Chapter 11: Citing Sources and Avoiding Plagiarism: Academic Integrity in Technical Writing

"If I have seen further it is by standing on the shoulders of Giants." — Isaac Newton, letter to Robert Hooke, 1675

Chapter Overview

A graduate student we'll call Priya — a master's candidate in environmental engineering, working her first literature review — sat across from her advisor with a printout between them. One paragraph was highlighted in yellow. She had written it the night before, exhausted, with a journal article open in another window. She had changed the wording. She had not copied and pasted. She believed, completely and sincerely, that she had done nothing wrong. Her advisor slid the source's paragraph next to hers, and the two read like the same sentence wearing slightly different clothes: same order of ideas, same structure, the author's distinctive phrases swapped one-for-one with synonyms, no quotation marks, no citation on that sentence. "This," her advisor said, not unkindly, "is plagiarism. And the worst part is that I know you didn't mean it to be."

That sentence — I know you didn't mean it to be — is the reason this chapter exists. The dramatic cases of academic dishonesty, the bought essays and the wholesale copying, are real but rare, and they are not your risk. Your risk is the honest mistake: the paraphrase that stayed too close, the fact you assumed was "common knowledge," the figure you redrew from a paper without crediting it, the AI draft you pasted in and forgot you didn't write. These are not failures of character. They are failures of technique — and technique is learnable. By the end of this chapter you will be able to cite a source correctly in any of the four major styles, tell honest paraphrase from patchwriting on sight, and defend a clear line between using a tool and stealing a thought.

Citation connects back to everything Part I and Part II have built. In Chapter 1 you learned that writing is thinking; citation is where that idea gets teeth, because the genuine paraphrase that protects you from plagiarism is only possible if you first understood the source — you cannot restate in your own structure an argument you never absorbed. In Chapter 2 you learned that audience is everything; citation style is an audience signal, which is why an IEEE reference and an APA reference, carrying the identical information, look nothing alike. And running underneath the whole chapter is a theme the book has not yet named directly: integrity. Clear writing is powerful, and power used dishonestly does real harm. The same skill that lets you explain a hard idea also lets you pass off a borrowed one as yours. Citation is the discipline that keeps the first from becoming the second.

In this chapter, you will learn to:

• Explain why citation exists — credibility, traceability, intellectual honesty — and stop treating it as a formatting tax.
• Pick the right style for your field (IEEE, APA, Chicago, ACS) and recognize the mechanics of each.
• Turn patchwriting into honest paraphrase with a citation — the core skill, and the chapter's spine.
• Name what counts as plagiarism, including the gray areas (common knowledge, self-plagiarism, mosaic plagiarism) that catch careful people.
• Set up a reference manager so that correct citation becomes the path of least resistance, and apply a defensible integrity rule to AI-generated text.

📕📗 This chapter is built for the Engineering/Science and Software/CS tracks, where you'll cite constantly — in lab reports, papers, lit reviews, README references, and design docs. Business/Professional readers (📘) cite less formally but are not exempt: a white paper that misrepresents a borrowed statistic or a proposal that lifts a competitor's language is the same integrity failure in a different suit. Everyone should read §11.3 (paraphrase vs. patchwriting) and §11.6 (AI and integrity) — those two sections are field-independent.

11.1 Why Citation Exists (It's Not About Avoiding Punishment)

Most people learn citation backwards. They learn it as a set of rules enforced by the threat of an honor-code violation: cite, or you'll get caught, or you'll lose points. That framing produces exactly the wrong behavior — citation done grudgingly, minimally, and only where you're afraid of being caught. To cite well, you have to understand what citation is for, and it turns out to be doing three jobs at once, none of which is "avoiding punishment."

First, citation establishes credibility. When you write "transformer models reduced translation error rates substantially after 2017," a skeptical reader has a fair question: says who? An uncited claim asks the reader to trust you personally. A cited claim says, don't trust me — check Vaswani and colleagues' 2017 paper and see for yourself. Paradoxically, citing other people's work makes your work more trustworthy, not less, because it shows you're not asking for blind faith. The density and quality of a document's citations is one of the fastest ways an expert reader judges whether the author has done their homework.

Second, citation enables traceability. This is the job most writers underrate, and it's the one that matters most in technical fields. A citation is a pointer. It lets a reader — including you, six months from now — follow a claim back to its source and check it, extend it, or challenge it. Science is a chain of verifiable links. Break the chain by failing to cite, and you have orphaned a fact: nobody downstream can confirm it, build on it, or correct it. When a researcher discovers that a widely-cited result doesn't replicate, the only reason they can sound the alarm is that the citation trail let them find the original and test it. Traceability is what makes a literature a literature and not just a pile of assertions.

Third, citation is intellectual honesty. Ideas have owners. Someone spent months, sometimes years, producing the result you're leaning on. Citing them is not a courtesy; it is an accurate accounting of where a thought came from. The default human tendency, especially under deadline pressure, is to let the boundary between "what I read" and "what I think" blur until borrowed ideas feel like your own. Citation forces that boundary to stay sharp. It's the writing equivalent of keeping clean books.

Here's a concrete pair to make the difference visible. Same claim, two versions:

❌ Before (uncited): "Studies have shown that code reviews catch most defects before release, and that smaller pull requests get reviewed more thoroughly." ✅ After (cited): "Empirical studies of code review report that reviews catch a large fraction of defects before release and that reviewer thoroughness drops sharply as change size grows [4], [5]." Why it's better: "Studies have shown" is the single most common fake-citation phrase in technical writing — it claims authority while pointing at nothing. The fixed version names that real evidence exists and tells the reader exactly where to find it. The first version asks for trust; the second offers proof. (Note: the bracketed numbers stand in for real references in the document's list — never write [4] for evidence you don't actually have.)

