ON THE MORPHOSYNTAX OF COMPARATIVE SEMANTICS
Recent research into the semantics of comparatives (Heim (2000, 2006), Schwarzschild and
Wilkinson (2002), Bhatt and Pancheva (2004), among many others) has converged on the
configuration in (1a) (Bowers (1975), Jackendoff (1977), etc.) in order to enable the degree
operator and the degree clause to move and take scope as a single constituent. In this paper I
will argue that this structure is incompatible with what is known about the morphosyntax of
comparatives, and the structure in (1b) (Abney (1987), Bowers (1987), Corver (1990, 1991,
1997a, 1997b)) should be adopted instead.
(1)
a.
AP
b.
Deg
0
Deg
more
[ CP than…]
DegP
A′
0
A
proud
PP
of her work
Deg
[ CP than…]
0
Deg
AP
0
PP
more A
proud of her work
The most obvious argument against the structure in (1a) comes from the fact that it does not
fit the surface word order (2), where the degree clause/phrase and the degree morpheme can
be adjacent only incidentally (see Bhatt and Pancheva (2004), Grosu and Horvath (2006)).
(2) a.
a more intelligent person than Einstein
b. * a more than Einstein intelligent person
c.
a smarter person than Einstein
Secondly, as shown by Milner (1978), Pinkham (1982), an overt pronoun may replace the AP
in the degree phrase (3). A pronoun is generally taken to replace a maximal projection (XP).
While in (1a) such a maximal projection is not available, in (1b) the pronoun can be taken to
replace the AP.
(3) a.
Il est aussi triste que
l'
était Jeanne hier.
he is EQ sad CMPZR PRED.CL was Jeanne yesterday
He is as sad as Jeanne was yesterday.
b.
Alice is incredibly tall, and Beth is even more so.
Most evidence, however, against the structure in (1b) comes from morphology: constraints on
the formation of synthetic comparatives as in (4) unambiguously show that they are created
by head movement, which is not compatible with the structure in (1b).
(4) a.
smarter, taller, trickier
English
b.
vriendelijk-er 'friendlier', klein-er 'smaller', Frans-er 'more French'
Dutch
The standard way of combining two syntactic terminals into a single word is head movement,
a syntactic process known to be heavily constrained (Travis (1984)): the landing site of head
movement must c-command the source position and no other head must intervene. Clearly,
the structure in (1b) doesn't satisfy the Head-Movement Constraint: Deg° doesn't c-command
A°, or vice versa. Although an alternative method of combining two syntactic terminals into a
single word, Local Dislocation, also exists (Embick and Noyer (2001), Embick (2007)), I will
demonstrate, contra Embick and Noyer (2001), Embick (2007), that it cannot account for the
properties of synthetic comparatives and superlatives. As a result, I claim, the configuration
in (1b), although compatible with the prevailing semantic treatments of comparatives, cannot
be maintained.
Local Dislocation (5) combines two syntactic heads into one under the condition of linear
adjacency. As such, it has to apply after Vocabulary Insertion and linearization. The reason to
postulate that synthetic comparatives and superlatives are not derived by head-movement is
that the traditional head-movement approach cannot account for the prosodic constraint on
synthetic comparative formation (6). Since head movement happens before vocabulary
insertion, no effect from the choice of the lexical root is expected.
(5) a.
[XP X [YP [ZP Z] Y]]
b.
[X *[ Z * Y]]
c.
[[Z0 Z+X] * Y]
2
(6)
a.
smarter, #more smart
b.
*beautifuller, more beautiful
However, given that Local Dislocation is a post-syntactic operation that applies only apply
after Vocabulary Insertion, it cannot derive comparative suppletion (7). As morphological
processes can only take place within a single syntactic head (Chomsky (1995:319)), before
Local Dislocation there is no context for suppletion, and once local dislocation has combined
the two terminals into a single syntactic head, vocabulary insertion has already occurred.
(7) a.
more + good → better
English
b.
plus + bon → meilleur
French
Embick and Noyer (2001) also claim that adverbial modification provides further evidence
against the head-movement analysis of synthetic comparative and superlative formation: if an
adjective is modified by an adverb, a synthetic form becomes impossible:
(8) a.
Mary is the most amazingly smart person.
b. * Mary is the amazingly smartest person.
I will demonstrate that adverb intervention effects are epiphenomenal: both morphologically
derived adverbs and adverbial PPs behave the same (9), showing that the intervention is not
due to phonology. I will present evidence (10, 11) to show that two base positions for adverbs
are available in comparative APs, above and below the comparative head, and the blocking
effect, when it exists, is due to the Head-Movement Constraint.
(9) a.
Jude is smarter to an amazing degree than Joe.
b.
Jude is smarter than Joe to an amazing degree.
(10) a.
This building is more structurally weak than that one.
b.
This building is structurally weaker than that one.
c. * This building is weaker structurally than that one.
(11) a.
I became a technically more proficient guitarist.
b.
I became a more technically proficient guitarist.
Further evidence against Local Dislocation or any alternative word-formation process relying
on linear adjacency comes from obligatory analytic comparative formation. Thus non-scalar
adjectives can form synthetic comparatives in Dutch but not in English, while metalinguistic
comparatives cannot be synthetic in either language (12, 13). I will argue that the coercion
process involved in the interpretation of (12a, 13a) can block head movement in one language
but not the other, but in the structure associated with the metalinguistic comparatives in (12b,
13b) head movement is impossible. Local Dislocation, on the other hand, is not predicted to
distinguish between the two cases.
(12) a.
My aunt is *Frencher/more French than Napoleon.
b.
The guy is *taller/more tall than gaunt.
(13) a.
Jan is Franser/
?meer Frans dan Piet.
Jan is French+cmp more French than Piet
Jan is more French than Piet.
b.
Jan is *dikker/
meer dik
dan vet.
Jan is thick+cmp more thick
than fat
Jan is more overweight than obese.
Further evidence for deriving synthetic comparatives and superlatives by head-movement and
therefore against the structure in (1b) comes from morpheme ordering, and the curiouser and
curiouser construction (Jackendoff (2000)). I will discuss two alternative ways of resolving
the issue: one, by modifying the semantics of comparatives to make it compatible with the
structure in (1b), and the other, by hypothesizing that the degree morpheme and the degree
clause form the complement of A°, which would preserve the desired constituency for DegP
and yet enable head movement to take place. Both approaches will be shown to be far from
straightforward.