The Nanbanjin Nikki

ザ南蛮人日記

Dare-ka and dare-mo

One thing that always bothered me in Japanese language pedagogy is how they teach quantifier words, such as dareka = “who-ka” or daremo = “who-mo”: They’re just translated atomically (in this case as “someone” and “everyone”), with no further discussion. However, it’s clear that those -ka and -mo particles perform a set role when added to question words:

dare-ka who-KA someone
nani-ka what-KA something
itsu-ka when-KA sometime
ikutsu-ka how many-KA some, a few
dare-mo who-MO everyone
nani-mo what-MO everything
itsu-mo when-MO always
ikutsu-mo how many-MO a lot

What’s more, there was something less obvious that kept nagging me in the back of my mind: I felt like these uses of ka and mo somehow aren’t so different from their usual roles – ka as the question particle, and mo as the “too” particle. I thought “who + too = everyone” made sense, somehow, but I couldn’t quite put my finger on why.

Turns out that this relationship which nagged me is widespread cross-linguistically; particles much like ka and mo are found in Hungarian, Russian, Sinhala, Malayalam and many other languages from entirely unrelated families, and they perform mostly the same roles (e.g. “who” + question particle = “someone”). There seems to be a flourishing research program on this topic, in keep with a tendency in many quarters to be skeptical of “words” as basic units, and to better analyze how meaning is created from word parts. This post is an informal appetizer adapted from examples by Szabolsci, Whang and Zu; readers who like technical discussions should check out the links in the bottom.

The fundamental insight is as follows:

But existential quantifiers can be thought of in terms of disjunctions or set unions: “there is a x” means “x₁ OR x₂ OR x₃ OR…”. Similarly, “for any x” means conjunctions or intersections: “x₁ AND x₂ AND x₃ AND…”. So it follows that:

Of course, mo isn’t the basic, neutral “and” in Japanese; this role belongs to to (mo is for distributive conjunctions). But let’s see some examples below to understand what’s meant by AND operations in the context of mo, and OR in the context of ka.

i. Tanaka ka Fujita ka

“Tanaka or Fujita”. KA marks a disjunction as we expected. (Notice however that Japanese may use two ka particles to point to a single disjunction operation.)

ii. Tanaka ga odorimasu ka.

“Does Tanaka dance?”. A yes-or-no question. How do we define a “question”, in semantic terms? In this case, perhaps something like:

a) The following are our possibilities: Either Tanaka dances, OR Tanaka doesn’t dance (think of the idiom: ka dō ka…).
b) Please choose one of the possibilities.

Notice that this definition, crucially, makes use of disjunction: one alternative OR the other. So it makes sense that KA particles are used for both straight disjunctions, and for yes-or-no questions.

iii. Dare ka

“Who-KA” = “Someone”. The question word “who” works like a variable, since in principle any of the things (persons) under consideration may be the “value” of who (the whodunit). With “someone”, we assert that one of those possibilities is true. So this is just like “there is an x in the set {Tanaka, Fujita, Takeda…} so that x”, i.e. {Tanaka OR Fujita OR Takeda…}

iv. Dare ga odorimasu ka.

“Who dances?”. This works much like a yes-or-no question (ii. above), except that “who” opens more alternatives to choose from (as in iii.): {Tanaka dances OR Fujita dances OR Takeda dances OR (Tanaka AND Fujita) dance OR (Tanaka AND Fujita AND Takeda) dance OR…}. Still disjunctions.

v. Gakusei no dare-ka

“Who-KA of the students” = “Some student” = “one of the students” = there is an x (who) in the set of students {Student1 OR Student2 OR…}

vi. Jūnin to ka “Ten people and KA” = “approximately ten” = either ten OR eleven OR nine OR twelve OR…

Now let’s check out MO:

vii. Tanaka mo odorimasu.

“Tanaka, too, dances”. This implies that 1) there is an unspecified someone who dances, and 2) Tanaka dances. In other words, someone (some x) dances AND Tanaka dances.

viii. Tanaka mo Fujita mo odorimasu.

“Both Tanaka and Fujita dance”. This is just like vii., but this time each MO satisfies the other: Tanaka AND someone else (=Fujita) dance, Fujita AND someone else (=Tanaka) dance.

ix. Dare mo odorimasu.

“Who MO dances” = “Everybody dances”. As one would expect, this is a conjunctive version of the general disjunction in iii above (dare ka). Equivalently, just generalize viii. to everyone in the set (because anyone can satisfy “who”): Tanaka dances AND someone else = Fujita dances AND someone else = Takeda dances AND…

It makes sense!

Japanese mo has a further, emphatic role as “even”, “as many as”; this extra role is marked specifically by its own prosodic stress. Some (many?) MO-family particles in other languages also have this extra sense (like Hungarian is and Mandarin dōu), in addition to the equivalents of the above. The “even” relation can be thought of as a particular case of conjunction: “Even Tanaka dances” does imply that Tanaka dances AND someone else dances, plus the additional fact that Tanaka isn’t expected to dance.

