This question pops up every time I write an article (in the computer science domain), and I am unable to find a British English style-guide providing a proper answer.

Where should citations be placed within the sentence and its punctuation? Where to put white space?

E.g., which options are preferable:

  1. More coffee is always better. [XY]
  2. More coffee is always better [XY].
  3. More coffee is always better[XY].
  4. More coffee is always better.[XY]
  5. According to [XY], more coffee is always better.
  6. According to XY[XY], more coffee is always better.
  7. According to XY [XY], more coffee is always better.
  8. According to XY, [XY] more coffee is always better.
  9. According to XY,[XY] more coffee is always better.
  10. According to XY, more coffee is always better [XY].
  11. According to XY, more coffee is always better. [XY]

In the IEEE editor's style manual it says on page 34:

References in Text: References need not be cited in the text. When they are, they appear on the line, in square brackets, inside the punctuation. Grammatically, they may be treated as if they were footnote numbers ...

I would interpret "inside the punctuation" as ruling out options 1, 4, and 11, even though the grammatical treatment of footnotes is not quite clear to me.

The Springer guide does not appear to comment on the question.

Any other hints / links / viable styles?


7 Answers 7


As others have mentioned, all sentences should be grammatically correct. The citation marker of the form [X] may play the role of a noun, or of silent word.

1, 4 and 11 are never correct - they put the citation into the next sentence, where it doesn't belong. Similarly, 8 and 9 put it into the wrong phrase of the sentence.

3, 6 and 9 combine words incorrectly. Spaces can't just be ignored.

That leaves 2, 5, 7 and 10. These are all used, except that the exact form of 7 will depend on the referencing style in use. You could have 'According to Smith [3]', but not 'According to Smith [Smith, 2017]', The latter would become 'According to Smith [2017]'. Also, for 10 you need to think about how long the second phrase is.

  • 1
    I have seen 4 and 11 at the end of paragraphs. I always took it as a signal that the whole paragraph is related to the reference and not just the immediately preceeding bit of information. Could that use be area specific? I am in engineering.
    – heuamoebe
    Oct 15, 2020 at 17:31
  • 1
    @heuamoebe I've only seen that once, and I don't like it, but it could well be area specific. My preference would be to say more explicitly at the start of the paragraph what you are doing, such as 'we recall the treatment of this topic given in [X]'.
    – Jessica B
    Oct 17, 2020 at 11:54
  • "The citation marker of the form [X] may play the role of a noun" - [ref] is a noun? I think 'is' here would be wrong. Sep 13, 2022 at 21:38
  • @VitaminE I can't quite follow your point. Mine was that you can form sentences like 'In [2] the authors prove that...'
    – Jessica B
    Oct 23, 2022 at 8:08
  • @heuamoebe The Chicago Manual of Style (17 ed., 15.26) prescribes 11 for a block quote. From the example therein: If you happen to be fishing, and you get a strike . . . then you may know you have been fishing in the Galapagos Islands and have taken a Golden Grouper. (Pinchot 1930, 123)
    – bongbang
    May 1 at 1:40

Since you specifically mention British English, one useful style guide for academic writing is New Hart's Rules The Oxford Style Guide. I bring this book up not because it is necessarily your authoritative guide, but because the book tries to describe varied styles in use and not just the their particular in-house style, so it may give you some general idea. Of course the most relevant style guide to follow is one used by your publication venue.

In the section about author-date references, i.e. (Author 2017) or Author (2017), the guide states

The reference is placed immediately after the statement to which it relates. If this happens to be at the end of a sentence the closing parenthesis precedes the closing point (but a reference at the end of a displayed quotation follows the closing punctuation).

This should correspond to option 2 in your list if the author's name is not mentioned in open text, or a variant of 7 if the author's name is part of the sentence.

For numbered references, which I believe is more common in computer science, the guide does not specifically mention the placement, but from the examples given, brackets are placed before the full stop, like this [1]. On the other hand, a superscript number (like a footnote cue) is placed after the full stop.2


This may depend on referencing style. According to the University of Manchester, for Vancouver-style referencing, which is very common in the UK:

If you do not mention the author’s name within your text, you should place the number in brackets at the end of the sentence, eg.

