Tutorial
How to read a Chinese paper without copy-paste
The relevant literature is there and the PDF will not cooperate: two columns, labelled figures and, sometimes, blocked copying.
Reviewed on August 11, 2026
How do you translate and read a Chinese scientific paper?
Start with the abstract, not the whole file. Translate and summarize only the abstract and decide in thirty seconds whether the paper is worth it — most papers get discarded at this step, and converting the full document beforehand is wasted work.
If the paper passes triage, the route depends on the PDF. With selectable text, a document translator converts the body in one pass. If it is scanned or copy-protected, there is no text to extract, and the route becomes screen-reading translation — what is visible can be recognized, even when selection is blocked.
In either case, figures are left out. Charts, maps and diagrams carry labels, axes and legends as pixels, and no document translator can reach them. Translate each figure on screen, so the labels stay in their original positions and the chart still reads as a chart.
The full route
1. Translate and summarize the abstract
Select just the abstract block. Read the translation and the generated summary and decide whether to continue. This cuts most of the pile.
2. Find out whether the PDF has a text layer
Try selecting a paragraph. If it selects, a document translator handles the body. If not, go through the screen.
3. Read section by section, not all at once
Capture paragraph by paragraph. Chinese uses no spaces between words, so a large block raises the chance recognition splits one wrongly and drags the error onward.
4. Translate each figure separately
Select the whole figure. Axis labels and legends stay in their original positions, which keeps the chart interpretable.
5. Verify numbers and units against the original
Machine translation sometimes reformats values and units. Before using a number in your own work, look at the original.
6. Capture the citation before closing
Mark the passage that mattered and keep the capture with the note. That is what stops you reopening the PDF three weeks later hunting for it.
Two caveats that genuinely matter
Machine translation is not good enough for literal quotation. Use it for triage and reading; to quote, return to the original and have the passage checked by someone fluent. Technical terminology is exactly where machine translation fails most, and the failure goes unnoticed because the sentence sounds right.
If the paper is confidential or unpublished, think about where the text ends up. A document translator uploads the whole file. Screen translation sends only the recognized passage. If not even that may leave the machine, the only route is extracting the text locally and translating on your own.
Frequently asked questions
Does it work on a PDF that blocks copying?
Yes. Protection blocks selecting, but the text is still visible — and what is visible can be recognized and translated.
What if the paper is scanned, with no text layer?
That is the classic screen-reading case. Quality depends on the scan resolution; zooming in before capturing helps considerably.
Can it translate the formulas?
Poorly, in any tool. A formula rendered as an image confuses recognition. Translate the surrounding prose and read the formula in the original.
Does Simplified versus Traditional change anything?
Not for translating. But Traditional characters carry more strokes and lose definition sooner at small sizes — if the result came out bad, zooming in usually fixes it.
Keep reading
Start with the abstract
Open the paper sitting in your folder and translate just the abstract. In thirty seconds you know whether the rest is worth it.
Download and try