Academic English: language for learning, research and inspectable thought

Academic English is the use of English to learn, investigate, explain, argue, evaluate, synthesise and communicate knowledge in settings where other people need to inspect how a conclusion was reached. It is not simply “formal English”. It is not a collection of impressive words. It is not a university accent, a five-paragraph essay template or a ban on first-person pronouns. Academic English is a system of language decisions shaped by evidence, disciplinary conventions, intellectual responsibility and the need to make thinking traceable.

A student can speak English fluently and still find academic study difficult. Everyday conversation allows shared context, gesture, immediate clarification and rapid repair. Academic communication often removes those supports. A reader encounters unfamiliar concepts in dense prose. A writer must represent sources accurately, distinguish their own position from other people’s claims, control definitions across thousands of words and make relationships explicit enough for a reader who was not present when the ideas were formed. A seminar participant has to listen, interpret, enter a discussion, refer to earlier contributions and qualify a claim in real time. A researcher has to turn a question into a method, evidence into a result, and a result into a conclusion that is no stronger than the data allows.

This is why academic English belongs inside SETC’s wider English system. The English Atlas maps the whole language, learning and performance territory. How English Works explains the mechanisms of sound, words, grammar, meaning and discourse. The Reading and Comprehension, Writing and Composition, Vocabulary in Use, Listening and Viewing and Speaking and Oral owners develop the component skills. This page owns the point where those components become a coherent system for advanced study, research and academic communication.

Academic English begins before university. A Primary learner explaining evidence from a passage is already practising a small form of academic reasoning. A Secondary student comparing viewpoints, defining a term or writing an evidence-based argument is building academic literacy. General Paper makes the transition especially visible because reading, knowledge selection, argument, evidence and evaluation must work together under time. University then changes the scale: longer sources, specialised disciplines, more independent research, explicit citation, greater uncertainty and more responsibility for choosing what to read, believe and write.

That progression matters. Academic English should not be taught as a sudden new dialect that appears after school. It is the increasingly independent use of language to handle knowledge.

What Academic English contains

Academic English is not the same as difficult English

Difficulty and academicity are different. A literary novel can contain extremely difficult vocabulary without being academic prose. A research abstract can use familiar words while expressing dense relationships. A legal clause can be technically demanding but serve a different institutional purpose from an academic argument. Academic English is defined less by how hard the words look than by the kind of knowledge work the language must perform.

This distinction protects learners from a common mistake: upgrading surface vocabulary while the reasoning remains unchanged. Replacing shows with demonstrates does not make a weak claim stronger. Replacing many with a plethora of may make the sentence worse. Academic maturity comes from better distinctions, better evidence, clearer structure, more accurate qualification and stronger control of source meaning.

Academic English is not only for multilingual students

Students who learned English as an additional language may need explicit support with vocabulary, grammar, pronunciation, idiom and cultural conventions. But first-language English speakers also need academic English. Everyday fluency does not automatically produce research reading, evidence synthesis, disciplinary argument or source management. Current university EAP provision reflects this reality: academic writing and communication support can be relevant across language backgrounds, because academic literacy is learned rather than inherited.

SETC therefore treats academic English as a capability system, not a learner label. The diagnostic question is not “Is English your first language?” It is “Which academic language operation is not yet reliable?”

Academic English and English for Academic Purposes

English for Academic Purposes (EAP) is commonly used for courses and teaching designed to help learners operate successfully in academic settings. EAP can include academic reading and writing, lectures, seminar discussion, presentations, vocabulary, study strategies, research, source use and the conventions of particular disciplines. Some programmes focus strongly on students using English as an additional language; others teach academic writing and communication to a much wider student population.

SETC uses Academic English as the broader public owner because the reader may arrive before, during or after formal EAP study. The mechanisms remain useful to a Secondary student entering more demanding texts, a GP student building disciplined argument, an undergraduate writing the first research paper, a postgraduate synthesising literature or an adult returning to academic study.

From school English to academic English

The transition is not a rejection of school learning. It is a change in scale and responsibility.

The most difficult transition is often not “harder English”. It is the movement from completing a supplied task to managing an intellectual process.

The Academic English operating loop

A useful general loop is:

Interpret the task → establish the question → locate knowledge → read selectively → represent sources accurately → build a model → decide the claim → organise the reasoning → draft → integrate evidence → qualify → revise → verify → submit or present → read the feedback → transfer.

Different academic genres rearrange this loop. An examination essay compresses research and drafting. A dissertation greatly expands them. A laboratory report gives the method and results unusual importance. A literature review makes source relationships central. A seminar discussion performs the reasoning orally. But the underlying control problem remains recognisable: can the learner move from question to evidence to communication without losing meaning?

The three diagnostic lenses

SETC uses three recurring lenses throughout this owner:

• Iona — reading and evidence: What does the source actually say? Which evidence supports the claim? Where does inference exceed the text?
• Maren — writing and architecture: What is the intellectual job of this section? How does the paragraph move the whole argument forward? What must the reader understand next?
• Leonie — execution and recovery: Can the process finish under the real constraint? Which decisions consume too much time? How does the learner recover when a source, paragraph or argument stops working?

These are not three separate subjects. They are three views of one academic performance system.

The aim: inspectable thought

The central purpose of Academic English is not to sound academic. It is to make intellectual work inspectable. A reader should be able to see what is being claimed, where the evidence came from, how the reasoning connects, how strong the conclusion is, what assumptions matter and what remains uncertain.

That standard improves language and thinking at the same time. Vague reasoning produces vague writing. Overstated evidence produces overstated language. Unclear definitions destabilise whole arguments. Better Academic English therefore comes from improving the relationships among language, knowledge, evidence and judgement—not from decorating sentences after the thinking is finished.

Quick Read

• Academic English is language for learning, research, explanation, argument, evidence and inspectable thought—not simply “formal vocabulary”.
• Academic reading identifies questions, claims, methods, evidence, definitions, assumptions, qualifications and limitations.
• Academic writing begins with task interpretation and argument architecture before sentence polishing.
• Quotation, paraphrase, summary and synthesis are different source-use operations; all require accurate attribution and preservation of source meaning.
• Academic language controls stance, certainty, evidence, disciplinary terminology and information density.
• Seminars, lectures, presentations and Q&A require real-time listening, interaction, repair and explanation as well as written competence.
• The release test is independent transfer: can the learner research and communicate a fresh academic problem after scaffolds or AI assistance are removed?

Academic reading: read for structure, claims and usable evidence

Academic reading is not ordinary reading performed more slowly. The reader usually has a job to do with the text. They may need to understand a theory, locate a method, compare two positions, extract a definition, judge the strength of evidence, identify a limitation, prepare for discussion or decide whether the source deserves a place in a paper. Purpose changes what deserves attention.

A student who tries to read every academic source from the first word to the last at the same speed will often spend too much time on low-value detail. A student who scans too aggressively may miss the relationship that makes the detail meaningful. Strong academic reading therefore changes speed deliberately.

Pass 1: orient before investing attention

Before close reading, establish the document’s basic coordinates. What kind of source is this? Who wrote it? What problem is it addressing? What field does it belong to? What is the likely function of the section in front of you? Which headings, abstract, introduction, conclusion, figures or references reveal the shape of the work?

Orientation does not replace reading. It creates a provisional map so later sentences have somewhere to go. Without that map, a student may spend ten minutes understanding a technical paragraph and still not know whether the paragraph contains the source’s main claim, a background definition or one limitation mentioned in passing.

Pass 2: recover the argument or information architecture

Most difficult academic texts become easier when the reader stops treating paragraphs as equal units. Ask what each section is doing. A section may define a problem, review previous work, describe a method, present evidence, interpret findings, challenge an earlier view, explain a mechanism, narrow a claim or identify uncertainty.

Write the function in the margin or notes using a verb: defines, contrasts, tests, reports, qualifies, explains, criticises. Functional labels are more useful than copying headings because they force the reader to interpret the intellectual action.

For argument-driven sources, build a simple model:

• What is the main question or problem?
• What is the central claim?
• What reasons support it?
• What evidence supports those reasons?
• What assumptions connect evidence to conclusion?
• What qualifications or exceptions limit the claim?
• What competing explanation is considered?
• What does the author say should follow?

For descriptive or empirical sources, the model changes:

• What was studied?
• Who or what was observed?
• How was it measured?
• What was found?
• How large or meaningful was the difference?
• What uncertainty remains?
• Which population, context or time period does the evidence actually cover?

Pass 3: close read the load-bearing sections

Close reading should be selective but deep. Slow down where a definition controls later reasoning, where a causal claim is justified, where a result is interpreted, where the author distinguishes two similar concepts, where a counterargument is answered or where a limitation changes how broadly the result can travel.

The question is not “Which sentence looks difficult?” but “Which sentence is carrying important meaning?”

Academic sentences are often dense because relationships are compressed

Academic prose frequently packs information into noun phrases, subordinate clauses and nominalisations. A sentence such as “The rapid institutionalisation of automated decision systems has generated concerns regarding the opacity of model-mediated allocation” may be easier to understand after unpacking it into propositions:

• Automated decision systems became institutionalised rapidly.
• These systems are used to allocate something.
• Models mediate the allocation.
• Observers are concerned.
• The concern is opacity: people cannot easily see how the allocation was produced.

Unpacking is not “dumbing down” the sentence. It is exposing the relationships. Once the reader understands them, the dense version can be reconstructed and evaluated.

Track reference across sentences

Academic texts frequently use pronouns, demonstratives and summary nouns such as this result, these conditions, such an approach or the latter problem. If the reader loses the referent, the text can appear to become difficult suddenly even when every individual word is familiar.

When meaning breaks, ask: What exactly does this expression point back to? The answer may be a noun, a whole proposition, a method described several sentences earlier or an entire preceding paragraph.

Read modality as evidence about commitment

Academic authors do not only state content; they signal how strongly they are committed to it. May, might, is likely to, appears to, is consistent with, demonstrates and establishes do not carry the same force.

A reader who strips those differences away can misrepresent the source. “The study proves X” may be false even when the study is excellent if the authors only claim that their results are consistent with X under defined conditions.

Iona therefore records not only the proposition but its strength. Her note may say: “Authors find association in sample; do not claim causation.” That small phrase protects later writing from accidental overstatement.

Separate source claim from reader inference

Academic reading becomes dangerous when a plausible inference is remembered as something the source directly said. Keep three layers separate:

1. Reported: what the source explicitly states.
2. Inferred: what follows reasonably from the source but is not directly stated.
3. Evaluated: your judgement about the significance, quality or limitation of the source.

Notes can mark these layers with simple labels. The exact notation matters less than preserving the boundary.

Definitions require boundary control

Academic arguments often depend on terms that look familiar but carry specialised meanings. Resilience, literacy, efficiency, equity, risk, validity and engagement can mean different things across disciplines and studies.

Do not assume the everyday meaning. Ask how the author operationalises the concept. What would count as an instance? What would not? How was it measured? Does another source use the same label for a different construct?

Definition control becomes especially important during synthesis. Two sources may appear to disagree when they are actually measuring different versions of the same broad word.

Read methods enough to understand what the evidence can carry

Not every reader needs to become a specialist statistician, but academic source use requires enough method awareness to avoid impossible conclusions. A cross-sectional survey and a randomised trial support different kinds of causal inference. An interview study can reveal rich patterns of experience but is not automatically evidence of population prevalence. A case study can expose a mechanism without proving that the mechanism is universal.

The method sets boundaries on the claim. Reading only the abstract and conclusion can hide those boundaries.

Read tables and figures as arguments

Data displays are not decorative additions. Check the variables, units, denominator, baseline, scale, time period, sample and uncertainty. Ask whether the visual comparison is fair. A truncated axis can amplify apparent change. An average can hide subgroup differences. A percentage can sound dramatic without the underlying count.

Connect the figure to the prose. Does the paragraph describe the same pattern the graph shows? Does the author move from correlation to causation? Is the “largest increase” large in practical terms or merely the largest among small changes?

Research reading is triage before depth

When a research question has hundreds of possible sources, the first problem is selection. Use titles, abstracts, keywords, references and citation trails to decide which sources deserve close attention. Triage asks:

• Does this source address my actual question?
• Is it recent enough for the claim I need?
• Is it a primary study, review, theory paper, commentary or reference work?
• Does the population or context match my problem?
• Is there a more authoritative or direct source?
• Would this source add something new, or repeat what I already have?

The goal is not to collect the maximum number of PDFs. It is to build the smallest source set that gives the question adequate conceptual and evidential coverage.

Source chasing: move backward and forward

Useful academic reading often moves through citation networks. Backward searching follows the references inside a useful source to earlier work. Forward searching finds later work that cited the source. This helps a reader discover foundational concepts, replication, criticism, extensions and newer evidence.

But citation counts are not quality scores. A source can be widely cited because it is controversial or historically important. Read the citing relationship.

Note-taking should preserve relationships, not sentences

Copying large passages creates a dangerous illusion of progress. The notes become a second unread library. Better notes transform the source into structures that can later support thinking.

A robust source note can contain:

• full citation or stable identifying information;
• the source’s question;
• central claim;
• method or evidence type;
• key finding;
• important qualification;
• one or two exact quotations only when wording itself matters;
• your interpretation or question, clearly separated from source content;
• connections to other sources;
• the possible role of the source in your own work.

Maren adds one more field: “What paragraph could this source help me write?” If there is no answer, she asks whether she is collecting the source because it is genuinely useful or because collecting feels safer than deciding.

Read multiple sources by relationship

Synthesis begins during reading. Instead of filing every source separately, group them by the relationship they have to the question:

• sources that define the problem;
• sources that support one explanation;
• sources that support a competing explanation;
• sources that use different populations or contexts;
• sources that disagree because of method;
• sources that show change over time;
• sources that identify a limitation or unresolved question.

This turns the literature from a pile into a model.

The comparison matrix

For a complex reading set, create a matrix with sources as rows and analytic dimensions as columns. Possible columns include definition, population, method, key claim, evidence, limitation, context and relevance to your question. The matrix makes differences visible that are hard to remember when sources are read days apart.

Do not let the matrix become a substitute for reading. It is a representation of your understanding, not the source itself.

Reading recovery when the text stops making sense

When a paragraph becomes opaque, rereading it repeatedly at the same level may not help. Diagnose the failure:

• Vocabulary: one or more terms are unknown or used technically.
• Syntax: the sentence structure hides who is doing what.
• Reference: a pronoun or summary phrase has lost its referent.
• Concept: the background model is missing.
• Argument: the relation between claim and evidence is unclear.
• Density: too many propositions are compressed into one sentence.
• Context: an earlier section established information you skipped.

Then choose the repair. Define the term. Unpack the sentence. Trace the referent. Read a simpler explanation of the concept. Return to the previous section. Draw the relationship. The correct response to difficulty is not always “read more slowly”.

Do not highlight everything

Highlighting is useful when the colour or mark has a defined job. A reader might mark the main claim, evidence, definition and limitation differently. Highlighting half the page merely changes the background colour.

A better test is retrieval: close the source and reconstruct the argument. What question was the author answering? What evidence mattered? What qualification changed the conclusion? If the reader cannot answer, the page was seen but not yet organised.

Annotation should support the next action

An annotation can record a summary, question, disagreement, connection, definition or future use. Good annotation reduces the cost of returning to the text later. “Important!” rarely does that. “Defines fairness as equality of opportunity, not equality of outcome—compare Source C” does.

Academic reading speed

Speed grows from several systems: vocabulary automaticity, syntactic parsing, background knowledge, genre familiarity, selective attention and confidence about what can safely be skipped. Forcing speed before these systems are sufficiently stable can turn academic reading into fast misunderstanding.

Train speed through repeated work on the decision process: orient faster, identify paragraph jobs, recognise signal language, locate definitions and evidence efficiently, and stop close reading when the task no longer requires it.

Reading for an examination and reading for research

Timed examinations give the reader a bounded source and a fixed output condition. Research reading gives much more control but creates a selection problem. The examination reader protects time against one difficult paragraph. The research reader protects weeks against endless source accumulation.

Leonie’s principle is the same in both contexts: define a stop rule. In an examination, the stop rule may be “I have enough evidence to answer this question.” In research, it may be “New sources are now repeating concepts already represented in my evidence set.”

Worked example: three students read the same research summary

Iona reads accurately but takes notes sentence by sentence. When asked for the central claim, she searches through a page of copied phrases. Her problem is not comprehension. It is representation. She needs to compress the source into an argument model.

Maren sees the main claim immediately but begins evaluating before preserving what the authors actually said. Her notes mix source and reaction. She needs stronger attribution boundaries.

Leonie understands both claim and method but spends too long investigating every unfamiliar reference. She needs a purpose-based reading budget and a source-triage rule.

All three students appear to have a “research reading” problem. The first repairable cause is different.

Academic reading practice ladder

1. Read one paragraph and state its function in five words.
2. Read one page and reconstruct the question–claim–evidence chain.
3. Mark every qualification in a short academic passage.
4. Compare an abstract with the full paper and identify what the abstract omits.
5. Read two sources and write one sentence explaining their relationship.
6. Build a source matrix for four sources.
7. Read a source under a fixed time budget and decide what deserves a second pass.
8. Return after several days and reconstruct the source from notes alone.
9. Use the source in a paragraph without reopening it, then verify against the original.
10. Repeat with a source from an unfamiliar discipline.

The progression moves from understanding to representation to retrieval to transfer.

Evidence that academic reading is becoming independent

Improvement is visible when the learner can enter an unfamiliar source without waiting for the teacher to identify the important paragraph; can explain the source’s claim without copying its wording; preserves qualifications; distinguishes evidence from interpretation; knows when method matters; records notes that remain useful later; compares sources by relationship; and stops reading when the task has enough evidence.

That is academic reading as a usable system rather than a heroic struggle with difficult pages.

Academic writing: build a structure another mind can inspect

Academic writing is not the act of putting knowledge into formal sentences. It is the construction of a traceable intellectual object. The reader should be able to identify the question, follow the writer’s line of reasoning, see how sources and evidence enter, distinguish report from interpretation, understand how claims are qualified and locate the basis of the conclusion.

This makes academic writing an architecture problem before it becomes a sentence problem. Strong grammar cannot rescue a paper whose sections do not serve the question. Sophisticated vocabulary cannot repair a paragraph whose evidence does not support its claim. Accurate citations cannot turn a literature dump into synthesis.

Interpret the task before planning the answer

Academic tasks often contain several decisions inside one sentence. “Critically evaluate the extent to which digital technologies improve educational equity” asks the writer to define at least three things: what counts as digital technology, what kind of improvement matters, and what equity means. It also asks for evaluation rather than description and for a judgement about degree rather than a binary yes/no answer.

Before researching or drafting, rewrite the task as an intellectual contract:

• What is the object of analysis?
• What operation is required: explain, compare, evaluate, interpret, analyse, propose?
• What boundaries are explicit?
• What important terms need operational definitions?
• What evidence would allow a defensible answer?
• What would count as a weak or incomplete response?

This prevents a common academic failure: producing good material for a different question.

The working question

Long academic tasks benefit from a working question that is narrower than the general topic. “Artificial intelligence in education” is a topic. “Under what conditions can generative AI feedback improve revision quality without increasing learner dependence?” is a question that can organise reading and evidence.

A working question is allowed to change. Research may reveal that the first version is too broad, impossible to answer with available evidence or based on a false assumption. Refining the question is intellectual progress, not failure.

The thesis is a controlled answer, not a ceremonial sentence

A thesis should tell the reader what the paper is going to establish and, often, under what conditions. Weak theses either repeat the topic—“This essay will discuss AI in education”—or make claims too broad for the available evidence—“AI revolutionises learning for everyone.”

A stronger thesis identifies the relationship the paper will defend:

Generative AI can improve revision quality when feedback requires learners to diagnose and rewrite their own work, but benefits are less reliable when the system replaces planning or produces finished prose, because assistance then reduces the cognitive work that later transfer requires.

This thesis does several jobs. It takes a position, includes a condition, identifies a mechanism and creates room for evidence that may support or qualify different parts of the claim.

Thesis strength should match evidence strength

Academic writing is damaged by verbs that exceed the evidence. Causes, proves, ensures and eliminates carry heavy commitments. Sometimes those commitments are justified. Often is associated with, contributes to, can improve, under these conditions or is consistent with is more accurate.

Qualification is not weak writing. It is precision about the boundary of what is known.

Build the argument before the paragraphs

Maren does not begin by deciding that the essay needs five paragraphs. She builds a reasoning model. For each major claim she asks:

• Why is this claim necessary to the thesis?
• What evidence could support it?
• What interpretation connects the evidence to the claim?
• What competing explanation must be considered?
• What does this claim allow the next section to do?

Only then does paragraph structure emerge.

Paragraphs need intellectual jobs

A paragraph should do something the whole paper needs. It may define a concept, establish a mechanism, present evidence, compare alternatives, test a counterexample, qualify an earlier claim or derive an implication. A paragraph whose only role is “talk about another example” may not yet have an intellectual function.

One useful diagnostic is to write a five-to-ten-word job label beside every paragraph. If two adjacent paragraphs have the same job, perhaps they should be combined. If no job can be named, the paragraph may not belong. If the job does not contribute to the thesis, the writer may be following the topic rather than the argument.

Topic sentences are control statements

A topic sentence does more than announce the subject. It establishes the proposition or relationship the paragraph will develop. Compare:

There are many problems with online learning.

with:

Online access can widen participation without reducing educational inequality when students differ substantially in device quality, study space and access to live academic support.

The second sentence gives the paragraph a specific reasoning job. It also tells the writer what evidence would be relevant.

Evidence must be interpreted, not displayed

Academic writing often fails in the gap between citation and reasoning. A source is introduced, a statistic appears, and the paragraph moves on as if evidence speaks for itself.

Evidence integration requires at least four decisions:

1. Why is this evidence relevant to this claim?
2. What exactly does the evidence show?
3. What inference does the writer draw?
4. What does the evidence not establish?

A quotation, study or statistic should change the reasoning in the paragraph. If removing it leaves the argument unchanged, it may be decorative.

The claim–evidence–warrant relationship

A warrant is the reasoning that connects evidence to a claim. Suppose a study reports that students who used a particular feedback method revised more errors. The evidence does not automatically support the claim that the method improves independent writing. The writer needs a warrant explaining why better revision on that task is relevant to independent writing, and perhaps further evidence showing that the improvement transfers when support is removed.

Making warrants visible reduces leaps in reasoning.

Use examples proportionately

Examples can clarify a mechanism, demonstrate a pattern or make an abstract distinction concrete. They become dangerous when vividness substitutes for representativeness. One dramatic case does not prove prevalence. One country does not automatically represent all systems. One successful classroom does not prove that a method generalises across ages and subjects.

Academic prose should signal the function of the example: for example, one illustration, a case that demonstrates the mechanism, evidence from this context. Do not silently turn an illustration into universal proof.

Counterargument is not a ritual paragraph

A useful counterargument tests the actual thesis. It may identify a competing mechanism, a subgroup for whom the claim fails, a cost that changes the evaluation, a methodological limitation or a value conflict.

Weak counterarguments are deliberately easy to defeat: “Some people disagree.” Strong counterarguments create pressure. The writer should be willing to narrow or revise the thesis when the objection is valid.

Evaluation needs criteria

Words such as effective, successful, fair, important and better require criteria. Effective for whom? Better according to which outcome? Over what time? At what cost? Under what constraint?

Without criteria, evaluation becomes preference. With criteria, the reader can inspect the judgement.

Comparative writing needs a common ruler

Comparison works when cases are evaluated through shared dimensions. If one educational policy is judged by cost and another by popularity, the comparison is unstable. Establish the common ruler first: learning outcome, access, cost, durability, implementation burden, distributional effect, or whatever the question requires.

Good comparison also preserves important differences. “Both systems use technology” may be true and useless if they use different technologies for different populations under different constraints.

Cause-and-effect writing requires mechanism and alternatives

Causal claims are among the easiest academic claims to overstate. Sequence is not causation. Correlation is not automatically causation. A plausible story is not enough.

Academic causal reasoning asks what mechanism connects cause and outcome, what alternative explanations exist, whether timing fits the proposed mechanism, whether the relationship survives comparison and whether the evidence design supports causal inference.

Writers should calibrate language accordingly. May contribute to and is associated with are not timid phrases when the evidence does not justify more.

Definitions must remain stable

A paper can appear sophisticated while its central term drifts. “Equity” may begin as equal access and end as equal outcomes. “Learning” may begin as test improvement and later mean long-term independent transfer. If the definition moves without acknowledgement, evidence collected under one meaning may be used to support another.

Define only what the argument needs, but keep those definitions stable. If evidence forces a refinement, say so explicitly.

Information order matters

Academic sentences and paragraphs manage old and new information. Readers generally process complex ideas more easily when sentences connect to what is already active before introducing the next load. Abruptly introducing a new technical noun phrase at the start of every sentence can make a logically sound paragraph feel disconnected.

