【摘 要】本文旨在解釋狀語附加語右嫁接的合理性。通過分析Larson和Kayne各自對狀語附加語左嫁接觀點解釋的方法，發現二者都存在不足。作者采用Ernst提出的Directionality Principle， Weight Principle等原則來分析狀語附加語的嫁接問題，在遵守普遍語法的經濟性原則的前提下，證明了右嫁接存在的合理性。

【關鍵詞】狀語附加語；右嫁接；左嫁接

Abstract: The complexities of the distribution of adverbial adjuncts in base structure have attracted many linguists’ attention. The main controversy on this issue is between the left-adjunction hypothesis and the Parameterized Direction Hypothesis with right-adjunction. The former represented by Larson (1988) and Kayne (1994) disallows right-adjunction in order to capture the general principle of UG. And the latter claims adjuncts normally and regularly adjoin to the right of maximal projections. With the theory proposed by Ernst， this paper is an attempt to provide arguments that the distribution of adverbial adjuncts in base structure can be best accounted for by PDH with right-adjunction.

Keywords: adverbial adjuncts;left-adjunction;right-adjunction

Introduction

Many linguists may be obsessed with the problem of the distribution of adverbial adjuncts in base position. Until recently， it was assumed in generative grammar that adverbials generally occur in adjoined positions which are on the edges of phrases.

Larson (1988)， Kayne (1994) and many others propose a current hypothesis that all languages have the same base structure and that head-initial/head-final word order patterns are derived by movement process. Therefore， all adjuncts adjoin to the left (or are in Spec position， which is to the left of heads) in base structure， any adjunct to the right of V in surface order must have obtained by a series of movement. I refer to this view as Left-adjunction Hypothesis.

But when encountered phrase like VP in (1)， the tree diagram below is quite natural.

(1) read this novel quickly

VP

VP AdvP

V NP

So many other linguists put forward that any postverbal adjunct to the right of some element X (a verb， argument， or other adjunct) is adjoined higher than X， just as preverbal elements are adjoined to the left， producing layered structure with the verb and its objects at the bottom and adjoined items attached upward on both sides. Thus this view supports the existence of right-adjunction. And this approach is called Parameterized Direction Hypothesis (PDH).

1 Views of Left-Adjunction

As far as adjuncts are concerned， views based on left-adjunction hypothesis come in two versions. The first dates from Larson (1988). And the second is a more promising one promoted in Kayne (1994) Alexiadou (1994).

1.1 The Defects of “Larsonian” Interpretation

There are some serious drawbacks to the Larsonian Interpretation of Left-adjunction making use of Logic Form Adjunct Raising (LFAR). It has no straightforward account of constituent structure， and its way of handling scope relations， LF-raising adjuncts， is fraught with stipulations and extra complications with respect to the traditional theory， which posits only one set of licensing conditions and no movements.

1.2 The Defects of Kayne’s Intraposition Analysis

Intraposition Analysis was originally motivated aiming at upholding a ban on right-adjunction and to eliminate the need to parameterize languages in terms of left-right relations.

On the theoretical level， the Intraposition Analysis is notably more complex and less motivated than the traditional right-adjunction approach.

If intraposition is semantically triggered， two of the problems carry over: there is no explanation for predicational adjuncts’ distribution， and the covariance of basic and adjunct orders is treated as an accident. In addition， the Intraposition Analysis encodes crucial syntactic restrictions into a semantic principle. Finally， the semantically based Intraposition Analysis forces an unnecessary bifurcation in the semantic licensing of adjuncts between base and LF structures.

2 Arguments for PDH with Right-Adjunction

Based on Ernst’s theory， I propose the analysis that maps adjuncts directly onto base structure in the positions they will have at the surface. The key to this approach is its basis in the C-complex (Complement-Direction) and F-complex (Functional-Direction)， each with an associated direction for XPs with respect to heads. These directions are invoked by a set of Directionality Principles to determine linear order， but they also have a role in deriving certain properties of movement. Moreover， the Directionality Principles make strong claims about the connection between the basic order of complements and that of adjuncts in a given language. Thus there are theoretical simplifications and restrictiveness to be found in PDH theory with right-adjunction as well.

The head-initial and head-final parameter plays an essential part in the discussion of the issue. Several set of data concerning the distribution of adverbial adjuncts needs illumination based on right-adjunction hypothesis.

(2a) Predicational adverbs are restricted to preverbal positions for their clausal readings in all languages.

(2b) Head-final languages generally require all adjuncts to be preverbal;

Head-initial languages in principle allow adjuncts on either side of V.

(2c) In head-initial languages， some adjuncts are restricted to preverbal positions， some to postverbal positions， and some may occur in either position.

In head-final languages， the inactivity of C-direction is very obvious， so only (2.1a) and (2.1c) bear some discussions.

2.1 Obligatory Left-adjunction of clausal predicational adverbs

The predicational adverbs straddle the syntax-semantics line: in syntax they may specify possible complements that are not semantically selected. In semantics the whole AdvP translates as a predicate ADJ (a “head”) taking its sister (complement) as an FEO argument. Now suppose that where syntax cannot mandate a direction for a nonhead (i.e.， where none of the syntactic requirements of the features [+Lex]， [+S]， or [+F] are in force)， C-Cir is always active for this purely semantic complementation， that is， where it is not realized according to the canonical syntactic X0-YP configuration. Predicational adverbs inside VP (manner adverbs) follow linearization， since the feature [+Lex] on V is at work， requiring the adverb to follow the same direction as complements.

2.2 Adverbial adjunction in head-initial languages

Within VP all adjuncts must be right-adjoined， because the relevant head X is of the category V (a lexical category bearing a C-complex feature)， the following tree diagram sheds light on my argument:

(2.2) IP

AdvP IP AdvP

DP INFL’

INFL PredP

AdvP PredP AdvP

Pred VP

ViVP AdvP

DP V’

ti

The ternary branching represents the two possibilities for binary branching and AdvP stands in for any adjunct. V is assumed to always raise to PredP， moving above direct objects in [Spec， VP]. F-Dir plays more important role in linearization， determining the leftward position of Specs (since they are [+F])， both landing sites in functional projections (as for subjects in [Spec，IP]) and (b) base positions of arguments of V within VP. Complements of functional heads (such as PredP) are to the right of their heads， since they bear [－F] and [+S] (a C-complex feature). Adjuncts within VP are necessarily to the right of V because they are [－F] and V is [+Lex] (a C-complex feature); combined with obligatory V-to-Pred movement， this in essence embodies the claim that the lexical VP is strictly head-initial in VO languages. Finally， adjuncts above VP are unspecified for direction， because neither they nor the heads of functional projections bear a C-complex feature. For head-initial languages (a) fix predicational adverbs in preverbal position， except for manner adverbs， which may also be postverbal， and (b) allow functional and participant adjuncts to occur on either side of V. However， in SVO languages， some functional adjuncts must be preverbal， and all participant adjuncts， along with some functionals， must be postverbal. In particular， it appears that the former group is made up of short， light adverbs， and the latter of larger， heavier phrasal categories — PPs， DPs， and CPs， of both participant and functional semantic types. This may be examined by Weight theory.

3 Left-Adjunction Versus PDH with Right-Adjunction

Given the following sentence like (3.1)， with the corresponding schema in (3.2)， the Left-Adjunction Hypothesis posits a structure like (3.3)， while the PDH theory with right-adjunction is illustrated in (3.4)

(3.1) She (often) had eaten (lightly) (on Sundays) (because of partying the night before).

(3.2) ADJUNCT1－ INFL － V － ADJUNCT2 － ADJUNCT3 － ADJUNCT4

(3.3 ) (3.4)

IP IP

A1IP A1 IP

NP INFL’ NP INFL’

INFL AuxPINFL AuxP

A2 Aux’ AuxP A4

Aux XP Aux XP

A3 X’XP A3

XVP XVP

A4 V’ VP A2

V V

(3.4) directly represents the surface order of the adjuncts in (3.1) and also correctly represents their scope relations in terms of c-command. (3.3) also represents the linear order directly but must resort to some other mechanism， such as raising adjuncts at the level of LF， to capture scope relationships. Taking the concentric phenomena of constituent structure， scope interpretation， and linear order into consideration， the right-adjunction hypothesis is obviously superior to the Larsonian Interpretation and the Intraposition Analysis because the Larsonian Interpretation has trouble to capture the layered constituent structure indicated by postverbal adjuncts correctly and the right-adjunction hypothesis appears less complex and more conceptually motivated.

Conclusion

I conclude that PDH with right-adjunction Hypothesis is superior to the left-adjunction hypothesis which disallows it at that level but then introduces movements to produce what amounts to right-adjunction on the surface. Despite proponents’ claims that Left-Adjunction are more restrictive than right –adjunction theories， I believe I have shown the opposite to be true with respect for intrapositions are examined closely and once a Directionality Principle account is elaborated for the right-adjunction theory， the latter turns more restrictive， simpler， and more highly motivated. If this held only for adjuncts， there might be a reasonable claim that other success of the left-adjunction should militate in its favor. Thus it seems that adjuncts are indeed allowed in principle to adjoin to the right.

References

［1］Alexiadou， A. (1997). Adverb Placement. Amsterdam: John Benjamins Publishing Company.

［2］Chomsky， N. (1995). The Minimalist Program. Cambridge， Massachusetts: MIT Press.

［3］Ernst， T. (2002). The Syntax of Adjuncts. Cambridge: Cambridge University Press.

［4］Larson， R. (1988). On the double object construction. Linguistic Inquiry 19: 335-391.

［5］Radford， A. (1997). Syntax: A Minimalist Introduction. London: Cambridge University Press.

［6］Stroik， T. (1990).Adverbs as V-sisters. Linguistic Inquiry 21:654-661.