[Skip navigation links]
Login

Match lang subtags Screen reader compatibility

Last updated: September 11, 2022

Expected Result: works in a screen reader

Actual Result: causes problems in 3 screen reader / browser combinations

In English 'active table' is pronounced 'act-if tay-bill'. In French 'active table' is pronounced 'act-eef tab-le'. In English 'radio' is pronounced 'ray-day-oh'. In French 'radio' is pronounced 'rad-yoh'. In German 'radio' is pronounced 'rah-di-oh'. In English 'dame' is pronounced 'daim'. In French 'dame' is pronounced 'damm'. In German 'dame' is pronounced 'dam-he'.

By default NVDA uses Windows OneCore voices, which only support language switching if the appropriate Windows language pack is installed. When the German (Germany) language pack is installed lang=de and lang=de-DE are voiced as German, but lang=de-AT is voiced as English. When the French (France) language pack is installed lang=fr and lang=fr-FR voiced as French, but lang=fr-CA is voiced as English. Changing NVDA settings to use the eSpeak voice allows language switching without additional language packs (so lang=de, lang=de-DE and lang=de-AT are all voiced as German, but sound more robotic than the OneCore voices).

WCAG Technique: WCAG 3.1.1 WCAG 3.1.2

Code used for this test:


        <html lang='fr'>
        <head> 
            <title lang='en'>Test for lang subtag matches</title>
            <link rel='stylesheet' href='SR-content-lang.css'>
        </head>
        <body>
            <h1 lang='en'>Following elements have different variants of German: de, de-DE and de-AT</h1>
            <p lang='de'>dame</p>
            <p lang='de-DE'>dame</p>
            <p lang='de-AT'>dame</p>
        </body>
        </html>
    

Reliability Trend

This graph shows reliability over time for this code in NVDA, JAWS and Voiceover. Other screen readers don't have enough historical data yet to plot trends.

100%80%60%40%20%0%2015201620172018201920202021202263%

Change History

Last updated: September 11, 2022

Screen readerBrowserModeNotesWhat the user hears
Fail NVDA 2022.2Chrome 105 Readinglang=de and lang=de-DE pronounced as German, lang=de-AT pronounced as English. Dam-he. Dam-he. Daim.
Fail NVDA 2022.2FF102 Readinglang=de and lang=de-DE pronounced as German, lang=de-AT pronounced as English. Dam-he. Dam-he. Daim.
Fail NVDA 2022.2Edge 105 Readinglang=de and lang=de-DE pronounced as German, lang=de-AT pronounced as English. Dam-he. Dam-he. Daim.
OK JAWS 2022.2207.25Chrome 105 ReadingAll pronounced as German. Dam-he. Dam-he. Dam-he.
OK JAWS 2022.2207.25FF102 ReadingAll pronounced as German. Dam-he. Dam-he. Dam-he.
OK JAWS 2022.2207.25Edge 105 ReadingAll pronounced as German. Dam-he. Dam-he. Dam-he.
OK VoiceOver macOS 12.5Safari 15.6 ReadingAll pronounced as German. Dam-he. Dam-he. Dam-he.
OK VoiceOver iOS 15.6Safari iOS 15.6TouchAll pronounced as German. Dam-he. Dam-he. Dam-he.

SortSite rules: AccPageLangMissing AccPhraseLangMissing

Test notes

All tests were carried out with screen reader factory settings. JAWS in particular has a wide variety of settings controlling exactly what gets spoken.

Screen readers allow users to interact in different modes, and can produce very different results in each mode. The modes used in these tests are:

In the "What the user hears" column: