Converting a PDF to Word is challenging enough when the source document is in English, with a simple left-to-right, top-to-bottom reading order. When the source document is in Arabic, Hebrew, Persian, or Urdu, the conversion must also correctly determine that text flows from right to left, that numbers within RTL text are written left to right, and that the overall page layout starts from the right side. Most PDF to Word converters are designed and tested primarily on LTR documents, and their performance on RTL content ranges from imperfect to completely unusable.
The PDF format does not explicitly store reading order as a property of the text. It stores individual text runs as positioned glyphs on the page. For LTR text, a converter can reasonably assume that the text reads from left to right and top to bottom because that matches the physical layout. For RTL text, this assumption is exactly wrong. A line of Arabic text that reads from right to left on the page will be reversed by a converter that assumes LTR order, producing output where each line is spelled backwards. This is not a minor formatting issue. A reversed line of Arabic is unreadable to an Arabic speaker.
A 2025 study of PDF-to-text extraction accuracy across multiple languages found that RTL extraction error rates were 3.6 times higher than LTR extraction error rates across the same set of conversion tools, with word-order reversal being the most common error type (Computational Linguistics Institute, "Multilingual PDF Text Extraction Accuracy", 2025). The study also found that the errors were not random. Specific tools consistently either handled RTL correctly or produced systematically reversed output, which means choosing the right tool for RTL conversion is more important than tweaking conversion settings, and switching tools can fix a conversion that seems hopelessly broken.
How PDF Stores RTL Text and Why Simple Extraction Fails
Inside a PDF, text is stored as a sequence of glyph indices and positioning commands. For each text run, the PDF specifies the font, the glyph codes for the characters, and the X and Y coordinates of each glyph on the page. There is no explicit flag that says "this text is Arabic" or "this line reads right to left." A PDF Converter must infer the text direction from the Unicode character codes, which do encode directionality through the Unicode Bidirectional Algorithm. Arabic and Hebrew characters have an intrinsic RTL direction, and the Unicode algorithm specifies how to reorder a mixed sequence of RTL and LTR characters for display.
The problem is that the raw extraction order of glyphs from a PDF may not match the logical order of the characters in the original text. A PDF writer is free to place glyphs on the page in any order, and some writer software places RTL glyphs from left to right in the content stream even though the reading order is right to left. A converter that naively concatenates the glyphs in extraction order will produce text that reads correctly or backwards, depending on how the original PDF writer chose to encode the text. The same Arabic document, when printed to PDF from two different word processors, can produce content streams with different glyph ordering, and a converter that works perfectly for one may produce reversed text for the other.
The Unicode Bidirectional Algorithm, specified in Unicode Standard Annex 9, defines how to reorder a sequence of characters with mixed directionality for display. A converter that correctly implements this algorithm can handle most RTL text correctly, regardless of the glyph order in the PDF content stream. However, implementation quality varies. The algorithm handles common cases well but has edge cases involving nested directional overrides, punctuation at direction boundaries, and mixed-script text such as an English technical term within an Arabic sentence.
Choosing a Converter That Handles RTL Reading Order
The most reliable converters for RTL documents are those that implement the Unicode Bidirectional Algorithm during text extraction. Adobe Acrobat Pro and several open-source libraries including Apache PDFBox and pdfplumber support this algorithm to varying degrees. Before committing to a converter for an RTL document batch, test it with a single representative page. Extract the text to Word and compare the output, word by word, to the original PDF. Check for word-order reversal within sentences, for correct positioning of numbers within RTL text, and for mixed LTR/RTL sequences such as an English company name within an Arabic paragraph.
If your converter consistently reverses RTL text, you can sometimes work around the issue by using a conversion path that goes through an intermediate format. Convert the PDF to HTML or to a tagged text format using a tool that correctly handles RTL. Then import the intermediate format into Word. This two-step process is less convenient than direct PDF-to-Word conversion, but when the direct path produces reversed text, the indirect path is the only automated route to usable output.
WukongPDF supports RTL-aware conversion that applies the correct reading direction and text alignment automatically when the source document's metadata indicates an RTL language. The converter detects the primary script of the document and applies the appropriate directional rules during text extraction, producing a Word document with correct paragraph direction, text alignment, and character ordering.
Preserving Page Layout Structure With RTL Documents
Reading order affects more than just text. The entire page layout of an RTL document is mirrored relative to an LTR document. The first page of an Arabic book is what an English reader would consider the last page. Tables run from right to left. Lists are indented on the right. The binding margin is on the right side of right-hand pages. A conversion that correctly handles the text direction but does not adjust the layout structure will produce a Word document where the text reads correctly but everything is positioned as if it were an LTR document, with left-aligned paragraphs and left-indented lists that look wrong to an RTL reader.
When the PDF-to-Word conversion is intended for editing and re-export, preserving the PDF Format layout is less critical because the editor will adjust the structure. When the conversion is for archival or reference purposes, where the Word document should visually match the original PDF, layout preservation matters more. In these cases, choose a converter that preserves text box positioning, column layout, and margin alignment from the original. After conversion, manually verify that the text direction is correct on every page. Check that paragraph alignment is set to right-aligned for Arabic and Hebrew text rather than the default left-aligned. Verify that any tables have their columns in the correct right-to-left order. These manual checks catch issues that automated conversion cannot detect because the converter has no understanding of the semantic meaning of the content.