The Documents Model Framework explains how readers build an integrated understanding from multiple documents while preserving where individual claims came from. Developed in 1999 research by Britt, Rouet and Perfetti and refined in later scholarship, it distinguishes a content-focused integrated model from an intertext representation of sources and their relationships. A reader can understand each document separately yet still fail to integrate them or correctly attribute claims.
Did you know that copying several articles into one summary can make an issue look simpler than the evidence really is? Two sources may disagree because they used different dates, populations or methods, while a third repeats an earlier report without independent verification. Vocabulary becomes a technology of evidence when readers know not only what each document asserts but which source supports it and how its claim relates to the others.
This guide explains document nodes, source–content links, source–source relationships, integrated mental models and the risk of a source-free ‘mush model’. It uses original examples and research from the 1999 Documents Model programme, 2013 multiple-documents literacy scholarship and later studies of intertext formation, with applications to Primary and Secondary English, Science, Mathematics, History, General Paper and translation.
Research note: Examples and classroom diagrams are original illustrations rather than new experimental data. Source quality and factual truth must be evaluated separately from the mere existence of a coherent document representation.
Contents: understanding several sources without losing their origins
Open 50 chapters and multiple-source exercises
- 1. Multiple documents need more than one summary
- 2. The 1999 theoretical foundation
- 3. A single-text situation model is not sufficient
- 4. A document is more than its body text
- 5. The documents model has related representations
- 6. An integrated mental model captures the issue
- 7. The intertext model represents source relations
- 8. A source node represents a document’s origin
- 9. Source–content links answer who said what
- 10. Source–source links express document relationships
- 11. Claim-level links are more precise than topic links
- 12. A three-document school example
- 13. The same claim can have several sources
- 14. A source can contain multiple claims
- 15. Source separation prevents a mush model
- 16. Remembering sources without integration is also limited
- 17. A tag-all representation can be demanding
- 18. Source nodes can carry motive and expertise
- 19. Publication date is often essential
- 20. Genre affects how claims should be read
- 21. Source prompts can strengthen the intertext model
- 22. The 2016 secondary-school research
- 23. The 2013 multiple-documents literacy review
- 24. Comparing sources requires matched claims
- 25. Factual disagreement can be genuine
- 26. Complementary sources can answer different questions
- 27. A conflicting source may change the research question
- 28. D-ISC investigates attention prompted by discrepancy
- 29. Braasch and colleagues tested discrepancy effects
- 30. Source monitoring is a related memory framework
- 31. Text–belief consistency can complicate integration
- 32. A coherent integrated model can still be inaccurate
- 33. An intertext model is not a causal network
- 34. Multiple-document reading is not simply summarising
- 35. Source information can be forgotten during drafting
- 36. Academic citation supports the model externally
- 37. Historical documents provide a natural example
- 38. Scientific literature uses different kinds of evidence
- 39. Numbers require claim-level attribution
- 40. The same author may publish different versions
- 41. Websites can change after publication
- 42. An author’s purpose can shape claims
- 43. Primary English can begin with two notices
- 44. Secondary English can compare two perspectives
- 45. Mathematics needs comparable data sources
- 46. Science requires methods and replication
- 47. General Paper needs a defensible synthesis
- 48. Translation must keep source identities distinct
- 49. A seven-column documents-model worksheet
- 50. The final task is an integrated answer with traceable claims
Quick Test: which connections belong in a documents model?
| Component | What it represents | Classroom example |
|---|---|---|
| Document node | A source and relevant metadata | Official notice, publisher, date and purpose |
| Source–content link | Which document made a particular claim | Notice A states nine o’clock |
| Source–source link | How the documents relate | Notice B supersedes the earlier schedule |
| Integrated mental model | The coherent overall issue | What opening time now governs the visit |
| Intertext model | Origins and cross-document relations | Which claims agree, conflict or repeat another source |
This table illustrates the representational distinctions. It does not assign automatic credibility or factual truth to any particular source.
1. Multiple documents need more than one summary
The Documents Model Framework explains how readers may build a coherent understanding when several documents discuss the same issue. Understanding each article separately is not enough if the task requires comparing their claims, sources and relationships.
A strong representation connects what the documents say with who said it. It also records where accounts agree, differ or supplement one another. This is why a pupil who remembers several facts without their origins can still struggle to explain conflicting evidence fairly.
