LINGUIST List 3.454

Tue 02 Jun 1992

Disc: Adjectives

Editor for this issue: <>


Directory

  1. Eric Schiller, Re: 3.446 Queries: Lists, Adjectives, Comma, Unhappier, Gopnik
  2. Richard Sproat, 3.446 Unhappier

Message 1: Re: 3.446 Queries: Lists, Adjectives, Comma, Unhappier, Gopnik

Date: Mon, 1 Jun 92 17:40:05 CDTRe: 3.446 Queries: Lists, Adjectives, Comma, Unhappier, Gopnik
From: Eric Schiller <schillertira.uchicago.edu>
Subject: Re: 3.446 Queries: Lists, Adjectives, Comma, Unhappier, Gopnik

Re: Adjectives
Jerry Sadock and I have been wrestling with this issue, and especially
the semantic side of it, for some time and he has come up with some
interesting thoughts on the semantic end which i hope will be
worked out well enough for presentation soon.

I have been concentrating on the syntactic end, and in my CLS paper
from April (co-authored with Barbara Need) we proposed a significant
revision ot X-bar theory which separates predicative adjectives,
which are projections of a head A, from pre-nominal modifiers which
are not, being an adverbial sort of category N1>>N1 (combine with
N-bar to form N-bar. I will not deluge the list with the argumentation,
but will happily send copies of the article to those who want it.

Applied directly to the question at hand, the point is that the
sets of lexical items which can be A[0] or [N1>>N1] are largely
overlapping, though not entirely. In some languages, such
as those of Southeast Asia, there is no morphological marking
to distinguish adjectives from stative verbs, and the NP has the
form head_noun - modifier - [numeral+classifier] - demonstrative
with the order varying widely among languages. Should the modifier
be analyzed as V or A? This is a common question. In my own work,
it is neither, but is rather of the category [N1>>N1] when it is
internal to the NP, and is a V when predicative, as there are absolutely
no tests to suggest that there are two categories (V or A) involved.
On the other hand, this assumes that words like 'very' subcategorize
on semantic, and not syntactic grounds.

When it comes to English, you might want to consider the Autolexical
view, which separates syntax from morphology from semantics in
the lexicon as well as in grammar. Thus the question of whether
something is an adjective gets different answers depending on which
of the three perspectives is adopted. I have found that teaching
English grammer gets much easier when you look at things this way.
Semantic operators in english include 'seem' (a verb), 'likely'
(an adjective) and 'probably' (an adverb). Don't blame me! This
is just a fact, and trying to map syntactic categories onto
semantic categories, as Croft (among others) does, just won't
work all of the time. The same problems arise when you try to
map morphology onto syntax. Sadock (1991: Autolexical Syntax.
University of Chicago Press) is a good introduction to this
multi-modular point of view. My 1989 BLS paper (Syntactic
Polysemy and Underspecification in the Lexicon) deals with
the part of speech problem in Southeast Asian languages.

The issues as they pertain to parts of speech are pretty much
theory independent and can be implemented in other frameworks
which are largely word-driven.

Eric Schiller
University of Chicago
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: 3.446 Unhappier

Date: Mon, 1 Jun 92 19:37:36 EDT3.446 Unhappier
From: Richard Sproat <rwsmbeya.research.att.com>
Subject: 3.446 Unhappier

Gregory Stump expresses four concerns about my analysis of _unhappier_
in LI 23 (1992:347-352). I think one of those concerns is exactly on
target and the other three are less clearly so.

1. Stump's first point questions whether the comparative form of a
scalar adjective is itself scalar. I guess it rather depends upon
definitions here, but if by `scalar' one means `scalar predicate' in
the sense of Horn (1972) or Hirschberg (1991), then surely _unhappier_
is scalar. Certainly it does not occur in the environments 'very
____', 'less ____' or 'as ____ as', but it isn't clear why one should
assume that these constitute an exclusive diagnostic set. (Note that
`very' fails for clear scalars like `very', so perhaps what is really
tested by `very' is gradability rather than scalar-hood. Thanks to
Gregory Ward for this and other points.) For what it's worth though,
earlier stages of English (as is well-known) did allow constructions
like _less_happier_, though what one wants to make of that is
unfortunately not clear to me:

 Or as a moat defensive to a house
 Against the envy of less happier lands;
 This blessed plot, this earth, this realm, this England,
 (Richard II)

One other point that I mentioned in the squib, which Stump did not
mention, is that _un-_, in its contrary interpretation, has the same
meaning as 'the opposite of ____'. Consistent with that, both 'the
opposite of happier' and _[un_[happier]]_ (assuming that structure)
have the same meaning.