🚪 Threshold Concept: Citation is connective tissue, not a penalty. Before you cross this threshold, citation feels like a tax you pay for the crime of using sources — an annoying formatting chore bolted onto the end of writing, something you do to avoid getting in trouble. After you cross it, you see citation as structural: it is the set of links that makes every claim in your document checkable, that connects your contribution to the conversation it joins, and that keeps your own thinking honest about what you knew versus what you learned. The shift changes your behavior at the draft stage. You stop "adding citations" at the end and start writing with your sources in view, because you now understand that an uncited claim isn't just unattributed — it's unverifiable, which in technical writing is nearly worthless. Once you see citation as connective tissue, you cite more, earlier, and more naturally, and the formatting becomes the trivial part it always should have been.

🔄 Check Your Understanding. A classmate says, "I only cite when I'm worried someone will think I copied. If I'm confident a claim is true, I don't need a source." Name two of the three jobs of citation this reasoning ignores, and why each matters even for a claim that's obviously true.

Answer

It ignores traceability and credibility (and arguably honesty too). Traceability: even an obviously-true claim needs a pointer so a reader can find the data behind it, check the exact numbers, or build on the original — "obvious to you" is not "verifiable by them." Credibility: an uncited claim asks for personal trust; a cited one offers proof, and expert readers judge a document partly by whether its claims are backed. The classmate is treating citation purely as self-defense against an accusation, which is the one job that isn't its main purpose. Citation serves the reader, not just the writer's reputation.

[📍 Good stopping point — you now have the why. The rest of the chapter is the how.]

11.2 The Four Major Styles, and How to Choose

Newcomers often ask which citation style is "correct," as if one of them is right and the others are wrong. None is correct in the abstract. A citation style is a convention — a shared agreement within a field about how to format references so that everyone in that field can read them at a glance. Asking which style is best is like asking whether kilometers or miles are correct: the answer is "whichever the people you're writing for use." Your job is not to have a favorite. It's to know which convention your audience expects and execute it cleanly.

Four styles dominate technical and scientific writing. Here is what distinguishes them and when to use each.

IEEE — engineering, computer science, electronics. The Institute of Electrical and Electronics Engineers style uses bracketed numbers in the text, assigned in order of first appearance: the first source you cite is [1], the next new one [2], and so on. The reference list at the end is ordered by those numbers, not alphabetically. In-text it looks like: "The protocol reduces latency by avoiding handshakes [3]." It's compact, which engineers like, and it keeps the prose uncluttered because a citation is just a small number rather than a parenthetical with names and years. If you're writing for an IEEE conference or journal, or in most CS venues, this is your style.

APA — psychology, social sciences, education, increasingly the life sciences and nursing. The American Psychological Association style is author–date: you cite by the author's last name and the year, in parentheses, in the text. It looks like: "Reviewer thoroughness declines as change size increases (Rigby & Bird, 2013)." The reference list is alphabetical by author. APA carries the year in the text on purpose, because in fields where findings age and get superseded, knowing at a glance whether a cited study is from 1998 or 2023 is itself important information. If you're writing in a social or behavioral science — or a journal that has adopted APA, which many in biology and medicine have — this is your style.

Chicago — history, the humanities, some business and the arts. The Chicago Manual of Style actually offers two systems. The notes-and-bibliography system uses superscript numbers in the text tied to footnotes (or endnotes), with a bibliography at the end; this is the humanities flavor, prized because a footnote can carry commentary, not just a reference. The author–date system looks much like APA and is used in some sciences. When someone says "Chicago style" in a technical context they usually mean the footnote system, which you'll meet in policy writing, historical technical work, and books. Its signature is the footnote: "...as the original specification required.¹" with the source at the bottom of the page.

ACS — chemistry and chemistry-adjacent fields. The American Chemical Society style is the convention for chemistry journals. It permits a few in-text formats (numbered like IEEE, or author–year), but its hallmark is meticulous handling of chemical literature: precise formatting of journal abbreviations, and a strong tradition of citing by superscript number. If you publish in a chemistry journal, you use ACS — and you check that specific journal's instructions, because ACS journals vary in detail.

Here is the same single source — a real, foundational paper — formatted in all four styles, so you can see the differences rather than just read about them:

The source: Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit,
Llion Jones, Aidan N. Gomez, Łukasz Kaiser, and Illia Polosukhin,
"Attention Is All You Need," presented at NeurIPS 2017.

IEEE:
[1] A. Vaswani et al., "Attention is all you need," in Proc. 31st Conf.
    Neural Inf. Process. Syst. (NeurIPS), Long Beach, CA, USA, 2017,
    pp. 5998–6008.

APA:
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez,
    A. N., Kaiser, Ł., & Polosukhin, I. (2017). Attention is all you
    need. Advances in Neural Information Processing Systems, 30.

Chicago (notes-and-bibliography, footnote form):
1. Ashish Vaswani et al., "Attention Is All You Need" (paper presented
   at the 31st Conference on Neural Information Processing Systems,
   Long Beach, CA, 2017).

ACS (numbered, superscript in text):
(1) Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.;
    Gomez, A. N.; Kaiser, L.; Polosukhin, I. Attention Is All You Need.
    Adv. Neural Inf. Process. Syst. 2017, 30, 5998–6008.

Look at what changes and what doesn't. The information is constant — same authors, same title, same venue, same year. What changes is the packaging: where the year sits, whether authors are spelled out or abbreviated to "et al.," whether you get brackets, parentheses, or superscripts, whether names are separated by commas or semicolons, what gets italicized. This is the same lesson as Chapter 2, applied to references: identical content, different audience, completely different surface. You don't memorize all four. You learn to recognize which one your venue wants, then you look up the exact mechanics — or, better, you let a reference manager (§11.5) generate them and you check the output.

💡 Tip — the one rule that beats memorizing any style: be consistent, and follow the target's instructions. Every journal, conference, and class publishes a style requirement; every style has an authoritative manual. Your job is not to have the rules memorized but to find the right rulebook and apply it uniformly. A reference list that's flawless APA except for three entries in IEEE format looks worse than one that's slightly imperfect but consistent. Pick the style your audience requires, then make every entry match its siblings.

🧩 Productive Struggle. Before reading the next paragraph, predict: why would engineering converge on numbered citations [3] while psychology converged on author–date (Rigby & Bird, 2013)? What about each field's writing makes one convention fit better than the other? Take thirty seconds and actually form a hypothesis before you read on.

