10 Citation
(Modified on 2026-08-11)
2,744 words
We stand on the shoulders of giants.
— Bernard of Chartres (1159)
One of the defining features of scholarship, is that new scholarly works (i.e., research papers) are tied into an existing body of scholarly literature. This assures that new knowledge is connected to, and understood in the context of, existing knowledge. The main mechanism for creating this connection is through citation. Citing previous, published, scholarship is essential to effective new scholarship; Lack of (or insufficient) citations will see your papers rejected. Conversely, citations are (for better or worse) an important way that the scholarly community measures the value (impact) of work—successful research is cited by others, getting a higher citation count.
Why do we cite?
Why do we cite? As always, it helps to think about citation in terms of rhetoric: the main point of citation is to enhance your argument. There are at least five different ways that citations improve our papers:
- By clarifying and contextualizing our ideas.
- By establishing our credibility.
- By justifying our assumptions and/or methods.
- By avoiding reexplanation of established ideas.
Most of your audience will be familiar with similar and related work in your field; citations should improve your readers’ understanding of your work by situating it relative to other work. Effective citations clarify and contextualize your argument, helping your readers understand what you are doing and why it matters.
Proper citation is also necessary as a way of establishing confidence in you as a researcher. Effective citations can bolster your credibility by illustrating that you are aware of—and understand—existing work, and how it relates to your work. If you fail to demonstrate knowledge of the field, reviewers may take this as evidence that you don’t understand the implications of your work or how your work is related to other work in the field. This erodes your credibility.
Citations can also be used to support or justify your assumptions, opinions, theories, or methodologies: If a previous scholar proposed a theory for classifying genres, you can cite their theory and generate new hypotheses from it; If a previous study used a specific type of statistical model to analyze similar data, you could use that same model and cite them; etc. Finally, citations can provide empirical “facts,” anecdotes, or conjectures for you to use or critique. A fact/idea/method that was peer-reviewed is (presumably) accurate/reasonable. This shifts the burden of quality and accuracy to the citation.
Finally, citing other research can allow us to avoid fully reexplaining everything we do from first principles. We point to a citation so we don’t have to explain everything ourselves. However, it is a bad idea to simply “offload” the work of explaining, or arguing, to your citations. Some readers won’t be familiar with the work you cite, or at least won’t have it memorized, and won’t bother tracking it down or reading it. If your argument relies on ideas from a citation, it is best to explain those facts as much as is necessary to understand your argument.
A common (bad) case in machine-learning papers is to simply say “we used so-and-so’s algorithm (So-and-So, 2019)” and then to never explain that algorithm. Of course, if your work is building off such-and-such algorithm, it makes sense to cite them and leave the details to them—you don’t need to completely reexplain the algorithm. However, you don’t want a reader who isn’t familiar with So-and-so’s work to be completely lost. It is better to at least quickly summarize the relevant aspects of the cited algorithm, so the reader can understand what your work is contributing. Trust me, no reviewer wants to have to read another paper in order to review your paper.
On a more human level, remember that reviewers are generally people who have invested their lives to a field of research, and they may be offended by lack of engagement with their field. Reviewers are typically selected specifically because they have worked on research related to yours, and will expect you to acknowledge the work of others in the field. There can also be ego involved: reviewers might be annoyed if you fail to cite them or work that they find particularly important.
Topical citations
In some cases, you’ll find citations that don’t fit neatly into the argument you are making or support a particular claim, but which are relevant to your topic. These are topical citations: works that are related to your topic, but not directly to your project. It is often appropriate to cite significant, well known papers on the topic, even if they are only tangentially related to your work. Topical citations may also include older, but well known papers, which have been superseded by later research. If your specific project doesn’t rely on them, you should consider carefully whether you really need these citations at all—only include them if they are necessary to clarify the nature of your work or enhance your credibility.
If you include topical citations, make sure to explain why they are related to your work. For example, you should never have a sentence like “Temperley (2012) performed a detailed study of the blues third in one song.” You need something like
Temperley (2012) performed a detailed study of the blues third in a single popular song; however, Temperley did his analysis “by hand,” which is not feasible for larger datasets, so our analysis methodologies are unrelated to his.
Topical references help to situate the current work in a network of past and ongoing research and demonstrates your awareness and engagement with the work of other scholars.
Still, this doesn’t mean you should go out of your way to cite every important paper and/or author in your field just out of respect or courtesy. Spurious citations, especially mentions to one’s own previous publications, should be avoided.
