I'm a freelance translator who charges per word; how do I count the words in a client's PDF?
Extract the file's text with One combined text, paste it into the word counter, and quote from the number it shows. Read the summary line first, though: pages with no text layer add nothing to the count, so count or estimate those separately and say so in your quote.
Why combined mode specifically?
Page by page mode puts a heading line such as --- page 1 --- before every page, and a word counter counts page and the number as two extra words each time. In an 80 page file that is around 160 words the client never wrote. Combined mode joins only the real text, with no headings or notes.
One sentence: Payment is due within 30 days of the invoice date. The word counter makes it 10 words.
The same sentence on two pages, each with its heading, goes from 20 words to 24.
In Arabic, a sentence the counter makes 11 words becomes 13 with a single page heading in front of it.
The steps
- Open the PDF text extractor and choose the client's file.
- Start with Page by page, press Extract text, and note any pages marked as having no text layer.
- Switch to One combined text. There is no need to extract again; the view changes on the spot.
- Press Copy all text and paste into the Word and character counter.
- Take the word count and add your estimate for any scanned pages.
Scanned pages: do not leave them to chance
Many contracts arrive with scanned pages at the back: a signed annex, a copy of a trade licence, a stamped letter. Those pages hold no characters, so the tool gets nothing from them, and it does no OCR. Quote on the extracted count alone, then discover three dense scanned pages, and you have translated them for free.
- State in your quote the extracted word count, the number of scanned pages, and how you will price those pages.
- To estimate a scanned page, count the words on a line or two and multiply by the number of lines.
- Or ask the client for a version exported from the original software instead of the scan.
Things that inflate or shrink the count
Headers and footers repeat on every page, and they really are text in the file, so they are counted. Agree with the client whether you will count them once. Arabic in some files comes out with letters broken apart, so a word may split into two in the count. Check a short sample before relying on the number, and treat it as the starting point of the agreement rather than a final figure.
Common questions
Is the character figure in the summary the same as the word count?
No. The summary shows the number of characters on the extracted pages. For words you need the word counter.
Is the client's file sent anywhere?
No. Extraction and counting both happen in your browser, which matters with confidential client documents.
The whole file is scanned. What now?
You will see a message that none of the pages has a text layer. At that point you need OCR software or a manual estimate.