Ok, so there seems to be some controversy over whether we can really say that there are 'の-adjectives', or whether we simply use a noun in an "attributive" way (a term which I don't actually really understand). But it's clear that there's something interesting going on here, a deviation from the "default" understanding of の (as marking the genitive case), and I'd like some clarification.

Let's look at what seems to be for whatever the classical example:

○ 「永遠の愛」 "eternal love"

It's clear that we can't just apply the pattern "X の Y ⇔ Y of X" here*. But what is really going on? How do we know that the usual pattern doesn't apply here - is it contextual? A matter of set-phrasing? Is it because 永遠, being abstract, would need to be reified to be used in the normal way?

* It's worth noting that the pattern doesn't even hold in English here, which does seem to be a special case.

"love of eternity" - a strange thing to talk about, but in English, reification is implicit so this works just as well as "love of gold".

But what is more strange is that with "love" in particular, this doesn't have the normal genitive-case meaning - an English speaker parses "love of gold" not as the love which is expressed by gold, but love such that gold is the thing that is loved. Similarly for "eternity", following the same role.

And it doesn't even work the same way with similar words... * "desire/lust of gold" - should be "desire/lust for gold". But regardless, a love which is eternal is not the same thing as a love which is expressed by (belongs to, really; but in what other sense can love be possessed than by expressing it?) eternity-seen-as-an-entity, which would be the default interpretation of the pattern.

What happens in Japanese with that example? 「金の愛」 - grammatical? How would it be interpreted? How about with explicit reification (「金のことの愛」・「金のものの愛」)?

And how literal is it to translate 「永遠の」 as "eternal"? Can we describe what's going on here more pedantically? What determines our ability to use a noun this way with の?

And am I getting too philosophical? :)

  • 3
    Don't expect a one-to-one correlation between languages. Languages are different.
    – user458
    Aug 21, 2011 at 0:03
  • 12
    Of course not; the point is to understand the difference more properly. Aug 21, 2011 at 0:21
  • 2
    According to dictionaries, the English preposition "of" has 17 different meanings and the Japanese case particle "の" has 20 (=14+1+1+1+3) differnt meanings. If you were taught the 20 meanings of "の" individually, would you be satisfied?
    – Gradius
    Jul 30, 2012 at 22:47
  • 2
    There is such thing in English as "attributive nouns", nouns that modify the nouns following them as if they were adjectives (look for "noun adjunct" on Wikipedia). "First-class flight", "communications system", etc. Grammatically there's no point to distinguish these nouns from adjectives in English because modern English grammar has no inflection for adjectives (adjectives never change form), but in languages like Japanese, or even English's close relatives like German, the grammatical difference must be taken note of. Dec 25, 2021 at 5:22
  • 2
    "Attributive" means it gives "attributes", quality, character, what have you, although strictly speaking an attributive adjective or noun must be close to the noun it modifies without being separated by a verb like "to be", in which case the adjective or noun is called "predicative". "Red" in "a red balloon" is attributive, but "red" in "the balloon is red" is predicative. Dec 25, 2021 at 5:26

3 Answers 3


As I understand it, the term “no-adjective” simply means “nouns which are typically translated to adjectives in English and other languages.” If we treat Japanese as a language in its own right, distinguishing them from nouns as different parts-of-speech is completely artificial.

The particle の makes a modifier of a noun. The exact relationship between the modifier and the modified noun can be almost anything; Daijisen lists fourteen relationships such as ownership, belonging, location of existence, location of action, time, and so on, and one of them is “attribute and condition.”

[瀕死]{ひんし}の[重傷]{じゅうしょう} a life-threatening injury
[縦]{たて}じまのシャツ a shirt with vertical stripes

(The examples are from Daijisen, the English translation of the first example is by FumbleFingers on english.stackexchange.com, and the translation of the second example is by me.)

The の in 永遠の愛 is the same thing. The noun [永遠]{えいえん} (eternity) is turned to a modifier 永遠の, and it describes an attribute of the love.

