Skip to content

Disciplinary Literacy

Principle · All principles
Evidence · 19 claims (10 for, 9 mixed) · 28 studies (12 causal, 6 review, 4 theoretical, 3 quant-synthesis, 2 qualitative, 1 design), q1–q4 · 4 of 28 report an effect size · 12 claims rest on one study

Conditional relationship

This page owns reading, writing and reasoning with a discipline's own texts as that discipline does: sourcing and corroborating documents in history, reasoning from sets of texts to an explanatory model in science, justifying claims in mathematics. For a learner who reads adequately in general but meets the texts of a discipline as a novice (they summarise a primary source as if it were a textbook, take an account at face value, or treat a scientific text as a list of facts), instruction that names the discipline's reading and reasoning practices, models them on the discipline's own texts and has learners practise them on new texts over a semester or longer is expected to improve comprehension of new texts in that discipline and the practised kind of disciplinary thinking (near-transfer), compared with textbook-led or usual instruction in the same subject.

What the evidence here tests. Two studies in the wiki compare a disciplinary-literacy curriculum with usual instruction (Disciplinary Literacy Instruction Improves Comprehension [+M]): a cluster-randomised trial in 24 high schools, in which a semester of text-based investigation in ninth-grade biology produced small gains in comprehension of new science texts (effect sizes 0.26 and 0.32) and no gain in students' explanatory essays; and a quasi-experiment in five urban high schools, in which a six-month document-based history curriculum improved eleventh graders' historical thinking, factual knowledge and general reading comprehension over textbook instruction (no effect size in the abstract read). A world-history curriculum evaluation found a gain in one of six historical thinking skills and comparable results on the other five (below). So the relationship is tested for adolescent learners in classroom settings, against usual instruction, at the end of a semester or school year.

What it does not test. No claim in the wiki compares disciplinary-literacy instruction with well-taught generic comprehension-strategy instruction, which is the comparison the page's name implies; none tests the teacher's disciplinary knowledge as a moderator, younger learners, adults, or disciplinary writing as an outcome (the one writing measure showed no gain). The default design below is therefore a proposal built around the tested curricula, with each step's evidence stated.

A limit worth keeping in view. Disciplinary practices are specific. In an expert–novice comparison recorded in the wiki, PhD historians judging live websites stayed on the page and were misled by logos and domain names, while fact checkers who left the site did better (One-session checklist instruction ... [~M]). Sourcing a historical document and judging an unfamiliar website are different practices, and expertise in one did not carry to the other. Expect gains in the practice taught, on the kind of text practised, not a general rise in "critical reading".

How the converted siblings sit beside this. Epistemic Cognition owns the standards for judging a kind of evidence in any domain; this page applies them to a discipline's own texts and genres. Evaluating Sources owns judging unfamiliar web sources by lateral reading, which is a civic practice rather than a disciplinary one. Scaffolding and Fading and the Cognitive Apprenticeship pattern own how modelled support is withdrawn; this page uses them for the hand-over and does not repeat them. Accessible Vocabulary & Syntax and Cognitive Load Management own how a difficult text is made workable.

Default design

The page's earlier guidance and the folded history-imagination page, made concrete as a default a designer can act on. Each step is a proposal unless a claim is linked beside it; the tested curricula support the package (explicit practices, the discipline's texts, a semester or more), not each step.