2. The 1999 theoretical foundation
Research by Mary Anne Britt, Jean-François Rouet and Charles Perfetti in 1999 helped establish this approach. Work included Content Integration and Source Separation in Learning from Multiple Texts and Toward a Theory of Documents Representation.
The authors extended accounts of single-text comprehension to address tasks where different sources provide competing or complementary content. These were theoretical and empirical contributions, not a universal recipe guaranteeing that students will evaluate all web pages perfectly. Later studies refined the model and tested instructional support.
3. A single-text situation model is not sufficient
A reader studying one article may construct a situation model of the events or concepts it describes. That representation can include people, times, actions and their causal relationships.
When two articles disagree, however, merging them into one unnamed story can destroy the disagreement. A multi-document representation needs to preserve the source of each proposition so that opposing accounts can coexist without being mistaken for one author making contradictory claims.
4. A document is more than its body text
The framework treats a document as a communication made by an identifiable source in a context. Relevant features may include author, publication, date, genre, qualifications, purpose and audience.
These features do not automatically decide whether a claim is true. Their value is that they help readers understand why a particular claim was made and what evidence supports it. A student should inspect the actual methods rather than assume a polished publication title guarantees correctness.
5. The documents model has related representations
Later formulations commonly distinguish an integrated mental model of the issue from an intertext model representing the sources and links between documents and their contents.
The first concerns what the reader understands about the topic. The second concerns which source advanced each claim and how the documents relate. Together they support a synthesis that remains coherent without erasing disagreement or source identity.
6. An integrated mental model captures the issue
A learner reading three explanations of a city transport proposal may form a representation of the project’s purpose, expected effects and unresolved questions. That is a content-level understanding across the available texts.
But the integrated model is not necessarily a verdict. It may contain several competing hypotheses when evidence remains incomplete. The reader should avoid creating an artificial single conclusion simply because one diagram looks tidier than several carefully qualified claims.
7. The intertext model represents source relations
An intertext model helps the reader record which document made which assertion and how that document relates to others. One report might support an earlier claim, another contradict it and a third provide a new measurement.
These are relationships among sources, not automatically among factual realities. Two articles can disagree because they use different definitions or refer to different years. Source links should preserve those differences until comparison establishes what is genuinely incompatible.
8. A source node represents a document’s origin
In the framework, a source or document node can represent information about the author, publication and circumstances of a text. It is not simply a decorative icon.
A learner might record that Document A is an original research article and Document B is a promotional summary. The different genres matter to interpretation, but they do not alone establish the correctness of every sentence. The source node offers context for further evaluation.
9. Source–content links answer who said what
A source–content link connects a particular claim with the document that advanced it. This prevents a familiar proposition from losing its origin when the reader synthesises several accounts.
For example, Document A might report one survey estimate and Document B a different estimate. The final summary should preserve which source produced each number. If it assigns both figures to Document A, it has a source-link problem even if both numbers were remembered accurately.
10. Source–source links express document relationships
A source–source link indicates that documents agree, disagree, support or otherwise relate to one another. The relation requires a stated basis, such as compatible measurements or explicitly conflicting conclusions.
The words agrees with and corroborates are not always interchangeable. Corroboration depends on relevant independence and evidence. Two articles copying one press release do not supply two independent observations merely because they share a conclusion.
11. Claim-level links are more precise than topic links
Two articles can discuss the same topic while making different claims. Knowing both concern education policy does not show that they disagree about a particular measure or causal mechanism.
A precise ledger identifies each proposition before comparing it. A reader should note the statement, relevant qualifiers and evidence, then draw the relation. Vocabulary skills matter because small words like some and always can change whether two claims genuinely conflict.
12. A three-document school example
Imagine three fictional documents about a library’s opening hours. An old printed notice says nine, a new official notice says ten from a specified date and a student newsletter repeats the old schedule.
A documents model retains all three as distinct communication events. It then marks which claim is current, which is outdated and which may have copied old information. The exercise illustrates source tracking, not an empirical experiment.
13. The same claim can have several sources
When a website quotes another publication, the claim appears in both places but may originate with only one. A reader should identify whether the second document supplies independent evidence or merely repeats the first.
This protects against an apparent consensus created by copying. A count of five matching headlines may still represent one underlying report. Source attribution and intertext links make the chain visible instead of treating repetition as five independent confirmations.