The above discussion is a gross oversimplification of the following papers:

Both articles compare examples from several languages. In the latest one, Szabolcsi starts by defining KA and MO as markers of lattice-theoretical “join” and “meet” operations (the generalizations of conjunction/intersection and disjunction/union), then proceeds to formalize them in terms of Inquisitive Semantics. Neat stuff.

Comments

Wow! I am astonished by how much it all makes sense.
My brain just exploded a little from reading this post. Thanks Leo!

By Claudio Freitas on .

I find this line of thought very interesting (and, like you, had the frustrating sensation of seeing some pattern behind the WH-ka/mo sets but not quite being able to elucidate it), but I would not really include sentence-final “ka” in the same basket.

Historically speaking though obviously there is a connection, and in fact that’s where I would focus the research. e.g. in the MYS “tare ka sumapamu” means “who lives/would live there?” So the question is, what happened between OJ and 2014 to allow the “WH-ka” quantifying forms to arise when they were originally genuine questions? Presumably it has something to do with the migration of question-ka to the end of the sentence, but…

By Matt on .

I’m still getting used to Classical (nevermind OJ) and most material focus on morphology and vocab, with little attention to syntax (kakari-musubi excepted, natürlich). Thanks for bringing this to notice, it’s a neat difference. Does this -ka work like the interrogative particle -(y)pe in Old Tupi, which attaches to whatever part of the sentence one is asking about? E.g.:

Peró-supé-pe endé ere-nhe’eng?
Portuguese-DAT-INTR you 2ND-speak?
“Did you speak to the Portuguese?

Peró-supé endé-pe ere-nhe’eng?
Portuguese-DAT you-INTR 2ND-speak?
“Did you speak to the Portuguese?

Peró-supé endé ere-nhe’eng-ype?
Portuguese-DAT you 2ND-speak-INTR?
“Did you speak to the Portuguese?

Regarding the sentence-ending -ka: Szabolcsi notes that in Hungarian the sentence-wide question morpheme, (-e), is distinct from the morpheme that takes the other KA roles (vagy). (This is also the case for the MO roles: there are two distinct particles, mind and is, which taken together work just like Japanese mo).

However, there’s a vagy-derived marker, vajon, used for certain “puzzlement” questions; and the question function does appear in many other KA-family particles, including Malayalaam -oo, Sinhala ə. and Tlingit (for wh-questions). So I’m still inclined to think that “it makes sense” to reuse the same particle to 1) build “someone” from “who”, and 2) mark questions (because all the kids are doing it). If you’re right in that non-quantifying -ka turned into quantifying ka at about the same time as sentence-final ka appeared, then Japanese seems to have evolved herself to fit this pattern!…

By leoboiko on .

Does this -ka work like the interrogative particle -(y)pe in Old Tupi, which attaches to whatever part of the sentence one is asking about?

Yeah, that was the usage I’m thinking of. Pardon my not differentiating otsu and ko syllables, but here are some examples:

– hitotu matu, iku yo ka henuru … (MYS 1042)
– simogumori su to ni ka aru ramu … (MYS 1083)
- iduku ni ka ware wa yado ramu … (MYS 0275)
– Yamato ni pa nakite ka ku ramu/ yobu kodori … (MYS 0070)

All of these in theory affect what they follow: “How many generations?” rather than “A few generations”. But this may be a bit fuzzy, e.g. does the last really mean “Is it calling that [the birds] come to Yamato?” or is it more naturally interpreted with a wider scope like “Do the birds come calling to Yamato?” with “ka” attaching to “nakite” as a representative of the whole clause?

Of course, “ka” was also used at the end of sentences (especially in combined forms like “ka mo” and “ka pa”, but not exclusively: quite a few poems end with “… mo aru ka”. But on the gripping hand, a lot of those final “ka”s aren’t actually questions but more general lamentations or exclamations, and it’s difficult to pin down whether the author genuinely, literally meant a rhetorical question like “Is it the case that (something melancholy)? [Yes, for lo it is so]” rather than just “Lo, (something melancholy)”.

Note also that mid-sentence “ka” is part of the kakari-musubi complex — so even far from the sentence’s end it was still messing around there somewhere. (In fact I assume it was the decline of KM that caused question-ka to migrate to the end of the sentence, if only because it’s usually a safe bet to blame any post-Heian changes involving the end of the sentence on the decline of KM.)

By Matt on .

Yeah, I also felt that it made sense to add “ka” with other stuffs to put an idea of unknown. So, if I wanted to mention a person who I don’t know who is, I put together a “question” particle (“ka”) or undefined reference prefix (“some”) together with the word for a individual (“dare”, still used mainly in questions, or “one”).
From a lay perspective, it is.