  1. Name one disciplinary practice at a time, and choose the texts for it. Pick the move the unit will teach (in history: sourcing, corroboration, contextualisation, continuity and change; in science: reasoning from several texts to an explanation; in mathematics: justifying a claim to a sceptic) and collect the discipline's own texts that call for it (primary documents, data, short research reports, proofs), not a textbook summary of them. Evidence for the package: Disciplinary Literacy Instruction Improves Comprehension [+M]. Teach and assess each practice on its own: sophistication in one second-order concept does not imply it in another (Dimensions of historical thinking are conceptually decoupled [~M], a review reporting British progression research).
  2. Model the practice aloud on a real text (10–15 minutes). The teacher reads a document or text set as an insider would, saying the questions asked (who wrote this, when, for whom, what else says so) and marking the text where each move happens (Demonstration, Annotating). Untested as a step; from the earlier page.
  3. Set an expert reading beside a novice one (10 minutes). Show two readings of the same text, one that takes it at face value and one that sources and corroborates it, and ask learners what the second did that the first did not. Make both readings neutral: written by the teacher or invented ("Reader A", "Reader B"), never a learner's own work shown to the group, so that no one is shamed by being the weak case; a learner's own reading is compared privately, with the teacher, if at all (untested proposal). Evidence for comparison, carried from general case comparison: comparing cases side by side improved learning and transfer by a moderate average amount over studying them one at a time (Comparing cases side by side ... [+M]; 57 experiments pooled, plus two experiments; not tested on disciplinary reading, and in one algebra study comparison raised procedural knowledge, not conceptual).
  4. Guided practice on a new text set (20–30 minutes per lesson). Learners apply the practice to two to four short texts in pairs, using a question card for the practice (for history, two question sets: probing questions that mine the source, check its creator's bias and seek corroboration, and general questions on what the author was thinking and what was left out), and recording each answer with a supporting quotation. From the folded page: the two-question-set strategy for imagining the past (a strategy page; no claim cited there).
  5. Anchor every interpretation in the sources (the folded page's recommendation). When learners reconstruct what happened or what someone thought, require that each part of the account cite a source and a place and time, so another reader could check it; an account with no link to evidence is returned as speculation, not marked wrong for imagining. Basis: theoretical only, from Collingwood as reported by one paper: the historian's imagined picture is legitimate when localised in real space and time and tied to evidence (Historical imagination ... constrained by evidence [+W]); the same paper argues this gives grounds for constructivist history teaching (Collingwood's ideas ... constructivist approaches [+W]).
  6. Argue from the texts in the discipline's genre. End each text set with a short written or spoken argument (claim, evidence from the texts, reasoning, a reply to a counterclaim) in the form the discipline uses (Argumentation). Evidence, carried from argumentation generally: online scientific argumentation activities had a medium positive effect on argument skill and content learning, and multiyear dialogic argumentation transferred to better individual essays on new topics (Argumentation Improves Reasoning [+M]). Plan explicit support for moving from discussion to independent writing: the science trial above found no gain in essays and its authors attribute that to too little such support.
  7. Plan the arc across a semester, not a lesson. A proposed arc (untested as a schedule; the tested curricula ran a semester or six months): weeks 1–3, one practice modelled every lesson on short teacher-chosen texts, with the question card and the teacher's think-aloud; weeks 4–8, learners lead the practice in pairs on longer or less familiar text sets, the teacher prompting only at the step where pairs stall; from week 9, learners choose or find the texts, run the questions without the card, and write the argument alone. Add the next practice once most learners apply the current one unaided on a new text set; move an individual learner on after two consecutive unaided text sets in which every interpretation is tied to a source. Withdrawing support follows Scaffolding and Fading.
  8. Assess on new texts, at the end of the unit and again later. If the brief gives its own rubric, use its strands as the assessment: score each strand separately on the new texts, map each disciplinary practice taught to the strand it serves, and do not add strands of your own (untested proposal). Otherwise, check comprehension and the practised skill on texts the learners have not seen, scoring each practice separately, and keep a writing measure separate from the comprehension measure. From the tested studies' designs; the horizon beyond a semester is untested here.

When a condition is not met: if the teacher cannot yet model the discipline's reading, start from a published curriculum's modelled lessons and annotated texts rather than improvising generic strategies under a disciplinary label (untested proposal; the earlier page warned that superficial "disciplinary-flavoured" activities without real epistemic work produce little benefit); if learners still labour over individual words, give fluency and vocabulary support alongside, not instead (row below).

Fitting the design to a situation

The three facts that most change the decision:

