You feed a PDF through an online translation tool and download the result. The text is now in the target language, but the document looks wrong. Headings have shifted. Bullet points are misaligned. Images have moved from their original positions. Worst of all, every hyperlink in the document is dead. The translation tool replaced the visible link text but stripped the underlying URL. The document is translated but broken.
Translating a PDF while preserving its formatting and interactive elements requires a translation approach that separates the text layer from the page structure. The challenge is that PDF text and PDF layout are tightly coupled. Changing the length of a translated sentence, which can grow by 20% to 30% when translating from English to German, French, or Spanish, pushes surrounding text and elements out of position. The result is text that overflows its original boundaries or overlaps adjacent content.
Why PDF Translation Often Breaks Formatting
Most Translate PDF tools work by extracting the text from each page, sending it to a machine translation engine, and then attempting to place the translated text back into the original page layout. The core problem is text reflow. English text translated to German expands by an average of 20% to 30% in character count. The same sentence that fit neatly into a text box designed for English now overflows the box. If the translation tool does not adjust the text box size or font size to compensate, the translated text is truncated, overlaps adjacent elements, or breaks the page layout.
Hyperlinks are lost because PDF links are stored as clickable rectangular regions overlaid on the page, separate from the visible text. When a translation tool replaces the text but does not update the link annotations to match the new text positions and lengths, the links either point to the wrong locations or are stripped entirely. The visible underlined blue text remains, but the clickable area behind it is gone.
The Layer-Based Approach: OCR, Translate the Text, Overlay on the Original
The most reliable way to preserve formatting during translation is to keep the original page images intact and add a translated text layer on top. This approach treats the original PDF as a visual background and places the translated text in transparent text boxes positioned over the original text. The original formatting, images, and links remain untouched because they were never removed. Only the text is translated and repositioned.
This approach works well for static documents where the visual layout is the primary concern and the translated text is for reading, not for further editing. The translated text can be adjusted in font size and box dimensions to fit the available space without affecting the background. WukongPDF supports this layer-based approach by performing OCR if necessary, extracting the text, translating it while maintaining the spatial relationship between text blocks, and placing the translation in a new layer on top of the original page images.
Preserving Hyperlinks During Translation
Hyperlinks are the most fragile element during translation because they are stored as annotations separate from the text. To preserve links, the translation tool must read each link annotation, record its destination URL, note which text it was associated with, translate that text, and reattach the link annotation to the new text position. Tools that do not explicitly support link preservation will strip the links.
A browser-based translation tool includes PDF Links preservation as part of the translation process. Before translation, the tool scans the document for hyperlink annotations and records each link's URL, visible text, and position. After translation, the links are reattached to the corresponding translated text blocks. External hyperlinks, those pointing to websites, are preserved with their original URLs unchanged. Internal links, those pointing to other pages within the same PDF, are remapped to account for any page number changes caused by text expansion. The result is a translated PDF where both the text and the navigation remain functional.
Handling Text Expansion and Contraction Across Languages
Different language pairs have different expansion characteristics. English to German typically expands by 20% to 30%. English to French expands by 15% to 25%. English to Spanish expands by 10% to 20%. English to Chinese or Japanese, by contrast, contracts by 20% to 40% because logographic characters pack more meaning into less horizontal space. A translation tool that does not account for these differences will produce documents where translated text overflows or leaves awkward gaps.
For expansion-heavy language pairs, the practical adjustment is to reduce the font size of the translated text by one or two points, or to allow the text boxes to expand vertically while keeping their horizontal width constrained to the original column width. Vertical expansion pushes subsequent content down the page, which is usually preferable to horizontal overflow that causes text to disappear beyond the page margin. For contraction-heavy pairs, the font size can be increased slightly to fill the space, though this risks creating a visual mismatch between the translated text and the original document's typography.
Quality Checking a Translated PDF
Machine translation is not perfect, and errors in translation can change the meaning of sentences, sometimes in subtle and consequential ways. After translation, have a human who reads the target language review the document. If no human reviewer is available, at minimum, spot-check a sampling of translated sentences by translating them back to the source language using a different translation engine and comparing to the original. This back-translation check catches obvious errors but is not a substitute for human review.
Check the formatting methodically. Scroll through every page. Verify that no text is truncated, overlapping, or shifted out of position. Click every hyperlink and confirm it navigates to the correct destination. Check that images, tables, and other non-text elements remain in their original positions. If the translation tool uses an overlay approach, confirm that the translated text adequately covers the original text, especially if the original text is in a larger font or darker color than the translation. Original text peeking out from behind the translation creates an unprofessional result.
For documents that will be regularly updated and retranslated, establish a translation memory. A translation memory stores previously translated segments and reuses them when the same or similar text appears in a new version. This reduces translation cost, improves consistency, and prevents the same formatting errors from recurring. WukongPDF's PDF Export pipeline can incorporate translation memory data to accelerate retranslation of updated documents.