The dataset has 80,000 messages -- a complete list of Tweets from India (and a few neighbouring areas) over a 1-week period. This includes the message, time posted, device information, user information and geographic information.
The project is an open-ended effort to understand the nature of Twitter usage in India.
Here's the blog post on the output: http://blog.gramener.com/544/the-language-of-tweets