Hello, I read that you were intrested in other corpuses than Wikipedia. I think that Swedish Wikipedia is a skewed source since so many articles are started by robots, and the frequency of odd formulations remain high even after they are manually cleaned up. The Swedish Gigaword Corpus contains one billion words from 1950-2015 analyzed with NLP and stored in XML format: https://spraakbanken.gu.se/en/resources/gigaword
A presentation: http://www.ep.liu.se/ecp/126/002/ecp16126002.pdf