All data is valuable, all evidence is statistical, all explanation is diachrony, all inquiry is permissible. Doing Uralic (pre)historical linguistics.
University of Helsinki
Joined January 2022
- Tweets10K
- Following158
- Followers1.3K
- Likes10.6K
Pinned Tweet
In my own recent work, I have gotten carried away into outlining a first sketch reconstruction of Proto-Surmic (w/ various tangents on language contacts)
protouralic.wordpress.com/20β¦
Wonder if I should start keeping some sort of a log of all the duplicate references I run into on Glottolog, because they're sometimes common indeed. Here's a language (Yemsa) with 48 references, of which by my count 9 duplicates (7 shown)
Instead of direct methodological development for deep-reaching reconstruction methods, I bet we also need to first work out some old families in detail with intermediate-step reconstructions, and use precedents from them of *well-known* 10kya-deep relationships as a control.
How common are, say, ChadicβCushitic parallels that are going to turn out to be trans-Sahel WanderwΓΆrter or disprovable accidents? Against these, what's going to remain actually standing eventually?
The current methodological edge in linguistic reconstruction might be: how deep reconstruction steps are reasonable? We can in fact e.g. reconstruct Uralic in two tiers of ca. 2500β¦3000 years each. Could we usefully push this to 4000 or 5000 years?
Places like the Americas quite probably have linguistic relationships of this shape lurking about that haven't been consistently identified, if at all. We probably know most of the 4000β5000-year-old relationships there but beyond that it gets much sparser.
Muted a guy in the global-sound-symbolism comments for overconfident AI slop. No objection to people using this to get ahead in independent research but people really need to understand that, in most fields, LLMs have no expertise and will suck up to any nonsense you might invent
Not fooling *yourself* is more critical than ever when you can summon robot lackeys to invent some plausible-sounding nonsense in agreement. This seems to be the main problem some people call "AI psychosis" which, most of the time, is not remotely actually psychosis.
Of course some other macroareal roots I think would be better approached as mainly just old areal borrowings, like "pan-African" β―kuru for 'turtle'; turtle shells are a trade item useful for crafts!
β & I wonder if we could loop into this somehow also Late PIE *kar- 'crab'.
This phenomenon keeps mystifying me, because yeah that doesn't look like a natural Wanderwort and even within families these roots tend to be somewhat irregular; but in languages with actual ideophones I've never seen one quite to this effect for 'fly, flap, buzz etc.'
(This Hesburger fact brought to you by a long-suffering Carrols loyalist. IYKYK)
As usual, a lot of references to wanting to find things like a "glorious" past or "uplifting" the nation. Why _does_ the past need to do this? Is there not anything, or enough things, people find in modern Hungary that would be worth taking pride in?
"We descend from illiterate hunter-gatherers" is, besides, not any kind of a specific embarrassment, it is the universal human condition. Civilization is a recent development; anatomically modern humans spent hundreds of millennia in wilderness before drifting into the Neolithic.