  • 17
    @Karl: Because の does not specify the exact relationship, we figure out the relationship from the meaning and the context. For example, 校長の話 can be both “story by the principal” or “story about the principal” depending on the context. Aug 20, 2011 at 23:43
  • 1
    @Karl: I would translate “love of gold” to 金に対する愛 or 金を愛すること. None of 金の愛, 金のことの愛, and 金のものの愛 is understandable. Aug 21, 2011 at 11:35
  • 5
    @Karl: I am not a language teacher and am just answering your question. At least these examples show that “XのY ⇔ Y of X” is not a good correspondence in either direction. Aug 21, 2011 at 18:37
  • 3
    @Karl Knetchel: "no" is ambiguous, there's nothing you can do about it. So is "of" in English: "the love of a mother" could mean either that "a child loves his mother" or that "a mother loves her child".
    – Axioplase
    Aug 23, 2011 at 2:02
  • 10
    @Karl Languages were not defined by rules. This is important. We will use an expression if it works in a certain context. Although languages are ambitious by nature, native speakers know commonly used patterns, for example, we never say "金の愛". This is the same as all languages including English. You probably counldn't figure out the meanings of fireball, firearm, firewall and fireball if you didn't know their meanings.
    – Gradius
    Jul 30, 2012 at 22:21

Two things to consider here I feel. The first one is analysing “〜の” as a genitive per sē rather than a generic adnominal marker. It should be noted that “〜の" can follow many more things than nouns, as “〜だ” can and it's commonly analysed as one of the two adnominal forms of “〜だ" the other being “〜な”. Indeed, we can also for instance say:

  • “食べただけの人” - “A person who merely ate”
  • “処刑するほどの罪” - “A crime worthy of execution”
  • “破壊神がベルス様の宇宙” - “The universe where Lord Beerus is the destroyer god”, note how here “〜の” follows an entire clause that has it's own subject, really showing how it's an adnominal form of “〜だ”
  • “子供の社長” - Typically interpreted as “a child company head” not “the company head of a child”

So it would be a mistake to begin with to assume that it's the non-genitive function of “〜の” that gives rise to the idea of “no-adjectives”. Some might however argue that it's merely a matter of translating it to adjectives in English and that they are nouns. That I don't agree with; they seem to grammatically behave as adjectives in Japanese. To use the common example of “病気の人”. The major thing that's going on here is that “病気” can be modified by adverbs. We can for instance say “少し病気の人". We can't say “少し子供の社長” This sounds as odd as “a company head who is slightly child”. “slightly child” does not make sense and neither does “少し子供”. “子供” is after all a noun, not an adjective, and it thus cannot be modified by an adverb and we need something like “少し子供っぽい”, but “病気” seems to be able to function as both a noun and an adjective.

  • I'm pretty sure that 12 years ago I was trying to understand the semantics of の itself, but this was interesting to me today, in that it gave me insight into why J-E dictionaries list some words explicitly as "no-adjectives" instead of or in addition to "nouns". Dec 3, 2023 at 10:51
  • @KarlKnechtel I kind of wish people never explained “〜の” to mean “of” and “〜だ” to mean “to be” to me. I can remember being so confused by how they actually functioned until someone told me to simply see the former as an abstract adnominal marker and the latter as an abstract conclusive marker and suddenly things such as “もういないかもだけど” made complete sense.
    – Zorf
    Dec 3, 2023 at 11:13
  • 1
    Well, that's fine for linguists, but... Dec 3, 2023 at 11:15
  • There are also sentence-initial constructions なのに…, なので…, (だ/です)から…, (だ/です)が…, (だ/です)としたら…, (だ/です)としても… etc., in which な, だ, です are only used as some "empty" verbs (with no semantic meaning), since they are followed by words, which require any preceding verb.
    – Arfrever
    Dec 3, 2023 at 14:16
  • 1
    病気 is very anomalous, in that it seems be an adjective to the extent it can be modified by adverbs (大変病気だ), but *病気な子供 is unacceptable---病気の子供 is the correct form. And it is a noun in 病気が悪化した. Further, unlike na-adjectives which can be adverbialized with -ni, 奇麗に, 平等に, there is no *病気に to mean 'in a sick manner'. 病気の人 is a relative clause, deriving from 人は病気だ, as explained. To make the structure explicit, 'a person who is sick', semantically equivalent to 'a sick person'.
    – N. Hunt
    Dec 4, 2023 at 1:43