Cohesion therefore includes reference, lexical repetition where useful, substitution, signalling phrases and controlled topic progression. The aim is not to add connectors such as moreover mechanically. The aim is to make the relationship recoverable.

Cohesion is not coherence

A paragraph can contain many linking words and still make no coherent argument. Cohesion is visible language connection. Coherence is the reader’s ability to construct a sensible overall model. A sentence can be connected grammatically to the previous one and still introduce irrelevant content.

Revise at both levels. First ask whether the ideas form a valid sequence. Then ask whether the language makes that sequence visible.

Academic voice

Academic voice is not created by removing personality or using longer words. Voice appears in the writer’s choices: which problem is framed, which distinctions matter, how sources are related, how evidence is weighed, where uncertainty is acknowledged and what conclusion is defended.

A source-rich paragraph can still contain a strong writer’s voice when the writer controls the synthesis rather than allowing citations to dictate the structure.

First person is a context decision

Some disciplines, genres and institutions accept or encourage first person; others prefer more impersonal structures in some contexts. There is no universal English rule that academic writing must avoid I. The correct question is what the genre and disciplinary convention require.

Avoiding first person can sometimes make agency less clear: “It was decided that…” hides who decided. Using first person can sometimes be unnecessary: “I believe the results show…” may weaken a sentence if the evidence can simply be described. Choose by function.

Nominalisation: compression with a cost

Academic writing often turns processes into nouns: decide → decision, evaluate → evaluation, fail → failure, expand → expansion. Nominalisation can package complex ideas and create cohesion, but excessive nominalisation hides actors and produces dense prose.

Compare:

The implementation of the intervention resulted in an improvement in the completion of tasks.

After teachers implemented the intervention, students completed more tasks.

The first may be appropriate if implementation itself is the object of discussion. The second is clearer when agency matters. Academic English should give the writer both options.

Passive voice: useful when the object or process matters

Passive voice is not inherently academic and active voice is not inherently better. “Samples were stored at −20°C” is efficient when the procedure matters more than the actor. “The researchers excluded 14 participants” may be clearer when the decision-maker matters.

Teach voice as information management, not a prohibition.

Hedging and boosting

Hedges reduce commitment: may, appears, suggests, in many cases. Boosters increase commitment: clearly, demonstrates, strongly, establishes. Academic judgement requires both.

Over-hedging makes well-supported conclusions vague. Over-boosting turns plausible findings into false certainty. The language should match the evidence.

Reporting verbs carry evaluation

States, reports, argues, claims, demonstrates, assumes, concedes, observes and speculates position a source differently. Choose reporting verbs according to what the source is doing and the stance you can justify.

Do not use proves merely to sound decisive. Do not use claims as a disguised accusation when the source simply reports a result.

Introductions orient; they do not perform ceremonial throat-clearing

A strong academic introduction helps the reader enter the problem. Depending on the genre, it may establish context, define the research gap, narrow the scope, state the question, identify a thesis and signal the route ahead.

Generic statements such as “Since the beginning of time, education has been important” consume attention without reducing uncertainty. Begin as close to the real problem as the reader can reasonably follow.

Conclusions close the reasoning, not merely the document

A conclusion should answer the question at the level the evidence permits. It may synthesise the main reasoning, state implications, acknowledge remaining uncertainty or identify conditions for application. Repeating topic sentences in new words is not synthesis.

The conclusion is also a final scope check. Has the essay moved from evidence about one population to a claim about everyone? Has a short-term finding become a long-term promise? Has the thesis changed? Final revision should catch those expansions.

Revision has levels

Do not begin revision by polishing commas if the argument architecture is unstable. Use passes:

1. Question pass: Does the paper answer the task?
2. Thesis pass: Is the central answer defensible and stable?
3. Architecture pass: Does every major section have a necessary job?
4. Evidence pass: Does every important claim have appropriate support?
5. Source pass: Is source meaning accurate and attribution clear?
6. Paragraph pass: Does each paragraph develop one coherent intellectual function?
7. Sentence pass: Are relationships clear, precise and readable?
8. Language pass: Grammar, vocabulary, spelling, punctuation and formatting.
9. Submission pass: Citation style, document requirements, figures, references and file integrity.

Later passes cannot reliably repair earlier ones. A beautiful sentence in the wrong paragraph is still in the wrong paragraph.

Reverse outlining

After drafting, write the job of each paragraph in the margin. Read only those job labels. Do they form an argument? Is one stage missing? Does the paper spend four paragraphs establishing background and one paragraph answering the question? Reverse outlining reveals the structure the paper actually has rather than the one the writer remembers intending.

Read aloud for sentence pressure

Reading aloud can expose overloaded sentences, broken rhythm, missing referents and accidental repetition. It does not determine academic correctness, but it gives the writer another channel for noticing processing difficulty.

Revision after feedback

Feedback should become a transferable rule. If a tutor writes “too descriptive” beside a paragraph, do not simply add the word therefore. Ask what operation is missing. Is the paragraph reporting sources without comparing them? Presenting a result without interpreting it? Giving an example without explaining its relevance?

After repairing the paragraph, find another paragraph with the same risk and revise it without waiting for the tutor. That is the beginning of transfer.

Worked example: one paragraph, three different failures

Iona writes a paragraph with accurate source representation but no thesis connection. Her repair is to make the relevance explicit.

Maren writes a beautifully structured paragraph but overstates a correlational study as proof. Her repair is evidence calibration.

Leonie has a strong claim and relevant evidence but spends forty minutes perfecting it in a timed assessment. Her repair is execution: draft the load-bearing reasoning, mark optional refinement and continue.

“Weak academic writing” is again an insufficient diagnosis.

Academic writing practice ladder

1. Rewrite vague questions as precise academic tasks.
2. Turn a topic into three possible working questions.
3. Write thesis statements at three levels of claim strength and identify which one the evidence could support.
4. Give each paragraph in a model essay an intellectual job label.
5. Insert evidence into a paragraph and write the warrant explicitly.
6. Qualify an overgeneralised paragraph without making it empty.
7. Combine two source summaries into one comparative synthesis paragraph.
8. Reverse-outline a draft and reorganise it before sentence editing.
9. Revise one feedback pattern across three different paragraphs.
10. Write a fresh response on a new topic using the same reasoning architecture.

Evidence that academic writing is becoming independent

A writer is progressing when they can interpret a task without waiting for a template, select a defensible thesis, design paragraph roles, integrate rather than display evidence, preserve source meaning, qualify claims, revise structure before surface language, detect recurring weaknesses and reproduce the same quality on a new question.

That is the point at which academic writing becomes a system the learner can carry rather than a document the tutor helped produce.

Source use: preserve meaning, ownership and proportion

Academic work is built inside a conversation that began before the current writer arrived. Sources provide concepts, methods, data, arguments, counterarguments and evidence. The writer’s responsibility is not merely to include them. It is to represent them accurately enough that another reader could trace the reasoning back to its origin.

This creates three simultaneous obligations:

1. Meaning: do not distort what the source says.
2. Ownership: make clear which ideas, words and findings came from elsewhere.
3. Purpose: integrate the source into your own intellectual job rather than letting it replace your reasoning.

Quotation: preserve the exact language when the language matters

Direct quotation is useful when the wording itself is evidence, when a definition requires precise treatment, when a distinctive formulation will be analysed or when paraphrase would destroy an important feature of the source.

Quotation is less useful when the writer simply wants to report ordinary information. Long quotations can shift the burden of selection and interpretation onto the reader. A quoted passage still needs framing: why this wording, why here, and what does the reader need to notice?

The Harvard Guide to Using Sources makes a similar distinction: quotation is especially appropriate when the exact language is itself important, while paraphrase and summary are often better when the main job is to represent ideas.

Paraphrase: reconstruct meaning in a new language form

Paraphrase is not synonym replacement. A weak paraphrase keeps the source sentence structure, swaps several words and creates a text that is neither independent nor necessarily accurate. A strong paraphrase begins from the meaning model.

Use this process:

1. Read until you can explain the idea without looking at the sentence.
2. Identify the propositions and relationships that must survive.
3. Close or move away from the source.
4. Reconstruct the idea for your own sentence and paragraph purpose.
5. Compare against the source to check meaning.
6. Cite the source.

Changing the syntax is important because genuine reconstruction usually produces a different sentence architecture. But surface difference is not the true test. The test is whether meaning has been preserved and the language is genuinely the writer’s own.

Paraphrase can accidentally strengthen a source

Suppose the source says, “The intervention may contribute to improved persistence among students who receive structured mentoring.” A careless paraphrase—“Structured mentoring improves student persistence”—has removed both the uncertainty and the population condition.

This is why Iona records qualifiers as part of the proposition, not as optional style words.

Summary: compress the information structure

Summary requires a decision about scale. What is central at the paragraph level may be detail at the article level. A 100-word summary of a paper and a one-sentence summary of the same paper cannot preserve the same amount of information.

Strong summary identifies the source’s organising structure before compression. For an empirical paper: question, method, main result, interpretation and important limitation. For an argument: problem, thesis, major reasons and qualification. For a theory: central concepts and relationships.

Weak summary follows the source sentence by sentence and deletes detail. Strong summary rebuilds the hierarchy.

Synthesis: explain relationships among sources

Synthesis is not writing Source A, then Source B, then Source C. It is reorganising sources around the problem you are trying to understand.

A synthesis paragraph may show that:

• two studies agree despite using different methods;
• two studies conflict because their populations differ;
• a newer study narrows an older claim;
• a theory explains the mechanism behind an observed pattern;
• one source provides a measurement method and another provides application evidence;
• several sources reveal a consistent gap in the literature.

The paragraph structure should be controlled by the relationship, not the order in which the writer discovered the sources.

Source synthesis is a mediation task

The writer is moving meaning from several independent documents into one new representation. That makes synthesis part of the mediation system. Meaning must cross the boundary without source identities disappearing.

Citation: make the route traceable

Citation systems differ in punctuation and format, but their deeper purpose is stable: identify the source sufficiently for a reader to know where the borrowed material came from and, where required, find it.

Citation also marks intellectual boundaries. A sentence can contain the writer’s analysis and a source-supported factual claim; the citation should make that relationship understandable.

Do not treat citation as a decoration added during final formatting. Source identity should be preserved from the first note. Retrofitting citations from memory at the end of a project increases the risk of lost attribution and false source matching.

Reference management begins during reading

Record stable source information when the source enters the project. Depending on the type, this may include author, title, year, journal or publisher, DOI, URL, access information and page location for quoted material.

The exact citation style can be formatted later. The identity cannot be recovered reliably if it was never preserved.

Patchwriting

Patchwriting occurs when a writer follows the source’s language or syntax too closely while making local substitutions or small grammatical changes. It can arise from weak paraphrase skill, limited language resources, fear of changing technical meaning or poor note-taking.

The repair is not “use more synonyms”. Return to the conceptual model. Explain the idea orally. Draw the relationship. Write from notes rather than from the open source. Then verify accuracy.

Common knowledge is context-sensitive

Not every statement requires citation. Widely known facts within a community may be treated as common knowledge, but what counts as common depends on audience and discipline. “Water freezes at 0°C under standard atmospheric pressure” is different from a specialised estimate or contested interpretation.

When uncertain, consider whether the statement is specific, disputable, derived from a source or central to the argument. Academic conventions and instructors may also set local requirements.

Primary, secondary and tertiary sources

Source type matters because it changes what kind of evidence the source provides. A primary research article reports original research. A review synthesises previous research. A textbook or encyclopedia may offer a tertiary overview. A policy document can be a primary source for what an institution officially states even if it is not independent evidence that the policy works.

Do not rank source types mechanically. Use the type that fits the claim. If the question is “What does the examination authority require?”, the official specification is the appropriate primary authority. If the question is “Does this instructional method improve long-term transfer?”, an official marketing page is not enough.

Authority and relevance are different

A highly authoritative source can be irrelevant to the specific question. A recent national report may say little about the subgroup in your study. A famous theory may not fit the mechanism you are analysing. Source evaluation asks both “Can this source be trusted for what it reports?” and “Does it actually answer the part of the question for which I am using it?”

Currency is claim-dependent

Some claims need current evidence. AI tools, examination specifications, software behaviour and policy can change rapidly. Other claims concern stable concepts or historical events where an older foundational source remains essential.

Do not equate newer with better. Ask what could have changed.

Bias does not automatically invalidate a source

Every source has a position, purpose and selection process. A company report may provide accurate information about its own published policy while being weak evidence for independent product effectiveness. An advocacy organisation may collect important cases while selecting them for a particular mission. A government source may be authoritative for official statistics while using categories that deserve analysis.