Deceptive citations
A deceptive citation is one that appears to support a claim, but doesn’t. Don’t do this; Always be honest (and accurate) in your citations! It is irresponsible to cite somebody who is speculating that \(X\) is true, but making it sound like they have provided empirical evidence that \(X\) is true. For example, David Huron speculates about the perception of “scale degree qualia” in his book, Sweet Anticipation (2006), but he doesn’t actually conduct a peer-reviewed research study. If I were to say “Learned statistical expectations create the perception of scale-degree qualia (Huron, 2006),” this would be misleading. I could say something like “Huron (2006) presents a theory which explains scale-degree qualia using an expectational framework.”
Of course, it is possible to accidentally miscite somebody, either by misunderstanding or confusing their work. This is where skimming can be dangerous: make sure to thoroughly read all citations that you actually use in a paper you want to publish. Misciting someone (i.e., misrepresenting their work) will at best annoy reviewers who recognize your mistake (especially if they are the person you are misciting). If a reviewer or editor thinks you are purposefully misrepresenting a citation, they’ll be really pissed.
Second-hand citations
Sometimes you find a paper which cites another paper which is useful to you. You might be tempted to perform a “second-hand citation,” using a phrase like “as quoted in,” “as cited in,” or “as discussed in.” Second-hand citations can be dangerous, because you are trusting that the more recent paper is accurately representing and interpreting the earlier paper. It is surprisingly common in published articles for references to contain a typo or mistake: Page numbers may be wrong, or the volume may be given as 23 instead of 32. Scholars can look very foolish when such errors are propagated through several publications. Arnold (1991) makes a typographical mistake in citing Novak (1985), and then Brown (1996), Cohen (2000), Donald (2005), Ellis (2012) all perpetuate the same typo. Evidently, none of the authors has checked the original publication! It is always preferable to look up the original work, read it for yourself, and cite it directly (assuming it really is relevant); This way you can be sure that your citation actually makes sense—you’ll also often discover more useful information in the original source!
On occasion, a second-hand citation is unavoidable. For example, an author (say Rossi, 2012) might cite or quote from Li (1965). You may be unable to find a copy of Li (1965) for yourself, or you may be unable to read the language in which Li (1965) is written. In these cases you may use a second-hand citation:
According to Li “All observed billbirds are nosiceptors” (Li, 1965, as quoted in Rossi, 2012).
It has been claimed that all observed billbirds are nosiceptors (Li, 1965, as cited in Rossi, 2012).
In most cases, you should only cite the original paper and not the more recent paper which cites it. Of course, if the more recent paper is itself relevant and adds more information than the original paper, you should cite it in its own right, as well. There is one other circumstance where you should always draw attention to a second-hand citation, even if you have looked up the original: Some scholars are especially conscientious and methodical in identifying pertinent research that other scholars have overlooked. If the reference is to a publication that has been generally neglected, you should also draw attention to the scholar who tracked-down the overlooked work. For example, Kim (2010) may have cited the work of Ferdinand (1937)—work that has otherwise not been recognized by other authors. In this case, it is appropriate to acknowledge the role of Kim in re-discovering this neglected research:
Ferdinand carried out early work on this issue (Ferdinand, 1937; as noted by Kim, 2010).
Quotations
Actual quotations from literature should be used extremely sparingly in scholarship, especially in science or engineering papers. There are two circumstances when quotation is warranted:
- The Elegant quote: When the author has expressed an idea in an especially elegant or memorable way.
- The Factual quote. When you want to demonstrate or establish a fact, such as that the author really did say something extreme, contentious, or questionable, or that the author was misquoted by another scholar.
The most common mistake is to resort to quotation because you don’t understand what the author is saying. Complicated quotations usually suggest that the author quoting the passage hasn’t been able to make sense of what was written. We might call this the The Mystifying quote. In effect, the mystifying quote says: “I don’t really understand what this person is saying, so I’ll just simply pass along the quote and hope that you, dear reader, can make some sense of this…”
How to Cite
The purpose of citation is to support a point. A proper citation cites a publication which supports the immediately preceding claim. An example would be:
In the 19th century, most minor-mode music is actually faster in tempo than most major-mode music (Post and Huron 2009).
Notice that the citation is not the point of the sentence. Write a sentence that expresses an idea and/or makes a point, then add citations that support the point you are making. The sentence is not about the citation itself!
