franka@mmintl.UUCP (Frank Adams) (07/23/85)
Some time ago, I noticed that when more than one adjective is applied to a noun in English, there is an invariable order in which they are applied. This order depends on what the adjective is specifying. For example, adjectives of size always precede adjectives of order; one says "the big red house", not "the red big house". I worked this out a bit further, although not systematically, and every adjective I could come up to could be fit into one of six or seven categories (I don't remember exactly, and I have lost my notes on it) which fell into an invariant order. This ordering was strong enough to seem like a grammatical rule; as in the examples above, reversing the order "doesn't sound right". I wondered about a couple of points. First, has anyone else ever noticed this ordering? I have never found any reference to such a thing. Second, does the same kind of ordering occur in other languages? I would assume that it does. Do they use the same ordering, or some other? Do languages which put the adjective after the noun reverse the order?
steiny@scc.UUCP (Don Steiny) (07/25/85)
> > > Some time ago, I noticed that when more than one adjective is applied > to a noun in English, there is an invariable order in which they are > applied. > one says the big red house", not "the red big house". > > I wondered about a couple of points. First, has anyone else ever noticed > this ordering? > Frank Adams YES! One thing I have spent some time working on is a transformational explaination. It is probably not the most favored course in this day and age, but there are compelling arguments. Consider a special type of "adjective" those adjectives that are formed with the "-ing" form of a verb or the past participle. I was sorting through words for another purpose when I noticed that the past participle form of intranstive verbs cannot be used as adjectives. Transitive verbs: need, fry. 1) He needed the books. 2) The needed books are here. 3) He fried the fish. 4) The fried fish were delecious. Intransitive verbs: "agree", "bark". 5) The dogs barked. 6) *The barked dogs . . 7) The doctors agreed. 8) *The agreed doctors . . Further evidence that intransitive verbs cannot be used in the "adjective" slot in NP's is homophones like "concentrate." There are two words "concentrate", one transitive and one intransitive, "to charge the soultion," "to concentrate on the problem." Notice that the first use is transitive and the second intransitive. 9) The concentrated solution was . . . 10) ?The concentrated question was . . . notice that when the word "concentrate" is used in that slot, it gets the transtive reading. #10 might not really be funny, if it is given the interpretation that the question was more concentrated than other questions, but not if it has the reading that the question is concentrating (note that the two verbs mean different things, the question is . . . which verb are we using?). Notice that with the "-ing" suffix intranstive verbs work fine: 11) The barking dogs . . . 12) the agreeing doctors . . . and that verbs that are homophones get the intransitive reading in that slot: 13) The concentrating student . . . 14) ?The consentrating question . . . These examples have lead me to hypothesize that the stucture that we are generating is not: DET ADJ NOUN but a relative clause that is reduced to the NP form we use. Thus, "the barking dogs" comes from "The dogs are barking", The "The fried fish" comes from "X fried the fish." This correctly predicts the reading of the homophones. X consentrated the solution => The concentrated solution . . . The student was concentrating. => The concentrating student . . . Basically, it seems that under some circumstances, the rule I) NP => DET ADJ[0] NOUN does not predict the surface structure as well as assuming that the string is generated from an underlying S and reduced to the left hand side of I. This proposal seems to help explain the words that double as verbs that occur in the ADJ slot in I, but what about words that are always adjectives? I have tried out, as a thought experiment, the idea that the I is NEVER generated and that all occurances of I come from reduced relative clauses. That would mean that: 15) The red barn . . . comes from: 16) The barn that is red. (Notice that there has to be some part of the transformation that arbitrarily switches the adjective and the noun, but English is backwards that way relative to the other Indo-European languages, and some rule like "English switching" does not strike me as too outragous). NP's that have multiple adjectives, like 17) The big red barn . .. comes from: 18) The barn that is red that is big . . . I think that each additional modification, i.e. "that is ..." has to make sense semantically. I think there are natural subset relationships that more carefully isolate the desired subset, and that these relationships are based on our experience. -- scc!steiny Don Steiny @ Don Steiny Software 109 Torrey Pine Terrace Santa Cruz, Calif. 95060