There are six distinctive conditions, which need to be satisfied, in order for a whistle-blowing case to be justified. (2)

Multiple Authors: Provide the numbers in brackets as they appear in the text after the sentence, eg.

Several drug trials proved that the antibody was released immediately. (2, 3, 9- 12)

Source: https://subjects.library.manchester.ac.uk/referencing/referencing-vancouver

The British Medical Journal (BMJ) likewise places the numbers after the full stop, but without brackets and in superscript, and without commas between reference numbers.

For example: https://bmjopen.bmj.com/content/bmjopen/10/2/e034056.full.pdf

The Lancet also puts the numbers after the full stop and in superscript, with commas without spaces and without brackets. They require you to:

• Cite references in the text sequentially in the Vancouver numbering style, as a superscripted number after any punctuation mark. For example: “…as reported by Saito and colleagues.15”

• Two references are cited separated by a comma, with no space. Three or more consecutive references are given as a range with an en rule.

Source: https://www.thelancet.com/pb/assets/raw/Lancet//authors/lancet-information-for-authors.pdf

You need to style according to where you intend to publish, or according to your institution (e.g., for a Master's thesis or PhD thesis) or employer.


One crucial difference between people's preferences on the matter, seem to be how we interpret a citation. Is it a word or is it a mark/reference? A word should be part of the sentence. A mark/reference may mark the sentence itself and should therefore not be a part of it. A citation has the purpose of referring to further reading elsewhere, and therefore sends you off to some other text. The citation is not a word, in my opinion, unless it is a subject or object as part of the sentence. In options 5-7 the citation is an object, while in options 1-4 it is not a subject/object, but a reference marking the sentence or the last word. Citation may be treated or read as a word in some formats, like APA, but not in formats where the citation is just an index in the bibliography. Footnotes are common behind punctuation (without white space), and I think references should be treated similarly. (Some formats even use superscript for citation, just like footnotes are marked.)

Option 4, 7 and 11 (without whitespace after punctuation) are preferably. Here's my reasoning: What is intuitive and what can cause confusion?

When citing/refering to a word or expression or person in a sentence, it makes sense to refer to it immediately after. with a list of different words to cite, it is easier for the reader to follow which citation refers to which word. Citing a person would look like options 6 and 7. Choosing between them would be a matter of aesthetics. Maybe option 11 is a good choice (but without the white space after full stop), as a citation mid sentence tends to make the sentence harder to read.

To quote user1101674 in a comment on this thread:

As mentioned, they [options 1 and 11] allow for the distinction between backing the entire sentence, or just a specific part of it. So if they were commonly understood in this way (apparently they are not), they would indeed reduce ambiguity.

Using the citation as a word in a sentence looks a bit untidy. Specially if you are going for the citation format of superscript number: According to ^1, more coffee is always better. (This is option 5.)

When paraphrasing a statement (like options 1-4) it makes sense to put the citation behind the punctuation (i.e. option 1 and 4, but I would choose option 4 because it connects the citation even tighter to the sentence). This is intuitive for three reasons. First of all: You first make a statement, which should end with punctuation. Then you cite it. In my view the citation is not a word. You don't read it out loud. The citation is not part of the sentence, but refers to where the reader can read more about the sentence.

Secondly it is less confusing and more consequent and neat if there is no difference between citation location for quotes "bla bla."[XY] and paraphrases: bla bla.[XY] And don't forget if you are quoting a sentence with citation: "bla bla.[AB]"[XY]

Thirdly it would be very confusing if readers cannot distinguish a paraphrased sentence from a referred word. If the last word in a sentence is a referred word, it would look like option 2 or 3. How can you tell it apart from the case where option 2 or 3 are paraphased sentences? Does your citation regard the word "better" or the sentence "More coffee is always better."? Therefore option 2 and 3 show confusing citation sites for paraphrased sentences. We are left with option 1 and 4.