Here's one strong answer. Engineering and CS papers often cite many sources in dense technical passages, and they care most about whether a claim is supported, not who supported it or when — so a tiny bracketed number that doesn't interrupt the sentence is ideal, and the prose stays clean even with six citations in a paragraph. The social sciences cite to situate a finding in an ongoing, contested conversation where who found something and when are part of the argument — a 2023 replication and a 1991 original mean different things — so carrying the author and year in the text surfaces information the reader needs to weigh the claim. Neither is "better." Each convention encodes what its field treats as important. That's the deeper point: a citation style is a compressed statement of a discipline's values.

11.3 Paraphrase vs. Patchwriting: The Skill That Keeps You Honest

This is the heart of the chapter, and the place where honest people get into trouble. Almost nobody reading this will deliberately copy a paragraph and claim it as their own. But a huge number of capable writers commit patchwriting without realizing it — and patchwriting is plagiarism, even when there's no intent to deceive.

Start with definitions, pre-loaded because they're the whole game.

A direct quotation reproduces a source's exact words, inside quotation marks, with a citation. You're saying: these are someone else's words, verbatim, and here's where to find them.

A paraphrase restates a source's idea in your own words and your own sentence structure, with a citation. You're saying: this idea came from someone else, but the expression is mine. An honest paraphrase requires that you have understood the idea well enough to rebuild it from scratch — not just swap words, but reconstruct the thought in your own architecture.

Patchwriting is the dangerous middle. It is what you get when you keep the source's sentence structure and most of its phrasing, and change only some words — substituting synonyms, reordering a clause, deleting a phrase — without quotation marks. The term comes from composition researcher Rebecca Moore Howard, who named the phenomenon to describe what writers do when they're working with a source they don't yet fully understand: they cling to its language because they can't yet restate the idea on their own. (This is a real, widely-attributed concept in writing-studies research; the point about why writers do it — engaging with hard, unfamiliar material — is itself the useful insight.) Patchwriting is plagiarism because the structure and language are still the source's, even though no single sentence was copied whole.

Now the spine of the chapter — a worked transformation. Here is a real, citable source sentence:

SOURCE (Strunk & White, The Elements of Style, Rule 17):
"Vigorous writing is concise. A sentence should contain no unnecessary
words, a paragraph no unnecessary sentences, for the same reason that a
drawing should have no unnecessary lines and a machine no unnecessary
parts."

And here is what a tired student produces at 1 a.m., trying to "put it in their own words":

❌ Patchwriting (this is plagiarism): "Strong writing is brief. A sentence ought to contain no needless words, and a paragraph no needless sentences, for the same reason a drawing should have no needless lines and a machine no needless parts."

Read the source and the "paraphrase" side by side and the problem is undeniable. The sentence structure is identical. The famous parallel — sentence/words, paragraph/sentences, drawing/lines, machine/parts — is reproduced beat for beat. The only changes are one-for-one synonym swaps: vigorous→strong, concise→brief, unnecessary→needless, should→ought to. The writer has done word-substitution, not paraphrase. There's no citation, and even with a citation this would be plagiarism, because the distinctive expression — the structure and the four-part analogy — belongs to Strunk and White. Crucially, the writer may have believed in complete good faith that changing the words made it theirs. It didn't.

Here is the same idea, honestly handled, three different ways:

✅ Honest paraphrase (your structure, their idea, cited): "Good prose earns its length: every word, sentence, and paragraph should do real work, the way a well-engineered machine has no spare parts (Strunk & White, 2000)." Why it works: The idea — that economy in writing parallels economy in design — is preserved and credited. But the architecture is the writer's own. The four-part list became a single compressed clause; the machine analogy is kept (it's the load-bearing concept) but rebuilt into new phrasing. You could not reconstruct Strunk and White's exact sentence from this. That's the test of a real paraphrase: it doesn't let the reader recover the original's wording.

✅ Direct quotation (their words, marked and cited): "Strunk and White put the principle memorably: 'Omit needless words... a sentence should contain no unnecessary words, a paragraph no unnecessary sentences' (Strunk & White, 2000, p. 23)." Why it works: Because the original phrasing is the point — its rhythm is part of why it's famous — quoting is legitimate here. The words are in quotation marks, attributed, and page-located. You're not pretending they're yours; you're showcasing theirs.

✅ Paraphrase + a short quote for the memorable bit (the most common real-world move): "The core of concise writing is that nothing should be present that isn't pulling weight — 'no unnecessary words,' 'no unnecessary sentences,' and the same for every line of a drawing or part of a machine (Strunk & White, 2000, p. 23)." Why it works: You restate the idea in your own frame but quote the few distinctive phrases that resist paraphrase, marked and cited. This is what experienced writers do most often — paraphrase the substance, quote only the irreplaceable wording.

What separates the honest versions from the patchwritten one isn't the citation alone — notice the patchwriting would still be plagiarism with a citation, because the expression is stolen even if the source is named. What separates them is that the honest writer understood the idea and rebuilt it, while the patchwriter clung to the source's words because they hadn't fully absorbed the thought. This is the deep connection to Chapter 1: patchwriting is a thinking failure that shows up as a writing failure. You patchwrite when you haven't yet understood your source well enough to say it yourself.

There's a reliable technique that prevents patchwriting almost completely, and it's worth adopting as a habit:

✏️ Try This — the close-the-source method. When you want to paraphrase a passage: (1) Read it until you genuinely understand the idea. (2) Close the source — minimize the window, flip the book face down. (3) Write the idea from memory, in your own words, explaining it as if to a colleague. (4) Then reopen the source to check you got the facts right and didn't accidentally reproduce distinctive phrasing. (5) Add the citation. The act of closing the source forces you to route the idea through your own understanding instead of copying off the page. If you can't write it from memory, that's not a writing problem — it's a signal you don't understand it yet, and you should reread before you write. This single habit is the most effective anti-plagiarism technique there is.

