LINGUIST List 3.471

Sat 06 Jun 1992

Disc: Adjectives and Morphology

Editor for this issue: <>


Directory

  1. Eric Schiller, Re: 3.465 Adjectives, Compounds
  2. Richard Sproat, even more _unhappier_

Message 1: Re: 3.465 Adjectives, Compounds

Date: Sat, 6 Jun 92 12:26:51 CDTRe: 3.465 Adjectives, Compounds
From: Eric Schiller <schillertira.uchicago.edu>
Subject: Re: 3.465 Adjectives, Compounds

Adjectives:

Robert Beard writes:
"1b. Comparative and Superlative forms are never lexically
negated because Comparison is an inflectional category and is
therefore formed after lexical (L-) derivation. Evidence: many
languages have an analytical comparative, _more happy_, _less
happy_, which require syntactic structure. Since inflectional
rules apply in syntax (please don't risk much of your career on
the assumption that it doesn't), Comparison must apply after all
L-derivation is complete. It follows that *un[happier] is impos-
sible for the same reason that *_unbetter_, *_unsicker_,
*_unfaster_, *_untaller_, etc. are."

I will continue to risk my career by taking the position that
morphology is morphology and syntax is syntax, and that inflection
is morphology and not syntax. In a non-derivational system such as
Autolexical Syntax there is no rule ordering, either explicit and
stated as applying between components, or implicit and covert.
There is much data in this set of posts, and I am not going to
try to deal with it all in a quick note. But once again I find that
major claims are made on the basis of data which involves grammaticality
judgements which are by no means universal. Not only do I find
My car is redder than orange.
acceptable but if there is any oddity for other speakers I suspect
it is semantic, since much money was spent on convincing us that
our clothes could become
Whiter than white!
by using a certain detergent.

My point is not that the data is flawed, or, as some would claim,
grammaticality judgements are useless, but rather that theories
which are built upon these judgements must have the flexibility
to allow for the kinds of variations we find, and not insist
that dialects which violate these supposedly universal rules
are somehow "abnormal".

If a theory can account for variation, either by using parameters or
by keeping variation in the lexicon, fine. But I do find it uncomfortable to
hear papers which imply that as a speaker of what might be described as
a liberal New York dialect (both readings fine!), I am some sort
of freak of nature with a defective language mechanism!

Eric Schiller
University of Chicago

Note: this is not directed at any particular framework. In fact,
my initial thoughts along these lines were inspired by Jim
McCawley's syntax courses as a 1st year student at Chicago many
years ago. Our judgements rarely matched on the complicated
data, and I was constantly trying to think up derivations which
allowed for my dialect. To the credit of his analytical framework,
I usually could, though not always...
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: even more _unhappier_

Date: Sat, 6 Jun 92 20:21:58 EDTeven more _unhappier_
From: Richard Sproat <rwsmbeya.research.att.com>
Subject: even more _unhappier_

In response to Greg Stump's and Robert Beard's postings on the
_unhappier_ issue, I want to clarify a few issues.

First, in response to Greg, I didn't think I was being glib in my
response to the issue of whether _happier_ is scalar. Rather, I was
trying to convey the sense of the term `scalar' that I was assuming.
Quite possibly I wasn't successful in that attempt; I am perfectly
prepared to accept the possibility that my assumption was too
simplistic and ultimately wrong; and I fully accept the further
possibility that I missed various subtle issues in Greg's response on
this point. But by presenting a simple answer, I do not think I was
being glib. And I agree that Greg's new examples do seem to suggest
that _happier_ should not be considered scalar, at least not clearly
so. As to whether this all means that my analysis of _unhappier_
becomes unworkable, I can only repeat for a third time my other piece
of evidence, namely that contrary _un-_ means roughly the same as _the
opposite of_, and that _John is happier than Bill, whereas Sam is the
opposite (of happier than Bill))_ -- where _the opposite_ must
presumably be `bracketed' outside _happier_, means that Sam is
unhappier than Bill. In other words, the bracketing I proposed for the
semantic interpretation of _[un [happier]]_ is at least consistent
with the possibility of interpreting _un-_ in the strongest way
possible, i.e., with the sense of _opposite of_.

In response to my quibble about use of phrases like ``cease to be
paradoxical'' Greg has the following to say:

 On the approach [Sproat] advocates, _unhappier_ is given a
 single structural analysis (viz. [ un [ happi er ]]), valid both for
 phonological and for semantic purposes; by contrast, cases like
 _uneasier_ and _slaphappier_ each have to have two different structural
 analyses ([ un [ easi er ]] and [ slap [ happi er ]] for phonological
 purposes, but [[ un easi ] er ] and [[ slap happi ] er ] for semantic
 purposes). Thus, the paradox remains, though in a somewhat disguised
 form: despite the obvious parallelism between the three expressions,
 _uneasier_ and _slaphappier_ have to be analyzed differently from
 _unhappier_.

But note that my squib did not say that all three *have* to be
analyzed differently: one would still get the right interpretation for
_unhappier_ assuming the semantic bracketing _[[un happi]er]_, and if
one can force that bracketing -- something Pesetsky was not able to
do, for example, but something that would follow from Beard's points
(see below) -- then there would be no problem. All I suggested was
that it was not necessary to assume that bracketing in order to get
the right interpretation. The problem with my assuming such complete
freedom of brackeing, as I conceded in the previous posting, is that
one could not explain why _uneasier_ does not mean the same as
_harder_ (consistent with the bracketing _[un [easier]]_), but only
has the interpretation consistent with the bracketing _[[uneasy] er]_.
I concur that this is a problem if one does not assume that, one way
or another, _-er_ must be interpreted `outside' _un-_; and however one
chooses to derive this constraint would also necessarily derive
_[[unhappy]er]_ as the only possible bracketing. Fine. Point taken.
But if one *were* to allow the other bracketing, then I would still
derive the correct interpretation for _unhappier_ (assuming my
semantic analysis holds).