The notion of 'no-adjectives' was conceived by Mio Isago in 1942; 三尾砂、話し言葉の文法. I would like to say it is 'ill-conceived' but Mio was writing at a time when Japanese linguistics was still in its infancy. Still, considering he was positing a category that doesn't even exist in traditional Japanese grammar, it is puzzling. The impetus may have come from pedagogy, the need to teach Japanese to foreigners, that was probably starting to expand. The misconception was taken up many years later by, for example, 森田千草 in 2010 ('The Internal Structure of Adjectives in Japanese'), and it seems some modern dictionaries, and dictionaries of grammar, for foreigners, teach it.

The idea is based on the lack of understanding of Japanese syntax in relation to the construction of NのN.

の is the form of the copula だ, when it is adnominalized (in Japanese, 修飾語化?). In English, we might have

An ancient mariner

and we can derive this from an underlying

The mariner is ancient.

Similarly, in Japanese, we can posit this derivation:

老齢の水夫 ← 水夫は老齢だ

but whereas in a normal relativization, the verb doesn't change form, e.g.,

鈴木さんはうそをつく ➞ うそをつく鈴木さん

だ obligatorily changes to の

水夫は老齢だ ➞ 老齢だ水夫 ➞ 老齢の水夫

This is fine for an underlying sentence where the copula is just that, 'is'. But we know that だ can act as a substitute for a whole predicate (奥津敬一郎 (1978) 「ボクハウナギダ」の文法).

Okutsu's analysis of his well-known sentence, 僕はうなぎだ, is:


where だ is substituted for the bracketed material. This substitution of だ for part of a predicate is called だによる述部代用化, translated as 'propredication' in English.

A canonical example of a 'no-adjective' is 瀕死. The frequently found expression,

瀕死の重傷, 'a fatal wound'

in fact has an underlying structure like

重傷は瀕死[を引き起こした] 'The serious wound has given rise to a near-death state.'

if we make a parallel with Okutsu's analysis, and after だ substitution and relativization:

重症は瀕死だ ➞ 瀕死だ重症 ➞ 瀕死の重傷

What is behind this is well-explained by the linguist John Haig:

'Japanese has a very general process of redundant information deletion which
 deletes information predictable from the linguistic or non-linguistic context.'

(Yes, 瀕死 is a noun: from a medical reseach paper, その他の動物に死亡あるいは瀕死は認められなかった. And 瀕死 means '(the state of being) near death', not 'fatal' which is 致命.)

In the construction NのN, where we do not have with a genitive (東京の名所, "Tokyo's sights"), or の marking the subject of a relative clause (ga-no conversion), then we have a relative clause made from an underlying sentence, which has an adjective-like function.

  • 1
    Analysing の as a form of だ just leaves me wondering why both の and な exist. As for the eel sentence, I think the natural interpretation is 僕は[注文したことが]うなぎだ. Since that leaves the predicate alone, and in particular doesn't require un-gluing from a copula and re-gluing to a particle. It's nice to see a concrete example of what you were calling "propredication" earlier; but I don't find the analysis convincing. Dec 3, 2023 at 10:57
  • 1
    In this way, you could even analyze all な-adjectives as nouns meaning "(the state of being) ...", e.g. 綺麗 = "(the state of being) beautiful"... (I agree that の (when it is not case particle or nominalizer) and な are forms of copulae, as are に, にて・で, と (when they are not case particles; one of と (now sometimes called "quotative particle") was also analyzed by Alexander Vovin as another defective verb).)
    – Arfrever
    Dec 3, 2023 at 12:36
  • Well, yes, in that na-adjectives (and generally 形動なり and 形動たり) are followed by a copula, you could see these phrases as relative clauses. In fact, na-adjectives were called 'copula nouns' in the earliest studies in Japanese linguistics, for this reason I suppose. My English example ('an ancient mariner') was a contrived, and badly so (should have been 'a mariner who is ancient'), but it was meant to show how a relative clause with a copula can function as a simple adjective.
    – N. Hunt
    Dec 3, 2023 at 23:38

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .