James Tauber
jtauber.com
James Tauber
@jtauber.com
Using computers to better understand languages, texts, and music

OG Web, Python, Corpus Linguistics, DataViz, Philology, Ancient Greek, Music Theory, Tolkien, Space, Health, Retro Computing

Perseus, Greek Learner Texts, @digitaltolkien.com
Tolkien is the most kiki/bouba of authors
December 7, 2025 at 5:15 PM
One is an elf, the other is a hobbit
December 7, 2025 at 5:11 PM
Digital Le Guin Project and History of Earthsea are in progress!
December 7, 2025 at 5:05 PM
Reposted by James Tauber
INITIAL RESULTS...

If we initially consider 168 Sindarin character names from glossary.digitaltolkien.com, we get 85 consisting of 3 or more syllables.

Of those, 62 (72.9%) have a light penultimate and every single one of those has a heavy ultima.

So 72.9% cretic among Sindarin character names!
December 7, 2025 at 10:04 AM
Reposted by James Tauber
Given the way Elvish accentuation works, as long as the penult isn't heavy (i.e. has a long vowel or coda) then you pretty much have a cretic on words with three or more syllables but, more narrowly, the ultima should probably be heavy too to count. I'll give stats on both definitions.
December 7, 2025 at 8:27 AM
Reposted by James Tauber
Fortunately I have a Python library which calculates all this.
December 7, 2025 at 4:02 AM
Reposted by James Tauber
One probably needs to consider "heavy or stressed" rather than merely "long" (which is an acceptable definition of cretic, I think).

For example, Aragorn works not because it's long-short-long but because it's stressed-light-heavy.
December 7, 2025 at 4:01 AM
Excellent work! I’ve signed.
December 5, 2025 at 7:41 PM
Yeah, the precision of exact days in a Leitner system is unnecessary
December 5, 2025 at 2:49 PM
One of the big “innovations” was that when an item was to be next shown in N days, it didn’t put it exactly at N days, but put it on the day with the fewest items within a window around N.
December 5, 2025 at 7:36 AM
a few years ago I started building a vocab flashcard tool that try to reduce that problem.
December 5, 2025 at 3:49 AM
Now I'm imagining Nigel Tufnel explaining it
December 4, 2025 at 7:36 PM
The TL;DR: both deny it's a philosophical difference, just Claude engineers think 1 is "good enough" and OpenAI engineers allow for more experimentation.

I wonder if there are any model-specific differences, though. Do different models exhibit different levels of "creativity" at the same temp?
December 4, 2025 at 7:32 PM
It's fun to ask them both :-)
December 4, 2025 at 7:25 PM