## Reading is not only getting the words right
A child can pronounce every word in a paragraph and still not understand the paragraph.
That fact is obvious to any teacher who has listened to a fluent reader finish a passage and then answer, “I don’t know,” when asked what it meant. The difficulty is that comprehension failure can hide. Word reading is visible. Meaning is assembled internally.
Consider this short passage:
*The reservoir fell below its usual level after months of weak rainfall. Consequently, restrictions were introduced. These measures reduced household demand, but they did not solve the underlying shortage.*
A reader needs more than the dictionary meanings of *reservoir*, *rainfall*, *restrictions*, *household* and *shortage*. They need to connect rainfall to reservoir level, recognise that *consequently* signals a result, understand that *these measures* refers to the restrictions, distinguish reduced demand from increased supply, and carry the idea of the “underlying shortage” across the final sentence.
Comprehension is built from connections.
This is why current literacy research is moving beyond the idea that reading comprehension is a bag of isolated strategies. Harvard Graduate School of Education’s June 2026 review of the science of reading comprehension emphasises that readers always read *something*: topic knowledge, vocabulary, language comprehension, syntax and connections across sentences, texts and experiences all matter. A 2026 study by Buchanan and Uccelli specifically examines “connective knowledge” and reading comprehension in upper-elementary students. The phrase is useful because it points to what strong readers do continuously: they join pieces into a coherent model.
The educational implication is important. If we teach vocabulary as isolated definitions, comprehension questions as isolated tricks and background knowledge as random facts, we may strengthen pieces without strengthening the connections that make those pieces usable.
Strong reading is not just word recognition plus memory. It is the construction of a connected meaning system.
## 1. A text does not contain all of its meaning on the page
Writers leave things unstated because language would become unbearable if every relationship were explained explicitly.
“The glass fell. Maya reached for the broom.”
The text does not say the glass broke. It does not say shards are on the floor. It does not say a broom is used to gather them. A competent reader builds those links from world knowledge.
“Jin forgot his umbrella. By lunch, his blazer was soaked.”
The text may never use the word *rain*. The reader supplies it.
As texts become more academic, the missing links become more demanding. A science textbook may describe evaporation, condensation and precipitation in separate paragraphs and expect the learner to understand the cycle. A history text may state that taxation rose, protests spread and political support weakened without explicitly spelling out every causal relationship. An argumentative essay may introduce a claim, qualify it several sentences later and then refer back to the whole idea with a phrase such as “this limitation”.
Reading comprehension therefore depends on what the learner can connect to what the writer provides.
This is not permission to guess wildly. Good readers build inferences constrained by the text and by reliable knowledge. The skill is not “make any connection”. It is “make the connection that makes the text coherent and remains consistent with evidence.”
## 2. Connective knowledge is different from simply knowing more facts
Background knowledge matters. A student who knows what a volcano is will understand a volcano passage more easily than a student encountering the concept for the first time. But knowledge becomes especially powerful when it is organised.
Suppose two students know the following facts:
– plants need light;
– chlorophyll is involved in photosynthesis;
– photosynthesis produces glucose;
– glucose can be used in respiration;
– leaves contain many chloroplasts.
One learner remembers these as five separate statements. Another can connect them into a model: leaves contain chloroplasts; chlorophyll captures light energy; photosynthesis uses that energy to produce glucose; the glucose becomes stored chemical energy and can later be used in respiration.
The second learner does not merely possess more facts. They possess more **relations among facts**.
That relational structure helps when a new text says, “A plant kept in darkness may survive temporarily by using stored carbohydrates.” The reader can connect darkness to reduced photosynthesis and stored carbohydrate to previously produced glucose.
This is the difference between a pile and a network.
Reading benefits from networks because new information has somewhere to attach.
## 3. Vocabulary is a network, not a row of dictionary cards
Vocabulary instruction often begins with a sensible move: word, meaning, example. The danger is stopping there.
