LINGUIST List 3.493

Sun 14 Jun 1992

Disc: Comparatives, Verb/Adj, X-Bar, Predication

Editor for this issue: <>


  • Larry Horn, Re: 3.472 Comparatives
  • , Re: 3.446 Queries: Lists, Adjectives, Comma, Unhappier, Gopnik
  • Eric Schiller, Re: 3.479 Queries: TV, Software, Predicates, Good, X-Bar
  • , Re: 3.479 Queries: TV, Software, Predicates, Good, X-Bar

    Message 1: Re: 3.472 Comparatives

    Date: Fri, 12 Jun 92 15:14:16 EDRe: 3.472 Comparatives
    From: Larry Horn <LHORNYaleVM.YCC.Yale.Edu>
    Subject: Re: 3.472 Comparatives

    Sorry I didn't have a chance to jump into the fray earlier, but I had a few thoughts on the status of comparatives as scalar predicates, and related matters. In his gauntlet-tossing posting of May 29, replying to Rich Sproat's recent LI squib on the a _unhappier_ bracketing paradox, Greg Stump argues contra Sproat that _happier_ and other comparatives (of scalar adjectives) are not themselves scalar values, or not clearly so. Sproat's June 1 reply maintains that comparatives are indeed scalar, although they don't (usually) appear in the frames Stump takes to be diagnostic for scalarity (very ____, less____, as ____ as). In his June 5 rereply, Stump reargues the nonscalarity of _unhappier_, on the basis of frames like Sandy is happier (than she was last year) if not actually happy. Sandy is happy if not actually happier (than she was last year). Sandy is not only happier (than she was last year) but happy. Sandy is not only happy but happier (than she was last year). --where clear instances of scalar predicates (good/excellent, warm/hot, widespread/universal) can only appear in the frames ...W... if not actually ...S... not only ...W... but ...S... with the weaker value in position W and the stronger in S. Thus, he concludes that "_happier_ doesn't behave as if it were a scalar adjective associated with the same scale as _happy_". By June 6, Sproat is ready to concede the point. In the 1972 dissertation cited by both Rich and Greg, what I argue is that comparatives are indeed scalar values, but the relevant scale is not defined by the absolute and comparative but by the equative and the comparative. Thus the relevant test sentences are Sandy is as happy as {Kim/she was last year} if not actually happier. Sandy is not only as happy as Kim, she's happier. --both of which are clearly irreversible. Similarly, Kim isn't even as tall as Sandy, much less taller. Sandy is as tall as Kim, {or even/and possibly (even)} taller. This is to be expected, given that A is taller than B unilaterally entails A is as tall as (i.e. at least as tall as) B. So too, the use of the weaker "A is as tall as B" will quantity-implicate that for all the speaker knows, A is not taller than B. (All of this is discussed in my thesis, and again in my 1989 book "A Natural History of Negation", pp. 386-8 and note 27 on p. 548, citing related work on comparatives as scalars by Fauconnier, Klein, Anscombre & Ducrot, Sadock, Cornulier, and especially Atlas's 1984 paper in Linguistics and Philosophy.) As Greg notes, it's not surprising that there's no scale of the form <happier, happy>, given in particular that neither of the two propositions _Sandy is happy_ and _Sandy is happier than X_, for an arbitrary NP or clause X, entails the other. Incidentally, the comparative is not necessarily the strongest value within its scale, since the superlative will unilaterally entail the correspond set of comparatives: Sandy is taller than Kim, {if not/and she may even be} the tallest on the team; This was a better year, if not the best. [again, the frames don't allow reversal] The relevant scale is <Aest, Aer than, as A as>. I'll come back to superlatives in a bit. The scale is thus defined by comparison of degree, and not by the semantics of the adjective itself. So it's not surprising that these scalar values don't satisfy the frames designed for scalar adjectives per se. This comparative failure is noted not only by Stump but by Jack Hoeksema in a more recent posting that cites substitution possibilities in the frames "Boy, is she ____" and "What a _____ guy", where _happy_ is OK but _happier (than Bill)_ is not. Clearly, too, _as happy as Bill_ is ruled out here, yet various considerations, some cited above, lead us to regard equatives as scalars. Incidentally, these diagnostics seem to allow through some non-obvious scalars, although arguably a scalar reading is then forced: Boy, is she ever {female/dead/perfect/unique}! Boy, is that ever true! Pure contradictories still seem a bit out of sort here: #Boy, is 7 ever odd! Anyway, my point here is that equatives and comparatives are no more at home in this environment, whether scalar (as I continue to insist) or not, than are such clear scalar values as quantificational determiners (some, most, all), cardinals, and ordinals. Now from all of this it does NOT follow that I would endorse Rich's parallel morphological/semantic analysis of _unhappier_ as [un[happy[er]]]. To the contrary, I find it exceedingly counterintuitive if not counterexemplified, partly by some of the arguments raised in earlier exchanges on this topic. Let me just add my own skepticism, fueled by the general consideration that the acceptability of an arbitrary instance of _unXer_ (and, I would add, of _unXest_) seems to be directly linked to the existence of the corresponding example with _unX_, a link which is not directly captured if the negative prefix is added last with the meaning 'opposite of'. Thus, even though we might expect _unsafer_, _uncleaner_, _unclearer_, and of course _unhappier_ to be blocked by the existence of the comparative of the lexicalized negative adjective in the same way that *_untaller_, *_unbetter_, etc. are, they don't sound all that bad, and the reason can only be that the basic uncompared adjectives (or the corresponding equative) are also unblocked, for whatever reason. If we're dealing with the comparison of negative adjectives rather than the negation of comparative ones, that's precisely to be expected. Even more troubling for the Sproat line, I would think, is the existence of un-superlatives. The ONLY way to explain the existence of _unwisest_ along the non-existence of _unsmartest_ and _unbrightest_ is by deriving such forms as superlatives of the un-adjective and not vice versa. Similarly with the unkindest (*unnicest) cut of all. And not only must the existence of _impurest_ be predicted (at least I can imagine referring to _the impurest sample_), but we must be able to predict that this particular superlative will take the marked iN- prefix. Presumably these cases would demand a conventional negation-first analysis, as indeed Sproat readily accepts for other examples that don't fit the new paradigm. But it begins to seem that there's less and less for the new approach to do. (I also have reservations about Rich's use of the notion oppositeness, but I better let those go undiscussed here except to note that (i) whatever its viability with the un-comparatives, it's all the more uncon- vincing for the un-superlatives: to say that Kim is the unhappiest he's ever been is not to say that he's the opposite of the happiest he's ever been, and (ii) it also seems less plausible for clausal un-comparisons, where e.g. _Sandy is unhappier than she's ever been_ comes out as 'Sandy is un [happier- than-she's-ever-been].) So I'd argue that while comparatives (and superlatives!) are indeed scalar, Rich originally claimed for the former, the constituency argument employing that claim remains highly questionable. --Larry Horn (LhornYalevm)