14. A source can contain multiple claims
An article might report a measured result, interpret its significance and propose a policy response. Those are different propositions with potentially different evidential support.
A good source–content map connects each specific claim to the author and marks its type. A student who quotes the interpretation as though it were a direct instrument measurement has changed its status. Vocabulary such as observed, suggests and recommends maintains the distinctions.
15. Source separation prevents a mush model
The literature describes a mush model in which readers merge topic information into a general understanding while losing which document supplied which claim. The content may feel coherent, but provenance has been neglected.
This is especially risky when documents conflict. The reader may combine two incompatible numbers into one invented average or attribute opposing claims to the same author. A source ledger preserves distinctions that unaided memory may not keep.
16. Remembering sources without integration is also limited
The opposite problem occurs when readers remember each document separately but cannot connect the evidence into a meaningful answer to the overall question. They may provide a list of summaries without synthesis.
A strong documents model needs both source separation and cross-text integration. The final account should explain what the sources collectively establish, where their claims differ and what remains unresolved. A bibliography alone is not an argument.
17. A tag-all representation can be demanding
Some discussions of the framework distinguish a highly detailed representation that tags almost every remembered fact to a source. That can be cognitively demanding, particularly for less experienced readers.
The educational goal need not be recording every sentence exhaustively. It is preserving the important claims and their origins accurately enough for the task. A selective, evidence-based documents model can be more useful than either a source-free summary or an unmanageable transcript.
18. Source nodes can carry motive and expertise
A scientist reporting an original experiment, a journalist summarising several studies and a retailer promoting a product have different relationships to the evidence. Source nodes can represent such characteristics.
These are contextual clues, not absolute rankings of truth. A scientific paper can be flawed and a news article can accurately quote a strong result. A reader must examine whether the source’s expertise and methods match the specific proposition under evaluation.
19. Publication date is often essential
A document may have been accurate when published but no longer describe current rules, prices or results. A later document can change the operative state without making the older source dishonest.
The source node should therefore include the relevant date or version. In a factual comparison, the reader needs to know whether the documents describe the same period. An outdated fact is not always a direct contradiction of a newer fact.
20. Genre affects how claims should be read
A systematic review synthesises studies, an original experiment reports a particular method and result, and an opinion article argues from selected evidence. These documents serve different communication purposes.
The learner should identify the genre before weighing claims. A promotional slogan may be rhetorically effective but provide no measurement, while an original study may contain narrow findings that cannot support a broad universal conclusion. Accurate vocabulary distinguishes assertion from evidence.
21. Source prompts can strengthen the intertext model
Research on multiple-document instruction has tested prompts asking readers to attend to author, publication and the links between claims and sources. Such prompts can help students construct a more detailed intertext representation.
That does not guarantee a permanent critical-thinking gain from one worksheet. The result depends on the design, learners and assessment. A teacher should check whether pupils actually attribute a novel claim to the correct document after instruction.
22. The 2016 secondary-school research
Research on source prompts for secondary-school students reading sets of websites has examined whether asking students to consider provenance improves their intertext-model formation.
The relevance to tuition is direct: pupils increasingly compare digital sources for coursework. A prompt can shift attention from only collecting facts toward understanding who made them. But source awareness must be followed by assessment of evidence, not simply copying author names into a table.
23. The 2013 multiple-documents literacy review
Britt, Rouet and colleagues’ scholarship on multiple-documents literacy emphasises strategic processing, source awareness and argumentation when students read conflicting documents. The model extends single-text textbase and situation-model representations.
The review describes document nodes, source–content links and interdocument relations. Its value is explaining why critical reading is not simply absorbing one author after another. Students need a structure that keeps perspectives distinct while integrating defensible content.
24. Comparing sources requires matched claims
If one report measures monthly traffic and another annual traffic, the numbers are not directly comparable without conversion and other relevant controls. If one uses respondents and another households, the populations differ.
The documents model helps retain those definitions under the correct sources. A learner should not call different measurements inconsistent merely because their numerical values differ. The relation among claims must respect units, samples and time.
25. Factual disagreement can be genuine
Two sources can make incompatible claims about the same event, with the same period and definitions, after relevant scope differences have been ruled out. In that case, a reader should preserve the disagreement rather than forcing an early compromise.