A learner who knows that *scarce* means “not enough” has some useful knowledge. A learner who also connects *scarce* to *scarcity*, *shortage*, *limited supply*, *abundant*, *ration*, *allocate* and *compete* has a richer semantic network.
That network supports comprehension because texts rarely repeat the same word. A passage may introduce “water scarcity” and later refer to “limited supply”, “the shortage” and “insufficient reserves”. A reader who treats each phrase as a new topic loses coherence. A reader who recognises the shared concept maintains the chain.
Word knowledge should therefore include several dimensions:
**Meaning:** What does the word mean here?
**Relations:** Which words are similar, opposite, broader or narrower?
**Morphology:** How does the word change form?
**Collocation:** Which words commonly appear near it?
**Register:** Is it conversational, formal, technical or literary?
**Context:** How does its meaning shift across subjects or situations?
Take *significant*. In everyday speech it can mean important. In statistics, “statistically significant” has a specialised meaning related to evidence under a model; it does not simply mean “large” or “important”. A student who has only one definition may misread the academic text.
Vocabulary depth is connective knowledge at the word level.
## 4. Small linking words can carry large logical loads
Students and teachers naturally notice nouns and technical terms. Yet some of the most important comprehension work is done by small words and phrases:
*however, therefore, although, in contrast, similarly, nevertheless, because, consequently, for example, in other words, despite, whereas, as a result.*
These are connective signals. They tell the reader how one idea relates to another.
Compare:
“Recycling reduces demand for some raw materials. **However**, collection and processing also use energy.”
“Recycling reduces demand for some raw materials. **Therefore**, collection and processing also use energy.”
Only one connective fits the intended logic. The second sentence has changed from contrast to consequence and no longer makes sense.
Academic reading requires students to notice these relational signals quickly. Harvard literacy researcher Paola Uccelli has long highlighted the importance of academic language features that go beyond technical vocabulary, including connectives and reference chains. These features can be invisible because fluent adult readers process them automatically.
Teachers can make them visible by asking one question repeatedly: **What relationship is this word signalling?**
Not “What does *however* mean?” but “What is being contrasted?” Not “What does *therefore* mean?” but “Which earlier idea is being used as the cause or reason?”
The connective only matters because it connects actual propositions.
## 5. Pronouns and reference chains can break comprehension quietly
Consider:
“Urban trees reduce surface temperatures by providing shade and through evapotranspiration. This effect can lower cooling demand in nearby buildings. It may also vary according to species, canopy density and local climate.”
What does *this effect* refer to? What does *it* refer to?
Strong readers resolve these references automatically. Struggling readers may keep moving with a fuzzy impression, then discover two paragraphs later that the meaning has collapsed.
Reference chains include pronouns and compressed phrases such as:
– this problem;
– these conditions;
– the change;
– such measures;
– the former;
– the latter;
– this process;
– the result.
The phrase may point back to a single noun, an entire event or a complex idea spanning several sentences.
A powerful reading routine is **link the pointer**. When students meet *this process*, make them underline the exact earlier process. When they meet *these factors*, list the factors. When a pronoun is ambiguous, compare possible referents and decide which keeps the text coherent.
This is not grammar for grammar’s sake. It is meaning maintenance.
## 6. Sentence comprehension is necessary but not sufficient
A student may understand each sentence individually and still fail to understand the paragraph.
Imagine:
1. A country imports most of its energy.
2. Global fuel prices rise sharply.
3. Domestic transport and production costs increase.
4. Consumers reduce spending on some non-essential goods.
5. Businesses delay investment.
Each sentence is straightforward. The comprehension challenge is causal integration. The reader must connect energy dependence to exposure to global prices, prices to production costs, costs to household budgets and demand, and demand to business decisions.
This is where many “comprehension strategies” become too shallow. A student can highlight a key sentence without building the chain.