    Message 2: Re: 3.446 Queries: Lists, Adjectives, Comma, Unhappier, Gopnik

    Date: Sun, 7 Jun 1992 20:59 MST Re: 3.446 Queries: Lists, Adjectives, Comma, Unhappier, Gopnik
    Subject: Re: 3.446 Queries: Lists, Adjectives, Comma, Unhappier, Gopnik

    To L.M. P. McPherson on Adjectives versus Verbs: There are Native North American languages that do mot appear to make a distinction between the categories "verb" and "adjective". The ones I've done fieldwork on Lakota (Siouan), Sarcee, Apache, and Navajo (Athapaskan), and Siberian Yupik (Eskimo-Aleut) did, on the whole, not make a clear distinction between verbs and adjectives. Rather adjectives are a sort of verb, and are inflected like verbs. Copulas are therefore very rare. In lakota, adjectives are really stative verbs; NPs containing "adjectives" do not really exist: you have to use either a relative clause, i.e. the dog that's big = the big dog, or say: dog-big where the big is a stative verb compounded with dog. There are subtle semantic and pragmatic differences between the two. Compounding is of course not typical of N- Adj sequences in lakota, because you can compound virtually anything with anything in lakota. In the Athapaskan languages, the situation is similar, except that there is a less basic distinction between stative and active verbs, and that NPs with attributive "adjectives" are more often relative clauses; compounding also exists, but the number of stems that can compound with a Noun seems to be a closed class, not an open class as in lakota. Of course, both in Athapaskan and in Siouan, there are some morphological (inflectional) and syntactic characteristics that do distinguish adjs as a subclass within the verbs. The situation is a bit different in Eskimo: most "adjectives" are verbs, inflected as such, thus "big" is a verb "to be big", a situation like the one in Siouan and Athapaskan. However, you also have a small but non- closed class of nouns which are adjectives, translatable as "big thing". The verbs can be used ( as in Athapaskan and Siouan) as attributive adjectives in realtive clauses; the nouns can be used attributively in apposition to another noun, agreeing in case and number with it (none of these languages have gender), thus 'dog big-thing' = 'big dog'. You can also use the nouns precicatively by verbalizing them with the derivational but productive N -> V suffix basically meaning 'to be', so 'dog' = 'the dog is big'. But you can also form the equivalent of NPs with attributive "adjectives" by using one of the N > N productive derivational suffixes that semantically correspond to "adjective", such as the ones meaning 'big', 'small', 'new', 'old', 'nice', 'cruddy', and a few dozen others. These are suffixes only, completely unrelated to nouns or verbs, and being suffixes, are also a closed class. I apologize to other authorities for inaccuracies herein; my goal being to compare in three very different Native American families, the things translatable with adjectives in English. Willem J. Reuse Dept. of Anthropology, University of Arizona Tucson, AZ 85721

