For European researchers preparing papers, theses and journal submissions in British English
European researchers writing in English often lose marks or face rejection not because their findings are weak, but because their writing does not follow the conventions that anglophone reviewers expect. This guide covers the structure of English-language academic papers (IMRaD and variations), source integration using Harvard and APA referencing, register and hedging conventions, and the most common interference patterns from Polish, German, French and other European languages. It is written for postgraduate students and early-career researchers who can write competent English but need to close the gap between “grammatically correct” and “publishable.”
Writing for an English-language journal is not the same as writing in your first language and translating the result. The reader sits in a different chair. They expect a particular shape of argument, a particular relationship between evidence and claim, and a particular flatness of tone. None of this is taught explicitly in most European doctoral programmes. You pick it up by submitting, getting rejected and reading the comments.
This guide skips the rejection. It covers what an English-language academic essay or research paper actually demands, where non-native writers tend to lose marks or reviewers, and how to handle the parts most guides skim over: structure, sources, referencing and the difference between sounding formal and sounding stiff.
An academic paper is a record of an argument, supported by evidence, written so that another specialist can check your reasoning and reproduce or extend your work. That is the whole job. Everything else — the conventions about referencing, formatting and tone — exists to serve that one purpose.
Two consequences follow. First, every sentence should earn its place by moving the argument forward or by supplying evidence that the argument needs. Padding is not just bad style. It is a signal to the reviewer and reader that you have not yet decided what your point is. Second, the reader is not your supervisor and not your friend. They are a stranger who knows the field but does not know your paper. Write for them.
Non-native writers often inherit a more ornamental style from their first-language academic training. German, French, Polish, Italian and Romanian scholarly prose tolerates, even rewards, long subordinate clauses and abstract noun phrases. English-language journals do not. The shorter your sentences, the more confident you sound.
Whether you are writing a coursework essay or a journal paper, you are answering a question. For a paper, that question is your research question. For an essay, it is the prompt your supervisor set. Anything in your draft that does not help answer that question can be cut. This rule is more useful than any other piece of writing advice.
At its most extreme, if you cannot say in one sentence what your paper is about, you may not be ready to write it. Sit with the question for another day.
Most weak drafts are weak because they were structured on the page rather than in advance. Before you write anything, list the claims you need to make in the order a reader needs to hear them. Each claim then becomes a section or a paragraph. If two claims belong together, merge them. If one needs three sub-claims in order to make its point, then give it three paragraphs. The outline is the work. The prose is the easy part once the outline holds.
An introduction sets out the question, explains why it matters and signals what the paper will argue. The body develops that argument one step at a time, with evidence at each step. The conclusion does not necessarily summarise the body. Instead it states what the reader should now believe, and states what remains open. Reviewers read introductions and conclusions first. Make sure they are strong and carry the paper.
Write the introduction last. You cannot accurately describe an argument you have not yet finished making.
If your paper has been drafted but the structure feels muddled, a native-English proofreader who works with academics can usually identify the problem in a single pass. FLC Poland offers this kind of structural review alongside line-level proofreading.
Three tests. Is it written by someone with standing in the field? Has it been through peer review or some equivalent quality check? Will it still be findable in five years? If the answer to all three is yes, cite it with confidence. If any answer is no, think harder.
Peer-reviewed journal articles and academic monographs pass all three. Working papers from established research groups usually pass. Preprints on arXiv or SSRN are fine to cite if you note their status. Government reports, statistical agencies and major international bodies are reliable for data, less so for interpretation.
Wikipedia is not a source. It is a finding aid. Use it to discover the real sources cited at the bottom of the article, then cite those. The same goes for any LLM output. A model can point you towards literature. It cannot replace the literature.
News articles, blog posts and op-eds are evidence of opinion, not evidence of fact. Cite them only when opinion is what you are studying.
You are not reading to absorb a field. You are reading to answer a specific question. Skim the abstract, conclusion and section headings before you commit to a full read. If the paper does not bear on your argument after that scan, move on. There is no virtue in finishing a paper that has nothing to tell you.
When a source does matter, take notes as you read. Record the page number for every claim you might cite. Distinguish your paraphrases from direct quotations in your notes by using quotation marks for the latter, every time, without exception. Sloppy note-taking is how accidental plagiarism happens.
Direct quotation is for cases where the original wording itself is the evidence. A legal definition, a famous formulation, a turn of phrase you intend to analyse. If any other words would do, you do not need a quotation. You need a paraphrase.
Keep direct quotations to under five per cent of your word count. Most journals expect far less. When you do quote, the quotation must match the original exactly, including any punctuation and any errors. If the original contains a mistake, add [sic] after it, so the reader knows the mistake is not yours.
A paraphrase expresses the source's idea in your own structure and vocabulary. Changing two adjectives in a borrowed sentence is not paraphrasing. It is plagiarism with extra steps. If you cannot restate the idea without leaning on the original phrasing, it means you have not understood it yet. Read it again, close the book, then write.
Every paraphrase still needs a citation. The idea is borrowed even when the words are yours.
Two referencing systems dominate European journals: Harvard (author-date, in parentheses) and APA (a stricter cousin of Harvard, with its own punctuation rules). Many journals describe their style as Harvard when they actually mean APA. Read the author guidelines before you start. The cost of getting this wrong can be a desk rejection.
The pattern is (Author, Year, p. xx). The page number is required for direct quotations and strongly preferred for specific claims.
Smith (2018, p. 42) argues that the model collapses under noisy inputs.
The model collapses under noisy inputs (Smith, 2018, p. 42).
For two authors, name both: (Smith and Jones, 2018). For three or more, use the first author plus et al.: (Smith et al., 2018).
APA looks similar but punctuates differently. An ampersand replaces and inside the parentheses, and a comma separates author and year.
Smith and Jones (2018) found a sharp threshold effect.
A sharp threshold effect was observed (Smith & Jones, 2018, p. 42).
p. is used for a single page: p. 42. pp. is used for a range: pp. 42–47. The dash between page numbers should be an en-dash, not a hyphen. APA 7 prefers an explicit end page wherever possible. A citation that says pp. 42–47 tells the reader exactly where to look. A citation that says pp. 42 ff. asks them to guess.
In Harvard and APA, you list only the works you actually cite. Alphabetise by first author surname. Order multiple works by the same author chronologically, oldest first. For two works in the same year, append a, b, c to the year.
Journal article (Harvard): Smith, J. (2018) 'The collapse of noisy models', Journal of Applied Linguistics, 42(3), pp. 100–120.
Journal article (APA): Smith, J. (2018). The collapse of noisy models. Journal of Applied Linguistics, 42(3), 100–120.
Book (Harvard): Smith, J. (2018) Modelling Language. Cambridge: Cambridge University Press.
Book (APA): Smith, J. (2018). Modelling language. Cambridge University Press.
Edited chapter (Harvard): Smith, J. (2018) 'Noise', in Jones, R. (ed.) Handbook of Modelling. London: Routledge, pp. 50–75.
Reference lists are the most checkable part of a paper and the part most likely to lose you marks or trigger a revision. A specialist proofreader catches inconsistencies you cannot see after weeks of work on the same document.
Formal English in an academic register is not Victorian English. It is plain prose with a few conventions:
Long words are not formal. Precise words are formal. 'Utilise' is not better than 'use'. 'Demonstrate' is not better than 'show', unless you mean it in the technical sense. If a short word does the job, use it.
The instinct most non-native writers have inherited from school is to avoid repeating the same word twice in close succession. Polish, German and many other European academic traditions reward a writer who finds three different ways to say researcher, scholar and academic across one paragraph. English academic writing does not work this way.
In English, if you have established a term, use that term. If you are talking about the model, call it the model every time — not the framework, then the approach, then the system. A reader has to assume that a different word means a different thing. Synonym variation in technical writing is read not as elegance but as imprecision.
Submit a PDF only if the journal asks for one. Most prefer .docx because it allows the typesetter to work with the file. Keep your own backup in both formats.
Plagiarism in 2026 is no longer mainly about copying paragraphs from books. It is about three things: failing to cite ideas you took from a source, passing off AI-generated text as your own original prose and recycling your own previously published work without acknowledgement. All three are caught more easily than they used to be.
The rules on AI assistance are still settling. Most major journals now require the disclosure of any substantive use of generative models in drafting. Using a model to brainstorm or to check grammar is generally accepted. Using a model to draft sections of the argument or to summarise sources you have not read is not. The safe rule: if you would be uncomfortable explaining what the model did, do not let it do that.
Citation is the cheapest insurance policy in academia. Cite everything you took from anywhere, including ideas, data, methods and phrasings. A paper with too many citations looks thorough. A paper with too few looks suspicious.
The first draft is for getting the argument onto the page. Editing is where the paper becomes good. Three passes is a reasonable minimum:
Structural pass. Read only the section headings and the first sentence of each paragraph. Does the argument hold up? Are sections in the right order? Is anything missing?
Paragraph pass. Read each paragraph as a unit. Does the first sentence promise something the paragraph delivers? Does the paragraph have one job, or three?
Line pass. Read it aloud. If a sentence makes you stumble, your reader will stumble too. This pass is where non-native writers benefit most from a native-speaker editor.
Leave at least a day between drafts. Two is better. The mistakes you cannot see today are visible after a weekend.
Hire a proofreader when you have finished your own editing, not before. A proofreader cannot fix a confused argument, and you do not want to pay one to do work you could have done yourself. Bring them in for what they actually offer: native fluency, an outside eye and a working knowledge of journal conventions.
The right proofreader for an academic paper is someone who reads in your discipline or close to it, writes in their first language and understands the kind of errors non-native writers tend to make.
Good academic writing in English is not a talent. It is a craft, and like any craft it rewards attention and repetition more than it rewards inspiration. Your first published paper will not be your best one. Your tenth probably will not be either. Keep writing, keep submitting, keep reading what good writers in your field do, and the prose comes along behind the work.
That is the part no guide can shortcut. The rest of it — the formatting and the referencing and the register — you now have in hand.
Need a specialist proofreader for your academic paper or journal submission?