Turning now to Beard's points, I agree that the assumption that
inflectional rules apply in syntax and therefore after the application
of derivational rules effectively rules out _[un [happier]]_ as a
possible bracketing. Of course, I had thought the jury was still out
on the issue of whether in fact inflection applies `in the syntax', at
least on the more obvious interpretations of that phrase: presumably
the fact that Anderson, Zwicky and others argue so strongly that it
does is some indication that there must somewhere be a contingent of
morphologists who aren't convinced. But as Beard suggests I won't
``risk much of [my] career on the assumption'' that inflection does
not apply in the syntax (an odd injunction, since it is possible to be
both well-respected and much-cited despite having made claims that are
far more transparently false than that). Indeed, I will sidestep the
issue slightly and say that I don't actually find very interesting the
question of whether inflection should be separated from derivation or
whether one should apply in the syntax and the other in the lexicon,
but that I am perfectly prepared to believe that there are
morphological operations like comparative -- call them inflectional if
you wish -- that are sensitive to particular syntactic contexts and
that therefore must apply after lexical operations like _un-_
affixation, which do not have the same kinds of syntactic sensitivity.
In other words, I concede Beard's point. (Still I am puzzled by
Beard's claim that "phonology has to work with [the structure
_[[unhappy]er]_] and, as long as we are not trying to salvage lexical
phonology, it can", since the paradoxicality of _unhappier_ never had
anything to do with any of the issues raised by lexical phonology,
unlike examples such as _ungrammaticality_.)

And I think Beard's claim about the affixes with which productive
`lexical' comparatives may be formed is an excellent one. It is
certainly true that, given that the only disyllabic adjectives that
productively allow _-er_ and _-est_ are (somehow) derived with _-y_
and _-ly_, adding _un-_ to the list does not cost much. In other
words, _-er_ and _-est_ will attach productively to monosyllables,
plus any monosyllables that (somehow) end in _-y_, _-ly_ or begin with
_un-_. It's not clear that this is any more esthetically pleasing a
solution than the claim that _-er_/_-est_ attaches to "all
monosyllabic stems and disyllabic stems ending on an open syllable
with a light vowel", which Beard dismisses, but at the very least it
may not be any more costly.

Still, I don't find everything that Beard says to be equally
agreeable. First of all, I don't know if the differences between
morphological theories are ``grander'' than I led people to believe:
indeed, I wasn't presuming to lead people to believe anything about
morphological theories in the large. And I don't think I said and I
certainly did not mean that bracketing paradoxes are literally that --
*bracketing* paradoxes -- in all theories of morphology. All I said
was that in all approaches to morphology where people have treated
bracketing paradoxes _qua_ bracketing paradoxes, the authors have
attempted to argue that their solution renders the constructions
non-paradoxical, and that IN THOSE THEORIES WHERE ONE AT FIRST GLANCE
SEEMS TO HAVE A PARADOX, the solution has always involved two levels
of structure, or two levels of analysis. Of course if you have a
theory where words don't have bracketings or anything equivalent,
then none of this is an issue. Fine. But in theories where words *do*
have bracketings, bracketing `paradoxes' are not a problem as long as
one can motivate two levels of analysis at which these words have
representations.

Beard also asks a more pointed question about the viability of the
notion that words have bracketings: it makes no sense, he claims, to
say that one can bracket a word derived by infixation, revowelling,
stem mutations, or (I infer) similar operations. The simple response
to that idea is to assume, following various people such as Marantz,
Lieber (i.e., her work on autosegmental morphology, and also her
latest book), McCarthy and Prince, and others that apparently
non-concatenative morphology can be understood if instead of mere
concatenation, one allows into the inventory of options for the
phonological spell out of morphology (its exponence, if you prefer)
other kinds of autosegmental associations. Then the only difference
between English prefixation and Tagalog infixation, for example, would
be at the level of phonological spell out, not at the level of real
morphological representation. I try to make exactly this point in my
book _Morphology_and_Computation_ (1992 MIT Press), but other earlier
work (including my own) has expressed this view. So while one may not,
for various reasons, be satisfied with this answer, at the very least
the idea of providing a bracketing for a morphological construct which
on the surface involves, say, infixation is not as non-sensical as
Beard implies.

As for _cook_ and _baker_, as Beard contends, it probably doesn't make
sense to assume parallel bracketings. But this seems like a desirable
result, since the relationship between _cook_ (noun) and _cook_ (verb)
is different from the relationship between _baker_ and _bake_ in one
regard: while _baker_ seems to be able to assign _bake_'s arguments to
an NP -- _baker of French pastries_ -- cook cannot: *_cook of French
cuisine_. Zero-derived deverbal nouns (putting myself at some risk by
using the term `zero-derived') usually don't seem to be able to assign
the verb's roles -- at least not with _of_ complements -- and perhaps
they should be formally distinguished from overtly derived forms like
_baker_.

To summarize this necessarily rather long-winded response:

1. Beard and Stump both present good reasons for thinking that the
bracketing of _unhappier_ must be _[[un happi] er]_ after all.

2. Nonetheless, and noting Stump's points about scalarity, there is
still some evidence that _unhappier_ should be interpreted correctly,
even with the other bracketing -- and hence from a purely semantic
point of view, it wasn't a particularly convincing example of a
bracketing paradox in the first place.

3. There was some misunderstanding of what I actually said in my
previous message, which I hope I have clarified.

4. Some of Beard's objections to bracketing-based theories of
morphology should not be taken uncritically.

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