Reading order

Overview

What reading order is

Reading order is the order in which a screen reader speaks the parts of your document. It follows the tag tree, the hidden structure layer inside the file, from the top of that tree to the bottom. It does not follow where the content sits on the page.

A sighted reader's eye knows to read the headline first, then the article, then the sidebar note off to the side. A screen reader cannot see the page. It reads the items in the order they appear in the tag tree, one after another. If that order matches the order a reader expects, the document makes sense. If it does not, the reader hears the right words in the wrong order.

What a reader loses when it is wrong

When reading order is wrong, a reader with a print disability hears the document in an order that does not match its meaning. A pull quote meant to sit beside the third paragraph is read first, before the article it was pulled from. A two column page is read straight across both columns, so the first line of the left column is followed by the first line of the right column, and the sentences break apart.

The words are all present. The reader can hear every one of them. But the sequence is scrambled, so the document is hard to follow or stops making sense. The order on the page looks fine to someone who can see it. The order in the file is the one that matters to a screen reader.

In depth

The tag tree decides the order, not the page

A tagged PDF has two layers. One is the visual layer, the page you see. The other is the tag tree, the structure a screen reader follows. The visual position of a block of text and its position in the tag tree are set separately, and they do not have to agree.

A designer places a heading at the top of the page and a sidebar in the margin. To the eye, the heading clearly comes first. But the software that created the PDF writes items to the tag tree in its own order, which may be the order the designer added them, not the order a reader would follow. If the sidebar was dropped onto the page last, it may sit last in the tag tree, and the screen reader reads it last, after everything else.

This is why reading order is a structure problem, not a layout problem. You cannot fix it by moving the box on the page. You fix it by ordering the items in the tag tree to match the order a reader expects. For how tagging and the tag tree work in general, see the topic on PDF/UA and the Matterhorn Protocol.

A logical order that matches what the reader expects

When the tag tree is in the right order, the screen reader speaks the document the way a sighted reader would read it.

Take a one page notice with a title, three paragraphs, and a closing line. In the tag tree, the title comes first, then paragraph one, paragraph two, paragraph three, then the closing line. A screen reader reads them in exactly that sequence. The reader hears the title, then the body in order, then the close. Nothing is out of place.

This is the case that needs no repair. It is worth naming because it is the target every other case is measured against: the order in the file matches the order the content means to be read.

A sidebar or pull quote that is read out of place

A common failure is a block placed visually at the top or side of a page but sitting late in the tag tree, so it is read after the main text instead of in its intended place.

Consider a report page with a pull quote in a coloured box near the top, lifting a striking sentence from the third paragraph. A sighted reader sees the box, understands it is a highlight, and reads the article normally. In the tag tree, the pull quote was added last, so it sits at the end. The screen reader reads the whole article first, then reads the pull quote on its own at the very end, with no context. The reader hears a stray sentence repeated after the article and cannot tell why it is there.

The fix is to place the pull quote in the tag tree where it belongs in the reading flow, or, if it only repeats text already in the article, to mark it so it is not read twice. A person has to decide which, because only a person knows whether the quote is new content or a repeat.

A multi-column page read across instead of down

On a page with two or more columns, the tag order can zig-zag between columns instead of finishing one column before starting the next.

Take a newsletter page with two columns of text. A sighted reader finishes the left column top to bottom, then moves to the right column. If the tag tree was built by reading across the page line by line, the screen reader reads the first line of the left column, then the first line of the right column, then the second line of the left column, and so on. The two stories interleave, and neither one can be followed. A sentence from one column is cut in half by a sentence from the other.

The fix is to order the tag tree so the whole left column is read first, top to bottom, then the whole right column. The columns look the same on the page either way. Only the tag order changes.

Where machine checking stops for reading order

Our tool compares the order of items in the tag tree against the order the content sits on the page. Where the two diverge, it flags that spot for a person to review. A divergence is a strong signal that the order may be wrong, because most of the time the tag order should track the visual flow.

A divergence is not proof of a fault, and a match is not proof of correctness. The tool cannot certify that the order is right, because only a person knows the intended reading order. A pull quote that is meant to be read out of normal flow may diverge on purpose. A two column page that happens to tag correctly may still hide a subtle slip the machine reads as fine. The machine can point to the places worth checking. It cannot judge meaning. For more on where automated checking stops, see the topic on what automated checking can and cannot find.

The reliable confirmation is a person listening to the document with a screen reader and following along. If what they hear matches what they would read, the order is right.

Reference detail

Standards mapping

Reading order sits under the Understandable principle in WCAG. For the four principles, see the topic on WCAG and the four POUR principles.

The relevant PDF tags

The order is the sequence of tagged items in the tree. Moving an item visually on the page does not change where it sits in the tree.

Common mistakes

The fix

Order the items in the tag tree to match the logical reading order a reader expects. Then have a person confirm it by listening to the document with a screen reader and following along. The visual layout does not need to change; only the tag order does.

Authoritative sources