Twitter’s linguistic landscape reveals the existence of global superdialects, according to a study by computational linguists from Northeastern University in Boston. By analysing 50 million geo-located tweets, they identified a new kind of global language usage, dubbed ‘superdialects’. These superdialects are characterised by words that are globally used but locally clustered. The study identified three superdialects: one in North America, one in Western Europe, and one that covers the urban areas of the rest of the world.
The linguists used a mathematical technique called ‘principal component analysis’ to identify these superdialects. This technique allowed them to find patterns in the vast amounts of data and to identify words that are used worldwide but have a local flavour. For instance, the word ‘you’ is used globally, but ‘uu’ is more common in the Netherlands, while ‘joo’ is preferred in Finland.
The study also revealed that superdialects are not static but change over time. For example, the use of ‘lol’ has decreased, while ‘haha’ has become more popular. The researchers believe that understanding these superdialects could help in machine translation and in the creation of more sophisticated language models. They suggest that this work could lead to the development of new tools for studying cultural and societal trends.
Go to source article: http://www.technologyreview.com/view/529836/computational-linguistics-of-twitter-reveals-the-existence-of-global-superdialects/