Comment by mahlerbox

Curious to hear the answer to this as well. Presumably it would be stronger on sentences with a high proportion of cognates/English terms with Latin roots? Does the English output over-represent words of Latin origin?