This paper describes an IMO rather interesting approach towards fake news, through their morphological content: the words from each piece of news (real and fake) were grouped into categories, then the researchers made a statistical analysis of the usage of those categories in real and fake news. And they found out that:

• fake news tend to use more foreign words, adjectives and nouns
• real news tend to use more W-words (who, what), determiners, prepositions and verbs

I think that their findings are damn useful. Perhaps not to detect fake news, but to understand how they work on a discursive level. For example, the usage of foreign words in fake news caught my attention - perhaps they’re used to mask the underlying meaning of the utterance? While real news are focused on describing events, and thus rely more on verb usage?