🔍 Why Does This Work? Why does closing the source prevent patchwriting so reliably, when "try to use your own words" with the source open in front of you so often fails? Think about it before reading on.

Answer

Because patchwriting is driven by proximity to the source's language. When the original is visible, its phrasing is the most available material in your working memory — your brain reaches for the words right in front of it and "paraphrases" by nudging them, which is exactly how synonym-swapping happens. You're editing the source rather than expressing the idea. Closing the source removes that crutch: now the only thing in your working memory is your understanding of the idea, so the words you produce are necessarily your own packaging of your own comprehension. It also doubles as a comprehension check — if nothing comes out when the source is closed, you've learned you didn't actually understand it, which is more valuable to know before you've built an argument on it. The technique works because it changes what your brain has access to at the moment of composing.

The default in technical writing is paraphrase, not quotation — and it's worth saying why, because it surprises people coming from the humanities, where quotation is common. In technical and scientific writing you're citing findings, methods, and data, not prose. Nobody needs the exact wording of how a result was phrased; they need the result. So you paraphrase: "Reviewer thoroughness drops as change size grows [5]," not a block quote of the authors' sentence. Reserve direct quotation for the rare cases where the exact wording matters — a precise legal or regulatory definition, a standard's mandatory language, a famously-phrased idea, or a claim so contestable you want to show you're representing it word-for-word and not paraphrasing it unfairly. If you find a technical paper stuffed with quotations, the author usually hasn't done the work of understanding and restating. Quote rarely; paraphrase well; cite always.

🔄 Check Your Understanding. A writer changes about 40% of a source paragraph's words to synonyms, keeps the sentence order and structure, and adds a citation at the end. Is this plagiarism? Explain your reasoning in one sentence, then state what they should have done.

Answer

Yes, it's plagiarism — specifically patchwriting — because the source's sentence structure and distinctive expression are reproduced even though individual words were swapped, and a citation credits the idea but does not license borrowing the form. What they should have done: close the source, write the idea from memory in their own architecture, then cite — or, if the original wording mattered, quote the borrowed phrases explicitly with quotation marks. Citation answers "whose idea?"; it does not answer "whose words?" — that's what quotation marks are for.

11.4 What Counts as Plagiarism — and the Gray Areas That Trip Up Honest Writers

Patchwriting is the most common trap, but it's not the only one. Let's map the full terrain, because the boundary of "plagiarism" is wider and stranger than most people assume, and the gray zones are exactly where careful, well-meaning writers stumble.

The clear cases. These are plagiarism by anyone's definition, and you already know them:

• Verbatim copying without quotation marks and citation — pasting someone's text and presenting it as yours.
• Patchwriting — the synonym-swap of §11.3.
• Uncredited paraphrase — restating someone's idea in genuinely your own words but without the citation. (The words are yours; the idea isn't, and you didn't say so.)
• Buying or commissioning work and submitting it as your own (contract cheating).
• Stealing a figure, table, chart, or dataset without credit — yes, visuals are plagiarizable. Redrawing someone's diagram in your own software, or replotting their data, still requires "adapted from [source]." (This connects to Chapter 9: a figure carries authored intellectual work, and a caption is the natural place to credit it.)

Now the genuinely hard parts — the gray areas, where the honest writer needs judgment, not just rules.

Common knowledge. You don't cite that water boils at 100°C at sea level, that Python is a programming language, or that the speed of light is roughly 3×10⁸ m/s. These are common knowledge — facts so widely known and uncontested within your audience that no reasonable reader would ask "says who?" But the boundary is audience-relative, which makes it slippery. "The transformer architecture uses self-attention" is common knowledge to an ML researcher and needs no citation in an ML paper; the same statement in a general-audience article might still warrant a pointer. The working test: Would your specific reader already know this without being told, and is it uncontested? If yes, it's common knowledge — don't clutter the text with a citation. If it's a specific finding, a statistic, a contested claim, or something you personally just learned from one source, cite it. When genuinely unsure, cite. The cost of an unnecessary citation is small; the cost of a missing one can be your integrity.

❌ Over-citing common knowledge: "Water is composed of hydrogen and oxygen (Smith, 2019)." — Now the writer looks like they don't know basic chemistry without a source, and the citation is noise. ❌ Under-citing a real finding: "Reviews catch about 60% of defects before release." — That specific number came from somewhere; without a citation it's an unverifiable, orphaned statistic. ✅ The judgment: Cite the specific, contestable, traceable claim; leave the universal, uncontested fact alone. Calibrate "uncontested" to your actual reader.

Self-plagiarism. This one shocks people: you can plagiarize yourself. Self-plagiarism is reusing your own previously-published or already-submitted work as if it were new, without disclosure. Submitting the same paper to two journals, recycling whole paragraphs from a published article into a new one without citing the original, or handing in one assignment for two courses — these violate integrity norms even though the words are genuinely yours. Why is it wrong if you wrote it? Because publication and submission carry an implicit claim of novelty and original effort for this context, and reusing prior work breaks that claim — it can also create copyright problems if you've transferred rights to a publisher. The fix is simple: disclose and cite your own prior work just as you would anyone else's, and check the specific venue's reuse policy. (There are legitimate forms of reuse — a thesis built from your own published papers, a conference paper extended into a journal version — but they're handled openly, with the relationship stated, not hidden.)

Mosaic plagiarism. A cousin of patchwriting: mosaic plagiarism stitches together phrases and sentences from multiple sources into a paragraph that has citations sprinkled around but whose actual language is borrowed throughout. It can pass a naive "did they cite something?" check while still being built almost entirely from other people's words. The tell is a paragraph where every sentence echoes a source's phrasing even though sources are named. The fix is the same close-the-source discipline from §11.3, applied across all your sources at once: understand them, then write the synthesis in your own architecture. (This becomes central in Chapter 15, where a literature review must synthesize sources into your own argument rather than mosaic them together.)

Citing a source you didn't read. You see a great claim in Paper A, attributed there to Paper B. You cite Paper B — but you never read Paper B. This is risky and, strictly, a small dishonesty: you're implying firsthand familiarity you don't have, and if Paper A misrepresented B (which happens constantly), you've now propagated the error under your own name. The honest move is either to read B, or to cite it as reported: "(B, as cited in A)." Track down the original whenever it carries real weight in your argument.

