Bai Ruixue

The study of the Double Object Construction has been one of the most heated topics in linguistics. Among the numerous issues related to the Dative and the Double Object Constructions, this paper focuses on the phenomenon of argument augmentation exemplified by the Dative & Double Object Construction (abbreviated as DOC) pair, “Mary baked a cake for John” and “Mary baked John a cake”, in which the verb “bake”, an inherently two-place predicate receives a new argument to appear as three-place in the DOC form. Some natural questions are: 1). How can we assert that some verbs are inherently three-place while some others are inherently two-place? 2). What is the motivation, or, how can it be possible for the inherently two-place verbs such as “bake”, which do not denote a recipient, to form a DOC which typically depicts a “transfer of possession”? 3). Are there any constraints on the verbs that can be augmented?

Drawing heavily on the materials and discussions from Larson (1988, 1990), I will argue that argument augmentation is primarily triggered by the syntactic construction of DOC itself, which has its own meaning as a whole and maps onto the verbs that originally do not convey a transfer of possession. This process of constructional mapping enables the two-place verbs to be augmented and thus to undergo the derivational alternation as proposed by Larson. In addition to this, some more constraints as well as motivating factors will be discussed in details.

0. Introduction

The following pair of sentences is quite common in English, in which the prepositional phrase “to Mary” is converted into the indirect object.

(1)a. John gave a book to Mary.

b. John gave Mary a book.

We call this syntactic alternation Dative Shift, the first sentence Dative form and the second one DOC or ditransitive form. The semantics of such sentences is quite clear: they interpret a transfer of possession, with “John” designated as Agent, “a book” as Theme, and “Mary”, Benefactive or Goal.

Given the obvious ungrammaticality of “John gave a book” as a sentence, we name the verbs like “give” as inherently three-place arguments to imply that they always require three NPs as their arguments, or, in other words, always assign three theta-roles. Its thematic grid can be illustrated in (2).

(2) Give {θAgent, θTheme, θBenefactive}

With alternating verbs like “give” in mind, we turn to look at the following examples:

(3)a. John baked a cake for Mary.

b. John baked Mary a cake.

In (3), the surface structure is just the same as in (1), in that the verb “bake” also undergoes Dative Shift to yield the DOC form. However, since it is safe to say “John baked a cake”, we can catch a significant difference between (1) and (3), that is, verbs such as “bake ” do not inherently denote a benefactive. In this connection, we assume that these verbs, prior to Dative Shift, undergoes another process known as Argument Augmentation, which augments a new argument, or, in other words, adds a surface valency to the verb. This process, as applying to “bake”, is symbolized as below:

[1] [2] [3] [4] 下一页  

(4) Bake1 {θAgent, θTheme}

↓

Argument Augmentation

↓

Bake2 {θAgent, θTheme, θBenefactive}

Larson defines this phenomenon in which a benefactive argument is added as Benefactive Augmentation, as differentiated from another type, Goal Augmentation. The latter can be illustrated by (5):

(5)a. He kicked a ball to her.

b. He kicked her a ball.

This type is restricted within motional events whereby the agent imparts a trajectory to the theme, as shown below:

(6)a. Kick1 {θAgent, θGoal}

↓

Argument Augmentation

↓

Kick2 {θAgent, θTheme, θGoal}

Larson discovers two types of augmentations as described above, Benefactive Augmentation and Goal Augmentation. They are the same in nature, except for the thematic roles assigned by different prepositions. Out of an economical reason, my discussion will be limited within a consideration of the Benefactive Augmentation only, and the case for the Goal Augmentation can be analogously derived along the same lines.

1. Definition of the inherently two-place verbs

So far we have touched upon the fundamental differences between two-place verbs and three-place ones. Nonetheless, it is not safe at all to fit a verb into either class just by semantic intuition or by examining whether it can take a single object, because, for instance, “He has given a book”, as a response to the question “What has he given to us?”, is acceptable for some people or under certain circumstances. Therefore, we have to find further evidences to demonstrate that verbs like “bake” inherently take two arguments.