The next task is evidence evaluation. One source may use a more direct method or have stronger corroboration. Balanced comprehension does not require judging both equally credible. It requires fair reconstruction of both claims before a justified comparison.
26. Complementary sources can answer different questions
A government record might establish the date of a decision, while a historical study discusses its longer-term effects. These documents may complement one another rather than compete.
A synthesis can use each source for the claim it is qualified to support. It would be misleading to demand that the date record also prove the later causal interpretation. Source–content links prevent a useful primary source from being assigned an argument it never made.
27. A conflicting source may change the research question
A student might begin by asking what happened and discover that the sources disagree. The more precise task becomes identifying the source of the discrepancy and which evidence could resolve it.
The documents model can accommodate that change by retaining the competing claims and their contexts. A good report should not conceal the unresolved issue. Clearly stated uncertainty can be the accurate result of multiple-document reading.
28. D-ISC investigates attention prompted by discrepancy
The Discrepancy-Induced Source Comprehension assumption predicts that encountering conflicting statements can prompt readers to pay closer attention to who made each claim.
This is a more specific processing hypothesis than the Documents Model Framework’s broad representational account. A reader may build source links for many reasons, including explicit prompts or task demands. The two theories should not be treated as interchangeable.
29. Braasch and colleagues tested discrepancy effects
Braasch, Rouet, Vibert and Britt published Readers’ Use of Source Information in Text Comprehension in 2012. Their experiments compared news reports with consistent versus discrepant assertions attributed to different sources.
They found enhanced source mention and memory in discrepant conditions, including eye-movement differences near source areas. That supports the specific hypothesis under the tested conditions. It does not establish that every contradiction automatically makes a reader a skilled fact-checker.
30. Source monitoring is a related memory framework
The Source-Monitoring Framework studies how people attribute remembered information to origins based on memory characteristics and judgement processes. It applies to perception, imagination, testimony and more.
The Documents Model Framework focuses on representing and integrating multiple texts while preserving document-level provenance. A reader can use knowledge from both. Yet confusing their theoretical mechanisms would obscure the distinct questions of memory attribution and cross-document comprehension.
31. Text–belief consistency can complicate integration
A reader may find a claim agreeing with a prior belief easier to accept and remember than a conflicting one. That tendency can distort the integrated model if the sources’ actual positions are not reconstructed fairly.
The learner should create a source ledger before forming a final conclusion. The source model helps prevent a familiar argument from erasing an opposing document; the evidence review then determines which interpretation is justified.
32. A coherent integrated model can still be inaccurate
Several articles can repeat one incorrect claim, making the assembled narrative appear consistent. The documents model may represent the agreement accurately while the underlying information remains false.
Factual verification therefore requires examining primary evidence, replication and source quality. Coherence and corroboration are not the same. Copying one error across several documents creates more text, not necessarily stronger evidence.
33. An intertext model is not a causal network
A source–source link indicating disagreement is a relation among documents. A causal link connecting two historical events is a relation among events in the world or story.
A pupil should label the arrow according to what it means. The fact that two reports disagree does not cause the reported event to occur. Diagrammatic precision is a vocabulary skill because one arrow can otherwise hide very different logical claims.
34. Multiple-document reading is not simply summarising
Writing a paragraph about each document separately may be a useful first step, but the final synthesis must explain their relationships. A sequence of disconnected summaries leaves the comparison unfinished.
Ask students where the documents agree, which claims conflict and how their evidence differs. Then formulate a response to the original research question. The documents model is valuable because it preserves source identity within this integrated answer.
35. Source information can be forgotten during drafting
A pupil may take notes from three websites and later combine them into a paragraph without retaining which sentence came from which. The final writing can inadvertently attribute an idea to the wrong publication.
A structured note system reduces this risk. Every claim-level note should include an identifier or link to the exact source location. A bibliography pasted at the bottom without claim-level connections does not repair uncertain attribution.
36. Academic citation supports the model externally
A citation such as author, year and page can serve as an external record of source–content relations. It helps a later reader verify whether the claim was drawn faithfully from the cited publication.
The reference must actually support the sentence. An impressive list of sources cannot justify a statement the cited papers never made. Proper citation is not merely formatting; it is a practical way to preserve the intertext model.
