2.2 Earlier accounts
2.2.5 The QP–DegP analysis – Lechner (1999; 2004)
Before turning to my proposal, let me briefly discuss the analysis provided by Lechner (2004), a revised version of Lechner (1999), which answers some of the questions that emerged in connection with the previous accounts mentioned
here and which provides important insights concerning the actual relations be-tween the various functional projections. This study is important first and fore-most because it reconsiders the syntactic relationship between the AP and the Deg head, in that it reflects the semantics of the Deg head much better than previous analyses.
Lechner (2004: 22) partially adopts the functional AP-hypothesis; that is, that the AP is embedded under a functional projection, the DegP, cf. Abney (1987), Bresnan (1973), Corver (1990; 1993; 1997), and Kennedy (1999). However, Lech-ner (2004: 22–23) assigns a different structure to the DegP, in that he proposes that the AP is base-generated in the specifier position of the DegP and not as a complement, in this respect recalling the proposal made by Izvorski (1995). At the same time, the complement position serves to accommodate the comparative subclause.
The structure – using the DegP in a string such asMary is younger than Peter is– is shown below (see Lechner 2004: 22, ex. 45):
(28) DegP
AP younger
Deg′ Deg
[+comparative]
than-XP than Peter is
An advantage of assuming that the AP is in the specifier of the DegP is that in this way, they can enter into a specifier–head relationship, and the [+compara-tive] Deg head can check off the features of the AP. Note that Lechner (2004:
23) claims that comparative morphology is base-generated directly on the A head, and therefore a string like younger cannot be syntactically decomposed intoyoungand the degree morpheme -er, contrary to Bresnan (1973), but in line with Izvorski (1995) and Corver (1997). As a matter of fact, Lechner (2004: 23) as-sumes that -ermorphology manifests a reflex of feature checking: this, however, selectively surfaces only on certain A heads, namely ones that are monosyllabic or bisyllabic. Hence, in the case of periphrastic forms (e.g.,more intelligent), the feature is claimed to be spelt out on Deg, resulting in the stringmore+ A.
This raises a rather compelling question in connection with periphrastic struc-tures, namely that if the comparative feature is spelt out on Deg in the form of more, then, according to the representation in (28), the string should actually be A
+more, e.g. *intelligent more, which is clearly not the case. Lechner (2004) leaves the derivation of the grammatical order unexplained. However, Lechner (1999:
25) originally proposed that in periphrastic comparatives the DegP is embedded under a QP. Thus, for a string likemore intelligent than Peter is, the structure in (28) should be modified in the way given in (29):
(29) QP
Q′ Q morei
DegP AP
intelligent
Deg′ Deg
ti
than-XP than Peter is
As can be seen, if there is a QP layer above DegP,morecan move up to the Q head position, thus resulting in the grammatical word order.
One advantage of the analysis given in (28), as Lechner (2004: 23) argues, is
“the dissociation of the surface position of -er from the location of its interpre-tation”. The problem of not separating these two becomes obvious when consid-ering theunhappier Bracketing Paradox, see Beard (1991), Pesetsky (1985) and Sproat (1992). This paradox lies in the observation that unhappier seems to be subject to two conflicting requirements. On the one hand, morpho-phonological rules would assign the following bracketing to the string (see Lechner 2004: 23, ex. 47a):
(30) [un [happier]]
The reason behind this is that -ermay only be attached to an A head that max-imally consists of two syllables, hence it must be attached prior toun-. However, this seems to produce the interpretation ‘not happier’ instead of ‘more unhappy’.
On the other hand, in order to derive the correct interpretation, the bracketing should be the one given in (31), see Lechner (2004: 23, ex. 47b):
(31) [[unhappy] er]
Note that in this case the morpho-phonological rules mentioned in connection with (30) are violated.
In order to overcome this problem, Lechner (2004: 23) proposes that the correct bracketing is the one in (30), but the interpretation of -eris not directly associated with its base position: it is a manifest of feature-checking, which involves the entire AP (unhappy).
With respect to the location of adjectival arguments, Lechner (2004: 26) makes use of some German data exhibiting such constructions to provide additional evidence for the structure he attributes to nominal comparatives. According to his analysis, the PP argument of an adjective is a complement of the adjectival head and it may be subject to right dislocation. Consider (Lechner 2004: 26, ex.
51):
‘since Hans is proud of his dog’
b. weil
‘since Hans is proud of his dog’
According to Lechner (2004: 26), the underlying order is the one indicated in (32a), building on the assumption that the AP is head-final; for such views, see for instance Haider & Rosengren (1998). As will be discussed later, taking such a stance is problematic not only in terms of maintaining a universal directionality of headedness (cf. Kayne 1994) but also because it may rather be the case that the German AP is in fact head-initial. Nevertheless, taking up the argumentation of Lechner (2004), (32b) is claimed to exhibit right dislocation of the PP argument.