Ask them to represent the relations:
**import dependence → price shock → higher costs → reduced spending → weaker investment**
A diagram, causal chain or short oral explanation can reveal whether the paragraph has become a system in the learner’s mind.
Strong reading often looks like invisible model-building.
## 7. Reading across paragraphs requires memory for the argument, not every sentence
Longer texts place a different demand on the reader. They cannot hold every sentence in working memory. They need to compress.
After a paragraph, strong readers retain something like a gist representation: “This paragraph explains why the first policy reduced demand but did not fix supply.” The exact wording fades while the structural role remains.
Students who do not compress may experience each paragraph as a fresh start. By page three, they have many local details but no global argument.
Teach **paragraph jobs**:
– introduces the problem;
– gives evidence;
– explains a mechanism;
– provides a counterexample;
– qualifies the claim;
– compares two explanations;
– shows consequences;
– proposes a solution.
Ask students to label the job in six or seven words. The purpose is not to turn every text into a template. It is to help them carry a manageable representation forward.
When a later paragraph says “This limitation”, the learner then has a compressed idea available to reconnect.
## 8. Prior knowledge helps only when the learner activates the right knowledge
Knowing many things does not guarantee that the relevant thing comes to mind.
A student may know that metals expand when heated but fail to activate that knowledge when reading about gaps in railway tracks. A learner may understand supply and demand in economics but not recognise the same structure in a passage about concert-ticket pricing.
Comprehension therefore involves **knowledge retrieval**, not just knowledge possession.
Teachers can support this by previewing a small number of concepts before reading. But previewing should not become pre-explaining the whole text. The aim is to activate enough relevant structure for the learner to make connections while reading.
Useful prompts include:
– What do you already know about this system?
– Which earlier topic might matter here?
– What does the heading make you retrieve?
– Which concept from last week could explain this diagram?
– What would you expect to happen next, and why?
Prediction is useful when it activates knowledge and creates a hypothesis to test against the text. Prediction is weak when it becomes guessing from the title.
## 9. Build knowledge in coherent domains, not as random “general knowledge”
If background knowledge supports reading, one tempting response is to collect trivia. That misses the point.
A coherent sequence is more useful than scattered facts because concepts recur and deepen. Students who read several texts about ecosystems encounter food webs, energy flow, adaptation, competition, biodiversity and human impact repeatedly. Each text strengthens a domain model that helps with the next.
Harvard’s work on knowledge-building literacy has reported promising transfer when curriculum deliberately develops content knowledge over time. The mechanism is intuitive: later reading becomes easier because the learner already owns some of the conceptual structure and vocabulary.
This suggests a curriculum principle: **choose text sets that build a world**.
Instead of five unrelated comprehension passages, use several texts that approach a substantial topic from different angles. A narrative, diagram, explanation, data table and argumentative article can all contribute to the same knowledge network.
The reading curriculum then does two jobs at once: students practise comprehension while becoming more knowledgeable.
## 10. Multiple texts reveal whether knowledge is truly connected
A single passage can hide shallow comprehension. Multiple texts force integration.
Give students two short accounts of the same event. Ask what both agree on, where they differ, and what new question appears when the texts are combined. Give a science explanation and a data chart. Ask whether the data support the explanation. Give a claim and two sources with different levels of credibility.
The learner now has to connect across boundaries.
This is increasingly important in the digital information environment. Real-world reading rarely means one clean textbook page. Students encounter search results, snippets, videos, charts, AI summaries, social posts and full articles. They must decide which pieces belong together and which contradict.
Connective knowledge becomes information literacy.
## 11. Discussion can expose connections that silent reading hides
A student can finish a page silently while holding a weak or mistaken mental model. Discussion makes the model audible.
Ask: “What caused what?” “What does *this* refer to?” “Which sentence changed your interpretation?” “How does this paragraph connect to the diagram?” “What earlier idea do we need here?”
When students explain, classmates hear alternative connections. One learner may notice a causal link; another may identify a contradiction; a third may supply background knowledge that clarifies a reference.
