<entry id="2308129" stat="R" corpus="jmdict" type="jmdict">
<ent_corp type="jmdict">jmdict</ent_corp>
<ent_seq>1547330</ent_seq>
<k_ele>
<keb>欲しい</keb>
<ke_pri>ichi1</ke_pri>
<ke_pri>news1</ke_pri>
<ke_pri>nf18</ke_pri>
</k_ele>
<r_ele>
<reb>ほしい</reb>
<re_pri>ichi1</re_pri>
<re_pri>news1</re_pri>
<re_pri>nf18</re_pri>
</r_ele>
<sense>
<pos>&adj-i;</pos>
<xref type="see" seq="1611550">欲しがる</xref>
<gloss>wanting (to have)</gloss>
<gloss>desiring</gloss>
<gloss>wishing for</gloss>
</sense>
<sense>
<pos>&adj-i;</pos>
<pos>&aux-adj;</pos>
<misc>&uk;</misc>
<s_inf>after the -te form of a verb</s_inf>
<gloss>I want (you, them, etc.) to (do)</gloss>
</sense>
<info>
<audit time="2011-05-07 14:57:47" stat="A" unap="true">
<upd_name>Nils Roland Barth</upd_name>
<upd_email>...address hidden...</upd_email>
<upd_detl>Link to verb</upd_detl>
<upd_diff>@@ -18,0 +18,2 @@
+&lt;xref type="see" seq="1611550"&gt;欲しがる&lt;/xref&gt;
+&lt;xref type="see" seq="1611550"&gt;欲しがる&lt;/xref&gt;</upd_diff>
</audit>
<audit time="2011-05-08 08:31:15" stat="A">
<upd_uid>jwb</upd_uid>
<upd_name>Jim Breen</upd_name>
<upd_email>...address hidden...</upd_email>
</audit>
<audit time="2013-02-17 22:49:16" stat="A">
<upd_uid>rene</upd_uid>
<upd_name>Rene Malenfant</upd_name>
<upd_email>...address hidden...</upd_email>
<upd_detl>-&gt;note</upd_detl>
<upd_diff>@@ -26,1 +26,2 @@
-&lt;gloss&gt;(after the -te form of a verb) I want (you) to&lt;/gloss&gt;
+&lt;s_inf&gt;after the -te form of a verb&lt;/s_inf&gt;
+&lt;gloss&gt;I want (you) to&lt;/gloss&gt;</upd_diff>
</audit>
<audit time="2015-05-22 10:42:46" stat="A" unap="true">
<upd_detl>the ratio seen in the n-grams is completely different from my experience. would have expected the kana form to handily outnumber the kanji variants.</upd_detl>
<upd_refs>n-grams
がほしい	49724
が欲しい	177030
してほしい	168004
して欲しい	155575</upd_refs>
<upd_diff>@@ -25,0 +26 @@
+&lt;misc&gt;&amp;uk;&lt;/misc&gt;</upd_diff>
</audit>
<audit time="2015-05-23 12:48:51" stat="A">
<upd_uid>jwb</upd_uid>
<upd_name>Jim Breen</upd_name>
<upd_email>...address hidden...</upd_email>
<upd_detl>Blame IMEs which dish out the kanji.</upd_detl>
<upd_refs>Google n-grams (2007 data) is similar.</upd_refs>
</audit>
<audit time="2021-11-16 06:05:02" stat="A" unap="true">
<upd_uid>Marcus</upd_uid>
<upd_name>Marcus Richert</upd_name>
<upd_email>...address hidden...</upd_email>
<upd_detl>Like 好き/嫌い - essentially impossible to gloss as an adjective.

