In Japanese phonology, sandhi is primarily exhibited in rendaku (consonant mutation from unvoiced to voiced when not word-initial, in some contexts) and conversion of つ or く (tsu, ku) to a geminate consonant (orthographically, the sokuon っ), both of which are reflected in spelling – indeed, the っ symbol for gemination is morphosyntactically derived from つ, and voicing is indicated by adding two dots as in か／が ka, ga, making the relation clear.
That was interesting to me, because I’d always wondered why the sokuon looked like a tsu. It was also confusing, since I’d never seen the mutation pattern they described (which explains the wonder), so I filed a mental note to keep an eye out for it. And only a few hours later, I got this word in a new lesson:
Which contains this helpful note, explaining the thing I was keeping an eye out for!
Just watch out for the しゅつ, which gets changed to しゅっ, as many つs do in this situation.
Confusion is gone!
But now I have a new wonder. Is there a better name for this phenomenon? I notice that rendaku is always called out (by name) in lessons when it frequently occurs. This new thing is noted to occur often, but it has no name!
As I understand it, rendaku is a name for the combination of (a) a mutation that occurs and (b) the conditions under which it occurs. The mutation itself a subset of “sandhi”, I’m not sure if the conditions are part of that.
IIUC, the つ → っ mutation includes both (a) the erasure of the tsu sound and (b) the lengthening of the following consonant. Part (b) is a subset of sandhi and can itself be described as a gemination, I think. But what do we call (a) plus (b), as well as the conditions under which the combo occurs (when a jukugo word has a tsu at the end of one component)?
The small つ is called 促音 (そくおん), so you could potentially refer to its effect of going from しゅつ to しゅっ as 促音化 (そくおんか).
化 is a suffix that functions like “-ification” in English, making it into a noun that describes a process of making something. E.g. アニメ化 (making something into an anime)
However, in discussions with learners and generally talking about Japanese, I’ve never heard someone refer to this as anything other than gemination (or simply mentioning 促音, the symbol) before.
I’ve always thought about it as ‘devoicing’. Wouldn’t ‘gemination’ mean that the vowel gets elongated? As I understand it any 3 kana combo should have the same length regardless if it contains つ or っ right?
Doesn’t gemination always refer to an elongated consonant in any language?
Sure, but the idea is that the following consonant kind of jumped into the space that was previously the つ spot, no? It’s why we write it with a double consonant in romaji.
Devoicing is a different concept altogether, where a typically voiced (voiced as in, using the vocal cords) sound becomes… not.
I’ve always thought about it as ‘devoicing’. Wouldn’t ‘gemination’ mean that the vowel gets elongated? As I understand it any 3 kana combo should have the same length regardless if it contains つ or っ right?
As a beginner (and the OP), here is my take. The example I gave above:
しゅつしゃ → しゅっしゃ
Yes, there are 3 morae (hopefully I’m using that word correctly) in each case. And two things are happening:
The tsu sound “disappears” (guessing this is what you’re calling devoicing, but I don’t think that’s the right word to use)
The sha sound becomes twice as long (stealing its extra mora from the disappeared tsu)
Gemination is the word for a consonant being lengthened (#2).
But I don’t think gemination requires (or even includes) the disappearance of a preceding sound (#1). Which, if I’m correct, means that gemination is NOT a complete description of this mutation (merely of half of it).
To be more clear, gemination alone would do: しゅつしゃ → しゅつっしゃ (the sha was geminated). But our mutation is more complex than that.
And the phenomenon itself is more complex than the mutation. It is “mutation-of-unknown-name occurring when a jukugo word has a tsu at the end of one component”. So I think both the mutation and the phenomenon need a name.
You can have a geminate consonant (促音) where there was nothing replaced, as in a loanword like バット or ラッシュ, so I suppose that’s just 促音 and 出社 is an example of 促音化, like I was describing.
I think it might be sufficient. Though maybe it doesn’t quite fully include in its definition the conditions under which it occurs. But if that is the only case under which it occurs, then maybe that inclusion is implicit.
I mostly updated the first post for posterity, since I was describing the problem wrong.
And also I was mostly curious if there was a well known name already in common usage, similar to rendaku. Sounds like maybe not? In which case I nominate your suggestion until a better one comes along.
I guess you’re saying that 促音化 might cover this, but that it might also cover other things, and so there could be a word for just 促音化 in jukugo?
I think things get kind of messy when looking at native Japanese words where things are more likely to just be referred to as 音変化 (おんへんか, sound change), even if it involved a つ getting converted to a っ. Generally speaking, the words we’re talking about, jukugo with 促音化, use borrowings from Chinese.