Source literacy means matching the claim to the source’s strengths and limits.

Represent disagreement fairly

When sources disagree, do not flatten the disagreement into “opinions differ”. Ask why. Do they define the concept differently? Study different populations? Use different time windows? Apply different methods? Begin from different values? Have access to different evidence?

Sometimes one source simply has stronger evidence. Academic fairness does not require pretending all positions have equal support. It requires representing each position accurately and explaining the evidential basis of the judgement.

Literature review: organise a field, not a bibliography

A literature review should help the reader understand the state of a question. It may identify major theories, methodological patterns, consistent findings, contradictions, blind spots and unresolved problems.

The organising structure should come from the literature’s relationships:

• chronological development;
• competing theoretical traditions;
• methods;
• populations or contexts;
• themes;
• mechanisms;
• areas of consensus and uncertainty.

A source-by-source parade places the synthesis burden on the reader. A literature review earns its value by constructing the map.

The literature gap

A gap is not merely “nobody has studied this exact combination of three nouns”. A meaningful gap can be:

• a population missing from otherwise strong evidence;
• a mechanism assumed but not directly tested;
• methods that cannot answer an important causal question;
• conflicting findings that need explanation;
• a construct measured inconsistently;
• a context where an established result may not transfer;
• a practical problem not addressed by existing research.

The gap should matter to the research question, not just provide an excuse for novelty.

Source identity inside notes

Never create notes in which copied source wording can later be mistaken for your own. Use quotation marks for exact words even in private notes. Record page or section locations. Separate “source says” from “I think”.

This habit makes academic integrity easier because ownership is preserved before drafting begins.

Source identity inside drafts

Writers often draft from notes and later forget where one claim came from. Use provisional citation markers from the beginning. Do not insert “citation needed” for everything and hope to reconstruct the origin later.

Academic integrity is larger than plagiarism detection

Integrity includes honest representation of sources, data, contribution and process. Fabricating a reference, citing a paper you did not read as if you had, hiding inconvenient evidence, misrepresenting a quotation, submitting another person’s work, manipulating data without disclosure or using AI in ways prohibited by the task all affect the trustworthiness of the academic record.

A student should understand not merely what is forbidden but why traceability matters: other people make decisions based on academic claims.

AI-generated citations

Generative AI can produce plausible-looking references that do not exist, misstate titles or attribute a claim to the wrong source. Every source supplied by an AI system should therefore be verified against a reliable bibliographic or primary source before use.

The same rule applies to AI summaries of research. If the claim matters, inspect the original source.

AI paraphrase and authorship

An AI system can rewrite a source fluently, but that does not remove the need for citation and may introduce meaning changes the learner does not notice. If the learner cannot verify the paraphrase against the original, the tool has created a new risk rather than solving source use.

Institutional rules on AI assistance vary. Students must follow the rules that govern their course, assignment and institution. SETC’s durable principle is simpler: preserve authorship, verify meaning and ensure the learner can reproduce the underlying capability independently.

Quotation, paraphrase and summary comparison

Worked source-use example

Three students use the same paper.

Iona quotes three sentences because she is afraid of changing the meaning. Her paragraph is accurate but the source dominates. The repair is meaning-first paraphrase followed by verification.

Maren paraphrases confidently but changes “may improve” into “improves”. Her language is independent, but her source representation is not. The repair is qualification tracking.

Leonie preserves meaning and attribution but adds five sources to a paragraph already supported adequately. Her problem is decision efficiency. The repair is a stop rule: another source must add a different mechanism, population, method or counterposition to justify the reading cost.

Source-use practice ladder

1. Mark which parts of a paragraph are source report and which are writer analysis.
2. Paraphrase one sentence from meaning, then compare every proposition with the original.
3. Summarise the same source at 100 words, 50 words and one sentence.
4. Quote one sentence and explain why exact wording matters.
5. Combine two agreeing sources without writing separate mini-summaries.
6. Combine two disagreeing sources and explain the reason for disagreement.
7. Build a literature matrix with method, population, claim, evidence and limitation.
8. Draft a synthesis paragraph from the matrix without reopening the sources, then verify.
9. Audit a page of writing for ownership boundaries and missing citations.
10. Verify every AI-suggested source against the original record before using it.

Evidence that source use is becoming independent

A learner is progressing when they can decide whether to quote, paraphrase or summarise; preserve meaning and qualification; record source identity during reading; integrate sources around claims rather than names; recognise source limitations; explain disagreement; distinguish evidence from writer inference; and complete these operations without a tutor reconstructing the source boundary after the draft is written.

That is source use as intellectual responsibility rather than formatting.

Research communication: turn inquiry into a traceable record

Research communication begins before the final paper. A research question, proposal, ethics application, literature review, lab notebook, method, data display, conference presentation and dissertation chapter are different records of the same intellectual process. Each genre makes some part of the inquiry visible to another person.

This is why Academic English cannot be reduced to the argumentative essay. Research communication includes explanation, procedure, description, measurement, uncertainty, source synthesis, data commentary and methodological justification. The writing must often answer not only What do you think? but How do you know?

Research questions

A research question should be specific enough to guide evidence but open enough to require investigation. “What are the effects of social media?” is too broad to determine a method. “How does late-night short-form video use relate to self-reported sleep onset among first-year university students?” defines a population, behaviour and outcome more clearly, although the final research design may still require refinement.

Question design controls later language. A descriptive question asks what is happening. A comparative question asks how groups or conditions differ. A causal question requires evidence capable of supporting causal inference. An interpretive question may ask how participants understand an experience. A design question may ask what intervention can meet defined constraints.

Do not write a causal conclusion to a descriptive question simply because the discussion becomes interesting.

The research proposal

A proposal persuades a reader that a question is worth investigating and that the proposed route is coherent enough to produce useful evidence. It commonly needs:

• the problem or context;
• a focused question or aim;
• relevant literature;
• the gap or reason for the study;
• a method;
• feasibility;
• ethical or practical considerations;
• the expected contribution.

The proposal is future-facing. Verbs, modality and certainty should reflect that the research has not yet produced results. “The study will prove” is usually an impossible promise. Better language describes what the method is designed to investigate, estimate, compare or explore.

SETC already maintains a detailed Research Proposal Writing guide; Academic English places that genre inside the wider research system.

Abstracts

An abstract compresses a larger research record into a small, highly selective representation. The reader may use it to decide whether the full work is relevant. That gives every sentence a high information load.

Depending on discipline and genre, an abstract may include background, question, method, result and conclusion. The proportion matters. An abstract that spends half its space on general background may leave insufficient room for what the study actually did and found.

Write the abstract after the main work is stable enough to represent accurately. Then verify that every abstract claim exists in the full work and carries the same strength.

Literature reviews as research infrastructure

A literature review can be a section inside a larger study or a substantial research product in its own right. Its central job is not to prove that the writer has read many papers. It is to organise what is known in relation to a question.

High-quality literature reviews make relationships visible:

• where evidence is consistent;
• where results conflict;
• which mechanisms have been proposed;
• which populations dominate the evidence;
• which methods create recurrent limitations;
• how definitions vary;
• what has changed historically;
• which questions remain open.

Review structure should therefore come from the intellectual map, not alphabetic author order.

Systematic and structured reviews

Some reviews use explicit search, screening and synthesis procedures so readers can inspect how the evidence set was created. The language of such a review must distinguish the search universe, inclusion criteria, excluded material, study quality and synthesis method.

SETC need not turn every student into a systematic-review specialist, but academic writers should understand the principle: evidence selection is itself a method and can introduce bias.

Methods: write what another competent reader needs to evaluate the process

A method section explains how evidence was produced or selected. The relevant detail depends on discipline. It may include participants, materials, procedures, instruments, measures, data sources, analytic methods, coding decisions, inclusion criteria and ethical approval.

Method writing should be precise enough for evaluation and, where appropriate, replication. Vagueness such as “participants were asked some questions” hides the process. Excessive detail can also obscure the decisions that actually affect validity.

Use tense and voice according to function. Completed procedures are commonly described in past tense. Passive voice can foreground the procedure; active voice can clarify who made a decision. Neither is universally superior.

Operational definitions

Research frequently turns abstract concepts into measurable variables. “Engagement” might be measured through attendance, time on task, self-report, participation or a composite index. Each operationalisation captures something different.

Academic writing should not silently move from the measured variable to the full abstract concept. “Students clicked more resources” does not automatically mean “students were more intellectually engaged.” The operational definition sets the boundary.

Results: separate observation from explanation

A results section often reports what the analysis found before the discussion explains why it matters. This separation varies by discipline, but the underlying distinction remains useful: observation and interpretation are different operations.

Results language should identify values, patterns, comparisons, uncertainty and statistical or qualitative evidence accurately. Avoid emotional adjectives when the data can carry the point.

Data commentary

Good data commentary does not narrate every number in a table. It selects the pattern the reader needs:

• overall direction;
• largest or most relevant difference;
• exception;
• comparison with baseline;
• uncertainty or variability;
• relationship to the question.

Then interpretation can explain possible mechanisms or significance without changing what the data actually shows.

Percentages, counts and denominators

Academic prose can mislead accidentally when percentages are reported without denominators. A “100% increase” from one case to two is different from an increase from ten thousand to twenty thousand. Report the information necessary for proportionate interpretation.

Similarly, averages can hide distributions. A mean improvement may coexist with subgroup decline. The language should match the data structure.

Uncertainty

Research does not become weaker when uncertainty is reported honestly. Confidence intervals, ranges, error bars, limitations, alternative explanations and cautious language tell the reader how much weight a result can carry.

The challenge is calibration. Do not use uncertainty as a ritual disclaimer after making absolute claims throughout the paper. Build it into the wording of the claims themselves.

Discussion: interpret without escaping the evidence

The discussion reconnects results to the larger question. It may explain patterns, compare with previous work, consider mechanisms, identify limitations, discuss implications and propose future research.

The main danger is expansion. A study with one population suddenly becomes a claim about all students. A short-term outcome becomes a long-term intervention promise. A correlation becomes a causal explanation. A statistically detectable difference becomes a practically important one.

Use the discussion to widen the view carefully, not to remove the boundaries created by the method.

Limitations: explain what they change

A useful limitation section does not recite generic weaknesses such as “small sample size” without consequence. Explain what the limitation affects. Does it reduce precision? Restrict generalisation? Create selection bias? Prevent causal interpretation? Make one variable difficult to distinguish from another?

Then identify what evidence would reduce the uncertainty.

Implications

Research implications may be theoretical, methodological, practical or policy-related. The distance from evidence to action matters. A laboratory finding can suggest a mechanism without justifying immediate large-scale policy. A qualitative study can identify important experiences without estimating how common they are.

Strong implication language preserves that distance.

Research reports and dissertations

A long research document creates a global coherence problem. The literature review, method, results and discussion must answer the same central question even though they use different genres of academic language.

Maintain a question–chapter map:

Every chapter should contribute to the same research object.

Signposting across long documents

Long academic texts need visible architecture. Headings, previews, summaries and transition paragraphs can help readers understand where they are and why the next section exists.

Signposting should not become repetitive meta-commentary. “This section will discuss…” is useful when the route is genuinely complex; unnecessary when the heading already makes the job obvious.

Research posters

A poster compresses a research project into a visual encounter. It combines academic writing with hierarchy, visual design and spoken interaction. The viewer may spend seconds deciding where to look.

The poster should make the research question, evidence and conclusion easy to recover. Dense paragraphs copied from a paper defeat the medium. Figures and short explanatory text often carry more meaning.

Conference presentations

A presentation is not an essay read aloud. Listeners cannot reread the previous paragraph while the speaker continues. Oral academic communication therefore needs stronger signposting, controlled density, strategic repetition and pacing.

Slides should support the explanation, not compete with it. If the audience is reading a full paragraph, they are no longer listening fully.

Question-and-answer sessions

Questions test interactional academic English. The speaker must understand the question, decide whether it challenges evidence, scope or interpretation, and respond without inventing certainty.

Useful response moves include:

• clarifying the question;
• acknowledging the valid part of a challenge;
• distinguishing what the data addresses from what it does not;
• explaining a methodological choice;
• stating uncertainty;
• identifying what further evidence would be needed.

“I do not know from this study” can be an academically strong answer.

Research notebooks and audit trails

Academic English also lives in private working documents: notes, memos, coding decisions, data dictionaries and research logs. These records help the researcher remember why a decision was made and can support transparency later.