For example, I don’t need to write
Madsen (1986) conducted an experiment that demonstrated that listeners are more likely to entrain to higher metric levels at faster tempos.
Instead, I can just write
Listeners are more likely to entrain to higher metric levels at faster tempos (Madsen, 1986).
Madsen, and their experiment, is not the point of the sentence.
Only make a sentence about a citation if you are actually making a point about the citation. For example, if you want to replicate (or criticize) a methodology or idea from a specific research paper, then you need to talk about that paper and its authors. For example, if I wanted to point out a flaw or argue a point about Madsen (1986), then I might need to make it about them:
Madsen (1986) conducted an experiment that claimed to demonstrate that listeners are more likely to entrain to higher metric levels at faster tempos. However, they used a very small sample of stimuli, all classical music, which may have biased their results. Madsen’s work has been widely cited, but never replicated.
You may also want to actually mention an author or paper to credit them with a theory or idea:
Huron (2006) proposed that tonal expectations are learned from the environment.
I would only do this because I want to be clear that this is not a “fact” or result—it is an idea that Huron proposed that I might want to test or criticize.
Framing citations
In some cases, you’ll cite something that doesn’t exactly “support” your point. Rather, a citation might
- be tangentially related to the point;
- make the same point, but using a somewhat different approach;
- make the same point, but only in passing—
- for example, the cited publication presents data that supports this point but the author(s) of the cited work don’t draw attention to this result;
- make the same point but provide a refinement;
- take an opposing or alternative point of view.
If the citation you are using doesn’t simply “support” your point, you’ll need to clarify how the citation relates to your point—put a frame around it. For example, when a cited work is contrary to a claim, it is common to add the phrase “although see:”
The amount of syncopation in American popular music increased steadily from 1850, peaking around 1970 (Smith, 1998; Smith & Jones, 1999; Jones & White, 2008; although see Black & White, 2006).
In some cases, you might want to simply point to citation for more information: “For a more detailed explanation, see Smith (2012).” In complicated situations, an explanatory footnote or endnote is appropriate.
In music analysis, it is common for different theorists to offer rather different analyses of the same musical work. Too often, one observes a regrettable tendency for music scholars not to cite other analyses of the same work simply because the author disagrees with another author’s analysis. This is simply bad scholarly practice. Always cite those who disagree with you, and with whom you disagree. Disagreements need to be kept professional rather than personal: failing to cite the work of people will simply make them angry and transform a civil disagreement into a personal dispute.
Please see my writing guidelines for additional examples and advice.
Literature in a paper
Research papers will have a large number of citations in their introductory sections, especially in the middle (“body”) of the introduction. As you introduce your ideas, you need to establish what previous research has done, how your research relates to it, and establish basic facts/assumptions. In a well written introduction, most of your important citations are naturally included in the introductory content. In some cases, it may be more natural and easy to understand if you have a separate subsection called “Background” or “Literature Review”—however, there will still usually be a significant number of citations in the introduction.
The goal of a literature review, whether it’s a separate section or folded into the introduction, is to support your argument (the validity of your research); A literature review is not simply a list of relevant citations. Each citation should relate to a point you are making that 1) helps us understand the ideas you are presenting and what you are proposing; 2) helps us understand how your idea relates to other work—including how your idea is novel (or isn’t); or 3) demonstrates that you know enough about your topic to make intelligent and relevant research decisions. At a larger level, each paragraph in a literature review should focus on a main point or argument, with each sentence making supporting points and citing literature along the way.
Let me say it once more for emphasis: A literature review is not simply a list of relevant citations.
Mantras
Scholarship is a community enterprise, and so you need to participate as a good citizen. Few experiences are more discouraging for a scholar than to read a publication that fails to cite one’s own research when the research is clearly pertinent. Acknowledge the hard work of others, in the same way that you hope others will acknowledge your own hard work. Make the extra effort to identify all of the pertinent research literature and to cite this work in your writing. In particular, don’t neglect historical literature and literature published in languages you don’t understand.
Scholarship is a community of colleagues and collaborators—not competitors. It’s natural to fear someone will “steal” your ideas, but that fear hurts everyone: if others are independently doing similar work, that’s good—you replicate each other and can collaborate. Learn from others (and cite them), ask questions, share your own experience, and join the community. Even in industry, collaboration is usually a net benefit.