37. Historical documents provide a natural example
Imagine one eyewitness describing a fire, another witness disputing the time and a later official record providing the incident date. A History student should preserve each source’s vantage point and what it can reliably establish.
The reader must also distinguish direct observation from later interpretation. A witness can be sincerely mistaken. The documents model helps organise competing accounts, but source criticism and corroboration decide what conclusion is warranted.
38. Scientific literature uses different kinds of evidence
An original experiment, a replication and a systematic review may all address the same hypothesis while making different claims about confidence and generality.
A student who treats them as three equal single observations loses the structure of the evidence. The source ledger should identify study design and whether documents report independent data. Vocabulary such as replicated, estimated and reviewed communicates those differences.
39. Numbers require claim-level attribution
One article might report a sample size, another an effect estimate, and a third a later meta-analytic result. Combining the numbers into a new statement without their proper relationships can create a false conclusion.
Write each statistic with its source, population, period and measurement unit. Ask whether the figures can be compared or synthesised. The documents model supports accurate evaluation precisely because content remains connected with its origin.
40. The same author may publish different versions
An original paper can be followed by a correction, response or revised analysis. A learner should not merge all versions without noticing that the evidence or conclusion changed.
Treat version and date as part of the source node. A new document may supersede the earlier calculation or clarify a limitation. The final account should state which result is operative and why, rather than quote an outdated version simply because it was read first.
41. Websites can change after publication
A live web page may be updated without changing its headline. Students should record access dates or archived versions where required by the assignment, especially for dynamic factual information.
The goal is not to create an elaborate citation for every casual reading task. It is to preserve enough provenance when claims will be used in graded, public or consequential work. A durable source link makes later verification easier.
42. An author’s purpose can shape claims
A company brochure may emphasise potential benefits, while an independent evaluation may document limitations. The two sources can describe the same product through different rhetorical purposes.
A reader should not assume that commercial purpose makes every factual statement false, nor that academic status makes a claim infallible. Instead, identify each document’s purpose, supporting evidence and omissions. Source awareness informs proportionate judgement rather than automatic dismissal.
43. Primary English can begin with two notices
A pupil reads two fictional announcements about a school event. One is an older draft; the other is the confirmed current notice. Ask who issued each, what the announcements say and which one governs attendance.
This exercise introduces source–content connections without demanding abstract research jargon. A child can point to the exact date and issuer, then explain why a familiar earlier notice is no longer decisive. The achievement is accurate real-world reading.
44. Secondary English can compare two perspectives
A student reads an article supporting a school policy and another questioning its effects. The teacher asks for faithful paraphrases of both, including the qualifications each author used.
Then the student identifies a shared premise, a genuine disagreement and the evidence each source provides. A strong answer need not give both positions equal credibility. It must represent them correctly before evaluating their relative support.
45. Mathematics needs comparable data sources
A mathematical modelling problem may supply statistics from different sources with different denominators. The pupil should not add, divide or average the figures until the definitions have been aligned.
A documents model can help organise metadata beside the data table. The calculation remains a mathematical task, but good source representation prevents invalid inputs. A precise number built from incomparable measures can still be wrong.
46. Science requires methods and replication
A Science assignment may compare a classroom experiment with a published research study. Both might discuss the same process but use different instruments or controls.
The student should identify what each experiment actually measured and whether the conditions are comparable. The documents model helps prevent treating classroom observations as a direct replication when important design features differ. Scientific evidence requires more than similar vocabulary.
47. General Paper needs a defensible synthesis
A GP student may integrate several sources to discuss the consequences of a policy. The essay should separate observed outcomes from stakeholder opinions and predictions.
Use reporting phrases that preserve attribution and uncertainty. The final paragraph can reach a clear conclusion when evidence supports one, while acknowledging unresolved limitations. A strong synthesis is neither an anonymous collection of claims nor an artificial balance between unequal evidence.
48. Translation must keep source identities distinct
A multilingual dossier can contain statements from several agencies or individuals. A translation that drops quotation markers or merges similar department names can change which entity was responsible for an action.
An intertext ledger helps preserve source and content links across languages. The translator should verify original author, date and claim scope before finalising the target-language summary. Fluency cannot excuse changed attribution.
49. A seven-column documents-model worksheet
Record document ID, author or organisation, date, principal claim, supporting evidence, links to other documents and limitations. A separate column can note the reader’s provisional conclusion.
