So we consider a text as a structured entity, or perhaps as an entity which is structured and yet at the same time that's the case with Roland Barthes.
Much of it is multi-structured (often in text format) and does not necessarily flow into neat rows and columns, such as one would find in a traditional database.
Last year we did hit approximately 1 zettabyte (one trillion gigabyte) of data in the public web and the speed by which new data is created continues to accelerate, including unstructured data in the form of text, semi-structured data from M2M communication and structured data in transactional business applications.