  • The learner's starting reading. A learner who cannot yet decode the text fluently has nothing left for disciplinary reasoning; one who reads well in general but treats sources as neutral is the learner the tested curricula served; one already reading as an insider needs harder texts, not more modelling. If unstated, ask: given one short primary text from this discipline, can the learner read it aloud, summarise it, and say who wrote it and why it matters?
  • Who can model the discipline's reading. The practices are taught by someone who uses them. If unstated, ask: who will read a text aloud as a historian, scientist or mathematician would, and do they have the texts and lessons to do it?
  • Which practice, on which texts, measured how and when. Gains appeared in the practice and the kind of text taught (one skill of six in one evaluation; comprehension but not writing in another). If unstated, ask: which disciplinary practice should learners be able to use unaided, on what new texts, assessed how, and when?
If the brief says… Then change… Basis
Learners are adolescent, reading adequately but new to the discipline's texts Use the default design as written, over a semester: one practice at a time, modelled, then text sets in pairs, then unaided arguments Disciplinary Literacy Instruction Improves Comprehension [+M] (ninth- and eleventh-graders, against usual or textbook instruction)
Learners still decode slowly or have low literacy Keep the discipline's texts but shorten them, gloss terms, read them aloud first; run fluency and vocabulary work alongside; reduce to one question at a time on the card Automatic word recognition frees resources for comprehension [~W] (a mechanism from theory and review, not tested as an intervention); text supports in Accessible Vocabulary & Syntax
Learners are child (primary age) Use one or two questions per source; let learners draw a picture or make a collage captioned with a quotation from the source; keep the rule that each picture points to its source untested; from the folded page's strategy (two-question-set strategy, no claim). The tested studies were with high-school students
Learners are advanced (upper-level undergraduates, graduate students) or already read as insiders Skip most modelling; give harder or contested texts, ask for critique of a published argument, and check with one unaided text set before deciding Instructional guidance that helps novices can become redundant or counterproductive as expertise grows. [~M] (cognitive-load studies, not disciplinary reading)
adult learners in work settings (nurses reading research, engineers reading standards, analysts reading reports) Use texts from their own work; model the profession's reading questions; keep sessions short and tied to a decision they must make with the text untested proposal; no claim here tests disciplinary literacy with adults
The goal is disciplinary writing (essays, lab reports, explanations) Add explicit steps from discussion to writing: a shared outline, a model paragraph in the genre, a draft with feedback on the use of evidence, then an unaided draft Disciplinary Literacy Instruction Improves Comprehension [~M] (comprehension gains, no gain in essays; the authors attribute it to too little writing support)
The goal is several historical thinking skills at once Teach and score each practice separately, give each its own lessons, and expect gains only where instruction targeted it Students learning from World History Project ... continuity and change [+M] and comparable competency in the other five skills [~M] (one evaluation of one curriculum); conceptually decoupled [~M]
Learners judge the past by present-day values Add a contextualisation step: before judging, learners list what people at the time could know and believed, citing a source for each untested as a move; the starting state is described in students approach the past with common sense, including presentism [~M] (a review, not a test)
The task is judging websites or online claims, not the discipline's own texts Teach lateral reading (leave the page, search what others say about the source) rather than document sourcing; use Evaluating Sources Civic Online Reasoning Instruction Improves Evaluation [+M] (high-school and university students, 2–8 lessons); historians' sourcing did not carry to websites (checklist claim [~M])
The teacher lacks deep disciplinary knowledge, or is adopting a new curriculum Start from the curriculum's modelled lessons; select a few activities rather than all; plan the scaffolds teachers will add, and give planning time in the first year Teachers find World History Project overwhelming in their initial adoption year [~M] (teacher interviews, one curriculum)
One teacher with a large class Model to the whole class; run text sets in pairs with one question card each; collect one written quotation-backed answer per pair per lesson and respond to the commonest gap the next day untested proposal
Scarce facilitator time (a volunteer tutor, short check-ins) Give each learner a fixed question card and annotated model text; the facilitator checks only one thing per visit: does each interpretation cite a source? untested proposal
Mixed modes (some remote, some in the room) or some learners phone-only Send the text pack before the session, both printed (posted or handed out) and as a single phone-readable file (short excerpts in large plain text, not scanned PDFs or wide tables); number the lines so anyone can point to a quotation by line number; no task requires shared editing, a laptop or a live link: phone-only learners answer each question card item by text message, a voice note or on paper photographed afterwards; pair a remote learner with a room learner by phone call for the argument step untested proposal
Two sessions only Session 1 (about 60 min): model one practice on one short text (15 min), neutral expert-beside-novice readings (10 min), guided practice in pairs on a two-text set with the question card (25 min), one quoted answer each (10 min). Between sessions: nothing required, or one optional question on a printed text. Session 2 (about 60 min): two-minute recap of the card, unaided practice on a new text set (25 min), a short argument from the texts (20 min), scored on the brief's rubric strands (10 min). Teach one practice, not three; claim only what session 2's unaided task shows untested proposal; the tested curricula ran a semester or more
Learners are tired (evening classes after work, apprentices after a shift) Shorter texts (half a page), one practice per session, talk before writing, the question card printed so nothing must be remembered; put the unaided task early in session 2, not at the end untested proposal
Single session only Model one practice on one short text (15 min), set an expert and a novice reading side by side (10 min), guided practice on one new text (20 min), one unaided question with a quotation (5 min); do not claim a lasting gain untested proposal; the tested curricula ran a semester or more
Fixed short time budget (e.g. 10 minutes a lesson inside a content course) Run one question from the card on one short source each lesson, with a one-line quoted answer; rotate practices every two weeks untested proposal
The brief names its own routine or rubric (a school's document-analysis sheet, a lab-report template, a rubric with named strands, its own categories) Keep it, and use the rubric's strands as the assessment, scored strand by strand; say which disciplinary practice serves which strand; change only what the evidence says matters: one practice targeted at a time, the discipline's own texts, every interpretation tied to a source untested; follows the tested package, not the brief's tool
Assessment criteria differ by subject (a cross-school rubric for communication or writing) Use the brief's own criteria, but let each discipline weight them in its own way, and score disciplines separately Disciplinary variation emerged in communication assessment [~W] (authors' interpretation of focus groups; STEM teachers stressing precision, humanities teachers audience)
Products or performances for a real audience (a museum label, a public history podcast, a science briefing) Keep the audience; require that every claim in the product is traceable to a cited source in a notes page the audience can see or a reviewer can check untested; see Authentic Audiences & Purposes
A whole-task performance the brief wants kept whole (a full document-based inquiry, an investigation report) Keep it whole; give the practices as checkpoints inside it (sourcing before interpreting, corroboration before claiming), not as separate worksheets untested proposal
Learners use AI tools to read or summarise texts Teach the discipline's standard for checking a summary against its sources: compare each AI claim with the text, name what it left out AI literacy requires discipline-specific calibration beyond prompt-writing skills [+W] (a review's endorsement of a framework; no data)
Private or sensitive material (family histories, community testimony, patients' records in a health course) Learners decide what is shared beyond the teacher; work from published or consented sources for class discussion; anonymise any personal source used untested proposal
High stakes (a document-based exam, a qualification) Practise in the exam's form late in the arc, but keep the practices taught in the discipline's texts earlier; judge readiness on unaided responses to unseen texts untested proposal