Comparing (2) and (4), we find that the represented two classes of verbs only differ in that the latter take an NP as non-argument adjuncts. That is to say, in order to justify our classification, we only need to prove that “to-datives”[1] as in (1) are arguments, whereas “for-datives” as in (4) are adjuncts.

In this connection, Mccawley (1988) notes that a “for-dative” expression in underlying structure is external to the V with which it is combined, in view of the fact that the latter behaves as a unit with regard to all relevant syntactic phenomena.

The “do so” substitution is one of the syntactic phenomena for testing. As a rule, “do so” substitutes for an entire VP, and thus includes the complement (for it is argument) but excludes the adjunct. This is made clear by the following instances, in which a “*” symbol implies ill-formedness.

(7)a. John ate a banana and Mary did so, too.

b. John ate a banana in the morning and Mary did so in the evening.

上一页  [1] [2] [3] [4] 下一页  

c. *John ate a banana and Mary did so an apple.

According to this test, the dative “to” phrase acts as an argument, while the “for” phrase does not, as shown below:

(8)a. John gave a book to Mary and Bill did so, too.

b. *John gave a book to Mary and Bill did so to Janet.

(9)a. John baked a cake for Mary and Bill did so, too.

b. John baked a cake for Mary and Bill did so for Janet.

Apart from this test, Maccawley also notes that “to” and “for” cases differ with respect to passivization: while the dative “to” phrase can be converted into the subject of a passive as seen in (10), the dative “for” phrase generally does not make a fully acceptable subject of a passive, as seen in (11) and (12).

(10)a. Honest Oscar sold a car to my brother for $200.

b. Honest Oscar sold my brother a car for $200.

c. My brother was sold a car for $200 by Honest Oscar.

(11)a. Fred bought a birthday present for Ethel.

b. Fred bought Ethel a birthday present.

c. ? Ethel was bought a birthday present by Fred.

(12)a. Beth Ann made dinner for Clove.

b. Beth Ann made Clove dinner.

c. * Clove was made dinner by Beth Ann.

These two tests provide reasons to classify verbs like “give” as inherently three-place which take the indirect object as arguments and those like “bake” as inherently two-place which take the indirect object as adjuncts. For the sake of convenience, we label them as Va and Vb respectively.

2. Derivation of DOC revisited[2]

Before further looking into the augmentation phenomenon, a brief review of the derivation of the DOC form is in demand.

Larson’s account for DOC, as well as many other derivational approaches, is based on Baker’s Uniform Theta Assignment Hypothesis (UTAH, 1988). It states that the two arguments, which fulfill the same thematic function with respect to a given predicate, must occupy the same underlying position in the syntax.

Larson puts forward the idea that the Oblique Dative form and the DOC are both Surface Structures derived from a third structure (see 13). This structure does not exist at Surface Structure, it is an underlying grammatical form that, through two different sets of transformational rules, becomes either the Dative or the DOC form.

(13) John the book gave to Mary

The formation of the Dative form is a straightforward case of V raising, while the situation for DOC is more complicated. In the Dative form, the preposition fulfills two syntactic functions: it is there to serve as a medium through which the verb can assign Dative Case and it also acts as an indicator for the thematic roles of Goal or Benefactive. Larson suggests that, in the case of verbs that undergo Dative Shift, “to”’s function of showing Goal / Benefactive theta-role is undermined by the semantic content of the verb. Any verb that may undergo Dative Shift will have, as part of its semantic make up, the ability to assign Goal or Benefactive roles to its indirect object. This means that in Dative forms the preposition “to” is “reduced” to a Case Marker. Then, in a transformational process similar to passivization, in which the subject theta-role (usually Agent) is absorbed by the Passive morphology, the Dative Case assigned by “to” is absorbed, then the indirect object must receive Case from the verb, thus yielding the DOC form “John gave Mary a book”.

[1] [2] 下一页