The pain and the passion behind building Asian language keyboard apps

This past weekend, I briefly mentioned the Perfect Tibetan Keyboard app, which allows Tibetan text input across a variety of platforms. Also, in December 2011, TechinAsia covered the launch of Phetasarth, a standardized Lao font developed by XY Mobile and adopted by the government of Laos.

Keyboard implementation varies widely across the many languages used in Asia, and even across different orthographies (writing systems) of the same language. So, for some unique insights into the development-side of these keyboard apps, I talked with Todd Kramer of Kramer Apps. Todd is the developer behind iOS keyboards for 16 Asian languages and three more globally.

Ken: Which languages have you developed mobile keyboards for so far? Have you worked on any other language-related products?

Todd: The apps I have made for Asian languages include Bengali, Burmese, Gujarati, Kannada, Khmer, Lao, Malayalam, Nepali, Oriya, Pashto, Persian, Punjabi, Sinhala, Tamil, Telugu, and Urdu. I also have an app called Laohu for learning Chinese, and Tamil Alphabet for helping speakers of Latin-based languages learning the Tamil script.

What is one of the largest challenges in mobile keyboard development?

Todd: The article you mentioned about Lao font development relates closely to one of the major challenges in making keyboards. Font development is critical to achieving more mainstream usage of these languages on computers and mobile devices, and having a unicode-compliant option really helps. As an example, Zawgyi is still the most popular Burmese font, and is considered to be a unicode font, but is not fully unicode compliant, which is why sites like Wikipedia don’t use it. However, many people have complained about how certain characters are rendered in unicode fonts, in Burmese and other scripts around the world, so it seems there is still work to be done here. I test different fonts for the languages I support, especially those for which I need to supply a font.

Another problem that comes up often is characters that don’t display correctly on iOS devices by themselves. This happens often with accent characters generally and with some vowels in Indic languages.

Which languages have you personally found most difficult to develop keyboards for?

Todd: Each language has its unique characteristics that have to be considered when building a keyboard. For example, the popular Tamil99 keyboard layout for Tamil makes typing faster by giving a dual role to the vowel keys. The form of the vowel depends on the letter that precedes it, so in building the keyboard I had to figure out the best way to ensure that the user always gets the correct result.

Although it’s outside of Asia, building a keyboard for Amharic (Ethiopia) was particularly challenging because there are nearly 300 distinct characters that must be supported (my keyboard also includes additional characters used in typing Tigrinya). Ethiopic script (Ge’ez) is an abugida – a script in which vowels and consonants are combined into a single unit – but unlike other abugidas, Ethiopic script has combined its vowels and consonants into indivisible units.

Finally, another challenge was ensuring the correct text orientation for languages written right to left (Pashto, Persian, Urdu). System elements default to the left to right orientation, and even when you set the orientation correctly, certain actions can trigger a reset back to left to right if you aren’t careful about explicitly managing the orientation.

What’s the latest news in the mobile keyboard development arena?

Todd: One thing that could be interesting is Apple’s rumored low-cost iPhone 5C, scheduled to be announced in a few weeks. I think this will give more people in Asia access to the iOS ecosystem and therefore access to the tools I create for languages in Asia and around the world.