Ten of the twenty-four rows cite a claim. Only the first rests on a claim that tests disciplinary-literacy instruction against usual instruction; the writing row uses the same claim's null result. The rest carry claims about history curricula, source evaluation, expertise, word recognition or AI literacy to this setting and say so; the remainder are untested proposals.

Observation, state and explanation

What can be observed is a learner's response to a text under stated conditions: which text, which question was asked, whether the question card or a model was available, what the learner wrote or said, and which sources they cited. "Reads like a historian" and "has the disciplinary practice" are inferred states, and they are inferred per practice: a learner can source well and contextualise poorly (conceptually decoupled [~M]).

Response Plausible explanations Observation that could change the interpretation
Summarises the document's content and never asks who wrote it The practice was not taught or not yet transferred; the learner treats all texts as textbooks; the text was too hard to get past its content Ask the sourcing question directly on a short, easy text; if they answer it there, the issue is transfer or load, not the concept.
Judges past people as foolish or immoral by present standards Presentism, a common starting state (novice presentism [~M]); missing knowledge of the period; the task invited a verdict Give one source on what people at the time believed, then ask the question again; compare the answers.
Writes a vivid reconstruction with no citations Imagination unanchored by evidence; learner did not know citing was required; could not locate supporting text Ask them to mark, in the source, where each part of their account comes from; the parts they cannot mark show which explanation holds.
Good discussion of the texts, weak essay The writing step was not supported; genre conventions unknown; time Give the same argument as a spoken answer, then a writing frame for the genre; compare.
Applies the history question card to a website, or the web checklist to a primary source Practice learned as a routine without its purpose; the two kinds of text were not distinguished Give one of each and ask which questions fit which, and why.

These rows are proposals, untested with learners as diagnostics. A probe is itself practice, so record it, and keep the remaining explanations open.

Evidence and qualifications

Learners, disciplines and outcomes differ across these claims (ninth-grade biology, eleventh-grade history, world-history work samples, web evaluation), so do not rank them by effect label. None measures a gain beyond a school year, and none compares disciplinary with generic strategy instruction.

The evidence by question

Further evidence, not yet read against this model

Claims the earlier page cited that bear only partly on the model:

Objective and learner-valued goal

The designer's objective is usually unaided use of a named disciplinary practice on new texts of that discipline, assessed by a stated task at a stated time. Ask separately what the learner wants from these texts: to pass the course's exam, to find out what really happened in a story that matters to their family or community, to read research in the job they want, to argue a case well. Agreement and divergence are both observations. A learner who wants the exam may want the practice reduced to a template; one who wants the story may resist sources that complicate it, and that resistance is where corroboration earns its place.

For example, a history class's objective may be sourcing and corroborating primary documents on a migration, while a student wants to know why their grandparents left. A text set that includes a family letter or oral-history excerpt (with the family's consent) serves both, and the sourcing questions apply to it as to any document; record whether the learner uses them there too.

What would revise this model?

