How do I read research papers efficiently when English is my second language?

Most reading advice assumes decoding is free and attention is the constraint. For a second-language reader the reverse is often true, and advice built on that assumption tells you to skim faster when the difficulty is that skimming requires recognizing which sentences matter, which is itself a language skill.

Structure is the way around it. Empirical papers put the claim at the end of the introduction, the design in the methods, the result in the first paragraph of the results, and the interpretation in the first paragraph of the discussion. Those four locations carry most of what you need, and going to them directly means reading four paragraphs closely instead of forty.

Decide what you want from the paper before you open it, because the reading order changes with the goal. Carey, Steiner, and Petri make this their first rule: a reader new to research should work through every figure panel, a reader entering a new field should weight the introduction's motivation and the conclusion's next steps, and a reader tracking a technique should go to the methods and the reason the method was chosen (Carey et al., 2020, pp. 1-2). Naming the goal first limits how much unfamiliar English you have to process in one pass.

The same authors suggest a staged pass instead of one front-to-back reading. Apply six questions about motivation, methods, context, results, interpretation, and next steps to the whole article, then apply them again to each figure, table, or experiment, which they say makes an intimidating paper manageable (Carey et al., 2020, pp. 2-3). Write a one-line answer to each question before you try to resolve every difficult word.

Figures and tables deserve more weight than they usually get. They are the least language-dependent part of the paper and often the fastest route to what was actually measured and found. Reading the figures first and the prose second is a legitimate strategy in any language.

Unpacking a figure has a fixed checklist, which is useful when the caption is written in dense English. Check the x- and y-axes, the variables, the colour scheme, the statistical approach, and how the data were collected, go back to the methods when the caption is not enough, and then state the figure's take-home message in your own words (Carey et al., 2020, p. 3).

Not stopping at every unknown term is what experienced readers actually do. In Hubbard and Dunbar's survey of researchers at one UK institution, 53% of those who gave reading advice named practical strategies such as reading a paper several times, allowing time for it, using selected sections to judge whether the paper is worth reading, and not getting bogged down in technical detail or terminology. Selective reading was the most common theme at 68%, critical reading appeared in 34% of responses, and reading with a specific question in 7% (Hubbard and Dunbar, 2017, p. 11). A workable rule follows from that: look a word up immediately only when it blocks the argument, the method, the result, or a figure, and mark the rest for a second pass.

Two limits on this evidence are worth stating. Hubbard and Dunbar surveyed 300 people in Biological Sciences at a single research-intensive institution, and they still concluded that PhD students and postdocs need explicit training, critical-reading activities in research groups, and supervisors who say what "go and read a paper" means at that career stage (Hubbard and Dunbar, 2017, p. 13). Neither their survey nor Carey and colleagues' rules compare second-language readers with native English readers, so treat the staged passes, the figure checklist, and the selective vocabulary rule as general reading structure that helps you, not as techniques shown to work specifically for L2 readers.

The last piece is accepting slower close reading without treating it as a deficiency. A paper you will cite deserves careful reading in any language, and taking longer over it is not a sign that you are behind. Speed comes from reading fewer papers closely, not from decoding faster.

Should I translate as I read, or read through it?

Translate selectively. Machine translation is reliable enough for deciding whether a passage matters and unreliable for technical terms, which often have a fixed meaning in your field that general translation misses. Translate to triage, then read the passages you will cite in the original.

The specific risk is that a translated technical term looks correct and is not. Words that carry a precise disciplinary meaning get rendered into their everyday equivalent, and a claim built on that translation can be subtly wrong in a way nobody catches until a viva.

For anything you will quote, work from the original text. For anything you are deciding whether to read, translate freely and move fast.

How do I build field vocabulary quickly?

Collect terms from your own corpus rather than from a general list. Keep a running file of the twenty or thirty terms your field repeats, with the definition as your field uses it. Most academic reading difficulty is domain vocabulary rather than general English.

Note where a term is contested, because many are. A word that means one thing to one school and something else to another is a source of confusion that reads as a language problem and is not, and knowing about the split is itself a piece of disciplinary knowledge.

Write the definitions in whichever language you think in. Nobody else reads this file, and the point is comprehension rather than presentation.

One more thing worth separating out: writing difficulty and reading difficulty are different problems with different solutions. If reviewers comment on your English, that is a drafting and editing question, not a reason to slow down your reading.