⚠️ Warning — translation is not laundering. Taking a source written in another language, translating it into English, and presenting the translation as your own original prose is plagiarism. So is running a passage through a paraphrasing tool or an AI "rewriter" and treating the output as yours. The intellectual content and structure still belong to the source. Changing the language — whether by human translation, a thesaurus, or a machine — does not change whose idea and whose structure it is. If the substance came from a source, cite the source.

🔄 Check Your Understanding. You wrote a strong "Related Work" paragraph for a course paper last semester. This semester, a different course's assignment overlaps, and that paragraph fits perfectly. You paste it in unchanged. Two questions: which integrity problem is this, and what's the honest way to reuse your own good work?

Answer

It's self-plagiarism — submitting work as new when it was already submitted for credit elsewhere, which violates the implicit claim that each submission represents fresh effort for that context. The honest move is to disclose it (tell the instructor you're building on prior work and ask whether that's permitted) and, if it's allowed, cite your earlier paper as a source the way you would anyone's. Many instructors permit reuse with disclosure; almost none permit it secretly. The principle generalizes: reusing your own work is fine openly, not silently.

11.5 Reference Managers: Make Correct Citation the Path of Least Resistance

You now understand citation conceptually. The practical problem remains: formatting references by hand is tedious, error-prone, and exactly the kind of work you'll cut corners on at 2 a.m. The solution is to stop doing it by hand. A reference manager is software that stores your sources, generates correctly-formatted citations and reference lists in any style, and inserts them into your document as you write. Used well, it makes correct citation easier than incorrect citation — which is the only sustainable way to stay honest under deadline pressure.

The two most common in technical fields are Zotero and Mendeley. Zotero is free, open-source, and broadly loved in academia; Mendeley (owned by Elsevier) is also free at the basic tier and integrates with that publisher's ecosystem. There are others — EndNote (paid, institutional), Citavi, Paperpile — but the workflow is the same across all of them, and it has three moves:

1. Capture. When you find a source — a journal article, a book, a standard, a web page — you save it to your library with one click (a browser connector grabs the title, authors, year, DOI, and other metadata automatically). The key habit: capture the source the moment you decide to use it, not later. Sources you mean to "add at the end" are sources you cite wrong or lose.
2. Cite while you write. A plugin in your word processor (or a \cite{} command if you write in LaTeX, via BibTeX/BibLaTeX) lets you insert a citation by searching your library. The manager drops in the in-text citation in your chosen style.
3. Generate the reference list. With one command, the manager builds the full reference list or bibliography from every source you cited, formatted in the target style — and reformats the entire document into a different style if your target journal rejects you and the next one wants APA instead of IEEE. That switch, which would take an hour by hand, takes seconds.

The payoff is enormous, but there is one rule you must not forget:

⚠️ Warning — the manager is a tool, not an oracle. Always check its output. Reference managers generate citations from metadata, and metadata is frequently wrong: a DOI lookup might capitalize a title incorrectly, mangle an author's name, list the wrong year, or drop a page range. The software will faithfully format the wrong information into a beautifully-styled wrong citation. You are responsible for every reference in your document, including the ones the machine generated. After you produce a reference list, read it against the actual sources. A garbled author name in citation 14 is your error, not the software's, the moment you submit it. The tool removes the tedium; it does not remove the responsibility.

A reference manager also quietly solves the patchwriting problem at the organizational level. Because it stores your sources and lets you attach notes to each one, you can write your own summary of each source in your own words at capture time — which is the close-the-source method (§11.3) baked into your workflow. By the time you draft, you're working from your own notes, not the source's prose, and patchwriting becomes much harder to commit by accident.

✏️ Try This. Before you write your next document that uses sources, install Zotero (it's free), add the browser connector, and capture three sources you already plan to cite. For each, write a two-sentence note in your own words summarizing what it says and why you'll use it. You've just done the hardest 80% of citation before drafting a word — and you've pre-empted patchwriting by routing each source through your own understanding at capture time.

🔄 Check Your Understanding. Why is "always check the reference manager's output" not just good practice but an integrity requirement — what could go wrong if you trust it blindly?

Answer

Because you, not the software, are accountable for every claim and pointer in your document. A reference manager pulls metadata that's often wrong, so blind trust can put a misspelled author, wrong year, or incorrect title — or even a citation pointing at the wrong paper — into your reference list under your name. A reader who follows a broken or incorrect citation finds your scholarship unreliable, and a citation that points at a source that doesn't support your claim looks like fabrication even when it was just an uncorrected lookup error. The tool reduces labor, but accountability is non-delegable: a wrong citation is your error the instant you submit it.

11.6 AI-Generated Text and Integrity (A Preview of Chapter 29)

You cannot write a chapter on plagiarism in this decade without addressing the elephant in the room: large language models can produce fluent technical prose on demand, and the integrity questions they raise are genuinely new. Chapter 29 treats AI as a writing tool in depth; here we cover only the integrity dimension, because it belongs in this chapter's territory.

Start with what's actually at stake, because the panic and the dismissal are both wrong. Using an AI tool is not automatically plagiarism, and it is not automatically fine. The honest question is the same one citation has always asked: is the work you're presenting as yours actually yours — your thinking, your judgment, your accountability?

Three distinct integrity problems arise with AI-generated text, and conflating them produces confusion:

Problem 1 — Authorship and disclosure. If you submit AI-generated prose as your own original writing where original writing was expected, you've misrepresented authorship — the same violation as submitting a friend's essay, regardless of how the text was produced. The rules here are contextual and evolving: a journal, a course, or an employer sets the policy, and policies range from "AI prohibited" to "AI permitted with disclosure" to "AI freely allowed." Your obligation is to know your context's policy and follow it, and when disclosure is required, to disclose honestly what role the tool played. The safe default when you're unsure: ask, and disclose.

