As one point of inaccuracy, Frellesvig says that EMC words ending with -t really were borrowed into Japanese with a syllable final -t; we can see this from the early Christian missionary sources, which romanize 発熱 as “fotnet” and 今日 as “connit”, for example. It’s harder to see in kana sources because there was no kana for -t, so the sound was usually written with a つ or ち but occasionally distinguished with hentaigana. The sound only later was lost from the language, before the writing system had a chance to catch up and give it a kana. (Compare syllable final -n, which existed in the language but shared a kana with む for some time before ん became the usual way to write it.)
2 Likes
That’s really interesting, thanks.
As an update, I managed to find a pdf of the book ChatGPT referenced, but it’ll take me some time to read over it as it’s 400 pages long 
1 Like
The other day I did a Google search for my favorite basketball team’s logo and the AI overview got the year it was introduced wrong by like 15. Wasn’t even trying to use AI and they have to go out if their way to be like “look how garbage this is”.
9 Likes
はつ for 初 is kunyomi, so the the IA reply is just missing the point.
Wiktionary says that the word follow an unusual accent pattern, so maybe that is the reason.
3 Likes
Was the sound really “lost”? Those romanizations just look like approximations of はつねつ and こんにち with the final syllable devoiced. It’s very common, especially in different dialects, for vowels to become devoiced (無声化). Eg: the す in です or します sounding like “s” and not “su”, or even names that end in き, like だいき, ending in more of a “k” sound than a clear “ki”.
I trust that Frellesvig knows the difference between “syllable final t” and “tu or tsu but devoiced”, so if he says there’s good evidence for a real final -t I believe that. Conveniently the relevant section of his book is quoted in an answer to this stack overflow question if you’re interested. There is also some evidence from how it affected pronunciation of the following word, which I don’t think would happen with mere devoicing.
3 Likes
True, I actually challenged it about exactly that after (albeit for another kanji lol).
I got confused. ChatGPT indeed said that, that はつ is not from Chinese lol I just woke up so my brain and memories were confused
I think it’s more likely that gemination happens when the sound was taken from Middle Chinese (and it used to have a checked stop in Middle Chinese). But because 初 はつ is indeed kun’yomi, it doesn’t geminate?
Anyways, I’ll see what I can find 
I should probably have googled “Frellesvig” before replying to you previously 
I had no context whatsoever for that name so I didn’t realize you were citing a source right there in your reply.
However, incredibly interesting read, thanks for the link! I immediately looked for where I could pick up a copy of his book only to find it’s nearly 9000JPY for a paperback… Maybe for my birthday this year…
3 Likes
That exact one no, but the general look of it is like the 明朝体 (Ming dynasty typeface) typeface, which is similar, in look (with serif like features) and use, to “Times Roman’“ for latin letters.
2 Likes
Thanks. The serifs are probably what’s tripping me up.