Title, language, bookmarks, tab order, security: the five document-level PDF settings every checker flags

Most guidance about accessible PDFs is about the tag tree: headings, tables, figures, reading order. But a surprising number of checker failures are document-level settings that take seconds to fix once you know where they live. This guide covers the five that come up most, names the rule that tests each one, and shows where the setting is in Acrobat, in Word, and in generation tools.

1. Title, and telling the viewer to show it

A screen reader announces a PDF by its title. If the title is missing, the file name is read instead, which is why so many documents are announced as "final_v2_approved.pdf". Two settings are involved. The first is the Title entry itself: WCAG technique PDF18 describes setting it in the document information dictionary, and PDF/UA-1 additionally requires it in the XMP metadata (Matterhorn 06-003, "XMP metadata stream does not contain dc:title", and 06-004, the human check that the title clearly identifies the document). The second is the viewer preference DisplayDocTitle: Matterhorn 07-001 and 07-002 fail a file whose viewer preferences dictionary lacks DisplayDocTitle or sets it to false. Acrobat's checker reports the pair as the Title rule and fixes both through Document Properties: enter the title under Description, then under Initial View set Show to Document Title.

• Word: File > Info > Properties > Title, before exporting with Save As.
• Acrobat: Document Properties > Description > Title, then Initial View > Show: Document Title.
• Generated PDFs: set the HTML <title> (Chrome, Prince, WeasyPrint) or the metadata option of the library; Matterhorn's dc:title condition is why a generic template title such as "Statement" is a defect.

2. Document language

Screen readers choose pronunciation rules and voices from the declared language. WCAG technique PDF16 sets the default language with the /Lang entry in the document catalog, supporting success criterion 3.1.1. Matterhorn checkpoint 11 tests it as 11-001, "Natural language for text in page content cannot be determined", and Acrobat's checker reports it as Primary language, fixable from the checker or under Document Properties > Advanced > Reading Options. Passages in a second language need their own tags with a Lang attribute (technique PDF19); the language tagging guide covers multilingual documents.

3. Bookmarks

Bookmarks, technically the document outline, let a reader jump between sections of a long PDF without paging through it, and WCAG technique PDF2 describes creating them. Adobe's checker states that its Bookmarks check "fails when the document has 21 or more pages, but doesn't have bookmarks that parallel the document structure", and offers to generate them from the structure elements. That threshold is Adobe's; PDF/UA-1's navigation checkpoint contains no test, as the Matterhorn guide explains, so a PAC report will not fail a long document for lacking bookmarks. Add them anyway. In Word, the Save As PDF options include "Create bookmarks using: Headings"; in Puppeteer the outline option generates them; in Acrobat the checker's Fix command builds them from headings.

4. Tab order

Keyboard users move between links, form fields, and other annotations with the Tab key, and the order is a page-level setting rather than a property of the tags. Adobe's checker notes that "it's necessary that the tab order parallels the document structure" and fixes it by setting each page's Tab Order to Use Document Structure (Page Properties > Tab Order). WCAG technique PDF3 covers both tab order and reading order. Forms are where this fails most visibly; see accessible PDF forms.

5. Security that blocks assistive technology

PDF permissions can prevent copying text, and assistive technology reads a PDF by extracting text. The PDF specification reserves a separate permission bit for extraction "in support of accessibility", and PDF/UA-1 requires it: Matterhorn 26-001 fails an encrypted file with no P entry, and 26-002 fails one whose tenth permission bit is false. In Acrobat's password security settings the option is labelled Enable text access for screen reader devices for the visually impaired, which Adobe's documentation notes applies to 128-bit and stronger encryption; Acrobat's checker reports the condition as the Accessibility permission flag. A document that must be locked against copying can still permit accessibility extraction. A document that must be archived under PDF/A cannot be encrypted at all; see PDF/A versus PDF/UA.

The rest of the document-level checks

Check all five in one pass

The free DocAccessible PDF checker reports title, language, and structural findings and separates what a person still has to decide. For a document you are about to publish, run it after the export, fix the document-level items first, and then move to the tag-level work in the 15-point checklist.

Frequently asked questions

Sources

The statements above rest on the primary sources below. Where a source is a law or standard, the version and date named in the text are the ones checked on the review date. This guide is general information, not legal advice.

1. Adobe: Create and verify PDF accessibility (Acrobat Pro) . The Accessibility permission flag, Title, Primary language, Bookmarks (21 or more pages), and Tab order rules as worded by Adobe.
2. PDF Association: Matterhorn Protocol 1.1 (PDF) . Failure conditions 06-003, 06-004, 07-001, 07-002, 11-001, 26-001, 26-002, and 27-001.
3. W3C WAI: PDF18, specifying the document title
4. W3C WAI: PDF16, setting the default language with the /Lang entry
5. W3C WAI: PDF2, creating bookmarks in PDF documents
6. W3C WAI: PDF3, ensuring correct tab and reading order
7. Adobe LiveCycle documentation: Using PDF security options . Wording and encryption-level note for the screen reader text access permission.