“The first million is hardest to get”: Building a Large Tagged Corpus as Automatically as Possible

Gunnel Kallgren