Problem 2 — Fabricated citations. This is the one that bites technical writers hardest, and it's specific to how these tools work. Language models generate text that is statistically plausible, which means they will cheerfully produce citations that look perfectly real — plausible authors, a plausible title, a real-sounding journal, a well-formatted DOI — for papers that do not exist. This isn't rare; it's a known, documented failure mode. A model asked for sources to support a claim may invent them wholesale. If you paste an AI-generated reference into your document without verifying it exists, you have committed citation fabrication — one of the gravest integrity violations there is — even though you didn't mean to. Every AI-suggested citation must be independently verified against the real source before it enters your document. Find the actual paper. Read enough of it to confirm it says what you're citing it for. If you can't find it, it probably isn't real.

Problem 3 — Outsourcing the thinking. This is subtler and connects straight back to Chapter 1. Even where AI use is permitted and disclosed, there's a deeper question: if the thinking was done by the model and you merely transcribed it, what did you learn, and what can you stand behind? The book's thesis is that writing is thinking — that the struggle to express an idea clearly is how you come to understand it. Outsource that struggle wholesale and you've outsourced the understanding too. You may end up unable to defend, extend, or correct prose with your name on it. Chapter 29 develops the operating rule in full, but here's the core of it: if you can't evaluate whether the AI's output is correct, you have no business presenting it as your work — you can't vouch for what you can't check.

💡 Tip — the verification rule, stated as a test. Before any AI-assisted text or citation goes into a document you'll put your name on, ask: Can I independently verify this is correct, and would I be comfortable defending it as my own judgment? If yes — you've understood it, checked the facts, confirmed the citations exist — the tool was an assistant and the work is yours. If no — you're transcribing output you can't evaluate — then using it isn't a shortcut, it's a liability with your name on it. The tool's fluency is not evidence of accuracy. Verification is your job, every time.

This isn't a counsel of fear. AI tools are genuinely useful for brainstorming, restructuring, and rephrasing your own thinking — Chapter 29 shows you how to use them well. The integrity point is narrow and firm: disclose per your context, verify every fact and citation, and never present as your own thinking work you can't actually evaluate. Those three rules keep AI on the right side of the line.

🔄 Check Your Understanding. An AI tool gives you a paragraph with three citations supporting your argument. They're formatted perfectly in APA. What is the single most important thing you must do before any of it goes into your paper, and why is "but they're formatted correctly" irrelevant?

Answer

Verify that all three sources actually exist and actually say what they're cited for — find each real paper and confirm it. Perfect formatting is irrelevant because language models generate plausible-looking text, and a fabricated citation is plausible-looking by construction: real-sounding authors, title, journal, and DOI for a paper that was never written. Correct APA formatting tells you nothing about whether the source is real; it only tells you the model is good at imitating the shape of a citation. Pasting an unverified AI citation risks citation fabrication — a severe integrity violation — under your own name. Format is cosmetic; existence and accuracy are what matter.

11.7 A Worked Example: This Book's Own Citation-Honesty System

Here is something a little meta, and genuinely useful: this textbook runs an explicit citation-honesty system, and it's worth showing you, because it demonstrates how a careful writer handles the real-world fact that you can't always cite with the same level of certainty. (You can see it in action in the Further Reading page of every chapter.) The book sorts every factual claim into three tiers:

Tier 1 — Verified. Landmark works the author is confident exist and can name precisely: Strunk and White's The Elements of Style, Williams's Style: Lessons in Clarity and Grace, Tufte's The Visual Display of Quantitative Information, the Plain Writing Act of 2010, RFC 2119. These get cited normally, with full confidence — and this chapter has done exactly that with the Vaswani et al. paper and the Strunk & White examples above.

Tier 2 — Attributed claim. Real, widely-held ideas the author can't pin to an exact page, year, or single source. The honest move is to signal the uncertainty in the prose itself: "research on the curse of knowledge suggests…," or "the patchwriting concept comes from writing-studies research by Rebecca Moore Howard." Notice what this chapter did not do: it did not invent a DOI, a precise year, or a journal name for the patchwriting research to make it look more authoritative. A real attribution with honest vagueness beats a fake citation with false precision, every time.

Tier 3 — Illustrative. Composite or fictional examples used to teach — Priya the graduate student, the "tired student at 1 a.m.," Dana's churn memo, Raj's README. These are labeled as what they are: "a composite example," "fictional but realistic," "anonymized." The reader is never misled into thinking an illustration is a documented case.

Why show you the book's internal machinery? Because it models a maturity about citation that's worth internalizing: honesty about your level of certainty is itself a form of integrity. Real writing doesn't get to cite everything to a verified primary source. Sometimes you know an idea is true and widely held but can't locate the canonical reference. The dishonest response is to manufacture a citation that looks authoritative — an invented year, a plausible journal. The honest response is to attribute what you can and flag the uncertainty in the text: "studies generally find," "the consensus among style authorities is," "this is well-attributed but I can't locate the original." That flag is not weakness. It's the writer being honest with the reader about the strength of the evidence — which is exactly the integrity that citation exists to serve.

This connects to a habit worth carrying everywhere: never let formatting polish substitute for actual verification. A perfectly-formatted citation to a source you haven't confirmed is less honest than an awkward "I believe this is right but couldn't verify the original," because the first looks authoritative while the second tells the truth. The three-tier system is just a disciplined way of always telling the reader the truth about how solid each claim is.

🔄 Check Your Understanding. You know a claim is true and widely accepted in your field, but after an hour of searching you cannot find the original source. Using the three-tier idea, what's the honest way to handle this in your document — and what's the dishonest shortcut you must avoid?

Answer

Handle it as a Tier 2 attributed claim: state it and signal the uncertainty in the prose — "research generally finds…," "it is widely held that…," "this is well-attributed in the literature, though I couldn't locate the original source." You attribute honestly without faking precision. The dishonest shortcut to avoid is manufacturing a citation — inventing or guessing a year, journal, or DOI to make the claim look authoritatively sourced. A fake-precise citation is worse than honest vagueness because it actively misleads the reader about the evidence and, if checked, reads as fabrication. Integrity includes being honest about how certain you are, not just about whose idea it is.

📐 Project Checkpoint

Across the portfolio so far you've built a charter (Ch 1), an audience profile (Ch 2), cut a paragraph for clarity (Ch 3), structured a document (Ch 4), and — in the chapters of Part II — sharpened sentences, paragraphs, visuals, and layout. Now you add the layer that makes your work trustworthy: sourcing it honestly. Every portfolio piece that makes a factual claim, leans on data, or builds on prior work needs citation done right, and this is the chapter where you wire that in.

This chapter's increment: build a source library and add honest citations to one portfolio piece — including one deliberate patchwriting-to-paraphrase fix.

1. Install a reference manager and capture your sources. Set up Zotero (or Mendeley). For the portfolio piece that uses the most external sources — likely your technical report, data memo, or blog post — capture every source you've drawn on or plan to. Aim for at least four. Capture each the moment you decide to use it, and write a two-sentence, own-words note on each (this is the close-the-source method, baked in).
2. Pick your style. Decide which style your piece's audience expects — IEEE if it's engineering/CS, APA if it's a social/behavioral or general-science memo, Chicago if it's policy/humanities, ACS if it's chemistry. Set your manager to that style. Write the style name at the top of the piece so it's a conscious choice, not an accident.
3. Do the patchwriting drill on your own work. Find one place where you paraphrased a source. Be honest: is it a real paraphrase, or did you stay too close — synonym swaps, the source's sentence structure? If it's patchwriting (most first attempts are), fix it: close the source, rewrite the idea from memory in your own architecture, add the citation. Keep both versions, labeled ❌ Before (patchwriting) and ✅ After (honest paraphrase), with a one-line note naming what you changed. This is the chapter's spine applied to you.
4. Generate and verify the reference list. Produce the full reference list from your manager, then check every entry against the real source. Fix the metadata errors the tool inevitably introduced. Confirm every in-text citation has a matching reference and vice versa.
5. Run the AI check (if relevant). If you used any AI assistance on this piece, verify that every fact and every citation it suggested actually checks out, and add a disclosure line if your context requires one.

Keep the before/after patchwriting pair with your portfolio — it's evidence you can tell the difference, which is the whole skill. Next chapter (Editing and Revision) is where all of Part II converges: you'll carry this piece, and every craft skill from Chapters 6–11, through a disciplined multi-pass revision — because clean sources, clear sentences, and honest paraphrase all get checked in the editing passes you're about to learn.

11.8 Common Mistakes & Practical Considerations

The principles are clean; practice is messier. Here are the mistakes that actually happen and the judgment calls the rules don't fully settle.

Mistake 1: Treating citation as an end-of-project chore. The single most common cause of bad citations is leaving them all for the last night. You forget which fact came from which source, you can't find the page number, you reconstruct references from memory, and patchwriting creeps in because you're working from half-remembered prose. Cite as you write, capture as you read. A reference manager makes this nearly automatic. Citation done continuously is easy; citation done retroactively is where errors and dishonesty breed.

Mistake 2: Believing a citation cures patchwriting. It doesn't. The most persistent misconception in this whole topic is that adding a citation makes any use of a source legitimate. Citation answers "whose idea?" It does not answer "whose words?" If you've borrowed the source's expression — its structure, its distinctive phrasing — you need either genuine paraphrase or quotation marks, and the citation. A cited patchwrite is still plagiarism. Re-read §11.3 until this is reflexive, because it's the error that catches the most honest people.

Mistake 3: Mismatching style to venue, or mixing styles. Submitting IEEE-formatted references to an APA journal, or — worse — a reference list that's three styles at once, signals carelessness to exactly the readers judging your rigor. Find the venue's required style before you draft, set your manager to it, and keep every entry consistent. When you don't know, the venue's "instructions for authors" or your assignment's rubric tells you; if truly silent, pick the dominant style for the field and apply it uniformly.

Mistake 4: Over-citing or under-citing. Two opposite failures. Over-citing — a source after every sentence, citations on genuine common knowledge — clutters the prose and makes you look unsure of basic facts. Under-citing — leaning on a source for a paragraph but citing it only once at the end, so the reader can't tell where your synthesis stops and the source's argument begins — quietly slides toward mosaic plagiarism. The calibration: cite specific, contestable, traceable claims at the point you make them; don't cite the universal and uncontested; and when you draw on one source across several sentences, signal it ("Rigby and Bird found that…") so the boundary is visible.

Mistake 5: Trusting metadata, the reference manager, or an AI tool without checking. All three generate citations from information that's frequently wrong. The responsibility for accuracy is yours, non-delegably. Read your reference list against the real sources. Verify every AI-suggested citation exists. The five minutes this takes is the cheapest integrity insurance you'll ever buy.

Mistake 6: Citing the abstract, the blog post, or the secondary source instead of the original. When a claim matters, cite the primary source — the paper that produced the result, not a news article or another paper describing it. Secondary sources misrepresent constantly. If you genuinely can only access the secondary source, cite it honestly as "(Original, as cited in Secondary)," but chase down originals for anything load-bearing.

🪞 Learning Check-In. Pause and assess honestly. Before this chapter, what was your most likely plagiarism risk — patchwriting from staying too close to sources, under-citing because you weren't sure what needed a citation, or trusting an AI/manager's output without checking? Name your specific risk in one sentence, then name the one technique from this chapter you'll actually adopt to address it (close-the-source method? capture-as-you-read? the verification rule?). Integrity isn't a trait you have or lack; it's a set of habits you install. Which one are you installing this week?

Frequently Asked Questions

Is paraphrasing without quotation marks plagiarism if I cite the source?

It depends entirely on how you paraphrased. A genuine paraphrase — the idea rebuilt in your own words and your own sentence structure — with a citation is correct and is the default in technical writing; no quotation marks needed. But if you kept the source's structure and just swapped synonyms (patchwriting), it's plagiarism even with the citation, because you borrowed the expression, not just the idea. The citation credits whose idea it is; quotation marks are what credit whose words they are. The test: could a reader reconstruct the original's wording from yours? If yes, you stayed too close — either rebuild it honestly or quote the borrowed phrases explicitly.