Write enough context for your future self. “Changed category” will be useless in three months. “Merged categories A and B because both refer to the same operational definition; retained original codes in archive” is recoverable.

Research collaboration

Collaborative academic writing adds negotiation. Authors must agree on terminology, structure, evidence standards and revision decisions. Shared documents make it easy to edit sentences but do not automatically create a shared argument.

Teams should establish the intellectual architecture before dividing sections. Otherwise each writer may produce a competent local chapter that does not combine into one coherent paper.

Peer review

Peer review asks a reader to evaluate the work rather than merely correct language. Good peer feedback distinguishes major from minor issues. A method flaw matters more than punctuation. An unsupported central claim matters more than one awkward sentence.

Writers need the reciprocal skill: interpreting feedback. Not every suggestion must be accepted, but every serious concern should be understood. If a reader misunderstood the argument, the writer should ask whether the text made the intended meaning recoverable.

Research genres are configurations of shared mechanisms

Proposal, abstract, literature review, method, report, poster and presentation look different because they prioritise different communicative jobs. They still draw from the same Atlas territories: reading, writing, evidence, mediation, register, data literacy, interaction and reasoning.

This is the benefit of the Academic English owner. The learner does not need a separate philosophy for every university genre. They need to understand how the common mechanisms are reweighted.

Worked research example

A student investigates whether retrieval practice improves vocabulary retention.

Iona checks the literature and notices that several studies measure immediate recall rather than delayed independent use. That changes the evidence map.

Maren refines the research question from “Does retrieval practice work?” to “Under what conditions does retrieval practice improve delayed retrieval and transfer of newly learned vocabulary?” The new wording creates a more useful literature review.

Leonie designs the workflow so reading does not continue indefinitely: foundational studies, recent reviews, transfer studies, then stop when new sources no longer add a distinct mechanism or population.

The quality of the eventual paper began with English decisions made before the first paragraph.

Research communication practice ladder

1. Turn a broad topic into descriptive, comparative and causal research questions.
2. Write a 150-word rationale explaining why one question matters.
3. Map five sources into consensus, conflict, method and gap.
4. Write a method description another student could follow.
5. Describe a figure without interpretation, then add interpretation separately.
6. Rewrite an overstated discussion paragraph so claim strength matches design.
7. Convert a 1,500-word report into a 200-word abstract.
8. Convert the same report into a poster outline.
9. Give a three-minute oral explanation and answer one hostile but fair question.
10. Write a final conclusion whose scope exactly matches the evidence set.

Each transformation is a mediation and register task. Mastery is demonstrated when the core meaning survives the change of genre.

Academic language resources: vocabulary, grammar, stance and disciplinary control

Academic English needs language resources capable of expressing complex relationships precisely. But the goal is not to sound difficult. The goal is to distinguish ideas that would otherwise collapse into the same vague wording.

A writer may need to distinguish correlation from causation, possibility from probability, description from explanation, necessary from sufficient conditions, average effect from subgroup effect, observation from inference and short-term change from durable trend. Academic vocabulary and grammar provide tools for those distinctions.

Academic vocabulary is relational

Learning the definition of significant is not enough. In statistics, statistically significant has a technical relationship to a test and threshold. In ordinary academic prose, significant may mean important or substantial. A writer must know the sense, collocation, discipline and evidential implications.

Academic vocabulary should therefore be learned through networks:

• word families;
• collocations;
• near-synonyms and distinctions;
• grammatical patterns;
• register;
• disciplinary meaning;
• common phrase frames;
• contrasting examples.

This aligns with SETC’s broader Vocabulary in Use system and Advanced English Vocabulary Mastery.

General academic vocabulary and disciplinary vocabulary

Some vocabulary travels across disciplines: evidence, factor, method, interpretation, consistent, derive, indicate, framework. Other vocabulary is highly specialised.

The distinction matters for learning priorities. A first-year student may gain broad value from general academic language while gradually acquiring the terminology of the field. A postgraduate specialist may need fine control of terms that rarely appear outside one discipline.

Do not replace everyday words without a reason

Academic writing does not require utilise whenever use is available. It does not require individuals whenever people is clearer. A more formal word is useful when it adds precision, carries the convention of the field or matches the genre. Otherwise it may add distance without meaning.

Collocation

Academic naturalness depends heavily on which words habitually combine. Writers draw a conclusion, conduct research, pose a question, address a limitation, provide evidence, raise concerns, account for variation and reach significance in specific technical contexts.

A thesaurus can suggest words but cannot guarantee collocation. Dictionary and corpus evidence help determine whether a phrase is typical in the intended register.

Phraseology

Academic writers use recurring phrase frames because recurring reasoning jobs need recurring language. Examples include:

• introducing a problem;
• defining a term;
• identifying a gap;
• comparing sources;
• reporting a result;
• qualifying a claim;
• acknowledging a limitation;
• explaining an implication.

Phrase resources can accelerate writing, but memorised phrases should remain subordinate to meaning. A phrase such as “It is important to note that” adds nothing if what follows is not important.

Grammar as academic meaning

Academic grammar is not a separate set of rules. It is ordinary English grammar being used to manage dense relationships. Several areas become especially important:

• clause relationships for cause, contrast, concession and condition;
• reference across long stretches of text;
• tense and aspect for research reporting;
• modality for certainty and obligation;
• voice for information focus;
• noun phrases for conceptual compression;
• relative clauses for definition and qualification;
• parallel structure for comparison;
• punctuation for hierarchy and scope.

Tense in academic writing

Tense choices depend on what the writer is doing. A completed method may be described in past tense. A figure in the present document may be described in present tense: “Figure 2 shows…”. A theory still treated as current may appear in present tense. Previous studies may be introduced through past or present perfect depending on discourse purpose and disciplinary convention.

Do not memorise one universal rule such as “literature review equals past tense”. Ask what time relationship and rhetorical status the sentence needs.

Modality

Modal verbs and related expressions carry epistemic and deontic meaning. May signals possibility. Must can express strong logical inference or obligation. Should can express expectation or recommendation. Could can express possibility, ability or a proposed option.

Because modality changes claim strength, it should be read and written carefully. A reviewer who changes may to will has altered the proposition, not merely improved style.

Conditionals

Academic reasoning often depends on conditions: an intervention may work if implementation reaches sufficient intensity; a conclusion may hold unless an unmeasured confound explains the pattern; a policy could have different effects depending on the baseline.

Conditional structures make boundaries visible. They are especially useful for preventing universal claims from escaping the evidence.

Concession

Concessive structures allow a writer to acknowledge evidence without abandoning the main judgement:

Although online delivery can increase geographical access, access alone does not ensure equivalent participation when device quality and study conditions differ.

The grammar makes the weighing relationship explicit.

Comparison

Comparatives require aligned dimensions. Grammar can expose invalid comparison: “The new programme had higher satisfaction than last year” compares a programme with a time period. Revision should identify the actual entities: “Satisfaction was higher among participants in the new programme than among the previous year’s cohort.”

Reference chains

Long academic paragraphs often contain this, these findings, such effects, the former, the latter and repeated conceptual nouns. Every reference should have a recoverable target.

“This demonstrates…” is dangerous when this could refer to the previous result, the method, the whole study or the author’s interpretation.

Lexical cohesion

Academic writers sometimes avoid repeating a key term because they were taught that repetition is poor style. They substitute near-synonyms and accidentally change the construct. If the paper distinguishes achievement, attainment, performance and learning, those words should not rotate merely for variety.

Repeat the technical term when conceptual stability matters.

Stance

Stance is how a writer positions themselves in relation to a claim. It includes certainty, evaluation, attitude, obligation and engagement with alternatives.

Stance can be carried through verbs, adjectives, adverbs, modal verbs, reporting structures and whole paragraph architecture. “Surprisingly, the intervention produced only a small effect” adds evaluation. “The intervention produced a small effect” reports more neutrally. Neither is universally better; the first requires a defensible reason for surprise.

Metadiscourse

Academic texts often guide readers explicitly: however, for example, in contrast, the next section, this distinction. These signals help readers navigate relationships.

Overuse can make prose mechanical. Use metadiscourse where the relationship might otherwise be costly to recover.

Nominal groups and information density

Academic noun phrases can become extremely dense:

the rapid post-pandemic expansion of institutionally supported hybrid assessment systems

This phrase packages multiple relationships before the verb arrives. Dense noun phrases are efficient for expert readers who share the concepts. They are difficult for novices and can hide ambiguity.

Writers should know how to both unpack and construct them.

Sentence length

There is no academic rule that sentences should be long. Sentence length should follow relationship complexity. A short sentence can carry a major conclusion. A long sentence can coordinate conditions that would become misleading if separated.

The quality test is whether the reader can recover the hierarchy without rereading unnecessarily.

Punctuation as structure

Commas, colons, semicolons, dashes and parentheses help academic writers mark boundaries and relationships. Punctuation should reduce ambiguity, not display sophistication.

A semicolon can join closely related independent clauses. A colon can introduce an explanation or list when the preceding clause licenses it. Parentheses can hold secondary information but can also bury something the argument actually needs.

Disciplinary differences

Academic English is not identical across fields. Humanities writing may foreground interpretation and textual evidence. Experimental sciences may use conventional report structures and compressed methods language. Social sciences often combine empirical evidence with theoretical framing. Mathematics relies heavily on symbolic representation but still requires definition, explanation and proof language. Design disciplines may integrate visual artefacts and reflective rationale.

Students should learn the English of their discipline by analysing authentic successful texts, assignment criteria and feedback—not by assuming one generic essay model governs the university.

Genre analysis

Take several successful examples of the same academic genre from the target field. Identify recurring moves:

• How do introductions establish the problem?
• Where does the research question appear?
• How are sources integrated?
• How much first person is used?
• How are limitations expressed?
• How do paragraphs begin?
• How are figures discussed?
• What does the conclusion do?

The goal is not imitation of one text but discovery of the genre’s option space.

Academic language and multilingual learners

Multilingual learners may have highly developed academic knowledge in another language. Teaching should avoid confusing unfamiliar English phraseology with lack of intellectual ability.

Use contrastive analysis where helpful. Show how English packages a relationship. Let the learner explain the concept fully before diagnosing the English form. Then practise the form until the idea can travel through English without losing precision.

Academic language and first-language English users

First-language speakers can also have weak academic phraseology, vague vocabulary or unstable formal grammar. Fluent conversation can mask these gaps because spoken interaction tolerates repair and shared context.

The same mechanism-based diagnosis applies.

Corpus and dictionary workflow

When uncertain about a phrase:

1. Check a learner or advanced dictionary for sense, grammar and labels.
2. Check authentic examples.
3. Use corpus evidence when typical collocation or disciplinary frequency matters.
4. Compare the examples with your intended sentence.
5. Do not copy a phrase whose meaning you cannot explain.

This turns reference tools into learning tools.

Academic vocabulary practice

For a target word or phrase, practise four forms of retrieval:

• define it in your own words;
• produce a common collocation;
• contrast it with a near-synonym;
• use it to express a real relationship from your subject.

Then return after a delay and use it in a new topic.

Language editing priorities

When editing academic language, prioritise meaning-changing errors:

1. unclear reference;
2. incorrect logical relationship;
3. wrong word sense or collocation;
4. tense/modality that changes claim status;
5. agreement or sentence boundary errors;
6. punctuation and style refinements.

This order is not absolute, but it protects the most consequential relationships first.

Evidence that academic language control is improving

The learner is progressing when they choose vocabulary for precision rather than prestige; preserve technical terms where stability matters; control claim strength through modality and reporting verbs; understand disciplinary variation; use reference tools independently; and revise sentences by asking what relationship must become clearer.

Academic language mastery is the ability to make complex meaning precise without making the prose unnecessarily difficult.

Academic listening, seminars and presentations: think with other people in real time

Academic English is not confined to pages. Lectures, seminars, tutorials, supervision meetings, presentations, viva-style questioning and collaborative research all require language to work under time. The participant cannot pause every sentence, consult a dictionary for every term or rewrite a question before responding. Listening and speaking therefore expose how available the underlying language system has become.

Lecture listening is structured prediction

A lecture is easier to follow when the listener tracks structure rather than trying to record every sentence. Speakers signal transitions: a new section, an example, a contrast, a definition, a return to the main point, an implication. The listener’s job is to identify the intellectual skeleton.

Before the lecture, activate enough background knowledge to recognise the territory. During the lecture, ask:

• What question is being answered?
• What is the main claim of this section?
• Which example illustrates the claim?
• Which term is being defined?
• Where does the speaker qualify or challenge the point?
• What relationship is the slide showing?