The teacher’s role is to keep discussion anchored to text and evidence. “What do you think?” is not enough. Ask, “Which words made you think that?” or “What knowledge are you using to connect those sentences?”
This turns comprehension from an invisible private event into something that can be inspected and improved.
## 12. Writing is a powerful test of reading connections
If students truly understand a text, they should be able to reconstruct its important relations in their own language.
Short writing tasks are especially revealing:
**Because–but–so:** Explain the idea using a causal connector, a contrast and a consequence.
**One-sentence chain:** Compress three linked sentences into one accurate sentence.
**Reference repair:** Replace vague phrases such as “this” with explicit referents.
**Concept map to paragraph:** Turn a network of concepts into prose with correct connectives.
**Paragraph to map:** Reverse the process and diagram the relationships in a paragraph.
**Explain the missing link:** Give two true statements and ask the student to write the sentence that logically connects them.
These tasks do not merely check memory. They reveal whether the learner has constructed a coherent model.
## 13. Comprehension questions should diagnose the connection that failed
A wrong answer to a comprehension question can arise from many sources.
The student may not know a word. They may misunderstand syntax. They may fail to resolve a pronoun. They may lack background knowledge. They may choose the wrong referent for “this problem”. They may understand all details but miss the causal relationship. They may bring in outside knowledge that contradicts the passage. They may fail to distinguish the author’s claim from an example.
Marking only right or wrong throws away diagnostic information.
Teachers can classify questions by the connective operation they demand:
– local word meaning;
– reference resolution;
– causal relation;
– contrast;
– sequence;
– inference;
– integration across paragraphs;
– integration with a diagram or table;
– comparison across texts;
– evaluation of evidence.
Patterns then become visible. A student who repeatedly fails reference questions needs a different intervention from one who lacks domain knowledge.
## 14. “Use context clues” is too vague unless we teach what context can do
Students are often told to infer unknown words from context. Sometimes this works. Sometimes the text does not provide enough information.
Context can signal:
– a definition;
– an example;
– a contrast;
– a cause or consequence;
– a category;
– emotional tone;
– grammatical role;
– a likely semantic field.
But context rarely gives a perfect dictionary definition. A learner may infer that *mitigate* means something like “reduce a bad effect” without knowing its full usage. That partial inference can be enough for comprehension, and the word can be confirmed later.
Students should also know when **not** to guess. If the unknown term is central to the whole argument, looking it up may be more efficient. If the passage contains no useful clues, invention is dangerous.
This is another form of connective judgement: use the surrounding network, but know the limits of what the network supports.
## 15. Multilingual learners may possess the concept before they possess the English label
A learner can understand an idea deeply in another language yet appear weak when the English wording is unfamiliar.
This matters because comprehension assessments can confound conceptual knowledge with English academic language. Teachers should ask whether the missing connection is conceptual, lexical or both.
Home-language knowledge can be an asset. A student who already understands migration, evaporation, democracy or ratio in another language has conceptual structure that can support English learning. The job is to connect the new label and academic phrasing to an existing idea.
Rich content should therefore not be withheld simply because a learner’s English is developing. Scaffolds can include diagrams, carefully selected definitions, cognates, multilingual discussion where appropriate, sentence frames and repeated exposure to academic phrases.
The target remains meaningful participation in the same knowledge-rich world.
## 16. Connective knowledge explains why rereading sometimes feels better without fixing comprehension
Rereading makes a text more familiar. Sentences become easier to process. The reader may feel that understanding has improved because the words flow more smoothly.
Sometimes it has. But familiarity can also conceal missing connections.
A better second read has a job:
– trace every pronoun and reference phrase;
– underline causal connectives;
– identify the claim and evidence;
– map the sequence;
– find where the author changes direction;
– connect the diagram to the paragraph;
– summarise each paragraph’s job;
– identify which prior concept the passage assumes.