Linguists might have hyper-specified words for these situations to avoid overlapping definitions (just imagining something like 音読性促音化 vs. 訓読性促音化, no idea if such things exist), so I guess you’d have to dive into linguistics papers to be sure.
rendaku is to voicing of consonants
as
______ is to sokuonka
But, voiced consonants do occur under many different conditions, so maybe that specialized name is required there. If sokuonka only occurs in one condition, it might not need another name that also covers the condition.
Exactly 6 months later (what a coincidence), I have answered this question. There was indeed a word to fill in the blank.
Backstory: I’ve been making more and more mind-blowing observations over the past few weeks about how useful ChatGPT is as a language learning tool. I realize some people have misgivings about using LLMs, but my experiences, to me, speak for themselves. And this is the latest one. I remembered this unanswered question from long ago, and decide to give ChatGPT a chance:
tl;dr the word is 促音便 (sokuonbin), and it differs from the suggested word above (促音化 or sokuonka) in that it only applies to the transformation if there are specific conditions surrounding it. So, sokuonka is the general transformation itself; sokuonbin is one specific instance of the transformation.
There are multiple references in the thread linked above. Two are:
I don’t know anything about 促音化 and 促音便, don’t have any background in linguistics, but I’m interested, so I’ve tried to read through this.
So, what are those conditions specifically?
I’ve read these several pages of ChatGPT output and still don’t understand what is, in its opinion, an example of 促音化 that is not 促音便.
It seems to me that it clearly contradicts itself within the same bullet point list.
A. “specifically lexically conditioned in compounds.”
B. “usually compound words”
So is in only compound words, or usually compound words?
Then it goes on to create a table that seems to suggest that 行った is 促音化 but not 促音便. It’s not clear if that’s the assertion it makes or not. If it is, why? Is this purely because 行った this is not a compound? Is it because it is not made for ease of pronunciation?
Let’s turn to linked sources. Japanese government source is unfortunately above my level to scan through and I’m a little lazy to OCR it at the moment to find specifically distinction between 促音化 and 促音便. But I can see that it’s dated 1983. Calling it “modern” is of course not technically wrong, it’s very much postwar, but ChatGPT should have provided publication year for context because terminology can easily change in 42 years.
Let’s consult a dictionary. 促音便: kotobank (translation mine, please point out and forgive any mistakes):
One type of 穏便, change for ease of pronounciation.
Specifically, changes of 「ち」「ひ」「り」sounds to っ in 連用形（conjunctive form), for example 待って（まちてー＞まって）。
More broadly speaking, also change to っ within nouns, for example 追手（おひて→おって）
That’s 2 more strikes against ChatGPT explanation:
ChatGPT’s table puts 連用形 outside of 促音便 scope
ChatGPT hasn’t even mentioned イ-row as a potential criteria
Let’s consult wikipedia for 促音便, obviously Japanese article, not English one.
Change of イ-row sound to っ is called 促音便.
…
促音化 of 「フ・ク・ツ・キ・チ」within 漢熟語 (Chinese-derived compound words) is sometimes mistakenly called 促音便. For example 一体（いち＋たい→いったい）
That’s one more strike against ChatGPT explanation:
3. ChatGPT said that all 促音化 are 促音便
In conclusion, as per usual, asking ChatGPT is great to find new keywords to search, but its explanations are not a cause to switch off critical thinking and ignore sources written by actual Japanese humans.
Thank you for your interest! I have a tendency to post topics digging into esoteric points that nobody wants to respond to (like this one). No judgment here, I realize my interests are weird.
Good question. Backing up a bit and making an analogy to make my point more clear (since I think it was also lost 6 months ago when I made the original “fill in the blank” question). Consider the difference between a “contraction” and an “apostrophe”. An apostrophe is a structural component. But a contraction is a specific structure that uses that component. The words can’t and John’s both use apostrophes, but only one of those is a contraction. Put another way: all contractions use apostrophes, but not all apostrophe usage occurs in contractions. In my question from 6 months ago, I was essentially looking for a word like “contraction”, but the answer I got was like “apostrophe”. More explicitly:
contraction is to apostrophe
(as) rendaku is to voiced consonant shifts
(as) sokuonbin is to sokuonka (aka gemination)
I posed that exact pattern 6 months ago with a ______ in place of sokuonbin.
As far as the actual conditions, I think they’re hard to describe succinctly in a way that captures every possibility (more on this later), but it is a fairly distinct and recognizable pattern that Wanikani lessons very often call out, in the same way they call out rendaku. Which is why it bothered me that they weren’t naming it in the same way they name rendaku. Read my OP way above, specifically the comments about www.wanikani.com/vocabulary/出社. You’ll see in the lesson where they call the pattern out, and note that it happens often.
