Developing for Luxembourgish, Nepali, Uzbek and Welsh

I’m currently working on some projects that examine the architecture of code construction when ‘accessibility’ is a key consideration in the design and build phase. Interestingly, if you broaden your definition of the term accessibility it can arguably encompass more than just ‘impairments’ per se – but also embody language and communication.

I’m currently working on some projects that examine the architecture of code construction when ‘accessibility’ is a key consideration in the design and build phase. Interestingly, if you broaden your definition of the term accessibility it can arguably encompass more than just ‘impairments’ per se – but also embody language and communication.

I’ve been fascinated in this subject since working with the RNIB sometime ago now on web accessibility and the need to drastically increase the quality of so much content that does exist online.

At the time of my last piece on this subject I looked into IBM’s Natural Language Processing work. This work comes out of Big Blue’s research division and uses various hybrid techniques to try and automate and interpret the myriad of idiomatic inconsistencies that exist in most world languages, dialects, creoles and pidgins.

What I have learnt so far is that for any multilingual or multi-accessible software to work effectively, there always needs to be a base. This may be a base of program code after which different versions of the final product can be produced. For example, develop your word processing application just so far then, at the appropriate watershed level, you can spawn versions for the visually impaired, deaf or even people with ‘motor’ impairments.

At the more cosmetic level, this theory also works. Take Microsoft Office 2007 (or an earlier version if you wish) for example. You wouldn’t develop all the way through to build this product in Luxembourgish, Nepali, Uzbek and Welsh. You can, using something they call a Language Interface Pack, take a base product in the closest available language and then lay a next-best option over the top of this so-called ‘suite’ that will simply change the drop down menus and dialog boxes without affecting the applications’ functionalities.

Note: For Luxembourgish it’s somewhat surprisingly English, Nepali is English or Hindi, Uzbek is unsurprisingly Russian and Welsh is also English.

I don’t know about you, but I think this is a great subject, it’s really interesting and it has a great human and social responsibility angle. As for myself, finding out that there really is a dialect called Luxembourgish was enough to keep me amused for hours.