Purposeful rereading turns exposure into analysis.
The learner is no longer simply seeing the same words again. They are strengthening the network among them.
## 17. A practical connective-reading routine
Teachers can use a five-step routine with demanding texts.
### Step 1: Retrieve
Before reading, activate two or three concepts likely to matter. Avoid a full lecture.
### Step 2: Track
During reading, mark connectives, pronouns, reference phrases and key repeated concepts.
### Step 3: Link
After each section, state how the new idea connects to what came before: cause, contrast, example, extension, qualification or consequence.
### Step 4: Compress
Write a short gist statement or draw a map. Preserve relationships, not every detail.
### Step 5: Transfer
Use the knowledge with a fresh text, question, diagram or problem.
The routine can be light. Not every page needs coloured annotation. The objective is to teach the invisible mental operations until students can perform them independently.
## 18. A worked example: from words to a causal model
Consider this passage:
*Coral reefs support diverse marine ecosystems. When sea temperatures remain unusually high, corals may expel the algae living in their tissues. This process, known as bleaching, removes an important source of energy. If stressful conditions persist, coral mortality can increase, reducing habitat for other organisms.*
A surface reader may remember: coral reefs, temperature, algae, bleaching, mortality.
A connective reader builds:
**high sea temperature → coral stress → algae expelled → less energy → higher coral mortality → less habitat → effects on other organisms**
They also resolve *this process* as the expulsion of algae and understand that *bleaching* names that process. They see that the final clause extends the effect from coral to the wider ecosystem.
Now a new text says that repeated marine heatwaves can reduce reef resilience. The learner has a model into which *repeated heatwaves* and *resilience* can attach.
That is the advantage of connective knowledge: new reading compounds instead of restarting from zero.
## 19. Assessment should ask whether students can build and use the network
Traditional comprehension questions can test this if designed carefully, but assessment can go further.
Ask students to:
– identify what a reference phrase points to;
– select the connective that preserves the logic;
– order events in a causal chain;
– explain why one sentence follows another;
– integrate information from text and diagram;
– compare two sources;
– write the missing inference;
– create a concept map and justify the links;
– transfer a principle to a new passage.
These tasks reveal more than whether a student remembers a sentence.
For older learners, ask them to detect when a text’s connection is weak. Does the evidence really support the claim? Does “therefore” overstate causation? Is a generalisation based on one example? Reading comprehension then becomes critical reading.
## 20. What parents can do without turning home into another classroom
Families can support connective knowledge through ordinary conversation.
After reading, ask “What does that remind you of?” Follow with “How is it similar?” or “What is different?” When a child meets a new word, connect it to a word family, opposite or real example. During documentaries or news, ask what earlier event caused the current one. In stories, ask what a pronoun or vague phrase refers to. When two family members interpret a passage differently, return to the text and compare evidence.
Reading widely also matters because knowledge accumulates. Museums, conversations, hobbies, travel, practical tasks, stories, science activities and documentaries can all create conceptual resources that later texts activate.
The goal is not to quiz constantly. It is to make connection-making normal.
## 21. The reading curriculum should leave students knowing more about the world
A narrow comprehension programme can accidentally teach students to treat every passage as disposable. Read, answer, forget, next passage.
A stronger curriculum allows reading to accumulate.
After a term of reading, students should not only be better at “finding evidence”. They should know more about ecosystems, migration, technology, government, culture, energy, health, history or whatever coherent domains the curriculum has built. They should possess richer vocabulary and more relations among concepts.
This matters because tomorrow’s comprehension depends partly on what today’s reading leaves behind.
Knowledge-rich reading creates a positive cycle:
**more knowledge → easier connection-making → stronger comprehension → more knowledge gained from reading.**
Students with less background knowledge can experience the reverse. Texts are harder, so they extract less, which leaves future texts harder again. Deliberate knowledge-building can interrupt that cycle.
## 22. Do not turn connective knowledge into another checklist
There is a risk in every educational insight: once named, it becomes a worksheet.
Students could be asked to circle five connectives, underline three pronouns and draw two arrows without understanding the text any better. The visible routine would replace the cognitive purpose.
The purpose is coherence.
Ask whether the student can answer:
– What is this text mainly building?
– How does this sentence connect to the last one?
– What does this phrase point back to?
– What knowledge is the author assuming?
– Which concept has appeared under a different word?
– What changed my model as I read?
– Where does the logic weaken?
Annotation is useful only when it helps answer these questions.
## 23. Examination reading becomes easier when students can see the hidden structure quickly
Under examination conditions, connective reading has a practical advantage: it reduces the amount of text a student has to hold as separate information. A connected model is more compact than a list of details.
Consider a comprehension passage in which a writer first describes a policy, then presents evidence of a short-term improvement, then introduces an unintended consequence, and finally argues for a modified version of the policy. A student who remembers four paragraphs as four separate blocks has to keep searching the page. A student who has built the structure — **proposal → benefit → limitation → revision** — can navigate questions more efficiently.
This does not mean students should force every passage into a predetermined pattern. The pattern must be earned from the text. But once the structure is recognised, details have addresses. Evidence belongs under the benefit or limitation. A reference phrase such as “this drawback” has an obvious candidate. The writer’s final position can be distinguished from the earlier description.
For inference questions, connected knowledge also protects against overreaching. Students often select an answer because it sounds plausible in the real world. The better question is whether the inference is licensed by the text plus reasonable background knowledge. An inference should be a bridge, not a leap.
For vocabulary-in-context questions, the network helps constrain meaning. If a word appears in a paragraph describing dwindling supply, rationing and competition, the semantic field narrows possible interpretations. If a connective signals contrast, the unknown phrase may be understood partly from what it is being contrasted with.
For summary, students need to preserve the main relations while removing examples and repetition. That is connective compression. A weak summary lists points. A strong summary shows how the points fit.
Teachers preparing students for examinations can therefore train three fast questions after each paragraph: **What is the paragraph doing? What does it connect to? What must I carry forward?** With practice, these questions become internal and no longer require visible annotation.
The goal is not a test trick. Examination pressure simply exposes what everyday reading already requires: the reader has to build a coherent representation quickly enough to use it.
## 24. A teacher can diagnose connective reading with one short paragraph
Long comprehension papers are not always necessary. A carefully chosen paragraph can reveal where meaning is breaking.
Give the learner a six- or seven-sentence passage containing one technical term, one causal connective, one contrast, one pronoun, one compressed reference phrase and one inference that depends on ordinary background knowledge. Ask the student to read it once and explain what it means. Then probe each connection.
Can they identify what the pronoun refers to? Can they explain what changed after *however*? Can they state the cause before the consequence? Do they know enough about the topic to fill the unstated link? Can they paraphrase the technical word without simply repeating the sentence? Can they summarise the paragraph without producing a list of disconnected details?
The pattern of answers matters more than the raw score.
A learner who understands the topic but repeatedly loses pronoun reference may need explicit work on reference chains. A learner who knows the words but cannot explain the causal structure may need sentence-combining and relationship mapping. A learner who reads fluently yet lacks the domain knowledge may need knowledge-building rather than another generic comprehension strategy. A learner who understands orally but struggles with dense syntax may need academic-language support.
This diagnosis also helps teachers choose text difficulty intelligently. If every sentence contains unfamiliar concepts, words and structures, the text may overload the learner so completely that no single repair is visible. If the text is too easy, it reveals nothing. The useful diagnostic passage has enough challenge to expose the weak connection while keeping the rest of the system available.
After instruction, use a fresh passage with the same underlying demand. Improvement should transfer. If students can trace reference only in the passage they practised, they learned the worksheet. If they can do it in new science, history and English texts, they learned the reading move.
That distinction is central to connective knowledge: the goal is not to master one text. It is to become better at building coherence wherever demanding text appears.
## 25. Knowledge checks can distinguish a reading problem from a knowledge problem
When a student struggles with a passage, teachers sometimes respond with another reading strategy. That is sensible only if strategy is the bottleneck. A quick knowledge check can prevent misdiagnosis.
Before asking the student to reread a passage about elections, for example, ask what they already understand about voting, representation, candidates and government. Before a passage about erosion, check whether weathering, sediment, water flow and landform are familiar ideas. If the conceptual floor is missing, the student may be doing all the “right” reading behaviours while having too little knowledge to build a coherent model.
The reverse also occurs. A learner may know the topic well in conversation yet fail to understand the written passage because academic syntax, reference chains or connectives are causing trouble. In that case, more background information will not solve the reading problem.
A useful two-minute diagnostic is: first ask the learner to explain the topic without the text; then give a short paragraph and ask them to explain how the sentences connect. Strong topic knowledge plus weak text integration points toward language and reading structure. Weak topic knowledge plus adequate sentence processing points toward knowledge building. Weakness in both requires a combined plan.
This distinction protects students from generic labels such as “poor comprehension”. Comprehension is an outcome produced by several systems working together. Diagnosis should tell us which system needs the next lesson.
## Conclusion: strong readers build systems of meaning
Reading comprehension is sometimes discussed as if words enter the eyes and meaning appears automatically. In reality, readers work continuously. They retrieve knowledge, resolve references, interpret connectives, integrate sentences, compress paragraphs, infer unstated relations and test new information against what they already know.
That work is connective.
It explains why decoding alone is insufficient, why vocabulary depth matters, why coherent knowledge-building helps, why small academic words can cause big problems, and why students may understand every sentence but still lose the argument.
The teaching response is not to abandon comprehension strategies. It is to place them inside a richer model. Teach words as networks. Teach connectives as relationships. Teach reference chains. Build coherent domain knowledge. Use multiple texts. Make students explain the links. Let writing expose whether the model holds together.
Strong readers do not merely know more words on the page.
They know how the page connects to itself, to other texts and to a growing world inside the mind.
That is why knowledge compounds.
—
## Related eduKateSG reading
– **How Prior Knowledge Works | What You Already Know Changes What You Can Learn Next**
How Prior Knowledge Works | What You Already Know Changes What You Can Learn Next
– **Context Cannot Rescue Every Weak Word: The Lexical Bottleneck in English Reading**
Context Cannot Rescue Every Weak Word: The Lexical Bottleneck in English Reading
– **How to Optimize Reading V1.1**
How to Optimize Reading V1.1
– **How Reciprocal Teaching Works | Turn Reading Into Predicting, Questioning, Clarifying and Summarising**
How Reciprocal Teaching Works | Turn Reading Into Predicting, Questioning, Clarifying and Summarising
## Sources and further reading
– Buchanan, B. & Uccelli, P. (2026), *Connective Knowledge and Reading Comprehension in Upper Elementary Students: A Growth Analysis*, Reading Research Quarterly 61(2), e70096; Harvard Language for Learning listing.
https://languageforlearning.gse.harvard.edu/publication/buchanan-b-uccelli-p-2026-connective-knowledge-and-reading-comprehension-upper
– Harvard Graduate School of Education (30 June 2026), *Science of Reading Comprehension*.
https://www.gse.harvard.edu/ideas/usable-knowledge/26/06/science-reading-comprehension
– Harvard Graduate School of Education (7 March 2025), *Lessons From the Science of Reading*.
https://www.gse.harvard.edu/ideas/education-now/25/03/lessons-science-reading
– Harvard Graduate School of Education (31 March 2023), *Building Background Knowledge in Science Improves Reading Comprehension*.
https://www.gse.harvard.edu/ideas/usable-knowledge/23/03/building-background-knowledge-science-improves-reading-comprehension