The simplest description of the condition that I can come up with is that it is when you have:
a compound word (formed by multiple kanjis) like 出社 and
the ending of one kanji changes from something like (tsu, ku… and many others) to small tsu, i.e. the 出 in the word above changes from しゅつ to しゅっ
Meaning, the reading of the kanji is essentially changing, but only for the sake of the compound word junction. And learning the readings of kanjis is the entire point of Wanikani, which is why I was so damn interested in this!
Note that the phenomenon above is not the same as a word like 行った. That word does use sokuonka (aka gemination), but it’s not sokuonbin because:
it’s not a compound word (rather it’s a conjugated verb)
no kanjis are being read differently
Note that your kotobank link seems in sync with all of what I said above.
Addressing your 3 strikes one by one.
I think you’re mistakenly believing her to be claiming that the examples in the table are exhaustive. They clearly are not, since (1) she said that the table showed “common endings” for which it occurred (clearly implying there were other less common ones she was omitting) and (2) she explicitly goes on to mention that there are less common ones, and she offers to list them if we want (see strike 2 below).
I think you missed the part where she explicitly said:
If you want, I can make a more exhaustive list of the most common kanji endings that usually cause sokuonbin, including less obvious ones like ひ → っ in compounds.
Sorry but you have perfectly inverted her words. Here is what she said (at the end, in the summary):
All sokuonbin instances are 促音化, but not all 促音化 is sokuonbin.
This is a weird thing to throw in, particularly after you made the human error of misunderstanding most of what she said (all 3 of your strikes are demonstrably wrong). FWIW, I’ve found that conversations with LLMs have helped my critical thinking. For example, all 3 of your mistakes above could have been sorted out by continuing the conversation with the LLM:
For strikes 1 and 2, you could have simply asked “is this an exhaustive list”? Are there other kana rows involved? And she would have immediately listed the other options (as she already offered to do in my linked conversation).
For strike 3, you could have simply asked: are you saying that all sokuonka are sokuonbin? And she would have gently reminded you that she just said the opposite.
The answers would have been immediate, and she’s a lot more polite than me! I’ve found that open-ended questions are best, and it’s best to ask lots of questions so that you can compare the results against each other, and to things you already confidently know (and also, obviously, to actual reference material, which will never become obsolete).
and ignore sources written by actual Japanese humans.
Nor have I done that. In fact, I think your kotobank link strengthens my understanding here. Thanks for that!
I’m sorry, I just don’t see how this is in sync with kotobank.
The dictionary provides 待って, 歌った, 売ったり as examples of 促音便 and you are saying 行った is not 促音便.
Could you please elaborate? It feels like I’m missing something very obvious here.
You’re right… I think my understanding is still incomplete. I just had a few more targeted question sessions with both ChatGPT and Gemini. I’ve found a third term to throw into the mix: Kango no Sokuonka (which Gemini claims is for compound kanjis, while they say sokuonbin is for verb conjugations).
Every robot seems to be in agreement that sokuonka is the most general term for gemination. But they seem to disagree (and also claim that linguists disagree) about the finer subcategories, i.e. whether or not verbs and compound kanji phenomenons are part of the same subcategory, or two distinct things. And they also say that some educational materials refer to everything as -bin. None of this should be surprising I guess; robots aren’t the only thing that are inconsistent. I just 10 minutes ago finished applying for an AUS travel visa, and the application told me on one page that it couldn’t be completed online, and on the next page it said “click here to complete online”.
Like you, my Japanese skills aren’t nearly good enough to read native Japanese materials, but I may need to start trying my luck with translated webpages. I’ll report back if I figure anything out. Ironically my partner was born in Japan and spent the first half of her life there, but she has no idea about any of this (just like I probably have no idea about the correct names for esoteric English grammar constructs ).
I can definitely imagine this being the case. There could be competing schools proposing their own terminology and classification, or there could be evolution over time where one terminology supplants another. Or some simplified materials for non-specialist use.
I’m interested in etymology and more generally linguistics although I’m very much a complete amateur in these fields. My impression from browsing both English-language and, to a lesser extent, native Japanese articles is that these very technical terms in Japanese tend to be used with more flexibility than you would expect based on what happens in European linguistics.
I mean if a term as simple and common as “ateji” can be used variably to describe multiple completely different phenomena, can we really hope than these much more niche concerns will have a well established and well defined term to describe them? I gave up on that personally.
I think the reasons are multiple: Japanese linguistics is a relatively new field that’s heavily influenced by the more developed European linguistics, for better or worse. Also historically the literary language of Japanese is Classical Chinese (or some Japanese version of it). That means that, until recently, intellectuals were more likely to spend time picking apart Chinese texts than the Japanese vulgate.
As a result in my experience the Japanese language tends to have a surprising lack of jargon to talk about itself and the way it works, especially when it’s about describing something that doesn’t exist in Chinese or the “Western European” Sprachbund and therefore can’t be loaned directly.
Or I guess you could use it as another example of Japanese’s love of ambiguity…
).