2. Stump's second point involves the observation that in adjectives with
inherently comparative meaning (_superior_, _inferior_), _un-_ is
interpreted as having contradictory rather than contrary reading,
suggesting that the interpretation in the case of the structure
_[un_[happier]]_ should by no means be expected to have the contrary
reading. So for

	Sandy wants to find someone uninferior to her at chess.

we get the interpretation that Sandy wants to find someone who is
(merely) not inferior to her at chess. It is interesting that Stump
picked _inferior_ as his example, because I believe that _superior_
displays the opposite behavior. I should point out that I don't find
either _uninferior_ or _unsuperior_ to be acceptable on any
interpretation (they are just ill-formed for me), but if I had to
force an interpretation for

	Sandy wants to find someone unsuperior to her at chess.

I would have to say that it means that Sandy wants to find a weaker
player than herself, not that she wants to find a player who is at
most as good as herself.

If this is right, then this suggests a possible reason for Stump's
results other than the conclusion that comparative-sense adjectives
are compatible only with the contradictory reading of _un-_. Two
properties of _un-_ have been noted in the literature dating back (at
least) to Zimmer 1964:

i) with scalar adjectives _un-_ typically has a contrary reading;
ii) _un-_ tends not to occur with bases that have a negative sense.

Zimmer gives a handful of exceptions to the second generalization:
_uncorrupt_, _unobnoxious_, _unmalicious_, _unvicious_, _uncruel_ ...
Interestingly, although the examples that I have listed here all seem
to have (at least potentially) scalar bases, it is by no means clear
that in these cases _un-_ is getting the contrary interpretation. That
is while I can surely refer to a person as _unkind_ and mean that they
are cruel, I don't think that I can felicitously refer to a person as
_uncruel_ and by that mean that they are kind. There is presumably
some pragmatic basis for this, possibly related to the ultimate reason
for the tendency expressed in (ii): conceivably, if you use a negative
adjective, you automatically implicate something negative about the
thing or person to whom the adjective is applied, and this negative
sense cannot be cancelled by the contrary reading of _un-_, hence
(following a suggestion of Horn) _un-_ reverts to the weaker
contradictory reading. This is not intended to be a well-thought out
analysis of the issues (neither do I sense that I am saying anything
particularly novel), but I think there is sufficient reason to doubt
the significance of Stump's example to the issue he intends to apply
it to.

3. Stump's third point is that even when both ADJ and un-ADJ coexist,
distinct phonological and semantic structures must be assumed. So,
while _uneasier_ must presumably have the phonological structure
_[_un_[_easier_]]_ for familiar reasons, its interpretation ('more
uneasy' and *not* 'less easy') suggests that it can never have the
*semantic* structure _[_un_[_easier_]]. I think Stump is right here.
To the list add _unwiser_.

4. Finally Stump points to his recent paper in Language as presenting
a framework in which _unhappier_ ceases to be paradoxical. The
problem here is not with Stump's assertion that _unhappier_ is no
paradox in his framework: this is true. What bothers me is the
repetition of the refrain that I have seen now more times than I can
remember; namely that bracketing paradoxes cease to be paradoxical if
you view them in such and such a way. The point is that, *all*
analyses of bracketing paradoxes have presented what seemed to the
authors at the time to be well-motivated theories wherein bracketing
paradoxes ceased to be such. Indeed, it would be bizarre if things
were otherwise, since presumably nobody who treats an apparently
paradoxical construction wants to argue that the construction in
question remains a paradox in their theory. Bracketing paradoxes
ceased to be paradoxical long ago, as soon as it was understood that
their solution involved the positing of at least two distinct levels
of analysis for words in some sense. The debate ever since has been
about what those levels might be in any particular case, not about how
to make Bracketing Paradoxes cease to be paradoxical.

Richard Sproat
Linguistics Research Department
AT&T Bell Laboratories			tel (908) 582-5296
600 Mountain Avenue, Room 2d-451	fax (908) 582-7308
Murray Hill, NJ 07974			rwsresearch.att.com
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue