The findings are specific for the germanic sub-group of the indo-European languages
I am glad that someone in the west finally pays some attention to mathematical linguistic analysis. For the part of the world which used to be on _OTHER_ side of the Iron Curtain a lot of this is yesterday's news.
First of all, they have been paying serious attention to mathematical linguistic analysis for many decades (*). In fact, when I was in high school, one of the reasons geeks bothered attending the CS and Math competitions was that they were co-located with the mathematical linguistics ones. Lots of lovely ladies from the grammar high schools with language orientation. Cough, cough, snigger, snigger... As a side effect, geeks like me attending the CS junkets ended up picking some bits about language by osmosis (facilitated by the rivers of wine and beer in the local restaurants after the competitions ended).
Second, the analysis of English (and Germanic language subgroup) humour comparatively to let's say Russian (as well as other Slavic languages) is something that has been done and dusted decades ago. The "funny bit coming at the end" is specific to English (and other languages from the same subgroup). In Russian it can come anywhere as instead of "situation" humour it is predominantly "multiple meaning/positional" humour. They still grok situation humour though, they just consider it second rate. As a result English humour can be translated into Russian (Jerom K Jerom translation is as good as the original) while Russian often cannot (Russians are rolling on the floor laughing at a Chekov play while English are crying in tears about "how tragic"). Things get even funkier once you go into Turk languages and other languages with different grammar structure compared to Indo-European.
In any case, applause and let's have more of that instead of throwing everything in the AI meatgrinder.
(*)The area has a lot of interesting results like for example you can use language analysis to figure out who influences whom. In fact, you can get a much better idea than from social network activity, especially if you combine it with modern statistical methods and modern number-crunching.