Appendix D — Writing Best Practices

(Modified on 2026-08-11)
2,270 words

This document provides a number of empirical/scientific writing guidelines and advice. Please follow this advice in writing for this class (and beyond). Following writing guidelines (sometimes arbitrary ones) and dealing with nitpicky editors is part of all professional work, both inside and outside academia.

Active/Passive voice

Many scholars revert to passive sentence construction in scholarly writing. Passive voice has the rhetorical effect of distancing the writers’ personal voice from the content. This creates the impression of greater objectivity; It makes it seem like the writing was done by an abstract, ideal scientist, not a flawed human—in essence, it is a veiled appeal to authority. However, this sort of rhetorical trickery is the opposite of what we want in scientific rhetoric!

You conducted scientific research. You posed the research questions. You defined variables and operational definitions. You gathered and analyzed the data. All of these steps are affected by our human fallibilities, and good, skeptical scientific rhetoric emphasizes these failings, not hides them.

  • Bad example: A 2x2 ANOVA was conducted…
  • Good example: I conducted a 2x2 ANOVA…

Don’t be mistaken: a scientific paper is not about you. You generally should not tell stories about yourself, or walk through a paper in a narrative:

First I did this, but then I thought that wouldn’t work, so I tried this instead… After that, we thought it would make sense to try this other approach.

This is (usually) not appropriate. Still, the goal is own and explain your decisions, usually using an active voice, but still focus on the research.

Adjectives

Adjectives and adverbs should be used sparingly, and only when really essential. Don’t use flowery language for the sake it of. When you are editing your writing, the first thing to do is comb through and remove adjectives that aren’t really needed—if you are anything like me, there will be lots of them!

Asides

Asides—like this—should only be used sparingly, and where they serve an essential purpose. If the aside can be split into its own sentence without breaking the flow of the logic, it probably should be.

There are three approaches to writing asides:

  • Commas:
    • “In previous research, conducted with the support of the National Science Foundation, we discovered X.”
  • Parentheses:
    • “In previous research (conducted with the support of the National Science Foundation) we discovered X.”
  • Em dashes:
    • “In previous research—conducted with the support of the National Science Foundation—we discovered X.”

Generally, parentheses should be reserved for shorter, less important asides, which just add a little bit of extra information; Parenthetical asides should be deletable without really changing the meaning of the sentence. Make sure to not overdo assides, because they will make it harder to read your writing. Use them sparingly, and mix up the different types.

Citation and References

Citations are used directly in the prose of your paper, either as a parenthetical citation (most common) or a narrative citation. All works cited in your prose must be listed in your references section—only include references (in your references section) that your paper actually cites.

Never refer to a citation by its title, unless that is relevant for some reason (which is extremely rare). Citations are always referred to by their author(s)’s last name(s) and the year of publication.

Parenthetical citation

Most citations should not play a grammatical role in the sentence they are used in. They are simply provided as back-up/support for the whole sentence (or clause). They are placed between parentheses at the end of the sentence, before the final period. If there are more than one relevant citations, they are all placed in the same parentheses at the end of the sentence, separated by semi-colons.

Many researchers have studied the cognitive load associated with metric ambiguity (X, 2003; Y, 2007).

For very long sentences with multiple clauses, you might have citations that are associated with specific ideas in the sentence. Place relevant citations after relevant ideas, especially before commas.

For example:

Many researchers have studied the cognitive load associated with metric ambiguity (X, 2003; Y, 2007) and tonal ambiguity (J, 2009; K, 2015).

Narrative citation

If you need to refer to a paper/study/author/algorithm in a sentence, you are doing a “narrative citation.” (This is in contrast with a “parenthetical citation,” which simply supports a sentence without being part of the sentence.)

If you are referring to a paper, your sentence should reference the author(s) by last name(s), followed immediately be the citation:

  • In APA style, only the year appears, so it looks like “Smith (1992) says that X is Y.”
  • In a numeric style, it might look like “Smith [4] says that X is Y.”

The keys is that the author(s) name is a noun in the sentence, not the citation. So something like “[4] says that X is Y” is not a complete sentence!

In some cases you might want to refer to a theory or algorithm or software from a paper. In this case, the name of the theory/algorithm/software should be used as a noun in the sentence, followed immediately by a full citation (name and year).

  • In APA style, it might be something like “The DeepGen model (Smith, 2016) is currently the state of the art.”
  • In numeric style, “The DeepGen model [4] is currently the state of the art.”

Once again, you must not do things like “The [4] algorithm is the state of the art”—this is not a complete sentence.

Citation placement

Citations are separated from surrounding words by spaces, but should not have spaces between them and punctuation. Citations should appear before closing punctuation, including final periods.

