Paraphrasing, quoting and summarising

When should you quote, paraphrase or summarise?

Many scholars ask how to paraphrase, but in many places the thesis doesn’t need a paraphrase at all. In a literature review, mostly you should be summarising or synthesising instead.

Keep one thing in mind. If you stay close to one source for long, its wording slips into your text without you meaning it. Summaries and syntheses force you to write in your own structure, because no one source has the structure you need.

Quoting well

Keep quotations short and give each one some job to do. After a quotation, discuss about it in a line or two. What is the quote assuming, and why did you pick it? Write that, and say whether your study is using the idea or testing it. A quotation with no comment looks like filler to examiners.

For long quotations each style has a different rule. APA 7 wants any quotation of 40 words or more as an indented block, and there you don’t put quotation marks. MLA sets off prose quotations that are longer than four lines. Chicago uses a block quotation for about 100 words or more. Your university may have its own rule also, so check the thesis guidelines once. For a comparison of the styles, see the citation styles guide.

How do you paraphrase properly?

Word swapping is quicker. Only this slower method actually works, and with practice it gets faster.

1. 1

   Read until you understand

   Read the passage, and some part around it also, till you can explain it to a colleague without looking.

2. 2

   Close the source

   Put the paper away or switch the window. Paraphrasing is not possible while the original is in front of your eyes.

3. 3

   Write from memory

   Write the idea the way you would explain it, in your own sentence structure and for your own purpose.

4. 4

   Compare and check

   Open the source again. Check that the meaning is correct and that no distinctive phrase of the author has come into your text.

5. 5

   Cite

   Add the citation, and a page number if your style asks for it. It is easy to forget this for paraphrased text, but it is needed there also.

Scholars mostly skip the second step. If the source is open, your sentence will follow its sentence without you noticing. When it is closed, you are forced to build the idea again, and then it comes out in your own structure.

Paraphrase for your purpose, not the author’s

The author wrote that passage for their own point, and you are using it for yours. So usually you need only one part of it. You will also frame it in your own way, maybe as evidence for your argument, as a contrast, or as a method you have borrowed.

What you can keep

There is no need to change everything. Technical terms such as “structural equation modelling” or “intention to use” you can leave untouched. Same for names of scales and models, standard names of statistical tests, and proper nouns. Putting a synonym in place of “Cronbach’s alpha” helps nobody. The checker will match these short terms anyway. That is fine, because examiners know the difference.

A worked example: patchwriting versus paraphrase

For this example we wrote a short passage ourselves, as if it was taken from a published study.

Among the 412 teachers surveyed, those who had received at least 30 hours of training in digital tools reported significantly higher confidence in online teaching. However, the difference in actual use of such tools in class was small, suggesting that confidence alone does not translate into changed practice.

Patchwriting (not acceptable): Of the 412 teachers who were surveyed, the ones who had got a minimum of 30 hours of digital tool training showed considerably greater confidence in teaching online. But the gap in real use of these tools in the classroom was minor, which suggests confidence by itself does not convert into changed practice (Author, Year).

There is a citation, but still this version is a problem. The sentence structure is the original’s only, clause for clause, with synonyms put into the slots. Rebecca Moore Howard, who does research on writing, gave this the name “patchwriting”. With enough words changed a checker may miss it. An examiner who has read that source will still recognise it. Some of the synonyms also shift the meaning. “Significantly” is a statistical claim, and “considerably” is not.

Paraphrase (acceptable): Training appears to change how teachers feel more than what they do. In a survey of 412 teachers, those with 30 or more hours of digital training felt more confident teaching online, yet used digital tools in class only slightly more than colleagues with less training (Author, Year).

Paraphrase used in synthesis (better still): Several studies suggest a gap between teachers’ confidence and their classroom practice. In one large survey, training raised confidence but tool use barely changed (Author, Year), and two smaller interview studies report the same pattern (Author, Year; Author, Year). Our study tests whether the same gap appears in rural government schools.

Examiners want to see this last version in a review chapter, because it uses the source for an argument which is your own. Synthesis is covered in more depth in our literature review guide.

Why do paraphrasing tools cause problems?

Many scholars are trying QuillBot to bring the similarity score down. Some use generic “article rewriters”, or they ask an AI chatbot to “rephrase this”. These tools create four kinds of problems.

• They produce patchwriting. Most of these tools keeps the source’s structure and just changes the words. That is the same pattern we showed in the example above.
• They change meaning. “Significant” becomes “substantial”, “correlated” becomes “linked to”, “may” becomes “does”. In research writing, each of these is a different claim.
• They read badly. No researcher in your field writes such sentences, and the examiner stops there. One example is “artificial intelligence” coming out as “counterfeit consciousness”. Cabanac, Labbé and Magazinov (2021) found this pattern in published papers. They gave it the name “tortured phrases”.
• They may raise an AI-writing flag. Turnitin and DrillBit now report AI-writing indicators along with similarity. When text is heavily rewritten, one number can go down and the other can go up. What those indicators can show, and what they cannot, is explained in our guide to AI tools in a PhD.

Actually a low similarity score was never the goal. The UGC 2018 regulations on academic integrity are about passing off other people’s work as your own. A rewritten passage still carries someone else’s idea and structure. So if the idea is theirs, cite it. If the wording has to be theirs, put it in quotes.

Common paraphrasing problems in Indian theses

Definitions

Many times a review chapter opens with five or six definitions of the main concept, each one reworded a little. Better to quote only the one or two definitions your study really depends on, with page numbers. The remaining ones you can sum up in one sentence, for example “Most definitions share two elements, X and Y (A; B; C).” Then say which definition your study is using and why.

Methods descriptions

Standard procedures like a laboratory assay, a sampling formula or a data-cleaning step are hard to paraphrase, and they don’t need it also. Say in short what you did, and cite the protocol or source for the details. Some overlap in the methods section is normal. Research offices usually judge it differently from overlap in the review or in the discussion.

Your own earlier papers

When a thesis chapter is drawn from a paper you have already published, the text will match. Cite the paper, and mention in the chapter that it is based on that paper. See in your university’s ordinance whether published papers can go inside the thesis. Rewording your own paper to hide the overlap does not solve anything.

Writing in a second language

Many scholars borrow phrases because they are not confident about their own English. We understand that. But using a tool will not fix it. First write simply in your own words, even if the sentences look plain, and get language edited afterwards. Plain sentences in your own structure are much safer than fine sentences borrowed from a source. Mostly they are clearer also.

A final check

• Every quotation is inside quotation marks (or set as a block quote) and is cited with a page number.
• No paraphrase is following the source’s sentence order clause by clause.
• No paragraph is built from a single source, unless you have said so and there is a reason for it.
• Technical terms and names of scales are kept as they are, and not replaced with synonyms.
• Every paraphrased or summarised idea has a citation, and this includes ideas taken from your own published papers.
• No text has gone through a paraphrasing tool or a “humaniser” tool.