How to prepare a Word document for translation: a checklist

Why preparation changes the quote and the result

Most professional translators work in a CAT tool (computer-assisted translation software) such as Trados, memoQ, Phrase or OmegaT. A CAT tool is not machine translation. It splits your file into segments, usually one sentence each, and saves every translated sentence for reuse. Clean files split cleanly. Messy files produce half-sentences, stray formatting codes and word counts that don't match the page.

The American Translators Association (ATA) says editable formats such as Word or Excel are easier for translators to handle. Twenty minutes of tidying on your side can save the translator hours, and save you a round of layout fixes when the translation comes back.

Step 1: Finalize the text first

Send the final version, not a draft. Edits made after the translator starts have to be tracked down sentence by sentence, and that time is usually billed.

• Accept or reject every tracked change. On the Review tab, open the Accept drop-down and choose Accept All Changes, or review them one at a time. Microsoft warns that the No Markup view only hides changes; they reappear the next time someone opens the file.

• Resolve or delete comments unless you want them translated. Accepting changes leaves comments in place.

• Check for hidden text. CAT tools treat it differently. Phrase's documentation, for example, says hidden text is not imported by default, and that tracked changes are imported as if accepted unless the setting is changed.

To see everything in one place, run Word's Document Inspector: File, Info, Check for Issues, Inspect Document. Microsoft says it finds comments, revision marks from tracked changes and text formatted as hidden. Run it on a copy, because Microsoft notes that removed content may not come back with Undo.

Step 2: Remove line breaks inside sentences

This is the most common problem in files that started as a PDF or were typed to "look right." If each line of a sentence ends with Enter or a manual line break (Shift+Enter), the CAT tool may split one sentence into several segments. The translator sees fragments, and earlier translations stop matching.

How to find them:

1. On the Home tab, select Show/Hide. Microsoft says it reveals hidden characters such as spaces, paragraph marks and tab marks.

2. A paragraph mark at the end of every line, instead of at the end of every paragraph, means the text was broken by hand.

3. In Word's advanced Find and Replace, the Special menu lets you search for paragraph marks and other special characters and replace them, for example with a space. Then restore the real paragraph breaks.

While Show/Hide is on, look for double spaces, rows of tabs used to line things up and empty paragraphs used as spacing. Remove them.

Step 3: Use styles, not manual formatting

A heading made bold, 16-point and centered by hand looks the same as one set with the Heading 1 style, but it behaves differently. Microsoft explains that when you change a style, all text with that style updates to match. Translated headings often run longer, and with styles you adjust one setting instead of 40 headings.

• Use Heading 1, Heading 2 and so on for headings.

• Use Word's own bullets and numbering, not typed hyphens or numbers.

• Set paragraph spacing in the style, not with empty lines.

• Keep formatting changes inside a sentence to a minimum. OmegaT's manual explains that formatting in Word files becomes tags around the text, and that tag mistakes can stop the translated file from opening. Fewer tags mean fewer chances for error.

Step 4: Keep text where the tool can reach it

• Text boxes and shapes: a box sized for English often cuts off a longer translation. Move body text into normal paragraphs where you can, and keep boxes for true sidebars.

• Tables: build them with Word's table feature, not with tabs and spaces. A table made of tabs falls apart as soon as the text length changes.

• Text inside images: words in a screenshot, logo or chart picture are not text, so the CAT tool can't extract them. Send the editable source of the chart or diagram, or list those images so the translator can quote them separately.

Step 5: Check fonts and direction

• Ask the translator whether your fonts suit the target language, especially for Cyrillic, Greek, Asian or Arabic scripts. Choosing a suitable font before translation saves a font swap at the end.

• For a language written right to left, such as Arabic or Hebrew, paragraph direction has to change. Microsoft notes Word's Right-to-left paragraph button appears only once a right-to-left language is enabled, so the translator may handle this step.

Step 6: Send the file with context

• The .docx itself, not a PDF made from it.

• Earlier translations, glossaries and product names that must stay in English. ATA says reference materials help a translator write correctly for the intended audience.

• The purpose and audience: a contract, a product sheet and a website page need different choices.

• A list of anything not to translate, such as addresses, codes or brand names.

• Your word count, with a note of how you got it. Word counts all text unless you select part of it. A CAT tool can count differently, so ask which count the quote uses.

If all you have is a PDF

If the original Word file is gone, converting the PDF is a separate job. Start with how to turn a scanned PDF into an editable Word document, and if the result looks wrong, see why PDF to Word conversions mess up formatting. Then run Steps 2 to 4 on the converted file: conversions almost always leave line breaks inside sentences and extra text boxes. Doing that by hand is the step EasyToDecode is being built to take over (see what's planned).

A quick pre-send checklist

• Final version, with every tracked change accepted or rejected

• Comments resolved or deleted, hidden text checked

• No line breaks inside sentences

• Headings and lists set with styles

• Tables built as real Word tables

• Text in images listed or supplied as editable text

• Reference files and a do-not-translate list attached

Hand over a cleaner file

When it launches, EasyToDecode will take a client PDF or scan and return a .docx built for CAT tools: line breaks inside sentences removed, headings and lists as Word styles, tables as real tables and text boxes kept to a minimum. It prepares the file; it doesn't translate it. Join the waitlist to hear when it opens.