"wanted;desired" etc. are passive voice, but this is really active, "_I_ want"
we could have a note on this only being used in reference to oneself or in questions posed to others</upd_detl>
<upd_refs>gg5 "want; desire; wish (for)" etc.  - all verbs</upd_refs>
<upd_diff>@@ -21,4 +21,2 @@
-&lt;gloss&gt;wanted&lt;/gloss&gt;
-&lt;gloss&gt;wished for&lt;/gloss&gt;
-&lt;gloss&gt;in need of&lt;/gloss&gt;
-&lt;gloss&gt;desired&lt;/gloss&gt;
+&lt;gloss&gt;wanting (to have)&lt;/gloss&gt;
+&lt;gloss&gt;wishing for&lt;/gloss&gt;</upd_diff>
</audit>
<audit time="2021-11-18 01:52:30" stat="A" unap="true">
<upd_uid>robin1354</upd_uid>
<upd_name>Robin Scott</upd_name>
<upd_email>...address hidden...</upd_email>
<upd_detl>Not sure about a note. The same rule applies to any adjective that describes a feeling.</upd_detl>
<upd_diff>@@ -21,0 +22 @@
+&lt;gloss&gt;desiring&lt;/gloss&gt;
@@ -28 +29 @@
-&lt;gloss&gt;I want (you) to&lt;/gloss&gt;
+&lt;gloss&gt;I want (you, them, etc.) to (do)&lt;/gloss&gt;</upd_diff>
</audit>
<audit time="2021-11-18 03:06:39" stat="A">
<upd_uid>Marcus</upd_uid>
<upd_name>Marcus Richert</upd_name>
<upd_email>...address hidden...</upd_email>
<upd_detl>Ah, that's true.</upd_detl>
</audit>
<audit time="2022-04-25 02:15:55" stat="A" unap="true">
<upd_name>Stephen Kraus</upd_name>
<upd_email>...address hidden...</upd_email>
<upd_detl>Aligning sense part-of-speech information for auxiliary adjectives</upd_detl>
<upd_diff>@@ -23,0 +24 @@
+&lt;pos&gt;&amp;adj-i;&lt;/pos&gt;</upd_diff>
</audit>
<audit time="2022-04-25 06:11:17" stat="A">
<upd_uid>jwb</upd_uid>
<upd_name>Jim Breen</upd_name>
<upd_email>...address hidden...</upd_email>
</audit>
<audit time="2024-07-10 02:22:58" stat="A" unap="true">
<upd_name>Non</upd_name>
<upd_detl>Arguing for a reversion of 2021-11-16 gloss change; it seems to have been based upon GG5's translation, which is made with the goal of conveying the meaning with a natural translation, not the most accurate one. JWN's entry supports the previous glosses.
Now, the current gloss of sense 1 has a verbal translation for an adjective - I think this by itself would be rather difficult and confusing for an inexperienced reader to reconcile as the english verb has its subject as the experiencer (the one who wants) while 欲しい has its subject as the stimulus (that which is wanted). 

In regards to voice in ほしい: I would not regard it as active nor passive; active and passive voices are features of verbs and the thematic roles of their participants in regards to which one takes the position of subject. 
Meanwhile ほしい is an adjective and its subject is fixed as the stimulus, with the experiencer being either omitted entirely or marked as a topic by は, or less commonly by に. 
In regards to the glosses themselves: they are participles, whether they are called passives in a particular instance depends partly on your choice of terminology. Nonetheless, be it stative passives or adjectival participles their semantic-to-syntax configuration matches that of ほしい (with the stimulus either as its subject or modified by it), which should make them more appropriate and clearer translations.

Desired thing = Participle modifies stimulus
That is desired = Stimulus as subject of participial predicate
ほしいもの = Adjective modifies stimulus
食べ物がほしい　= Stimulus as subject of adjectival predicate</upd_detl>
<upd_refs>JWN</upd_refs>
<upd_diff>@@ -19,3 +19,3 @@
-&lt;gloss&gt;wanting (to have)&lt;/gloss&gt;
-&lt;gloss&gt;desiring&lt;/gloss&gt;
-&lt;gloss&gt;wishing for&lt;/gloss&gt;
+&lt;gloss&gt;wanted&lt;/gloss&gt;
+&lt;gloss&gt;desired&lt;/gloss&gt;
+&lt;gloss&gt;wished&lt;/gloss&gt;</upd_diff>
</audit>
<audit time="2024-07-15 01:58:44" stat="A" unap="true">
<upd_uid>jwb</upd_uid>
<upd_name>Jim Breen</upd_name>
<upd_email>...address hidden...</upd_email>
<upd_detl>This one is not simple. Of the 550+ sentences we have linked to this entry, 120+ have "want" associated with it and 5 have "wanted". I think attempting to align the glosses with the POS is obscuring things. 
My preference is either to leave it as it is. We could change it to be more like the JEs but then a note such as M&amp;T's might help.</upd_detl>
<upd_refs>JE gloss summaries:
GG5: want; desire; wish [care, long] for…; lust 「after [for]….  (Ex: 欲しいもの a want; a thing desired; an object of desire)
中辞典: want; desire
ルミナス: want, would like. hope 
Makino and Tsutsui (B) pp144-5. - "expresses a person's desire for some object".</upd_refs>
</audit>
<audit time="2024-07-15 17:15:27" stat="A" unap="true">
<upd_name>Non</upd_name>
<upd_detl>Agreed that it is not simple, adjectives of subjective emotion are tricky - but I propose this not to match the glosses to the part of speech but because I believe 'wanted' clarifies what property is being attributed to the ガ各(or modified noun) and facilitates a more accurate translation. 

I believe this is actually reinforced by the example translations that have it as 'want' in english: while translating 食べ物がほしい with a gloss of "wanted" would render "Food is wanted" which is semantically equivalent to a given example of "I want food", doing so with a gloss of "want" gives us "Food wants" which is not. In the latter case, one must account for the mismatch by either shifting the functions of the particles or the meaning of the gloss itself, at which point I think it has failed as a gloss.

If this proposal is still to be rejected I would be in favour of at least changing it to 'to want' rather than 'wanting' - I struggle to see any benefit the latter has over the former and any translation with it sounds extremely unnatural compared to the alternatives; still, I would rather have only the participle or both.

A note would be all around useful regardless of what is chosen, it is welcome clarification and also helps with the edge case of something like 私が欲しいもの in which the experiencer is actually marked by が.</upd_detl>
</audit>
<audit time="2024-07-24 11:33:08" stat="A">
<upd_uid>jwb</upd_uid>
<upd_name>Jim Breen</upd_name>
<upd_email>...address hidden...</upd_email>
<upd_detl>I've been mulling this over, and hoping that someone else will join in.
What I plan to do it set the glosses for sense 1 back to the ones Marcus set in 2021. I think they work well. Yes, the JEs all have the verb-like "want; desire; wish", etc. but they really don't work well with an adjective. Making them "to want", etc. is going even further. The first gloss really avoids having a note.
I'll reopen it for a while.</upd_detl>
<upd_diff>@@ -19,3 +19,3 @@
-&lt;gloss&gt;wanted&lt;/gloss&gt;
-&lt;gloss&gt;desired&lt;/gloss&gt;
-&lt;gloss&gt;wished&lt;/gloss&gt;
+&lt;gloss&gt;wanting (to have)&lt;/gloss&gt;
+&lt;gloss&gt;desiring&lt;/gloss&gt;
+&lt;gloss&gt;wishing for&lt;/gloss&gt;</upd_diff>
</audit>
<audit time="2024-07-24 11:33:24" stat="A" unap="true">
<upd_uid>jwb</upd_uid>
<upd_name>Jim Breen</upd_name>
<upd_email>...address hidden...</upd_email>
</audit>
<audit time="2024-07-25 03:10:28" stat="A" unap="true">
<upd_name>Non</upd_name>
<upd_detl>Minimising this mismatch is precisely why I suggested the past participle; 'to want' is definitely off, I rambled about it in my first commentary - but so is this. 
'Wanting' is the present participle, it preserves all the verbiness of 'want' and with it the same problems: 食べ物が欲しい to "Food is wanting (to, of, for)" yields us the same semantic discrepancy, and sounds awkward as it is not normally used with stative verbs of emotion like this.
It somewhat works only if we take this to be its nominal use: "Food is (a) wanting (to, of, for)" but this is incumbent upon the reader to correctly select this interpretation, which I do not think is very likely, I do not see what suggests that they choose it over the former and needing to fill out the prepositional phrase afterwards makes it rather odd as a translation since 欲しい needs no further specification.
Given all this, I do not see how it avoids any note, both possible interpretations are awkward and one is outright wrong.

Meanwhile 'wanted' has only one possible interpretation in which it correctly assigns the thematic roles, has same number of arguments as 欲しい and would only need a note if we were to specify the one case where it behaves anomalously - which is not intrinsic to it and would need an additional note if explained with any other gloss.</upd_detl>
</audit>
<audit time="2024-07-25 08:59:49" stat="A" unap="true">
<upd_name>Alan Cheng</upd_name>
<upd_email>...address hidden...</upd_email>
<upd_detl>To throw in my two cents:

From a syntax perspective, in most situations in Japanese, the case marker corresponds to the grammatical function (e.g., nominative が marks the subject, accusative を marks the direct object). But I agree with the view that 欲しい (along with 好き and others) is an exception where the nominative case marker が actually indicates the direct object (Hasegawa 2014). 

For what it's worth, other examples of such "case alternation" occur with potential verbs, where the direct object can be marked with either が or を, and with motion verbs, where the location of motion--not the direct object--is marked with を.

From a cognitive perspective, I would argue that 欲しい *feels* transitive, so I agree with Marcus's 2021 comment that it should have an active gloss. This mainly a feeling from personal experience, as I don't feel that the glosses "wanted"/"want-able"/"desirable" accurately convey the semantics of 欲しい compared to the more transitive "want". 

If I say 肉がほしい, I mean "I want fish", and not "Fish is wanted" (which is semantically incorrect, as the want-er is definitely the speaker). In my interpretation, this is a sentence whose subject is not directly referenced and whose predicate ほしい acts on a direct object 肉 with the case marker が. One could argue for the interpretation "Fish is wanted (by me)", but this feels stilted to me.

There are also plenty of examples in both literature and spontaneous speech where native speakers replace が with the "nonstandard" を when using 欲しい, which may suggest a similar stance that 欲しい is an transitive predicate. For instance, Murakami in his novel, The Wind-Up Bird Chronicle (1994): 「十分だけでいいから時間を欲しいの。」(see Schaanning 2019 for more examples).

So I'm in favor of "want" or "wanting" rather than "wanted" for the glosses. I see the potential value in using "wanted" as a gloss since it's easier to explain to a learner why 欲しい takes the case marker が and makes POS alignment between Japanese and English easier, but I think it's less faithful to how 欲しい is used and perceived. 

All told, there is no definitive solution to this: Hasegawa admits, "At one time, I [...] equated nominative [が] with subject, but I am now convinced that accepting a nominative-marked direct object is more appropriate [...] The reader can likewise select either stance because both are equally persuasive and equally problematic."</upd_detl>
<upd_refs>Hasegawa, Y. (2014). Japanese: A linguistic introduction. Cambridge University Press.

Schaanning, J. G. (2019). Case-alternation with Japanese adjectives: A cognitive transitive perspective (Master's thesis).</upd_refs>
</audit>
<audit time="2024-07-25 16:30:33" stat="A" unap="true">
<upd_name>Non</upd_name>
<upd_detl>I would like to provide some counterpoints.

On case alternation other than 欲しい, etc.
The translative を does not qualify, it is used specifically with intransitive motion verbs to mark a path traversed - it is semantically distinct from accusative を. Here we have a particle taking upon another function that is not present elsewhere; with 欲しい, what is being proposed is that a particle suddenly takes the function of another particle for seemingly no reason. 
Translative and accusative を have much better parallels in locative and dative に.
What happens with the (ら)れる・できる potential forms is much more similar, but it can still be inferred from their historical usage and etymology that they were originally structured with が over that which is possible to be acted upon, and the actor with に - they do not originally take objects and I think it a mistake to call the が格 an object.

On transitivity, semantics and syntax.
"Fish is wanted" and "I want fish" are semantically equivalent. You seem to be conflating semantic/thematic roles and syntactic roles, they are related but not the same; across those two sentences the thematic roles remain the same, it is only the syntactic roles that change in regards to who is the subject.
That 欲しい feels transitive, I suspect comes from it having two thematic roles that you would expect of a transitive verb: stimulus and experiencer.
In the two sentences above, the fish is the stimulus and I/me is the experiencer; in European languages, we commonly have the experiencer as the subject and the stimulus as the object of a transitive verb, that is the case of the english 'want' - but Japanese does not do this, it has the stimulus as the nominative resulting in an intransitive word while the experiencer is either omitted, topicalised, or marked by the dative.

As for the case alternation in 欲しい、嫌い and 好き.
I advocate for disregarding this entirely. Though you can occasionally find someone who uses を欲しい most do not. Acceptability tests also consistently rank the use of accusatives with those adjectives lower than the standard nominative; and if you go asking about you will find someone who opposes those structures after not too long - particularly so for を欲しい - but I do not think you could find someone who considers the usual が as ungrammatical.
Those uses are an anomaly and should be held to be an anomaly, the grammar of adjectives does not license their existence.

Lastly, が as a grammatical object.
It is unfalsifiable. Japanese lacks things such as number and gender agreement on predicates, which deprives us of any features we could use to determine what is and is not the subject. The only recourse we have is the が格.
So, you could say that in that one case it does indeed represent the object - but you cannot prove it and neither can I disprove you.
However, if we say that it becomes the object, then that becomes a peculiar exception to the grammar: we have essentially created a new rule. Similarly, we would need another one that says that に marks the subject whenever it is used every blue moon or so.
I argue, that since neither can be proven nor disproven, we take it to always be the subject and choose the gloss accordingly as that reduces the number of rules and exceptions we have to work with, making the process more consistent and eliminating potential confusion by the reader: if all else is equal, the simpler alternative should be better.</upd_detl>
</audit>
<audit time="2024-07-30 23:55:55" stat="A">
<upd_uid>jwb</upd_uid>
<upd_name>Jim Breen</upd_name>
<upd_email>...address hidden...</upd_email>
<upd_detl>Thanks for the discussion. Short of having extensive information about the use of the term in context (as in Makino and Tsutsui) we really have no alternative but to have a small set of succinct glosses. I'm comfortable with the current set.</upd_detl>
</audit>
<audit time="2024-07-31 08:17:10" stat="A" unap="true">
<upd_name>Non</upd_name>
<upd_detl>Apologies but I must insist that, for the reasons previously listed, it is the worst set out of the three.
I have thus far tried, without apparent success, to focus on the alternative glosses' merits. 
However, I think this point can also be made by trying to find the current gloss' merits instead.

For example, I once again bring up this: Assume all three candidates(wanting, want, wanted ) are all equally valid in regards to semantics and syntax and then try to translate a few sentences with ほしい using each of them - I think you will agree that the translations with the current gloss sound significantly more awkward than either of the rest.
This prompts the question: if they are all equally meritorious but 'wanting' has the singular demerit of being the most unnatural, what is it that makes it the most suitable choice over the others, which harbour no such fault?</upd_detl>
</audit>
<audit time="2024-08-06 05:20:48" stat="R">
<upd_uid>Marcus</upd_uid>
<upd_name>Marcus Richert</upd_name>
<upd_email>...address hidden...</upd_email>
<upd_detl>I've only skimmed the discussion but personally I still think "wanting" is better and more helpful than the original "wanted". As Jim and Alan seem content with this version as well, I think it's time to let this rest.</upd_detl>
</audit>
</info>
</entry>