    Message 3: Re: 3.479 Queries: TV, Software, Predicates, Good, X-Bar

    Date: Thu, 11 Jun 92 12:06:49 CDRe: 3.479 Queries: TV, Software, Predicates, Good, X-Bar
    From: Eric Schiller <>
    Subject: Re: 3.479 Queries: TV, Software, Predicates, Good, X-Bar

    re: that good of In my recent CLS Paper (Eric Schiller & Barbara Need: The Liberation of Minor Categories: Such a Nice Idea!) I set forth an autolexical analysis of a variety of constructions, concentrating on predeterminers and to a lesser extent degree terms. this construction can be handled in the same way as 'too good a', if one wants to assume that 'that' is a degree term (in our system, something which combines with an adjective to form a predeterminer) and that 'of' is acting as an article (syntactically combines with N-bar to form NP). The latter assumption is reasonable, given the a/of variation seen in sorta/sortof, and the 'of' could be motivated here by some form of hypercorrection.

    I am not sure this is the best analysis, but I toss it into the arena so that interested parties can play with it. No intellectual or theoretical investment undertaken - it is just a thought.

    Eric Schiller University of Chicago

    Message 4: Re: 3.479 Queries: TV, Software, Predicates, Good, X-Bar

    Date: Thu, 11 Jun 92 15:52:42 EDRe: 3.479 Queries: TV, Software, Predicates, Good, X-Bar
    From: <pesetskAthena.MIT.EDU>
    Subject: Re: 3.479 Queries: TV, Software, Predicates, Good, X-Bar

    Ronnie Wilbur writes:

    >In ASL, predicate nominals have no overt verb: > ____hn > JOHN DOCTOR >[...]] >the predicate nominal has a head nod on it (without the head nod, the >sequence is supposed to mean "John's doctor"). The facts about ASL predicate >nominals appear to be very similar to those of Russian, as far as I am >able to determine. Essentially, you can negate the predicate nominal ("John >is not a doctor") or you can modify the nominal itself ("John is a good doctor" >) but attempts to modify the "missing" verb (with tense-aspect-modality, for >example) are unacceptable without an overt verb. > >My question concerns what the analysis of e.g., the Russian case would be. Is >there a V slot that is unfilled?

    One datum that might become an argument for an empty verb is that Russian, like Black English as described by Labov, does not allow the empty verb before a movement site. My Russian is becoming somewhat rusty but I think relevant examples are things like the following:

    *Masha -- xoroshij lingvist, kakov Misha tozhe. Masha good linguist, which Misha also

    Masha -- xoroshij lingvist, kakovym Misha tozhe byl. which Misha also was

    [I used to give these examples with WH-words other than 'kakov', but the examples in something like this form were suggested to me by Elena Paducheva, a Russian linguist. Native speakers, please correct any errors.]

    As Labov noted in the BEV context, this is a familiar (and still unexplained) property of contracted auxiliaries in English ("*...which Misha also's"). If the explanation for the contraction facts should turn out to presuppose the existence of a verb position, then presumably the same will carry over to Russian.

    You might want to look at an MIT dissertation from 5 years ago or so by Tova Rapoport, however, which analyses the zero copula in Hebrew as simply not there. I do not know whether the Hebrew zero copula otherwise acts like contraction in English.

    > >[...] >Please respond to me directly ( and I'll summarize >the results if there is sufficient interest. Thanks. ANY SYNTACTIC APPROACH >WELCOME!

    I've been wishing that someone would sort these things out for a long time, so I took the liberty of posting my remarks myself. Russian syntax is so woefully underdeveloped for a language with a rich descriptive tradition! Maybe someone will read this and write something interesting.

    -David Pesetsky