paper. If you do not have the central claim, the main method, the principal result, and one limitation by twenty minutes, the paper is either above your field literacy or unusually badly written. Either is information.

Genre conventions

• The preprint vs the published paper. Preprints (bioRxiv, arXiv, SSRN) have not been peer-reviewed; published papers have. At C2 you should distinguish them and weight accordingly.
• The replication paper does not introduce new findings; it asks whether prior findings hold up under repeated testing. These are crucial and underappreciated.
• The review paper synthesizes a field; it does not present new data. Look for systematic review, meta-analysis, narrative review, scoping review in the title.
• The conflict-of-interest section at the end of biomedical and clinical papers is not boilerplate. Read it. Pharma funding does not invalidate a paper, but it is information.
• The retraction watch. Papers can be retracted. A paper from 2018 that has been retracted should not be cited as evidence in 2026. Retraction Watch is a public database.

Common Russian-speaker reading challenges

1. Treating the IMRaD section order as the required reading order. Russian academic training tends to encourage linear reading; the professional move in English-language scholarship is non-linear. Read the abstract and discussion first.
2. Missing the calibrated hedging. Russian academic prose often hedges through different markers (perfective aspect, modal particles) than English does. The English-language hedging vocabulary (suggest, may, appears, is consistent with) must be read deliberately for a C2 reader from the Russian tradition.
3. Reading the limitations section as a confession of weakness. In English-language scholarship, a thoughtful limitations section is a strength, not a weakness. The paper acknowledging its limits is the more reliable paper. A Russian academic instinct to skim past the what we did not do section, or to read it as throat-clearing, will mislead.
4. Translating field-specific terms via Russian cognates. Significant is not значительный; it is a statistical term (P less than the chosen alpha). Power is not мощность in the colloquial sense; it is the probability of detecting a real effect. Robust is not крепкий; it is insensitive to assumption violations. At C2 you must recognize these terms in their disciplinary senses.
5. Skipping the citation graph because the names are unfamiliar. Russian academic training in a different national literature tradition may produce a habit of reading without tracking citation networks. In English-language scholarship the citation graph is the conversation; reading without it is reading half the paper.
6. Reading humanistic papers as if they were empirical, expecting IMRaD. A humanities paper does not have a methods section in the empirical sense; its method is the close reading or the historical analysis. Looking for IMRaD where it is not there produces frustration and missed argument.
7. Underweighting the role of theoretical-framework labels. A Foucauldian reading, a Bourdieusian analysis, a Lacanian frame are not name-dropping; they signal the paper’s intellectual lineage and constrain its argumentative moves. Recognizing the framework tells you what the paper will and will not do.

Summary

• Read non-linearly: abstract, discussion, figures, introduction, methods. Twenty minutes for a first non-specialist read.
• IMRaD for empirical papers; gap-claim-close-reading for humanistic papers. Each genre has discoverable structure.
• Hedging is calibration. Suggests, may, is consistent with are precise terms, not vagueness. Read them as carefully as the claims.
• The limitations section is a strength signal. A serious limitations section means the authors thought through what their study cannot show.
• Citations are moves. Building on, in contrast to, drawing on a small sample, claimed — read citations as positioning, not as decoration.
• Distinguish results from interpretation. The figures show what the study found; the discussion is the authors’ argument about what the findings mean.

Next lesson: Recognizing satire and parody — The Onion and beyond.