Unique
Microsoft SwiftKey entirely developed the language model with the input of Yuri Michielsen. It is unique that a large IT company does something this important for the Limburg dialect, because most Limburgish projects are carried out by volunteers. 'There wasn't any money for a Limburgish language version. Funds were available for Frisian, but not for Limburgish. This despite the fact that Limburgish is far more common on digital platforms,' says Dr Yuri Michielsen-Tallmann, the key initiator behind this project. He has served as the president of the Limburgish Academy Foundation for several years now and was appointed project coordinator of the Limburgish Corpus Dictionary this summer. Both activities fall under the auspices of the Language Culture in Limburg chair of prof. dr. Leonie Cornips at Maastricht University.

Limburgish as a Microsoft language
There has always been a debate surrounding the correct spelling of Limburg dialects. Elderly people tend to be proponents of Veldeke 2003 spelling or Raod veur 't Limburgs (Limburgish Language Council) in more conventional media. Younger people tend to display considerable variation on social media with regard to spelling. Thanks to Microsoft, it will soon be easier to write digital dialects because this leading IT company is the first to add the Limburgish language option to mobile applications. All Microsoft SwiftKey keyboards will come with the Limburgish language option as standard. The ultimate goal is to make diacritics such as ë, ò, é, äö, oë, oeë, àè, ieë, ieè, eë, ië, aeë, èë, èw, àèë, àèw and aoë easier to type. Users will also be able to type faster.

Spelling
The Limburg keyboard and spellcheck model for the Microsoft SwiftKey beta app has a large glossary of Limburgish words. Many of the dialects adopt Veldeke 2003 spelling, but the program also includes Limburgish spelling variations borrowed from internet and social media channels. Not all spelling variations could be included, but the existing spelling variations will be expanded over time, as each individual user can add his or her preferred spelling. The software on which this language model was based learns from the spelling preferences of individual users. If users routinely use a specific spelling, their preference will be automatically displayed over time. The software also adds new words and spelling preferences to the collective Limburgish glossary. The more often users spell a specific word in a specific way, the more likely this word will be presented as an option to individual users.

Prediction
The predictive model of the Limburgish keyboard and spellcheck function works as follows: the software uses the first few letters to predict the user's word and preferred spelling. For example, if a user types sto the program will offer stoon ('stand') as an option. If a user types sjta the program will offer the word sjtaon ('stand'). The software can also predict which word is likely to follow the previous word and can present several options.

Future
After conquering Microsoft, Yuri hopes to get other IT giants on board. At least one company is showing real interest. More details about this development will follow shortly. This achievement was the result of the love of the Limburgish language. While no funds were made available, the project did lead to an international partnership with Dr Ligeia Lugli (King’s College London), Michael Anthony Schuler, BA (Harvard) and Dr Jean Robert Opgenort in San Francisco.