Why Arabic Breaks When You Convert PDF to Word (and How to Fix It)

You convert an Arabic PDF, a contract, an official letter or a paper you need to translate, to Word. The result is wrong. Words run backwards, letters that should join stand apart, or the page is full of symbols from no known language.

If you work with Arabic documents, as a translator, student, expat or office, you have probably seen this before. The cause is usually inside the PDF itself, not you and not only the converter. Once you know what to look for, you can tell in a minute whether a file will convert cleanly, and what to do if it won't.

Text PDF or scanned PDF?

Start here, because everything else depends on it.

• A text PDF was exported from software such as Word or generated by a system. It contains real characters you can select, search and copy.
• A scanned PDF is a picture of paper. It may look like crisp text, but to a computer each page is just an image.

From a text PDF, a converter extracts the characters already there. A scanned PDF has none, so the page must first be read with optical character recognition (OCR). Skip OCR and you usually get a Word file full of page images with nothing to edit.

Why words come out backwards

Arabic is written right to left, but computers store text in reading order: first letter first. This is called logical order. When the text is displayed, software applies the Unicode Bidirectional Algorithm (bidi) to lay it out right to left on screen.

A PDF, however, is closer to a drawing than a document. Many PDF generators write Arabic glyphs in visual order, left to right as they appear on the page. The page looks perfect, but when a converter reads the text in stored order, every word arrives reversed.

Reversing it by hand flips the numbers and Latin words in the line too, so "2026" becomes "6202". The correct fix is to rebuild the logical order with the bidi algorithm itself, as this technical write-up on DEV Community explains. Good converters do this. Not all of them do.

Why letters fall apart

Each Arabic letter has up to four shapes, depending on whether it stands alone or sits at the start, middle or end of a word. The rendering engine picks the right shape and joins the letters. This step is called shaping.

A PDF doesn't store letters. It stores glyphs, the drawn shapes from a font. To know that a particular glyph is, say, the medial form of ha, a program relies on a lookup table inside the file called ToUnicode, which maps each glyph back to its character. That is where things go wrong:

• The table is present and correct: characters can be recovered and Word joins them again.
• The table is missing or wrong: the converter has to guess, and you get symbols, meaningless Latin letters, or correct letters that refuse to connect.
• Each glyph is placed on its own: some files position every glyph separately, the converter copies that layout faithfully, and Word can't reconnect them.

Fonts that were never embedded

A PDF can embed its fonts or only name them. When a font isn't embedded, the viewer substitutes another one, and a converter may have too little information to interpret the glyphs.

Older documents can be harder still. Some pre-Unicode publishing systems used fonts that put Arabic shapes in the slots of Latin letters. The page looks Arabic, but the text underneath is Latin characters, and OCR is the only practical route.

A one-minute test before you convert

Before trying converter after converter, check the file:

1. Open it in your browser or any PDF reader.
2. Select a line of Arabic (click and drag, or long-press on a phone).
3. Copy it and paste it into Notepad, TextEdit or a notes app.

Then compare:

• You can't select anything: it's scanned. You need OCR.
• The pasted text is correct: the file is healthy and conversion should go reasonably well.
• Words are reversed: visual-order storage. A converter that rebuilds bidi order can usually handle it.
• Letters are separated: glyph storage issues. Results depend on the tool; OCR may work better.
• Symbols or Latin gibberish: missing mapping or a legacy font. OCR is the realistic fix.

This tells you whether you are fighting the tool or the file.

What actually works

If you can get the original Word file, use it

The best conversion is the one you skip. If the PDF came from a colleague or client, ask for the source document. When you need a PDF again, export from Word or use Word to PDF instead of printing and scanning, so it stays a text PDF.

If the file is scanned, run OCR first

Use OCR PDF to turn page images into searchable, copyable text, then convert to Word.

OCR quality depends heavily on the scan: straight pages, even lighting and enough resolution. Stamps, handwriting and heavy diacritics lower accuracy, and a study on arXiv found that open-source OCR struggles more with lines that mix Arabic and Latin script. Always proofread names and numbers.

Fix paragraph direction in Word

If the words are fine but paragraphs run left to right and punctuation sits on the wrong side, fix the paragraph direction:

• Select everything with Ctrl + A (Cmd + A on Mac).
• On the Home tab, in the Paragraph group, click Right-to-Left Text Direction.
• On Windows, Ctrl + Right Shift does the same.
• Set alignment to right.

If the button isn't there, add Arabic as an editing language in Office's language settings. Note what this does not do: it won't repair a word whose letters are reversed internally, and it won't reconnect separated letters.

Switch to a proper Arabic font

Empty boxes often mean the font lacks Arabic support. Select the text and pick one that covers Arabic, such as Arial, Simplified Arabic, Sakkal Majalla or the free Noto Naskh Arabic. This fixes boxes, not text stored incorrectly in the PDF.

Need just the text? Skip Word

To copy a paragraph or a quote, PDF to text is simpler. In Safha it runs in your browser, so the file never leaves your device, and it attempts to put Arabic back into reading order. You lose the formatting but get clean text quickly.

Realistic expectations

No converter can promise perfect Arabic output for every file. The result depends first on how the PDF was made, then on the tool. Even the biggest names have struggled: in a thread on the Adobe Community forum a few years ago, an expert noted that Acrobat's export and OCR features did not support Arabic at the time.

As a rule of thumb:

• Modern text PDFs exported from Word usually convert well.
• Complex tables and multi-column layouts may need manual tidying.
• Full diacritics, stamps and signatures may not carry over intact.
• Scanned files always need a proofread after OCR.

PDF to Word in Safha runs on our server, as does OCR: your file is sent to our server over an encrypted connection, processed, and deleted right after processing. For more on Arabic files specifically, see Arabic PDF to Word.

Test the file, pick the route that fits it, and budget a few minutes for review. It beats trying ten tools on a file that never held clean text.

FAQ

Why does Arabic text appear reversed after converting PDF to Word?

Many PDFs store Arabic glyphs in the order they appear on screen rather than in reading order. When text is extracted in stored order, words come out backwards. The fix is a converter that rebuilds bidi order, not manual reversing.

How can I tell if a PDF is scanned?

Try to select and copy a word. If nothing can be selected, or the whole page highlights as a single image, the file is scanned and needs OCR before conversion.

Will Word's right-to-left button fix disconnected letters?

No. It sets paragraph direction and alignment only. Separated letters or words reversed internally are extraction problems and need a better conversion or OCR.

Do you keep my files after converting to Word?

No. The file is sent to our server over an encrypted connection, processed, and deleted right after processing. If you only need the text, PDF to text runs in your browser without uploading anything.