  • Bad example: Duple is the most common musical meter(Honing, 2007).
  • Bad example: Duple is the most common musical meter. (Honing, 2007)
    • Good example: Duple is the most common musical meter (Honing, 2007).
  • Bad example: According to Honing(2007) , who studied this matter extensively, duple is the most common meter.
    • Good example: According to Honing (2007), who studied this matter extensively, duple is the most common meter.

Commas

Use Oxford commas.

  • Bad:
    • “Timing, rhythm, meter and tempo.”
  • Good:
    • “Timing, rhythm, meter, and tempo.”

Dashes

There are three types of dashes:

  • Hyphen: -
    • Used to connect words.
      • Multi-word adjectives should be connected by hyphen, but not nouns. For example:
        • Bad: “Hypothesis driven research.”
          • Good: “Hypothesis-driven research.”
        • Bad: “This research is hypothesis-driven.”
          • Good: “This research is hypothesis driven.”
        • Bad: “We optimized fit using least-squares.”
          • Good: “We optimized fit using least squares.” (“Least squares” is a noun.)
        • Bad: “We optimized fit using a least squares estimate.”
          • Good: “We optimized fit using a least-squares estimator.” (Now “least squares” is used as an adjective.)
  • En dash: –
    • Two dashes in plain text: --
    • En dashes are used only for numeric ranges, like 3–10 (3--10 in plain text).
      • It is incorrect usage to use hyphens for numeric ranges!
      • Bad: “All participants had 3-10 years of musical experience.”
        • Good: “All participants had 3–10 years of musical experience.”
  • Em dash: ⸺
    • Three dashes in plain text: (---)
      • Don’t put spaces on either side of the em dash!
    • Em-dashes are used either as a strong parenthetical or as a break—sort of like a semi-colon, but with less independent sentences

Definitions, Jargon, and Acronyms

All acronyms should be spelled out the first time they are used—always—with the acronym indicated in a parentheses. Only then can you use the acronym to save space. It doesn’t matter how well known the acronym is:

  • Automatic tempo detection is a major topic in the field of music information retrieval (MIR). MIR researchers have proposed dozens of algorithms…
  • The fast Fourier transform (FFT) is used frequently in this field.
  • We measured tempo in beats-per-minute (BPM).

However, if you only refer to the acronym concept once or twice, don’t bother with the acronym—just spell the whole thing out each time.

Jargon should be introduced in the same way, but using italics (only the first time used).

Human perception of the timing of musical events is largely driven by the phenomenon of musical meter.

If you are using the jargon/term in a non-standard way, or if there is no widely agreed on understanding, you should immediately define the term as you are using it.

Pitch in Western music is organized in a system called tonality—a hierarchical organization in which some pitches are more “structurally” stable than others.

The italicized emphasis tells the reader that this is a word being used in a specific technical way. Why is this important? Consider the following example:

  • Bad example: In this paper, we explore the application of multi-level linear models to the prediction of pitch sequences.

When I say “multi-level linear models,” are you the reader supposed to understand what that means? You probably know what the words “multi-level” and “linear” mean, so you might guess what a “multi-level linear” model is. But no! Multi-level linear models are a specific type of standardized statistical model. So what I should have written is:

  • Good example: In this paper, we explore the application of multi-level linear models to the prediction of pitch sequences.

This way, it is clear to the reader that “multi-level linear model” is a specific type of model that I am referring to. I could then provide a definition, but even if I don’t, at least the reader knows that they could go look up what a “multi-level linear model” is.

Scare quotes

Do not use “scare quotes” to introduce technical/precise jargon. Scare quotes are used in a different way: to indicate that you regard the term with skepticism—that you don’t want to “own” the term. You are basically saying “some people use this term, but it might not be appropriate.”

So you can use scare quotes, for example,

  • Slang words;
  • Words that you’ve heard people use, but you’re not sure how widely the term is used;
  • Words that are used, but that you think are inappropriate (for various reasons).

For example, as a musician I’ve heard people use the term “laid-back groove,” but I’m not sure if everybody uses that term, and I don’t want to commit to it.

In this study, we investigate the rhythmic correlates of the “laid-back” feel in groove-based music.

The difference between writing “‘laid-back’ groove” and “laid-back groove” is that in the second case, I am owning the term, and suggesting to the reader that we use it as a technical term. Scare quotes invite the reader to consider the term with a grain of salt; Emphasis (italics) asks the reader to accept the term as a specific technical concept.

Flow

Make use of words and punctuation to connect sentences.

  • Use colons whenever you are “setting up” an inevitable next sentence.
  • Use semicolons (sparingly) to connect two complete sentences which are closely related.
  • Start sentences with “however,” “thus,” “therefore,” etc. so the reader can anticipate how each sentence relates to the previous sentence.
    • However, don’t do this for every sentence—it will get tiresome—and don’t use the same connector twice in a row.

Connections between paragraphs should also be smooth. This should mainly be accomplished by ordering paragraphs in logical ways, and having the final sentence of one paragraph lead to the first (topic) sentence of the next paragraph. Paragraphs, however should not start with words directly referring to the previous paragraph: You shouldn’t start a paragraph with “but” or “however”—the paragraph should stand on its own.

Paragraphs and topic sentences

In a paper, each paragraph should focus on one coherent idea, introduced in the first (topic) sentence. Paragraphs might vary in length a lot: one “coherent” idea might only require a three-sentence paragraph—another might require 20 sentences. Don’t insert a paragraph break if you haven’t finished with an idea. Conversely, don’t insert multiple ideas into a single paragraph.

Each paragraph should be a somewhat contained environment. Paragraphs should never have pronouns like “it” or “these” referring to content from another paragraph. Paragraphs shouldn’t start with words like “however” or “thus”, because those words are immediate continuations of what came before.

It can be very helpful to write (for yourself) a single sentence which expresses what each paragraph is saying.

Quotes

Always use double quotes. The only time to use single-quotes is when you have a quote within a quote:

  • “According to Huron, the ‘statistical learning’ is the most important concept in music cognition.”

Quotes at the end of a sentence should encompass the final period. However, exclamation or question marks should only go in the quote if they are part of the quote.

Referential language

Only use words like “this” or “it” (or other words), if “it”/“this” is the unambiguous main subject of the previous sentence, and you use “it”/“this” near the beginning of the sentence.

  • Good example:
    • “Empathy is an essential feature of musical collaboration. It can be observed in many musical interactions.”
  • Bad example:
    • “Empathy is an essential feature of musical collaboration, driving the interactions between improvisers which generate maximum creativity and expressively. We find that it can be observed in many musical interactions.”

If the previous sentence has multiple clauses, the main subject is probably too ambiguous.

Don’t be afraid to repeat yourself—often that is the clearest way of communicating. If “it” refers to the concept “empathy,” there is no reason to use “it”—just repeat “empathy”.

If “it” is a longer concept, like “inter-musician dynamic oscillation patterns” you don’t want to repeat that over and over. However, rather than just using “it,” you can do a partial repetition.

  • Bad example:
    • “One of the defining features of jazz improvisation is inter inter-musician dynamic oscillation patterns. They have been observed in all genres of jazz performance.”
  • Better example:
    • “One of the defining features of jazz improvisation is inter inter-musician dynamic oscillation patterns. These patterns have been observed in all genres of jazz performance.”

If the reference is getting further away from the original statement, you could use longer repetitions like “these dynamic patterns” or “the inter-musician dynamics” without repeating the whole mouthful.

Sentences

Try to keep your sentences shorter—use long sentences only where they are needed to express an idea. Make sure to punctuate long sentences properly, to make them readable. You should also read long sentences aloud to yourself to see if they make sense.

However, don’t only use short sentences. Good writing makes use of a mix of longer complex sentences and short simple, punchy sentences. If you write too many short sentences, it can be choppy and hard to follow the flow/thread of thought.

Spell check

Use spell checks! No reviewer (or instructor) wants to waste time correcting spelling. Use correct (i.e., American, duh) spelling—none of this “colour” or “analyse” crap!

Spell-checks won’t catch word-substitute errors—like “their” and “they’re”—so be careful! Double check your sentences before submitting anything. It can be very good to read your writing out loud to catch spelling and grammar errors.

Tentative language

Scientists, and the scientific community, are humans, and can and do fall prey to rhetorical tricks and invalid claims. Bad scientific rhetoric makes dramatic, claims.

  • “Science is objective.”
  • “This is a fact.”
  • “The evidence speaks for itself.”
  • “This proves our hypothesis.”

There is also a trap to fall into using impressive language: Using obtuse jargon; Hiding behind complex equations; Even the cliche (mostly fictional, thankfully) of reporting numbers to high-decimal precisions (“it will take 29.324395 seconds.”).

Each claim suggests implies that the idea is inevitable and objective. These are invitations to turn off your critical thinking. They are the opposite of scientific rhetoric!

Be sure to use the tentative language of empirical research. Say “we will see that the results are consistent with the notion that…” Of course, we never say “the results prove that…” or “the results establish that…” Don’t say “we will demonstrate that…” or “our results show that…” Simply say “the results are consistent with…”