Which citation style should I use?

Whichever your audience requires — there's no universally "correct" style. Use IEEE for engineering and computer science (numbered, [1]), APA for social/behavioral sciences and much of the life sciences (author–date), Chicago for humanities, history, and policy (often footnotes), and ACS for chemistry. Your journal, conference, or assignment specifies one; follow its "instructions for authors" or rubric. When genuinely unspecified, pick the dominant style for your field and — the rule that matters most — apply it consistently across every reference.

Do I need to cite common knowledge?

No — facts that are widely known and uncontested within your audience don't need citations (water boils at 100°C; Python is a programming language). But the boundary is audience-relative and slippery: what's common knowledge to specialists may not be to a general reader, and a specific finding, statistic, or contested claim always needs a source even if it feels familiar. The working test: would your specific reader already know this without being told, and is it uncontested? If yes, skip the citation; if it's a particular result or number, cite it. When unsure, cite — an unnecessary citation costs little; a missing one can cost your integrity.

Is using ChatGPT or another AI to write my paper plagiarism?

Not automatically, but it raises three real integrity issues. (1) Authorship: submitting AI prose as your own original writing where original work is expected misrepresents authorship — follow your context's policy, and disclose when required. (2) Fabricated citations: AI tools routinely invent realistic-looking citations for papers that don't exist, so every AI-suggested source must be independently verified before it enters your document. (3) Outsourced thinking: if you can't evaluate whether the output is correct, you can't legitimately present it as your work. The rule: disclose per your context, verify every fact and citation, and never put your name on reasoning you can't defend. Chapter 29 covers using AI well in depth.

What's the difference between a reference list and a bibliography?

A reference list contains only the sources you actually cited in the text — every entry corresponds to an in-text citation, and vice versa. This is what IEEE and APA use. A bibliography (in the Chicago sense) can be broader: it may include works you consulted and drew on for background even if you didn't cite them directly. In most technical and scientific writing you want a reference list — tight, with a one-to-one match between citations and entries — because traceability is the goal: every reference should point at a claim, and every claim that needs support should point at a reference.

Chapter Summary

Key Takeaways

• Citation does three jobs: establishes credibility (don't trust me, check the source), enables traceability (a pointer a reader can follow), and maintains intellectual honesty (an accurate accounting of whose idea it is). "Avoiding punishment" is not on the list.
• A citation style is a convention, not a correctness contest. IEEE (numbered, engineering/CS), APA (author–date, social/life sciences), Chicago (often footnotes, humanities/policy), ACS (chemistry). Same information, different packaging — choose by audience, apply consistently.
• Patchwriting is plagiarism, even with a citation, even without intent. Swapping synonyms while keeping the source's structure borrows the expression. Citation credits the idea; only paraphrase-or-quotation handles the words.
• Paraphrase is the technical-writing default; quote rarely. You cite findings and methods, not prose. The close-the-source method (understand it, close it, write from memory, then check and cite) prevents patchwriting almost completely.
• The gray areas need judgment: common knowledge (audience-relative — don't cite the universal), self-plagiarism (you can plagiarize yourself; disclose reuse), mosaic plagiarism (borrowed language across many cited sources), and citing sources you didn't read.
• A reference manager (Zotero/Mendeley) makes correct citation easier than incorrect citation — but you must verify its output, because it formats wrong metadata flawlessly.
• AI integrity rests on three rules: disclose per your context, verify every fact and citation (models fabricate realistic-looking sources for papers that don't exist), and never present as yours reasoning you can't evaluate.
• Honesty about your certainty is integrity too (this book's three-tier system): cite verified sources normally, attribute uncertain claims honestly in the prose, label illustrative examples — never fake precision to look authoritative.

Action Items

1. Install Zotero or Mendeley and capture sources as you read, with an own-words note on each.
2. Run the close-the-source method on every paraphrase: understand, close the source, write from memory, check, cite.
3. Pick your audience's required style and apply it to every reference consistently.
4. Verify every reference against its real source, and every AI-suggested citation's existence, before submitting.
5. When you can't locate a source for a true claim, attribute it honestly ("research generally finds…") — never manufacture a citation.

Common Mistakes

End-of-project citation; believing a citation cures patchwriting; mixing or mismatching styles; over- or under-citing; trusting metadata/managers/AI without checking; citing the secondary source instead of the primary one.

Decision Framework — "How should I handle this source?"

Spaced Review

A few questions reaching back, to strengthen retention.

1. (From Chapter 9) Chapter 9 argued that a caption should interpret data, not just label it ("Revenue fell 23% in Q4 (Figure 3), driven by…," not "Figure 3 shows revenue"). Now connect it to this chapter: if you adapt a figure from another paper, where is the natural place to credit the source, and what does that tell you about figures and plagiarism?
2. (From Chapter 10) Chapter 10 was about using white space and visual hierarchy so a document is easy to navigate. A reference list is a dense block of small text. Using Chapter 10's principles, name one design choice that makes a long reference list more usable — and why citation is a design problem as well as an integrity one.
3. (Bridging, Ch 10 → Ch 11) Both chapters serve the reader by making information findable. Chapter 10 made it findable on the page (hierarchy, white space); this chapter makes a claim findable in the literature (the citation as a pointer). State the shared principle in one sentence.

What's Next

You can now source your writing honestly — pick a style, paraphrase without patchwriting, cite with a tool, and stay on the right side of the line with AI. Chapter 12: Editing and Revision is where Part II comes together. Every skill from Chapters 6 through 11 — clean sentences, flowing paragraphs, interpreted visuals, navigable layout, honest citations — becomes something you check on purpose, in order, through a disciplined multi-pass revision. The patchwriting fix you just learned, the reference list you just verified: those are editing passes. Chapter 12 teaches the full hierarchy — content before structure before sentences before proofreading — and the truth Part I promised and Part II has been building toward: revision is where the writing actually happens.

Practice: Exercises · Quiz Go deeper: Case Study · Case Study 2 Review: Key Takeaways · Further Reading