The expectation should weaken if comparisons with well-taught generic comprehension-strategy instruction, on the same texts and outcomes, show no advantage for disciplinary instruction; if gains on new texts vanish at a delay beyond the course; or if teachers without deep disciplinary knowledge, using a published curriculum, produce the same gains (then teacher knowledge is not the boundary the earlier page assumed). If disciplinary writing improves only when explicit writing support is added, the default design's writing step becomes required rather than recommended. If the expert–novice reading contrast adds nothing to modelling alone, drop it. Do not protect the model by calling every null result "superficial implementation" after the fact.

Comprehension of new texts, the practised disciplinary thinking skill, disciplinary writing, factual knowledge and use outside the course are separate claims. The present evidence establishes no dose per practice and no criterion for moving on.

Claims

  • Annotating — annotation schemes can be discipline-specific (e.g., sourcing marginalia in history), making expert reading moves visible
  • Cognitive Apprenticeship — the core instructional model: experts model disciplinary reading and reasoning, then fade support
  • Case Studies/Case-Based Learning — cases supply the authentic disciplinary materials learners need to practice expert heuristics
  • Accessible Vocabulary & Syntax — scaffolds language demands so learners can engage with authentic disciplinary texts
  • Active Learning — disciplinary literacy requires learners to do epistemic work, not just receive disciplinary content
  • Epistemic Cognition — the standards for judging a kind of evidence, in any domain; this page applies them to a discipline's own texts
  • Evaluating Sources — lateral reading of unfamiliar web sources, a civic practice distinct from sourcing a discipline's documents
  • Cognitive Apprenticeship pattern — modelling, coaching and fading inside authentic work, the hand-over this page uses

Examples

Illustrative

Historical Document Analysis (Stanford History Education Group) — The Reading Like a Historian curriculum models sourcing, corroboration, and contextualization with primary documents, then has students apply the same heuristics to new documents. Breakstone, Smith & Wineburg report gains in historical thinking relative to textbook-based instruction.

Civic Online Reasoning curriculum — Teaches the lateral-reading and source-evaluation strategies that professional fact-checkers use, with documented improvements in learners' evaluation of online information [+M].

Science notebooks and claim–evidence–reasoning frameworks — Learners record observations, construct claims, and justify them with evidence in the genre conventions of scientific writing, practicing disciplinary argumentation rather than generic summary writing.

Mathematical justification routines — Learners critique and construct mathematical arguments ("Convince a skeptic"), reading and producing proofs as disciplinary texts rather than only solving exercises.

Annotated journal-article reading in undergraduate STEM — Students read a primary research article with instructor annotations explaining how a scientist reads (methods critique, results interpretation), then apply the same moves to a second article — a Demonstration-then-practice structure.

Key Sources

  • Shanahan, T., & Shanahan, C. (2008). Teaching disciplinary literacy to adolescents: Rethinking content-area literacy. Harvard Educational Review, 78(1), 40–59. doi:10.17763/haer.78.1.v62444321p602101
  • Moje, E. B. (2008). Foregrounding the disciplines in secondary literacy teaching and learning: A call for change. Journal of Adolescent & Adult Literacy, 52(2), 96–107. doi:10.1598/JAAL.52.2.1
  • Wineburg, S. S. (1991). Historical problem solving: A study of the cognitive processes used in the evaluation of documentary and pictorial evidence. Journal of Educational Psychology, 83(1), 73–87. doi:10.1037/0022-0663.83.1.73
  • Goldman, S. R. (2012). Adolescent literacy: Learning and understanding content. The Future of Children, 22(2), 89–108. doi:10.1353/foc.2012.0011
  • Shanahan, C., Shanahan, T., & Misischia, C. (2011). Analysis of expert readers in three disciplines: History, mathematics, and chemistry. Journal of Literacy Research, 43(4), 393–429. doi:10.1177/1086296X11424071
  • Lemisko, L.S. (2004). The Historical Imagination: Collingwood in the Classroom. Canadian Social Studies. https://www.quasar.ualberta.ca/css. doi:10.29173/css130
Page metadata
Field Value
type principle
id disciplinary-literacy
aliases recognize-imagination-in-learning-and-knowing
title Disciplinary Literacy
description For adolescents who read adequately in general but meet a discipline's texts as novices, a semester or more of instruction that names, models and has them practise the discipline's own reading and reasoning moves on its own texts is expected to improve comprehension of new texts in that discipline and the practised kind of disciplinary thinking; two studies test this against usual or textbook instruction, none against generic strategy instruction, and the one writing measure showed no gain.
status review
canonical True
generated by claude/unspecified on 2026-10-14
trust unverified — no one has confirmed this page's content
sources 1 — lemisko-2004

Frontmatter is the page's machine-readable record — see the Schema & Guide.