Hebrew and Arabic Accessibility

Introduction

This document covers the issues involved in making Hebrew sites compliant with accessibility standards based on WCAG.

Since my personal expertise is with Hebrew I would be delighted to hear from anyone involved in Arabic web accessibility where similar issues apply.

The Concerns

The first problem with Hebrew web sites in the past was the lack of support for true bi-directional (BILI) pages. Web designers therefore had to use a visual architecture, where the text was fed to the screen backwards, and the layout was set to align right. These pages were inherently inaccessible. Fortunately today there is a good standard of support from browsers for BILI and web authors can now encode pages correctly using a logical architecture. There are many good sites that explain how to build a BILI logical architecture.

The second and more controversial issue is Hebrew and its vowels:

Hebrew Vowels and Accessibility

Hebrew orthography has two main components: letter characters that represent the consonant sounds, and diacritics representing the vowels and variations in pronunciation. The diacritics are written above, below, or within the letter characters.

Regular Hebrew and Arabic text is written without (most of) the vowels or diacritics, which causes ambiguity. The average number of readings of each word is about 2.4, with some words having up to 8 different readings. Arabic words can have up to 15 different meanings.

The only way to disambiguate the words is using context.

Children learn to read first with the vowels and then as they become more experienced readers, they stop reading phonetically and rely on visual memory. At this point they no longer need vowels. Hence text aimed at older children and adults normally do not have vowels inserted.

Who it affects

When text is written without diacritics there are two groups of people who will be unable to access the content.

• Visually Impaired
  Visually impaired people often use software called screen readers. A screen reader takes what is displayed on the screen and directs it to a speech synthesis for audio output. However without vowels the screen reader may mispronounce the words making the content incomprehensible.
• Cognitively Disabled, Learning Disabled and Low Readers:
  You can not “sound out” words written without vowels phonetically as some of the sounds are missing. The skill of reading without text is dependant upon a combination of context, experience, and visual memory. If any of these functions are impaired or missing the user may be unable to read the words.

Scenarios where people may have difficulty or be unable to read without vowels include: Users with dyslexic and/or a low visual memory, new immigrants, people who received inadequate schooling and cognitively disabled people.

Examples

For example take a look at http://www.letlive.org.il/letlive/index.asp It is a cute site called úðå ìçéåú ìçéåú
(transliterated:"Ten lichayot lichyot) or "give life to animals" (it is an animals right charity) The word "to animal" and "life" has the same word spelled the same way twice, each one is pronounced differently, due to vowels - they have the vowels in as a picture in the heading, but miss it out of the text.

Where we could solve the problem

The problem could be addressed in two ways. The users could have a tool that automatically adds vowels to any site or the web authors could put the vowels into their content.

The User End:
Software could be installed on the users computer that uses natural language processing to guess the vowels of each word.

Devices like this are very expensive, and not entirely accurate. At the moment such screen readers require the user to copy the text into the program’s dialog box. In other words this solution cannot be used by blind people.

Further research is currently being carried out at The Technion University and others which will hopefully result in a more affordable and useful solution. However, at the present time we are years away from this type of end-user solution.

Author’s side:
Solving the problem from the author’s side is immediate and cost effective. In other words, the problem can be solved today. One can add the vowels for free without tools in Unicode, and one can buy word processors where it is possible to put in the vowels. Or you can use a free JavaScript tool to make it attainable by anyone. Mac comes with a similar free tool.

However, acknowledging that for a large site this is an enormous amount of work, they can buy Nakdan from Melingo. This automatically puts in the vowels, and then uses a word processor to correct any errors placed by the automatically inserted vowels.

The Guideline Dilemma

Vowels are an extremely important item when promoting Hebrew accessibility. However if becoming accessible is considered an enormous task then web authors will not attempt to become accessible. Should Accessibility guidlines require Hebrew vowels in web content or not?

Barriers to entry for a guideline that require diacritics:

• Time involved
  Each letter can have vowels attached, making vowelation a time consuming process.
• Skill involved
  Adding in vowels correctly requires knowledge of the root and the structure of words and verb tenses. Most people will not want to appear ignorant by entering sequences of the wrong vowels.

Perspective
Providing enough vowels for automatic and phonetical vocalization, is not as daunting as one perceives, only one or two vowels per word (as in English). However there are a considerable amount of ‘dots’ that need to be added around the letters if one wishes to be correct and complete.

An example is as follows: a newspaper intended for low readers will use partial vowels. This implies that you can read it phonetically, with as much care in pronunciation as there is when the average person speaks but it is not "correct". One cannot detect the grammatical history of the word. (Sometimes a normal newspaper will contain occasional vowels on long and hard to pronounce words. For example a word of six letters may have one vowel specified.)

A Hebrew Bible would have the correct vowelation (about double the amount as the newspaper) including vowels that add a slight change to the letter that most people are unaware of. They also have additional cantorial marks (how to stress or sing words)

Full and correct vowels are not required for accessibility purposes I have heard people complaining that most times when people do put in the vowels (even on official notices) they do so incorrectly. However, those people complaining tend to have doctorates in Semitic languages.

The burden - real or imagined?

How much of a burden is it truly to put in the vowels?

Software such as Melingo can now do much of the work for the web author. True they often make mistakes but authors can touch up the page themselves using a word processor editors.

Touching up a page is very different then doing it from scratch. Essentially all one needs to do is read over the page with the automatic diacritics and when the diacritics have been put in erroneously - correct it.

If one runs Melingo before the site is proof read or tested, then the added step of touching up the diacritics becomes part of the proof read.

At the end of the day, I am convinced that a Hebrew site requires vowels to be accessible. There is not all that much point in having accessibility guidelines that are widely endorsed and implemented if they do not result in true accessibility.

All this and more too

There is good news - adding vowels helps make a site accessible to the disabled as well as new immigrants and people who have difficulty reading. At the end of the day, one will have a website that provides information to as many people as possible. That is the power of the web and the aim of accessibility.