He may be referring to what I suspect too, that even though it is unlikely that any one Germanic language has the key to this, let alone the many similar linguistic mysteries, deeply diving into each of the languages to graph connections will likely provide the best chance at solving such things.
I believe that not only do we have a hard time with this matter because our conception of what “try” means today is not original, but also that effectively “try and” and “try to” are effectively different concepts of the mind that we simply do not understand anymore, homographs whose separate original meanings and appropriate uses have been lost over time.
I look forward to what LLMs will be able to unlock with this kind of historical and linguistic analysis. I think we will start unlocking some interesting this this way. I could see us being able to, to some degree, rewind linguistic history and also then apply it to historical context. Imagine an LLM that can not only graph all languages and meanings in those languages, but will also be able to infer meanings and origins going back into history by cross referencing all related languages … at least with some degree of accuracy greater than what we can do today. Now we just need all historical texts in digital form and to not erase all life on the planet.
