You are here

Analysing Language in Arabic Tweets about the Arab Spring

The final paper at AoIR 2011 is presented in absentia of the original authors, who were led by Muhammad Abdul-Mageed, and focusses on the use of Twitter during the continuing Arab Spring uprisings. It examines the linguistic features of the forms of Arabic used in these tweets, as well as the topics and sentiments expressed. The authors examined some 2000 tweets sampled at random from some 233,000 tweets gatered between November 2009 and February 2011. Tweets were coded for topic across a range of thematic categories, for language (standard vs. non-standard Arabic), and sentiment (objective, subjective; positive, negative, neutral, mixed).

There were many instances of paralinguistic and prosodic cues, as well as emoticons; not much mixing of Arabic and Latin characters; and not many abbreviations. Some 66% of tweets were in standard Arabic, the rest in non-standard variations); over 50% of tweets were subjective, and 41% of those tweets were positive. Culture, politics, and interpersonal communication dominated. This is one of the first studies of morphologically rich languages, then (as opposed to English).


I am just wondering what is the best analysis tool that can be used in such arabic tweets?


Hi Khalid,

good question - I'm not sure. We've been using Leximancer and QDAMiner/Wordstat in our work (on tweets in English and other languages using Latin script); don't know if either of them handle Arabic script, too.

More general statistical and network analysis methods should be the same regardless of language, though...


I'm studying in Turkey about Arab Spring.I was gonna ask you that is it possible to achieve this paper anyway? I searched it but I couldnt find it.Waiting for ur response.Thanks

Hi Ece, sorry, but it's not my paper; I've only blogged about it being presented at the conference. Your best bet is to get in touch with the authors... Axel