Notes should preserve hierarchy

Linear transcription hides structure. Use indentation, arrows, headings, abbreviations and spatial grouping to show relationships. A note should make it possible to reconstruct the lecture later.

For example:

Problem: low transfer after guided practice
→ possible cause 1: cue dependence
→ cause 2: surface overfitting
Evidence: delayed unfamiliar task
Implication: vary examples + remove prompts gradually

This note is shorter than the spoken explanation but preserves the reasoning.

Do not write while understanding is still collapsing

Some learners attempt to record a difficult phrase before they have understood it. Attention becomes trapped in transcription and later content disappears. If the speaker is moving quickly, preserve the concept first and mark the uncertain term for later verification.

Listening notes are not a transcript. They are a working model.

Use slides strategically

Slides can reduce listening load by providing names, diagrams or structure. They can also create divided attention when the listener copies text while the speaker explains something different.

Decide what the slide already preserves. If a definition is visible and will be available later, listen for the example or qualification the speaker adds orally.

Academic listening recovery

Missing one sentence should not destroy the lecture. Use recovery:

• mark the gap;
• keep following;
• listen for restatement or example;
• use the next section to infer what role the missed point played;
• verify after the lecture using slides, reading or a peer.

This is the same performance principle used in examination listening: protect the future signal.

Listening to questions

In a seminar or presentation, questions often contain several layers. A participant may give background, state an objection and finally ask the actual question. Do not answer the first recognisable phrase. Identify the communicative job.

Useful clarification moves include:

• “Are you asking whether the result would generalise beyond this sample?”
• “Do you mean the mechanism, or the evidence for the mechanism?”
• “If I understand correctly, the concern is about the baseline comparison.”

Clarification is academic control, not weakness.

Seminar discussion

A seminar is collaborative reasoning. Strong participation is not measured by airtime alone. A participant can contribute by introducing evidence, asking a precise question, connecting two earlier points, identifying an assumption, offering a counterexample, clarifying a definition or summarising where the group has reached.

Enter the discussion by relationship

Instead of waiting for an entirely new idea, connect to what is already active:

• Extend: “That mechanism may also explain…”
• Qualify: “That seems to hold when…, but perhaps not when…”
• Contrast: “The second study gives a different result because…”
• Clarify: “Are we using ‘equity’ to mean access or outcome?”
• Synthesise: “Both points seem to depend on the same assumption…”

These moves make interaction intellectually cumulative.

Disagreement

Academic disagreement should target the proposition, evidence or reasoning. Language can separate critique from personal attack:

I agree that the effect is visible in the sample, but I am not sure the design supports the causal interpretation.

The sentence acknowledges one point, identifies the exact disagreement and gives a reason. “I disagree” alone does not advance the discussion.

Turn-taking

Interaction requires social timing. Enter too late and the topic moves on. Enter too aggressively and discussion becomes competition. Notice pauses, completion signals, moderator cues and invitations.

For multilingual or hesitant speakers, preparing a small set of entry moves can reduce cognitive load without scripting the content.

Seminar preparation

Do not prepare a speech for a discussion. Prepare resources:

• one central claim from the reading;
• one piece of evidence;
• one question;
• one connection to another source;
• one point you are uncertain about.

This makes participation flexible when the conversation develops differently from expectation.

Academic presentations

A presentation converts a written or researched argument into a real-time experience. The audience cannot control pace in the same way as a reader, so the speaker must manage cognitive load.

A useful presentation architecture is:

1. establish the question;
2. explain why it matters;
3. give the audience a route;
4. develop a small number of load-bearing points;
5. use evidence that can be understood in the available time;
6. state the conclusion at the correct strength;
7. prepare for questions.

Slides are external working memory

A slide can hold a diagram, definition, figure or structural cue while the speaker explains. It should reduce the audience’s memory burden.

Slides become harmful when they contain the full script. The audience must choose between reading and listening. Use text for what needs to remain visible; use speech for explanation.

Data presentations

When showing a graph, orient the audience before interpreting:

• What is on each axis?
• What groups or conditions are shown?
• What time period?
• What is the main pattern?
• What exception or uncertainty matters?

Then explain why the pattern matters to the question.

Pronunciation and intelligibility

Academic speaking does not require imitation of one prestige accent. It requires enough pronunciation control for the audience to recover important words and relationships.

Stress becomes important in technical vocabulary. Pausing helps mark information structure. Intonation can signal contrast or completion. Rate should adjust to concept density.

A multilingual speaker with a noticeable accent can be an excellent academic communicator. The criterion is intelligibility and control, not accent erasure.

Fluency under intellectual load

A speaker may be fluent in conversation and hesitant during academic explanation because the reasoning load is higher. This does not necessarily indicate weak general English.

Reduce load by practising conceptual chunks: definitions, comparison frames, evidence descriptions and qualification moves. The goal is not memorised speeches but faster assembly of recurring academic operations.

Academic discussion vocabulary

Useful spoken academic language includes ways to:

• seek clarification;
• refer to a source;
• build on a previous point;
• signal uncertainty;
• challenge a claim;
• distinguish two meanings;
• summarise discussion;
• propose a next question.

Practise these functions across different subject matter so the language becomes generative.

Presentation rehearsal

Rehearsal should test the system, not merely increase familiarity. Record a run and check:

• Can a listener identify the main question in the first minute?
• Are transitions audible?
• Does every slide have one main job?
• Are technical terms explained before they become essential?
• Does evidence remain visible long enough to interpret?
• Does the conclusion match the evidence?
• Is the presentation within time without rushing the final third?

Timing

Leonie divides presentation time by intellectual load, not slide count. A complex result may need three minutes; a title slide needs seconds. Build checkpoints: if the presentation has not reached the main evidence by a defined time, compression must occur before the conclusion is damaged.

Question recovery

When a question is difficult:

1. pause;
2. identify what the question targets;
3. clarify if needed;
4. answer the part supported by your evidence;
5. state the boundary where evidence ends;
6. return to the larger claim if useful.

Do not fill uncertainty with invented detail.

Supervision and tutorials

One-to-one academic discussion is another genre. Students benefit when they arrive with questions and evidence of work rather than “I do not know what to do.” A useful supervision note can contain:

• what I tried;
• where the reasoning broke;
• which alternatives I considered;
• what decision I need help making.

This makes feedback more diagnostic.

Peer discussion and three-student tutorials

SETC’s three-student model is especially useful for academic communication because the group creates multiple reasoning paths while remaining small enough for each one to be visible.

Iona may challenge source interpretation. Maren may challenge structure. Leonie may challenge feasibility or timing. Each student hears a question they did not generate themselves and must adapt.

The teacher can then distinguish prepared knowledge from responsive control.

Academic speaking practice ladder

1. Explain one concept in 30 seconds without notes.
2. Explain the same concept to a beginner and then to a specialist.
3. Summarise a source orally in one minute while preserving its qualification.
4. Respond to a clarification question.
5. Build on another speaker’s point without repeating it.
6. Disagree with one claim and give the reason.
7. Explain a graph in 90 seconds.
8. Present a three-minute argument from one slide.
9. Answer an unseen question after the presentation.
10. Repeat on a new topic with reduced preparation.

Evidence that academic listening and speaking are becoming independent

Progress appears when the learner can follow structure rather than transcribe, recover after missing information, ask clarification questions, enter discussions by relationship, distinguish disagreement from contradiction, explain data without reading slides, respond to unseen questions, state uncertainty honestly and preserve intellectual control under time.

Academic communication becomes mature when spoken language can carry reasoning without a written script doing the thinking in advance.

Learning Academic English: multilingual transfer, AI, diagnosis and independent performance

Academic English is learned through repeated encounters with difficult knowledge, language and tasks. Explanation is useful, but explanation alone does not create availability. A student can understand what synthesis means and still fail to synthesise when four unfamiliar sources arrive under a deadline. The learning system must therefore move from explicit understanding into retrieval, variation, integration and independent performance.

Multilingual academic learners

Many university students study through English while also thinking, reading or communicating in other languages. That multilingual repertoire can support academic learning. A learner may understand a theoretical concept deeply in another language before acquiring the English phraseology required to discuss it. They may compare how two languages express evidence, certainty or agency and become more aware of the choices English requires.

The full SETC owner for that territory is Multilingual English & Mediation. Academic English applies those principles to research and study.

Do not confuse language form with conceptual knowledge

When a multilingual student gives a simple English explanation, ask whether the concept is simple or only the current language. Invite diagrams, examples or explanation in another available language where appropriate. If the conceptual model is strong, the teaching target may be English expression rather than the underlying idea.

Conversely, fluent English can conceal weak conceptual understanding. Elegant language should not substitute for a model.

Cross-linguistic transfer in academic writing

Languages and academic cultures differ in preferred rhetorical patterns. A writer may be accustomed to delaying the central claim, using different evidence conventions or signalling writer stance differently. The response should not be “your first language is wrong”. Show the expectations of the target academic community and explain what reader problem the convention solves.

This turns adaptation into repertoire expansion rather than identity correction.

Translation and academic work

Students may translate notes, concepts or draft ideas across languages. Translation can support learning if the learner verifies technical meaning and recognises that terms rarely map perfectly. A translated technical phrase should be checked against authentic usage in the target discipline.

Do not cite a translated wording as if it were the source’s original language unless that relationship is made clear.

AI in Academic English

Generative AI can produce fluent academic-looking language instantly. This changes the diagnostic environment. A polished paragraph no longer proves that the student can construct the reasoning, select the evidence or control the source independently.

Academic English teaching therefore needs a distinction between product quality and learner capability.

Useful AI roles

Depending on institutional rules, AI may be useful for:

• generating additional practice examples;
• simulating a critical reader;
• offering alternative explanations of a difficult concept;
• comparing possible paragraph structures;
• creating questions for retrieval practice;
• testing whether a definition has clear boundaries;
• providing provisional language feedback;
• simulating seminar questions;
• helping a learner inspect register differences.

The educational value comes from what the learner does with the output.

High-risk AI roles

Risk increases when AI:

• writes the assessed argument;
• creates source claims the student does not verify;
• generates citations;
• paraphrases material the learner does not understand;
• changes evidence strength invisibly;
• produces a literature review without source-level reading;
• substitutes for the planning and revision operations the task is meant to assess.

Institutional rules can prohibit or permit different uses. Students must follow the rules governing the actual assessment.

The independence test for AI

After AI-assisted practice, remove the system. Give a matched but unseen task. Can the learner still interpret, plan, read, write, cite and revise? If not, the tool may have improved the product without yet improving the learner.

SETC treats this as the same transfer test used for model answers and tutor scaffolds.

Verification

AI output is a claim source, not an authority. Verify factual information through appropriate sources. Verify quotations against originals. Verify citations. Verify numerical claims. Verify disciplinary terminology. If the AI says “research shows”, ask which research.

AI and academic voice

Repeatedly accepting generated revisions can make a learner’s writing smoother while weakening ownership of the decisions. A better exercise is comparative: ask the system for two revisions, identify what each changes, reject one choice and explain why, then write a third version independently.

The learner should remain the decision-maker.

Cognitive load in academic work

Academic tasks become difficult because many operations compete for limited attention. A student reading a research paper may simultaneously manage unfamiliar terminology, syntax, background knowledge, note-taking and the research question. A writer may juggle thesis, sources, paragraph structure, citation and sentence accuracy.

Teaching can reduce unnecessary load by isolating one operation before recombining it. Practise paraphrase on familiar content before paraphrasing a difficult theory. Practise figure description before interpreting a complex dataset. Practise seminar entry moves before asking the learner to discuss an unfamiliar paper.

Then restore authentic complexity. Isolation is a temporary training environment, not the final task.

Time management for coursework

Long deadlines create a different timing problem from examinations. A student can spend three weeks reading and discover that no argument exists. Use milestone decisions:

• question defined;
• initial evidence map built;
• source set bounded;
• working thesis;
• outline;
• first complete draft;
• structural revision;
• source verification;
• language and submission check.

Each milestone should produce an artefact that makes progress inspectable.

Procrastination can be diagnostic

Delay is not always a motivation problem. A student may avoid a task because the question is ambiguous, the reading set feels infinite, the paragraph has no job or the next decision is unknown.

Ask “What decision are you avoiding?” before prescribing more discipline.

Academic feedback

Feedback can target product, process or self-regulation.

• Product: this paragraph lacks evidence.
• Process: you are selecting sources before defining the paragraph claim.
• Self-regulation: before drafting the next section, test whether each source has a named role.

The third form increasingly transfers responsibility to the learner.