friesen@psivax.UUCP (Stanley Friesen) (07/29/85)
In article <521@mmintl.UUCP> franka@mmintl.UUCP (Frank Adams) writes: > >Some time ago, I noticed that when more than one adjective is applied >to a noun in English, there is an invariable order in which they are >applied. This order depends on what the adjective is specifying. > >I worked this out a bit further, although not systematically, and every >adjective I could come up to could be fit into one of six or seven categories > This ordering was strong enough to seem like a >grammatical rule; as in the examples above, reversing the order "doesn't >sound right". > >I wondered about a couple of points. First, has anyone else ever noticed >this ordering? I have never found any reference to such a thing. > Yes, I saw it in an early book on a non-transformational approach to grammer. I forget the title right now, but next time I'm at UCLA I will try and remember to look it up. -- Sarima (Stanley Friesen) {trwrb|allegra|cbosgd|hplabs|ihnp4|aero!uscvax!akgua}!sdcrdcf!psivax!friesen or {ttdica|quad1|bellcore|scgvaxd}!psivax!friesen
franka@mmintl.UUCP (Frank Adams) (08/06/85)
This is a digest of the responses I received to my posting on adjective
order. I received a request to post my old notes on the subject; alas
they are long gone, and I do not have the time to try to reconstruct them.
I hope the information here is useful to some of you.
From pwa-b!philabs!allegra!tektronix!tekchips!sheldonn
Terry Winograd mentions the adjective ordering in his book
"Understanding Natural Language", 1972. He certainly wasn't
the first to notice it, however.
cheers,
--Sheldon
...!tektronix!tekchips!sheldonn
From pwa-b!philabs!ihnp4!gargoyle!sphinx!mmar
The question has been noted before, but not with any very convincing
answers. The suggestions I've seen were pretty vague, and turned on the
idea of inherent vs accidental properties.
Thanks,
Mitch Marks @ U Chicago
... ihnp4!gargoyle!sphinx!mmar
From pwa-b!philabs!prls!eva
I believe that the late Henry Lee Smith Jr. had some analysis of this.
Whether he ever wrote anything up is another matter; his style of
Descriptivism was very much overshadowed when I had contact with him,
so he may not have bothered to publish. He collaborated with George Trager,
so you might look for some combination of the two names.
Brian Phillips
From pwa-b!philabs!ihnp4!uw-beaver!ssc-vax!bcsaic!michaelm
Yes, it has been noticed before, and I agree that it is an extremely
interesting problem! I don't have any references to "linguistic"
discussions of it, but the book "A Grammar of Contemporary English" by
Randolph Quirk, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik
(published by Longman's at the price of --eek-- $45 or so; but then,
it's 1100+ pages) discusses it.
This book, by the way, is a grammar of English for non-native speakers,
and is one of the most complete grammars of English (or any language)
that I've ever seen. That doesn't mean that it's complete, though...
At any rate, the relevant sections are 5.41 and 13.65-13.68. Quirk et al
note, e.g. that if the adj means something like "consisting of," "involving,"
or "relating to," it must be adjacent to the noun:
the extravagant (pleasant, only, London) social life
*the social extravagant... life
a serious (city, mere,...) political problem
*a political serious (...) problem
I suspect that one might be able to find exceptions, and of course one
would like a better characterization of what "consisting of etc." means.
Quirk et al have several other classes, and discuss their relative ordering.
Incidently, if your library doesn't have this book, there
is an abridged version by the same authors. I don't have it, so I don't
know how much it discusses this adjective question.
Re other languages, Spanish is a language that (commonly) has adjectives
following the noun, rather than before, but I don't have any intuitions
that I would trust about adj order. Any native Spanish speakers reading
this? English sometimes has long adj phrases after the noun, e.g. "a man
angry at the world," but they don't seem to "stack" very well.
--
Mike Maxwell