When a paper in a second language takes forever, the cause is usually not your vocabulary. It is reading from the first line to the last at a constant depth. Papers are not written to be read that way. Changing the order alone cuts the time per paper substantially.
Know the structure first
In most fields, papers follow this order.
| Section | What it holds | Read it first? |
|---|---|---|
| Abstract | The whole thing, condensed | Yes |
| Introduction | Background, and the gap this work fills | Yes |
| Methods | How the work was done | Later |
| Results | What came out | Figures only |
| Discussion | How the authors read the results | Yes |
| Conclusion | The claim | Yes |
Methods exists so that others can reproduce the work. If you are not reproducing it, skip it on a first pass. It is also the section where most people stall and give up.
The order to read in
Reading in this order means that stopping partway still leaves you with something.
- Title and abstract. Decide whether this paper is for you. If not, stop here.
- Figures and their captions. The claim of a paper lives in its figures. Seeing them first means the prose explains something you already have in mind.
- The last paragraph of the introduction. The sentence beginning “In this work we…” is the claim in one line.
- Discussion and conclusion. The interpretation, and the limits.
- Methods and the body of the results, if needed.
Steps 1 to 4 tell you what the paper claims. Go on to step 5 only when you plan to use the same method, or when you want to check whether the results hold.
Looking up fewer words
Ten dictionary lookups per paragraph means you will not finish. Two habits cut that down.
Do not look up every unfamiliar word. Finish the paragraph first, and look something up only if the paragraph did not make sense. Often two unknown words still leave the meaning intact.
Learn twenty field-specific terms first. The same technical terms recur across a field’s papers. Writing down the terms you looked up in your first paper halves the lookups in your second. Technical terms pay back faster than general vocabulary.
Using machine translation well
Machine translation is fine to use. It has a place.
Where it helps: getting the sense of a paragraph quickly. Translating an abstract to decide whether to read the paper is a good use.
Where it misleads: technical terms. Fields have settled translations, and a general-purpose engine may not use them. Where the translation does not make sense, check the term in the original.
What to watch: reading only the translation can leave you believing something the original did not say. Check sentences that carry the conclusion against the original.
Choosing a first paper
Make your first paper a review article. Reviews summarise a field, and they are easier for three reasons.
- They explain the background as they go.
- They contain little experimental detail, so Methods does not stall you.
- Their reference list is a reading list for what comes next.
Starting with a recent primary paper means the background is assumed rather than explained, so you stop repeatedly. One review first makes the same primary papers much easier.
Recording what you read
When you finish, write down three things. A long summary is not needed.
- What this paper claims, in one sentence
- The figure or table that supports it
- How it relates to your own work, in one sentence
With those three you will not need to reread. Without them, by the tenth paper you will have lost the first.
Keeping papers in one place and notes in another adds friction to writing those three lines. In Litlas you can save a paper and write notes on it in the same place, and follow citations from a review to the primary papers without leaving the page. The free plan needs no card and works right away.
Going further
A three-pass method is described in reading a paper in three passes, and choosing which papers to read is covered in how to find prior research.