Responding to feedback

Do not treat feedback as a list of corrections. Cluster comments by mechanism:

• task interpretation;
• argument;
• evidence;
• source representation;
• paragraph architecture;
• academic language;
• citation;
• editing.

Then identify the earliest recurring cause. Ten sentence comments may all originate from one unstable reference pattern.

Feedback log

Maintain a small table:

The diagnostic matrix

Do not diagnose from one document

A poor essay may reflect unfamiliar content, deadline pressure, misunderstanding or one unusual task. Look for repeated patterns across work. Ask the learner to explain the process. Compare supported and independent performance.

Secondary English to Academic English

Secondary students already practise many academic operations: inference, evidence, explanation, comparison, summary, situational awareness, argument and editing. The bridge is to make these operations more independent and cross-disciplinary.

A Secondary student who can support comprehension answers with evidence can learn to support a research claim. A student who can write a clear expository paragraph can learn literature synthesis. A student who understands audience and purpose in situational writing can adapt register to a laboratory report or university email.

The mechanisms transfer; the context and responsibility increase.

General Paper to Academic English

GP is a particularly strong bridge because it trains issue interpretation, knowledge selection, argument, evidence, evaluation, source comprehension, summary and application. But university academic work changes several conditions:

• research is often deeper and more specialised;
• source selection becomes the student’s responsibility;
• citation becomes central;
• disciplinary conventions matter more;
• long projects require project management;
• methods and data may need explicit treatment.

The Advanced English / General Paper route therefore feeds naturally into Academic English without being identical to it.

First-year university transition

New students often receive feedback such as “be more critical”, “develop the argument”, “use the literature”, “improve academic style” or “engage with sources”. These phrases compress complex operations.

Translate them:

• Be more critical: test assumptions, evidence, criteria, alternatives and limitations.
• Develop the argument: explain why each reason supports the thesis and what follows.
• Use the literature: organise source relationships around your question.
• Academic style: improve precision, structure, stance and register, not just vocabulary formality.
• Engage with sources: interpret, compare and evaluate rather than inserting citations.

Postgraduate Academic English

Postgraduate work often increases the demand for disciplinary identity. The learner must not only understand the field but position work inside it. Literature sets become larger, methodological choices more consequential and claims more specialised.

The language challenge shifts toward synthesis, precision, research voice, uncertainty and long-document coherence.

Academic English for expert learners

Advanced users may no longer need broad grammar instruction. Their improvements can come from finer control:

• definition boundaries;
• claim calibration;
• source synthesis;
• disciplinary phraseology;
• compression without ambiguity;
• clearer figures and data commentary;
• stronger oral defence;
• editing for reader effort.

Three-student Academic English tutorials

A three-student tutorial creates an unusually useful academic environment because each learner can externalise a different part of the system.

Iona reads the source and defends what it actually says. Maren builds the paragraph or argument. Leonie tests whether the process is efficient and survives a constraint. Roles rotate. Each student must eventually perform all three operations.

Peer differences become evidence. One student’s overclaim lets the group practise evidence calibration. Another’s clearer synthesis shows what changed. The tutor can observe individual reasoning while preserving interaction.

Scaffolding

Scaffolds can include source matrices, paragraph job labels, question templates, synthesis frames, revision checklists and model analyses. They are useful when they reduce unnecessary load and direct attention to the target mechanism.

They become harmful when the learner cannot perform after removal.

Fade deliberately

A scaffold sequence may move:

worked model → annotated model → partial framework → checklist → self-questioning → independent task.

Do not remove everything at once. Remove the support that the learner can now internalise.

Transfer tasks

After learning a mechanism, change the surface:

• new topic;
• new discipline;
• new source type;
• new genre;
• new time constraint;
• new audience;
• reduced support.

If the capability survives, it is becoming portable.

Academic English practice week

A balanced week for an advanced learner might include:

• Reading: one difficult article with argument mapping.
• Vocabulary: five to ten discipline-relevant phrases retrieved in context.
• Source use: one paraphrase and one synthesis exercise.
• Writing: one paragraph built from claim–evidence–warrant.
• Listening: one lecture segment with structure-first notes.
• Speaking: a two-minute explanation plus unseen question.
• Review: revisit one old feedback pattern on a fresh task.

The exact volume should match the learner. The principle is integration without overload.

Measure progress beyond grades

Grades are important evidence but can be noisy and course-specific. Also track:

• time needed to read a comparable text;
• quality of source notes;
• ability to reconstruct an argument without the source open;
• frequency of source distortion;
• thesis stability;
• paragraph-role clarity;
• revision independence;
• presentation timing;
• response to unseen questions;
• performance after support is removed.

When Academic English tuition should reduce

Support should reduce when the learner can diagnose common problems, plan and research independently, interpret feedback, revise future work, verify source claims and recover from difficulty without waiting for the tutor.

The goal is not permanent expert supervision of every assignment. The goal is a learner who increasingly knows how to learn, write and communicate inside an academic discipline.

The final transfer test

Give the learner a fresh academic problem outside the practised topic. Require independent reading, source representation, argument or explanation and final verification. Delay the task long enough that short-term memory of the model is less useful.

Ask:

• Can the learner define the task?
• Can they locate useful evidence?
• Can they represent sources accurately?
• Can they organise reasoning?
• Can they choose language that matches claim strength?
• Can they finish under the constraint?
• Can they explain why they made the important decisions?

If yes, Academic English is becoming a capability rather than a collection of university conventions.

Disciplinary Academic English: one system, different knowledge cultures

Academic English changes across disciplines because disciplines ask different questions, recognise different kinds of evidence and communicate through different genres. A student who succeeds in a literature essay can still need explicit support when writing a laboratory report. A scientist who writes precise methods may find reflective or interpretive writing unfamiliar. The underlying English system remains connected, but the balance of language, reasoning and evidence changes.

Humanities

Humanities writing often gives substantial weight to interpretation, conceptual framing, historical context and close analysis of texts or artefacts. Evidence may include quotations, archival material, primary documents, visual objects or theoretical sources.

The writer needs to distinguish observation from interpretation. “The narrator repeats the phrase three times” is an observation. “The repetition creates ritualistic inevitability” is an interpretation requiring explanation. Strong humanities writing shows how features of the evidence support the interpretive claim rather than assuming the quotation is self-explanatory.

Definitions can be contested. Terms such as modernity, identity, genre or representation may have theoretical histories. The writer should position the use rather than pretend one dictionary sentence settles the field.

Social sciences

Social-science writing often coordinates theory, empirical evidence, population, measurement and causal or explanatory reasoning. Writers need strong control of generalisation. A finding among one group at one time should not silently become a universal statement about “people”.

Method vocabulary matters because design affects inference. Sample, variable, association, effect, construct, validity, bias and confound are not decorative terms. They define what the evidence can support.

Natural sciences

Scientific Academic English often rewards concise method description, precise measurement, data commentary and clear distinction between results and interpretation. The language may appear impersonal, but the deeper requirement is traceability.

Numbers need units. Comparisons need baselines. Observations need conditions. Figures need labels. Claims need to distinguish what was measured from what is inferred. A beautifully written science report with an ambiguous variable or missing method detail is not academically strong.

Engineering and computing

Engineering and computing writing frequently combines technical description with design decisions, constraints, requirements, testing and trade-offs. A system may be “better” on speed and worse on energy consumption. A model may improve benchmark performance while increasing latency or reducing interpretability.

Writers should make evaluation criteria explicit. Technical readers need to know what was optimised, what constraints were accepted and which failures remain possible.

Mathematics

Mathematics uses symbolic language extensively, but prose still carries definitions, assumptions, proof structure, interpretation and explanation. “Clearly” is not a proof step. A variable should be defined before use. A conclusion should follow from the stated conditions.

Explanatory mathematical English helps readers understand not only which manipulation occurred but why the move is valid.

Education

Education research often crosses psychology, sociology, policy, measurement and classroom practice. Writers need to distinguish achievement, learning, engagement, participation, attainment and performance rather than using them as synonyms.

Interventions are especially vulnerable to causal overstatement because classroom conditions are complex. Describe implementation, comparison conditions, population and outcome carefully.

Law

Legal Academic English may involve close interpretation of statutes, cases, doctrine and competing authorities. Precision of scope and definition is unusually important. A word can have a technical legal meaning different from everyday use.

SETC can explain the language mechanisms—definition, argument, source hierarchy, qualification—while recognising that substantive legal interpretation requires the appropriate legal expertise and jurisdiction.

Medicine and health sciences

Health research writing often needs careful risk language, population description and differentiation between statistical and clinical significance. Claims can have real-world consequences.

A result in a controlled study should not automatically become medical advice. Academic language should preserve the research boundary.

Business and management as academic fields

Academic business study differs from ordinary workplace communication. A university management paper may evaluate theory, analyse cases, interpret financial or organisational data and cite research. The academic owner remains SETC because the task is scholarly English. When the task becomes actual workplace communication, career documentation or professional role application, the federation hands off to eduKateSG.

Interdisciplinary writing

Interdisciplinary work creates translation problems between knowledge cultures. The same word may mean different things across fields. Evidence standards may differ. A computational model, ethnographic account and historical archive do not produce the same kind of claim.

The writer must establish enough shared language for readers to see how the different forms of evidence relate without flattening them into one method.

A complete worked Academic English cycle

Consider a student project asking:

Under what conditions does AI-generated feedback improve student revision without increasing dependence on automated writing support?

Stage 1: Interpret the question

Maren identifies the load-bearing terms: AI-generated feedback, revision, improve, dependence. Each needs a working boundary. Is feedback sentence correction, conceptual critique or full rewrite? Is improvement measured by immediate draft quality or future unaided writing? Dependence cannot be inferred merely because the tool is used often.

The question is conditional rather than binary. The final answer must identify conditions.

Stage 2: Build the evidence map

Iona creates source categories before collecting papers:

• research on automated or AI feedback;
• research on formative feedback and revision;
• research on learner agency and self-regulation;
• research on transfer and scaffold dependence;
• studies that compare assisted and unassisted performance.

This prevents the search from becoming a list of papers containing the keyword “AI”.

Stage 3: Screen sources

One source measures user satisfaction. Useful for acceptability, but not evidence of revision quality. Another compares drafts before and after AI feedback but has no unaided retest. Useful for immediate improvement, weak for independence. A third includes delayed unaided writing. That source is highly relevant to the dependence question.

Relevance is determined by the research model, not the title.

Stage 4: Preserve source meaning

Iona’s note for a study records:

Students using automated feedback corrected more surface errors on revised draft; no measure of later independent writing; authors frame system as revision support, not proof of learning transfer.

That final qualification will later prevent Maren from writing “AI feedback improves writing ability.”

Stage 5: Build the synthesis

The sources suggest three distinct mechanisms:

1. AI can increase the amount and immediacy of feedback.
2. Feedback helps only when the learner can interpret and act on it.
3. If the system performs the revision itself, product improvement may coexist with weak transfer.

The literature review can now be organised by mechanism rather than software brand.

Stage 6: Form a working thesis

Maren writes:

AI-generated feedback is most likely to improve durable revision capability when it identifies problems, prompts learner decision-making and is followed by unaided retesting; evidence of improved assisted drafts alone is insufficient to establish independent learning, particularly when systems generate replacement prose.

The thesis is strong enough to guide the essay but leaves room for evidence that may refine the conditions.

Stage 7: Design paragraph jobs

• Paragraph 1: why feedback timing and volume could help.
• Paragraph 2: why interpretation and learner action mediate the effect.
• Paragraph 3: evidence separating revised-product quality from unaided transfer.
• Paragraph 4: conditions under which generated language may increase dependence.
• Paragraph 5: implications for teaching design and assessment.

Every paragraph is now a reasoning stage.

Stage 8: Draft evidence with warrants

Maren does not write:

Smith (2026) found fewer errors after AI feedback.

and move on.

She adds the inference and boundary:

The reduction in surface errors indicates that automated feedback can improve the immediate revision product, but because the study did not include a later unaided task, the result does not establish that students internalised the editing decisions.

The evidence now has a job.

Stage 9: Control time

Leonie gives the project a source stop rule and drafting deadline. When another paper repeats the immediate-revision result without adding transfer evidence, it is recorded but does not delay the first complete draft.

She protects a full revision cycle because discovering a structural problem after the deadline is more damaging than missing one additional source.

Stage 10: Reverse outline

The first draft reveals that the paragraph on “AI dependence” contains three different meanings: frequency of use, inability to perform unaided and emotional reluctance to work without AI. The term drift would make the conclusion unstable.

Maren chooses the operational meaning relevant to learning: reduction in independent capability when support is removed. Other meanings are acknowledged but separated.

Stage 11: Source audit

Iona checks every major factual claim:

• Does the source actually support it?
• Is the population preserved?
• Is the result causal or correlational?
• Did the writer remove a qualification?
• Is a citation attached to the right proposition?

Stage 12: Language edit

Only after structure and evidence stabilise does the sentence-level edit begin. Several phrases are simplified. “Utilisation of generative artificial intelligence technologies” becomes “use of generative AI” where no technical distinction is lost.

Two absolute claims become conditional. One long nominalised sentence becomes an active explanation because the responsible actor matters.

Stage 13: Final independence test

The student closes the draft and explains the argument orally in three minutes. If the writer cannot reconstruct the line of reasoning without reading the essay, the document may contain language that has not become an owned model.

Then the student answers an unseen question: “Would the same argument apply to AI-generated brainstorming rather than feedback?” The response requires transfer.

This is Academic English as a full cycle, not a formatting style.

Academic assessment and deadline recovery

Academic performance can fail even when the learner possesses the underlying capability. Long projects create planning risk; timed examinations create execution risk.

Timed academic essays

A timed essay compresses task interpretation, retrieval, planning, writing and checking. The learner needs decision thresholds.

Use an opening budget:

• decode the question;
• define the key term if needed;
• choose the thesis;
• select two to four load-bearing reasons;
• identify evidence;
• begin before planning becomes avoidance.

Protect the final minutes for global checking: unanswered question, contradiction, missing conclusion, severe reference ambiguity and obvious language errors.

Coursework recovery

When a long assignment is behind schedule, do not respond by polishing the introduction. Rebuild the minimum complete architecture:

1. question;
2. working thesis;
3. section jobs;
4. minimum evidence for each job;
5. complete rough draft;
6. one structural revision;
7. source and language verification.

Completeness creates something that can be improved. Perfection in the first third creates an unfinished paper.

When a source cannot be understood

Use a recovery ladder:

1. read abstract/introduction/conclusion for the map;
2. identify the one section relevant to your question;
3. define essential terms;
4. find a reliable introductory explanation of the concept;
5. return to the source;
6. if the source remains unusable and is not essential, replace it rather than losing the project.

When the argument stops working

Do not force more evidence into a broken thesis. State the evidence pattern in plain language. What does it actually seem to show? Rewrite the thesis to fit the evidence. Academic writing is allowed to discover a different answer from the one the writer expected.

When feedback arrives late

Triage feedback by expected return:

• thesis/answer problem;
• structural problem;
• evidence/source problem;
• paragraph reasoning problem;
• sentence/language problem;
• formatting problem.

Repair the highest-level issue that can still materially change the submission.

Academic English FAQ

Is Academic English just formal English?

No. Formality is one register dimension. Academic English is a larger system for reading, reasoning, source use, evidence, research, writing, discussion and communication. Some academic genres are relatively conversational; others are highly formal.

Do I need difficult vocabulary to sound academic?

No. You need vocabulary precise enough for the distinctions in your field. A simple accurate word is better than an advanced word used unnaturally.

Should I avoid first person?

Not universally. Conventions vary by discipline and genre. Use first person when it clarifies agency or is accepted by the task; avoid it when it adds nothing or violates a specific convention.

Is passive voice more academic?

No. Passive voice is useful when the process or object deserves focus. Active voice is useful when the actor or decision-maker matters. Choose by information structure.

How many sources should an essay have?

There is no universal number. The question, assignment, discipline and evidence burden matter. Ten redundant sources are weaker than five well-chosen sources that cover the necessary mechanisms, evidence and counterpositions.

Can I use AI to improve Academic English?

Possibly, depending on institutional rules. Use it in ways that preserve authorship, source verification and independent learning. Do not treat generated citations or factual claims as verified evidence.

How do I become more critical?

Replace the vague instruction with operations: test definitions, evidence quality, method, alternative explanations, generalisation, criteria, trade-offs and uncertainty. Critical thinking is not automatic negativity.

How do I improve academic style?

Begin with meaning. Stabilise the question and reasoning. Then improve precision, information order, reference, cohesion, stance and sentence economy. Style editing cannot repair a missing argument.

How can I read academic papers faster?

Improve orientation, genre knowledge, vocabulary and selective close reading. Do not force uniform speed. Read load-bearing sections deeply and low-value sections according to purpose.

How do I know whether my paraphrase is safe?

Check that the language and structure are genuinely reconstructed, every important proposition and qualification remains accurate, and the source is cited. Compare against the original after writing.

What is the difference between summary and synthesis?

Summary compresses one source or body of information. Synthesis reorganises multiple sources around relationships relevant to a question.

What should I do when sources disagree?

Identify the source of disagreement: definition, population, method, evidence, time period, theory or values. Then weigh the evidence rather than writing only that “there are different views”.

How do I develop an academic voice?

Voice develops through repeated decisions about framing, evidence, definitions, source relationships, qualification and judgement. It is not created by inserting opinion phrases or removing every personal pronoun.

Evidence and further learning

Current university Academic English provision supports the broad architecture used here. University College London’s English for Academic Purposes modules explicitly integrate academic reading and writing with listening, speaking, academic registers, coherence, rhetorical strategies, voice and discipline-relevant communication. UCL’s preparation courses also connect critical reading, writing, note-taking, summarising, referencing, discussion, presentations, feedback and independent study.

The University of Edinburgh’s Academic Communication courses likewise combine listening and note-taking, reading and text analysis, discussion, writing and presentation, with advanced work on source synthesis, referencing and responsible AI use. The Harvard Guide to Using Sources distinguishes quotation, paraphrase and summary according to the writer’s purpose and emphasises clear source attribution.

Useful external reference routes include:

• UCL English for Academic Purposes
• UCL Preparation for Academic Studies in Higher Education
• UCL Academic English
• University of Edinburgh: English for Successful Academic Communication 1
• University of Edinburgh: English for Successful Academic Communication 2
• Harvard Guide to Using Sources: Summarising, Paraphrasing and Quoting

These external resources are not SETC’s curriculum. They are evidence that strong Academic English work is multi-skill, source-aware, discipline-sensitive and concerned with independent academic communication rather than vocabulary formality alone.

Continue through SETC

• English Atlas — locate Academic English inside the whole language, learning and performance system.
• How English Works — return to the language mechanisms beneath academic communication.
• Reading and Comprehension — deepen reading, inference, evidence and source interpretation.
• Writing and Composition — continue into genres, structure and revision.
• Multilingual English & Mediation — translation, paraphrase, synthesis and meaning across boundaries.
• English Proficiency — understand how academic performance fits into a wider ability profile.
• Advanced English / General Paper — the pre-university bridge into academic reasoning.
• Advanced English Vocabulary Mastery — develop C1–C2 lexical precision, collocation and phrase control.
• Three-Student English Tutorials — see the teaching model used to make reasoning and language decisions visible.

The Academic English standard

A learner has not mastered Academic English because they can make prose sound formal. Academic control is visible when they can enter an unfamiliar intellectual problem, locate and understand appropriate sources, preserve source meaning, distinguish evidence from inference, construct a defensible line of reasoning, communicate it in the required academic genre, respond to critique, verify the record and reproduce the capability when the topic, discipline or support changes.

Read accurately. Think proportionately. Represent sources honestly. Build reasoning another person can inspect. Communicate it in the form the academic community can use. Then prove that the system still works when nobody is standing beside you.

Academic English curriculum: build the system in the order the learner needs

Academic English does not need one universal twelve-week course, because learners arrive with different profiles. A strong reader may need source-integrated writing. A fluent writer may need research methods language. A multilingual postgraduate may need seminar interaction rather than basic grammar. The sequence below is therefore a curriculum architecture, not a rigid calendar.

Entry diagnostic

Collect evidence from several modes before deciding what to teach:

• one difficult academic reading with a short reconstruction of its argument;
• one source note showing claim, evidence, method and limitation;
• one paragraph integrating at least two sources;
• one short academic explanation or argument without a model;
• one spoken explanation followed by an unseen question;
• one feedback sample from recent school or university work where available.

The purpose is not to generate a score. It is to locate the first bottleneck.

Phase 1: academic orientation

Build the distinction between topic, question, claim, evidence, source and argument. Teach the learner to identify what a task is asking and to distinguish academic communication from simply sounding formal.

Exit evidence: the learner can take an unfamiliar assignment question and explain the intellectual job in plain language.

Phase 2: reading architecture

Train orientation, paragraph function, claim–evidence mapping, qualification and recovery from dense sentences. Use sources difficult enough to create real reading decisions but not so specialised that missing domain knowledge overwhelms the target skill.

Exit evidence: the learner can reconstruct a source’s question, central claim, evidence type and limitation without copying the abstract.

Phase 3: notes and source ownership

Replace transcription with structured notes. Preserve source identity, distinguish source report from learner interpretation and practise meaning-first paraphrase.

Exit evidence: notes remain usable after a delay and do not confuse source language with the learner’s own wording.

Phase 4: paragraph reasoning

Build paragraphs around intellectual jobs. Train claim, evidence, warrant, qualification and connection to the larger thesis.

Exit evidence: the learner can explain why every sentence belongs and what would be lost if the paragraph were removed.

Phase 5: synthesis

Move from one source to source relationships. Compare method, population, definitions, evidence and conclusions. Write by relationship rather than author order.

Exit evidence: a synthesis paragraph uses several sources without becoming several mini-summaries.

Phase 6: academic language precision

Target the grammar and vocabulary revealed by real work: modality, reporting verbs, reference, cohesion, collocation, nominalisation, conditionals, disciplinary terms and information order.

Exit evidence: the learner can make a dense paragraph more precise without merely making it longer or more formal.

Phase 7: argument and evaluation

Train criteria, counterargument, causal reasoning, comparison, definition stability and evidence calibration.

Exit evidence: the learner can narrow a claim when evidence is insufficient and explain why the revised claim is stronger academically.

Phase 8: research genres

Apply the system to proposals, literature reviews, reports, methods, results, discussion, abstracts or other genres relevant to the learner’s field.

Exit evidence: the learner can identify which common Academic English mechanisms have been reweighted for the new genre.

Phase 9: spoken Academic English

Build lecture notes, seminar entry, clarification, disagreement, presentation and question recovery.

Exit evidence: the learner can explain a difficult concept and respond to a question that was not rehearsed.

Phase 10: integrated research task

Assign a small research problem requiring source search, reading, notes, synthesis, writing and presentation. Support only the bottleneck.

Exit evidence: the learner coordinates several Academic English systems without quality collapsing at the integration point.

Phase 11: feedback-to-transfer

Use external feedback to identify a recurring mechanism, repair it and test it on a different task.

Exit evidence: improvement appears where the original feedback wording is absent.

Phase 12: scaffold removal

Remove matrices, sentence frames, tutor-selected sources and planning prompts progressively. Give a fresh task after delay.

Exit evidence: the learner can define, research, communicate and verify the work independently.

Choose the first route by the evidence you already have

What a premier Academic English system should eventually make possible

A learner should be able to enter a new discipline and learn its language from authentic texts. They should know how to use dictionaries, corpora, specifications and disciplinary examples without waiting for a permanent vocabulary list. They should be able to read a source sceptically without becoming cynical, use evidence without surrendering their own reasoning, and change their mind when the evidence changes.

They should be able to write for a reader rather than for a template; speak with enough control to think aloud with other people; translate difficult knowledge into clearer forms without distorting it; and use AI, search and digital tools as resources without outsourcing the intellectual work those tools are meant to support.

Most importantly, they should be able to tell the difference between not knowing the English, not knowing the concept, not understanding the task and not yet being able to perform under the constraint. Those are different problems and deserve different repairs.

Academic English becomes powerful when it makes the learner more capable of entering future knowledge independently. The final product is not one excellent essay. It is a person who can encounter a new intellectual world, learn how that world communicates, and participate in it with increasing accuracy, judgement and responsibility.

Apply Academic English in real university and research genres

Use the guide above for the complete system, then move into these existing SETC applications when the task becomes a specific academic genre.

• Academic Writing — structure, evidence, sources, clarity and formal style.
• Research Writing — questions, sources, notes, synthesis, citation and drafting.
• Literature Review Writing — synthesise themes, methods, gaps and critical relationships across sources.
• Research Proposal Writing — question, literature, methods, feasibility and contribution.
• Abstract Writing — compress purpose, method, results and conclusion without distorting the larger study.
• Annotated Bibliography Writing — citation, summary, evaluation and research relevance.
• Statement of Purpose Writing — academic direction, research fit, evidence and future direction.