Choosing between option 1 and 4 is for me a matter of aesthetics, so I opt for option 4. Maybe someone can come up with a case where it is natural to have citation first in the sentence? If so, there may be a semantic reason for option 4 as well.

Language exists in order to understand and be understood. Therefore language and grammar ought to be designed so that confusions are avoided. That's the purpose of language.


What does your prefered journal say and do?

(I'm not in Computer Science, so I'm attempting a general, non-field specific answer.)

Each journal is going to have its own style guide as to how to format the citations. There are likely to be generalities, but each journal and publication venue is likely to have slightly different preferences.

Taking a look at the "instruction to authors", as you have with the IEEE journals, is a good start, but as you mention, there's likely to be a bit of ambiguity in how to interpret those instructions.

The way I would approach this is to take some of the articles by big names in your field in the journal you're considering, and see how they do things. (This would be in the actual finished, formatted papers as published in the journal, after it's been through the copy editors - not preprints.) If you're not targeting a specific journal, then pick one of the big name journals in your field, preferably one which exercises good editorial control, and see how papers there are formatted. You would likely want to survey a number of journals and a number of authors, to see how much is field-specific style versus journal-specific or author-specific preferences.

There really isn't a formal way how citations are treated "grammatically", as they're not really grammatical elements. Claims otherwise are really just post hoc rationalizations. Instead, the treatment of citations is a stylistic one, rather than a "grammatical" one, much like the treatment of the Oxford/serial comma, or whether to put punctuation inside of quote marks or outside of it.

  • "... big names in your field in the journal you're considering, and see how they do things. " -- this is not a sound advice. This is the blind copying the blind. Many 'big names' write poorly because they are not native speakers. Further, they lend their name to many papers, and I know many big names do not read such papers. Sep 13, 2022 at 21:40
  • @VitaminE Hence the recommendation to take a look at a number of papers from a number of authors. The concept isn't that big names have some sort of special connection to "the one true way" of formatting (this answer rejects the concept that such a thing exists), but instead that examining such papers will give you a good impression of the established conventions in the field. Because if the major names in the field don't follow them, can they really be claimed to be conventions?
    – R.M.
    Sep 14, 2022 at 2:02

As explained, 1, 3, 4, 6, 8, 9, and 11 are incorrect, which leaves 2, 5, 7 and 10. Let's consider those.

5 I dislike, it reads as, According to [manuscript].... I favour 7, According to [authors [manuscript]].... But prefer 10, According to [authors], more coffee is always better [manuscript], because the citation no longer obstructs my reading. Whether to opt for 2 or 10 depends on the message you want to convey:

  • More coffee is always better [XY],

suggests you agree, whereas 10 does not,

  • According to XY, more coffee is always better [XY],

it clearly attributes the claim to XY.


I would go for 1, 7 and maybe 11, since the citation is used to back the whole sentence.

Type 2 citations seem to refer to the end of the sentence, especially in more complex sentence structures like: "Research has shown that coffee is dark, hot and always better [XY]." In such a sentence, I would assume that only the "always better" statement is backed by the source.

  • 7
    I disagree: I have never seen options 1 or 11 and believe them to be wrong - the reference should fit within the rules of grammar wherein you would never have any words outside of a sentence.
    – Phil
    Feb 22, 2017 at 14:05
  • 1
    Since the rules of grammar for citations are still under debate (academia.stackexchange.com/questions/49487/…), I would still opt for the less ambiguous alternative. Feb 23, 2017 at 11:26
  • 2
    I'm not sure it's less ambiguous in practice. Maybe the usage varies between fields, but I don't recall ever having seen a citation outside of the punctuation like this. To me, it comes across as sloppy, rather than seeming more precise. Feb 27, 2017 at 5:53
  • 1
    While I agree that options 1 and 11 are rare (but I have seen them), I disagree that they are sloppy. As mentioned, they allow for the distinction between backing the entire sentence, or just a specific part of it. So if they were commonly understood in this way (apparently they are not), they would indeed reduce ambiguity. Feb 28, 2017 at 5:40

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .