I am working on how entities take a new sense over time. I am trying to find out a large english corpus (free to download) which should have the time annotation of the origin of the text. I suppose some newspaper corpus and/or blog corpus should be fine for my work. Appreciate if someone could point to such corpus.
There is a corpus of internet discussion boards that is split up into weekly files that spans 4 years. It is freely downloadable here: http://www.psych.ualberta.ca/~westburylab/downloads/usenetcorpus.download.html
Why do you need to download such corpus? The corpus managers enable you to search for these corpora better than to work with them on your own computers...
You can see dozens of corpora in Sketch Engine which some of them are freely available, e.g. 62-million-word English corpus ACL Anthology Reference Corpus (ARC) available at https://the.sketchengine.co.uk/open/
Further corpora are accessible with free 30-day trial account, e.g. EEBO corpus (English from the 15th to the 19th century), 18-billion-word Timestamped English corpus and also next 17 languages (2014-2016).
Timestamped corpora with every day updated data are accessible with paid subscriptions.