An Academic Phrasebank for Each Part of a Research Paper
An academic phrasebank of copyable lines for each part of a research paper, from the introduction through the abstract, with hedging and citations.

An academic phrasebank is a set of lines for the jobs a research paper repeats. A useful line names the job, leaves a blank where your own material goes, and drops out once the sentence can stand without it. A line that could be pasted into any paper is not finished.
The lines in this guide were written for the guide. None of them comes from the University of Manchester Academic Phrasebank, from a published paper, or from a lab notebook. The passages are models. There is no study behind them. Copy the shape, then put in the topic, the procedure, and the finding from your own draft.
When a sentence already carries your finding, your source, or your question, and only the wording is stiff, Rephrase AI can rephrase that sentence. Rephrase AI is designed to preserve the meaning. Keep the new sentence only if the finding, the source, and the question are unchanged. Your instructor's rules on AI use still apply.
1. What an academic phrasebank is for
The University of Manchester Academic Phrasebank describes itself as a general resource for academic writers. The homepage says the resource aims to give examples of the phraseological nuts and bolts of writing, organized by the main sections of a research paper or dissertation, with other phrases under general communicative functions. The top menu lists introducing work, referring to sources, describing methods, reporting results, discussing findings, and writing conclusions. A side menu lists general functions, including being cautious, being critical, and signalling transition.
The same page says the resource should be particularly useful for writers who need to report research. The phrases and the headings can help a writer think about content and organization, or they can be incorporated where that is appropriate. In most cases, the page says, a certain amount of creativity and adaptation will be necessary. The items are mostly content neutral and generic. The homepage says that using them is therefore not stealing other people's ideas and does not constitute plagiarism. Some entries include specific content words for illustration, and those words should be substituted. The page says the resource was designed primarily for academic and scientific writers who are non-native speakers of English, and that native speakers may still find much of the material helpful. The homepage also says recent data suggest that the majority of users are native speakers, and it does not give those figures. The site was created by John Morley. The homepage says Academic Phrasebank is the intellectual property of the University of Manchester, and the page prints a 2023 copyright line.
The about page says the resource draws largely on an approach pioneered by John Swales in the 1980s. In the page's summary, Swales used genre analysis to identify rhetorical patterns in research-article introductions, and he defined a move as a section of text that serves a specific communicative function. The about page cites that work as Swales 1981 and 1990. The page says this unit is one of the main organizing sub-categories of the Academic Phrasebank, and that much of the language Swales was interested in was phraseological. The page also says much of the language people use is stored and retrieved as pre-formulated constructions, and that the resource tries to make examples of that dimension explicit.
The phrases, the about page says, were taken from authentic academic sources. The original corpus was 100 postgraduate dissertations completed at the University of Manchester. Phrases from articles across a broad range of disciplines have also been incorporated, and the page says that incorporation continues. In most cases the phrases have been simplified and sifted from their particular content. Where content words remain as illustrations, the page says they are substitutions and should be replaced when the phrase is used.
The about page says four questions are asked before a phrase is added. Does the phrase serve a useful communicative purpose in an academic text? Does the phrase contain collocational elements, formulaic elements, or both? Are the content words generic? Does the combination sound natural to a native speaker or writer of English? A later heading on the same page turns to when reuse is acceptable. The page reports a 2013 survey by Davis and Morley of 45 academics at two British universities, followed by interviews. The characteristics the page says emerged are these. A reused phrase should not have a unique or original construction. It should not express a clear point of view of another writer. It may be up to nine words, and past that length acceptability declines. It may contain up to four generic content words, meaning nouns, verbs, or adjectives that are not bound to one discipline. The page treats disciplinary phrases as a separate, usually shorter category, often noun phrases, and says acceptability for those depends on how far members of that academic community use and understand them. The sentences below are full example sentences, not short fragments inside that nine-word note. A writer who wants to stay inside that note should keep only a short generic piece and write the rest from the draft.
This guide follows the paper's parts rather than the Manchester menus. The guide adds the research question, the abstract, and the rebuttal, because those jobs sit inside or beside the sections, and the guide keeps citing and hedging together because both of them travel through every section. The lines are originals for this page. A sentence that copies another writer's point of view is outside the reuse the about page describes, and a paraphrase of a source still needs a citation.
2. How long an academic phrase should be
Use two different lengths, and do not mix them up.
The short length is the reusable piece the Manchester about page describes. That page, reporting the 2013 survey, says acceptability declines once a reused phrase runs past nine words, and the page caps generic content words at four. That note is about a fragment you drop into a sentence you wrote. It is not a word count for the sentence, the paragraph, or the section.
The long length is the section. None of these pages states one word count for every paper. Use the limit in the assignment or the journal.
| Piece of writing | What the page says about length |
|---|---|
| A reused generic phrase | The about page says a reused phrase may be up to nine words, and that acceptability declines beyond that. The page also says the phrase may contain up to four generic content words. |
| A dissertation or thesis introduction | The introducing-work page says these introductions tend to be relatively short compared with the other sections, and functionally complex. |
| Methods | The describing-methods page says the section tends to be much more extensive when the methods are new, unfamiliar, or controversial, or when the audience comes from many disciplines. The page gives no word count. |
| A conclusion | The writing-conclusions page says conclusions are shorter sections. In a research article they may be optional when the discussion already consolidates the study. |
| An abstract | UNC calls an abstract short and self-contained. The definition this guide uses does not set one word count for every paper. |
| Any full section | No page here replaces the limit on the assignment in front of you. |
A model sentence in this guide is often longer than nine words because the sentence includes the model topic. If you keep the sentence and only swap one noun, you have copied a line, not adapted a fragment. Cut the line down to the job, then write the details from your notes.
3. How to build a sentence from a phrase
Start from the job, not from a favorite line.
- Name the job in the margin. Examples are state the topic, ask the question, mark a gap, report the procedure, report the finding, point at a table, explain with a limit, close, summarize for an abstract, answer an objection, cite, or hedge.
- Write the sentence from your notes, in ordinary words, before you look at a phrase list.
- If the ordinary sentence is unclear, borrow the shape of one line below and replace every detail. The topic, the comparison, the source, and the number have to come from your draft.
- Read the result against a paper on a different topic. If the sentence still fits that other paper, the details never arrived.
- Keep a citation on any sentence that uses a source. Keep a hedge on an explanation you have not shown. Do not put the hedge on a finding you did show.
The order of jobs inside a section is a typical order, not a rule. The Manchester introducing-work page says its own category order is typical, far from fixed, and that not every element appears in every introduction. The same caution fits the other sections. A short assignment can state the purpose and stop. A paper with no empirical finding does not need a results sentence.
4. Academic phrases for each part of a paper
Each group below is a set of frames. Replace the slot with the material from your draft. One short passage after each group shows a frame in a model. The model is not a finding.
Introduction
A short introduction can name the topic, say why the paper takes it up, and preview the order. A very short assignment can state the purpose and stop. For a full opening paragraph, see introduction paragraph examples.
State the topic
- This paper examines X.
- This paper compares X and Y.
- This paper asks whether X.
Say why the topic is in this paper
- X is the change this study makes.
- This paper takes up X because the method measures it.
Preview the order
- The sections that follow cover X, then Y, then Z.
This paper compares a short dip and a longer dip on the same kind of paper. Dip time is the change the method makes. The sections that follow cover the question, the procedure, the finding, and the limit on that finding.
Why this example works. The topic is the comparison. The reason is the design. The preview names the order and does not praise the result. The dip is a model. A real introduction uses the assignment's topic.
Research questions
A research question should show the purpose, stay narrow enough for the assignment, and stay open enough that the answer needs more than a yes or no when the assignment asks for analysis. For a longer set of versions, see research question examples.
Ask what the paper can answer
- How does X change Y?
- Which X do the groups name, and do the groups differ on Y?
- What do participants report doing during X?
A question that is too thin, and a revision
- Thin version. Did X matter?
- Revision. How does X change Y, and does Z change with it?
How does dip time change the distance the dye travels on this paper, and does the color change with it?
Why this example works. A yes or no question does not say what would count as mattering. This question names the paper and the two outcomes. It is still a model. A lab report that the assignment wants as a yes or no comparison should follow that assignment.
Literature review
The review shows what is already known, places the sources in relation, and says what they leave open. For the move from one source to the next, see transition words for a literature review.
Place sources in relation
- Previous studies have reported that X.
- Accounts of X differ on Y.
- A study of X and a study of Y do not support the next sentence in the same way.
Mark what those sources leave open
- What these studies leave open is X.
- The studies report X. They do not report Y.
- Few of these studies ask X.
Previous studies have reported that a tutoring account often starts once the draft is open. What those accounts leave open is the minute before the draft is open. The paper asks what tutors report doing then.
Why this example works. The first sentence places the sources. The second names the missing piece. The third hands that gap to the question. The tutors are a model. A real review names the studies and cites them.
Methods
The methods section says how the research was carried out, in enough detail that another experienced person could repeat it. Most of these sentences use the simple past. For a fuller methods passage, see methods section examples.
Say what was done
- Participants completed X.
- X was measured after Y.
- The records were read for X.
Name the comparison
- X and Y were the two outcomes.
- The same X was used so that Y would be the change under comparison.
Strips cut from the same paper were dipped for a short time or for a longer time. Distance and color were the two outcomes. The same paper was used so that dip time would be the change under comparison.
Why this example works. The verbs are in the simple past. A second person could not yet repeat the model, because the model never states the times, the dye, or the paper stock. A real methods section adds those details from the notes.
Results
Report the pattern, point to the display, and keep an outcome that showed no difference. The statistics in the frames below are made-up examples. They are not results from a study. For a full results passage, see results section examples.
Report the pattern
- X was higher than Y.
- More of group A reported X than group B.
- The repeated action in these records was X.
Point to a display
- Table 1 lists X for each group.
- Figure 1 shows the same comparison.
Report a result that was not significant
- X and Y did not differ significantly, t(df) = x.xx, p = .xx.
- The groups did not differ significantly, t(df) = x.xx, p = .xx.
Replace the made-up values with the output from the analysis. Do not leave the letters in the paper.
On these strips, the longer dip left the dye higher than the short dip did. Figure 1 shows the two strips. The two dip times did not differ significantly in the color of the dye, t(df) = x.xx, p = .xx. Those test values are a made-up example.
Why this example works. The pattern is stated, the figure is named, and the outcome that was not significant stays in the section. The test values show the shape only. The sentence does not explain why the dye traveled farther.
Discussion
The discussion returns to the finding, offers one explanation, and keeps that explanation inside what the study did not test.
Attach an explanation and its limit
- X is one account of Y. The study does not test Z.
- The result describes X. It does not, by itself, show that Y should change.
Return to the question
- The question asked about X and Y. The study shows a change in X and no clear change in Y.
Hand the limit forward
- Whether X would hold for a different Y is left open.
- A rule for every case of X is not what these records can support.
More contact time is one account of the higher dye line. The model does not test moisture, paper type, or dye brand. The question asked about distance and color, and the color did not visibly change. Whether the distance pattern would hold on a different paper is left open.
Why this example works. The finding is restated, one explanation is offered, and the explanation names what was not tested. The color outcome stays in the passage. The last sentence is a limit, not a new result.
Conclusion
A conclusion looks back, stops the claim where the evidence stops, and can name a next step that follows from that limit. For endings that are not research papers, see conclusion examples.
Look back
- The study found X, and Y did not change.
Stop the claim
- The claim about X stops at Y.
Name a next step
- A later study could test X while holding Y constant.
On the strips in this model, a longer dip moved the dye farther, and the color stayed the same. The dye claim stops at these strips and these two dip times. A later trial could time the dip in smaller steps on the same paper.
Why this example works. The first sentence looks back at both outcomes, including the one that did not change. The second sentence is a boundary. The third sentence is a next step that the boundary suggests. The passage does not call the model important.
Abstract
An abstract is a short account that can stand alone. It reports the study. It does not evaluate the study. For the difference between a descriptive abstract and an informative one, see abstract examples.
Cover the study
- The study compares X. It finds Y, and Z does not change.
Leave the evaluation out
- The abstract reports X. It does not call X a success.
The model compares two dip times on one kind of paper. The longer dip left the dye higher on the strip, and the two times did not leave a visible color difference. The account of what the distance might mean is in the paper, not in this abstract.
Why this example works. A reader can see the comparison and the two outcomes without the rest of the paper. The last sentence points at the discussion instead of judging it. A real abstract uses the paper's outcomes and the length the journal set.
Rebuttal and counterargument
State the objection fairly, then answer it without erasing it. For full rebuttal paragraphs, see rebuttal examples.
State the objection
- A reader could say that X, rather than Y, explains the result.
Answer without erasing the objection
- The objection asks whether X was the only difference. The methods name Y. They do not name Z, so the objection stays open on Z.
A reader could say the longer dip only looks farther because that strip stayed wetter. The methods name the paper and the dip time. They do not name a drying step, so the objection stays open on drying. The distance difference is still what the strips showed.
Why this example works. The objection comes before the answer, and the answer does not call the objection silly. The methods gap is real inside the model, because the methods lines never mentioned drying. The last sentence keeps the finding and moves the objection onto the cause.
Citing and hedging
Citing tells the reader which work supports the sentence. Hedging tells the reader how sure the explanation is. A sentence sometimes needs both.
The pointer may be the author's name, a bracket, or a number, and the choice follows the discipline. For notes and author-date, see Chicago citation examples. For a numbered medical style, see Vancouver citation examples. Copy the entry from the work you used and from the guide you were assigned.
Keep the citation on the sentence that uses the source
- The sentence that uses a source carries the citation in the form the style sheet names.
- When the author is the subject, the year or the number the style sheet requires is still part of the citation.
- A summary of a source still cites that source.
Hedging is for an explanation, a recommendation, or a claim about cases the study did not cover. The cautious-language page says discussion sections tend to use these devices where the writer explains a finding. Do not use a hedge to hide a result that showed no difference. Report that result in the results frames.
Hedge the explanation, not the finding
- X may not appear when Y changes.
- The result describes these cases. It does not, by itself, show that Z should change.
- The account that blames X stays provisional until a comparison holds X apart from Y.
On these strips, the longer dip left the dye higher than the short dip did. The higher line may not appear on a different paper. More contact time is one account of the difference, and that account stays provisional until moisture is held apart from dip time.
Why this example works. The first sentence is the finding, and it is not softened. The second sentence limits how far the finding reaches. The third sentence hedges the cause. A citation would be added to any sentence that relied on someone else's study. The model does not cite anyone, because the model has no sources.
When the line already says what you mean and only the wording is stiff, use the public paraphrasing tool to rephrase that sentence in your own words. Leave the finding, the source, the number, and the limit as they are. The tool changes wording. It cannot supply a result, a citation, or an objection that your draft does not already contain.
A weak paragraph and a revision
Weak version
This paper discusses an important topic and reviews many studies. The methods were appropriate and the results were significant. Various findings show that more research is needed. In conclusion, the issue remains complex.
Revision
This paper compares a short dip and a longer dip on the same kind of paper. Strips cut from that paper were dipped for one of the two times. On these strips, the longer dip left the dye higher, and the color did not visibly change. More contact time is one account of the higher line, and moisture was not tested. The dye claim stops at these strips.
Why the revision works. The weak version praises the paper and hides the comparison. The revision states the topic, the procedure, both outcomes, a limited explanation, and a boundary.
A last check before you submit
- The line matches a job. Topic, question, gap, procedure, finding, explanation, limit, abstract, objection, citation, or hedge.
- The details are yours. A sentence that still fits a paper on another topic is not finished.
- A reused fragment stays generic. The about page's nine-word note is about that fragment. It is not permission to paste a paragraph.
- The finding is plain and the cause is hedged. A result that showed no difference is still reported.
- The citation is on the sentence that uses the source. The style sheet chooses notes, author-date, or a number.
- A rephrased sentence still matches the notes. The finding, the source, and the limit survive the wording change.