However, if the AP is an attribute in a nominal expression, see (33),dislocation is not possible (Lechner 2004: 26, ex. 54):
(33) a. weil
‘since Hans met a woman proud of her dog’
b. * weil since
Hans Hans
eine a.f.acc
stolze proud.f.acc
Frau woman
getroffen meet.ptcp
hat has
[PPauf of ihren
her.m.acc Hund]
dog
‘since Hans met a woman proud of her dog’
As can be seen, the extraposition of the PP is ungrammatical; this leads Lech-ner (2004: 27) to conclude that extraposition is not permitted from a DegP that is an attribute within a nominal expression. The same is not true for the com-parative subclause: this can apparently be extraposed. Lechner (1999; 2004) in-troduces a special mechanism for it, by way of which the (original) comparative subclause ends in such a position that it is coordinated with the (original) matrix clause. Since this is clearly a kind of syntactic process that would go against stan-dard minimalist assumptions and also a problematic proposal inasmuch as com-paratives can hardly be considered coordinated structures (see Bacskai-Atkari 2010a), I will not present this part of Lechner’s analysis here.
Even if one disregards the problems related to the movement of the compara-tive subclause, further ones arise in connection with the analysis given by Lech-ner (1999; 2004). First, the treatment ofmore is highly disputable as it does not take into consideration that it is built up ofmuchand the degree morpheme. It is therefore also not straightforward how strings likeas many (books)should be analysed, whereas manyobviously cannot be considered atomic.
Second, the status of the QP is not clear either. Though on the basis of Lech-ner (1999) it ought to be geLech-nerated in periphrastic structures, neither LechLech-ner (1999) nor Lechner (2004) assume its presence in morphological comparatives. It appears that these contain merely DegP projections. On the one hand, this is a problem for a unified analysis of degree expressions as the maximal projections would be different, that is, either a QP or a DegP, without even implying any syn-tactic difference. More importantly, the absence of a QP layer leaves the question of where modifiers are located unanswered.
Last but not least, the treatment of PP arguments is far from being uncontro-versial, especially because Lechner (2004) takes it for granted that the AP is head-final and the PP underlyingly precedes the A head. The opposing view is quite substantially present in the literature; see for instance Webelhuth (1992). How-ever, there are serious problems with Lechner’s examples as well in the sense that the data as such are misleading. Consider:
(34) a. [PPAuf
‘Hans should be proud of his dog.’
b. % [PPAuf
‘Hans should be proud of his dog.’
c. Stolz
‘Hans should be proud of his dog.’
The data show the possible movement patterns of APs containing PP comple-ments in main clauses. The most typical order is the one in (34a), where only the PP moves to a position preceding the verbsollte. However, it is also possible to move the entire degree expression. In that case, the natural order is A + PP, as in (34c). If the PP precedes the A head, as in (34b), the clause is not accepted by all speakers, and speakers who allow it remarked that the adjectivestolz‘proud’
must be stressed, which indicates that the position of the adjective on the right is most probably due to information structural requirements (and is therefore not a neutral order). This is already problematic for Lechner (2004: 26), but the problem only increases with the speakers who do not accept (34b) at all, while Lechner (2004) would predict (34b) to be the unmarked case.
The apparent contradiction between (32) and (34) can be explained if we con-sider some basic facts about German clause structure. In simple terms, subclauses show the underlying word order SOV, the VP (and the TP) being head-final (Haider 1985: 34), whereas in main clauses the inflected verb moves to the top-most C (see Fanselow 2004: 30, following den Besten 1989, Richter & Sailer 1998:
133–134). The moved verb comes second in the clause; it tolerates only one pre-ceding constituent. This condition is satisfied in (34c), where the A head precedes the PP complement; however, in (34b) the word order is either the result of mov-ing two constituents before the verb (ungrammatical) or of the PP movmov-ing into a position above the AP (speaker-dependent), which is tolerated normally (by all speakers) only if the AP is contained within a nominal expression, as in (33b).
I will return to the question of why degree expressions differ in predicative and in attributive structures – for now, suffice it to say that the core problem concern-ing the data provided by Lechner (2004) is that they only seemconcern-ingly support his claim, but the desired word orders arise merely because he uses subclauses.
Apart from (34), the possibility of intervening modifiers also indicates that the order PP + A head cannot be the underlying one. Consider the examples in (35):
(35) a. Lisa
‘Liz is (really) proud of her husband.’
b. Lisa
‘Liz is (really) proud of her husband.’
In (35a), the adjectivestolztakes a PP complement and may optionally be mod-ified by an adverb such as wirklich ‘really’. In (35b), the adjective and the PP complement appear in the reverse order; since the adverbwirklichcan intervene between the two, it is obviously not the underlying order. This raises the question of where modifiers could be located in the analysis provided by Lechner (2004), indicating that his structural representation is far from complete.