The worksheet makes multiple-document reading inspectable. It should distinguish actual agreement, copying and independent corroboration. It is a pedagogical tool rather than a literal record of a participant’s mental nodes, but it can support reliable academic reasoning.
50. The final task is an integrated answer with traceable claims
After reading several sources, a student should be able to explain the topic coherently, specify which document supports each important proposition and identify where the evidence is incomplete or disputed.
The Documents Model Framework provides a vocabulary for that accomplishment: integrated content, intertext relations and source–content links. Understanding becomes reliable when the final synthesis preserves both meaning and provenance instead of losing either in a smooth, anonymous story.
Full worked case: three documents and one library schedule
A fictional scenario contains an old printed notice saying the library opens at nine, a current official notice changing opening to ten from a specified date and a student newsletter repeating the older hour. The student must understand the topic and also retain which document supplied each time.
| Document | Claim | Relation and status |
|---|---|---|
| A — old notice | Opening at nine | Historically accurate for an earlier schedule |
| B — current official notice | Opening at ten after the effective date | Controls the present visit under the scenario |
| C — student newsletter | Opening at nine | Repeats the older value; not independent confirmation |
The correct conclusion is not an average of the times. It is that the current authorised notice governs the visit, while the older reports remain understandable historical documents. The exercise preserves both content integration and provenance.
Eight-stage multiple-document synthesis routine
- Identify the research question and the precise claim to be compared.
- Create a source node for every document with author, date and genre.
- Record each important proposition under the document that actually states it.
- Compare definitions, time periods, units, populations and measurement methods.
- Label source–source relations such as agreement, genuine conflict or copying.
- Construct an integrated explanation without erasing unresolved differences.
- Evaluate evidence quality separately from mere source familiarity.
- Write a final answer whose important claims remain traceable to supporting documents.
Frequently asked questions
1. What is the Documents Model Framework?
A theory of comprehension across documents that preserves both integrated content and the origins of claims.
2. Who developed the original model?
Researchers including Britt, Rouet and Perfetti in 1999.
3. What is an integrated mental model?
A coherent representation of the topic or situation discussed across documents.
4. What is an intertext model?
A representation of the sources and how document claims and origins are connected.
5. What is a document node?
A representation of a source with relevant information such as author, date, genre and purpose.
6. What is a source–content link?
The relationship connecting a specific claim to the document that made it.
7. What is a source–source link?
A relationship of agreement, disagreement or other relevant connection between documents.
8. What is a mush model?
A merged understanding of content that loses important source attributions.
9. Does multiple-document comprehension require treating all sources equally?
No. Faithful reconstruction comes first, followed by justified weighting of evidence quality.
10. What should students practise?
Produce a coherent synthesis in which important propositions can still be traced to their sources.
Original scholarship and connected vocabulary reading
- Charles Perfetti: Academic Publications — primary bibliography for the 1999 Documents Model chapters.
- Multiple-Documents Literacy: Strategic Processing, Source Awareness, and Argumentation (2013) — the intertext model and source–content links.
- Measuring Multiple Text Integration: A Review (2018) — detailed review of Documents Model research.
- Toward a Typology of Integration: Examining the Documents Model Framework — types of document representations and integration.
- Fostering Secondary-School Students’ Intertext Model Formation: The Effectiveness of Source Prompts — classroom-relevant experiment.
- Investigating the Roles of Document Presentation and Reading Interactions on Multiple Document Comprehension — later empirical application.
- Braasch, Rouet, Vibert and Britt (2012): Readers’ Use of Source Information in Text Comprehension — discrepancy-related sourcing experiments.
- Vocabulary | What Is the Source-Monitoring Framework? — the distinct psychology of remembering origins.
- Vocabulary | What Is the Text–Belief Consistency Effect? — how prior beliefs affect conflicting-text comprehension.
- Vocabulary | What Is Discourse Validation? — checking propositions for consistency.
- Vocabulary Learning Hub — language comprehension, source awareness and reliable reasoning.
The Documents Model Framework shows that understanding multiple texts requires both coherent integration and reliable source separation. A strong reader can explain the shared issue, identify who advanced each claim and compare evidence without silently merging contradictions. Vocabulary becomes useful when a persuasive synthesis remains transparent enough for others to verify.