The Analysis

Given the environment which licenses the use of -dé- and -dào in the previous chapter, we are now able to answer the two main issues in this paper fully. The first issue is the interpretation shift with some V-part. construction plus a bare noun phrase as the object of the verb. This is illustrated in (46) (originally noted in (2a) and (3a) in chapter 1):

(46) a. Ābǐ chī-wán nà-kuài dàngāo le Abi eat-finish that-Cl cake Asp.

'Abi ate up the piece of cake'

(Implication: There is no more cake left.) b. Ābǐ chī-wán dàngāo le

Abi eat-finish cake Asp.

'Abi finished eating cakes.'

(Implication: It's unknown if there is any cake left in the context.)

As illustrated above, the interpretation of the RVP -wán shifts from the the one in (46a) as expressing the total consumption of the piece of cake to the one in (46b) where -wán is more

likely used to indicate the termination of the activity. Whether there is cake left on the plate is unclear in this case.

The explanation of this resultative interpretation shift is rather simple. With the hypothesis of -dé-, we can use -dé- insertion as a test to figure out what is actually occurring with these two sentences:

(47) a. Ābǐ chī-dé-wán nà-kuài dàngāo Abi eat-dé-finish that-Cl cake.

'It is possible for Abi to eat up the piece of cake' b. ?? Ābǐ chī-dé-wán dàngāo

When sentences in (47) are the adjusted version of the sentences in (46) with -dé- inserted, the distinction is clear. It has been acknowledged that the V-wán construction, which takes the RVP of allowing potential mode, can acceptably occur with -dé- as shown in (47a). The two degrees required by -dé- (i.e., dx and ds)can be provided in the expression. Precisely speaking, when the degree of Abi's stomach capacity for a piece of cake (i.e., dx) is at least as much as the size of the specific piece of cake determined by the quantized NP (i.e., The size of the piece of cake is the ds), the sentence saturates the requirement of using -dé-.

On the other hand, the inaccessibility of (47b) to be with -dé- comes from the uncertainty of the value of ds. From the aspect of pragmatics, we cannot be sure that Abi's stomach capacity for a piece of cake by eating can be always as much as an uncertain size of an unspecified

piece of cake. The piece of the cake can be of many sizes. Therefore, when the size of the cake (i.e., ds) surpasses Abi's stomach capacity for the piece of cake (i.e., dx), the reading is incompatible with -dé-. Besides, the sentence will be interpreted as nothing but the termination of the action with cake left on the plate after Abi has finished his eating. Taking it from the other side, if the size of the unspecified piece of cake happens to be the same as or less than Abi's stomach capacity, (47b) would be compatible with -dé- and thus (46b) will be read as there being no cake left on the plate. This can be even further supported if we can adopt some strategies to clarify the size of the cake such as giving a context with fix-sized cakes or using grammatical manner to specify the piece of the cake as in the context as below:

(48) Abi's Father: Ābǐ chī dé wán zhèxiē dōngxī ma?

Abi eat dé finish these things Comp.

'Can Abi eat up all of these?' Abi's mother: Zhìshǎo tā chī dé wán dàngāo ba

at_least he eat dé finish cake Comp.

'He can eat up the cake at least.'

In (48), with the given context, the utterance spoken by Abi's mother uses V-wán + BNP (i.e., the underlined "cake" ), yet it still produces the reading of total consumption, which makes it compatible with -dé-. In addition to this, we also can adopt a grammatical method such as topicalization as below:

(49) dàngāo tā chī dé wán, zhèngcān jiù búyīdìng le

cake he eat dé finish, prandial_meal would hard_to_say Asp.

'Cakes! He always can eat them up. But it's hard to say for a regular meal.'

With topicalization, the specificity of the bare NP (i.e., underlined dàngāo ) is raised and thus is considered as some unique one in the context with a definite size as ds. As the ds may be the same or smaller than the eating capacity of the subject here, the sentence is licensed to use -dé-.

To give a short summarized answer to the question, the interpretation shift can be explained as follows: In most cases, V-part. constructions with RVPs like -wán have two degree points.

One is the dx, and the other is the referential ds. The reason that causes the interpretation shifts when the V-part. construction occurs with -wán and a bare NP is because the bare NP would make the degree of ds an uncertain one. Yet, the uncertainty can be recovered via contextual or grammatical methods to make a V-part. + Bare NP construction to have the same interpretation as those with a quantized NP object.

The second issue to be examined is the compatibility of potential mode. We have seen in many examples above, in which some V-part. constructions have potential mode when others do not. There are two major distinctive features between these two types of RVPs.

First of all, potentiality-dé- is an infix. It was shown in chapter 2 that some V-part.

constructions (i.e., V-guāng) should not be considered as compounds, and these non-compounding V-part. constructions would certainly be incompatible with the potentiality-dé- insertion. When this reason is rather descriptive than explanatory, the other motivation is solid.

By following the summary of section 5.2, we can find that events which are incompatible with potential mode should be categorized as CR. In this case, the event structure of V-part.

constructions without potential mode would be different from that of V-part. constructions with potential mode. In other words, V-part. constructions that do not allow -dé-/-bú- insertion have their event structures as one independent event consisting of a causing phase and a resulting phase. This kind of event structure is similar to what was illustrated with 'to paint the door yellow' in English with (41a) and/or the Mandarin fāhuáng ('to turn yellow') in (45a).

On the other hand, V-part. constructions that have potential mode with -dé-/-bú- insertion consist of two independent events with a causal relation. These V-part. constructions are similar to what we have seen with VR constructions like 'to push the door open' in English (40a) and/or wèibǎo ('to feed someone till his repletion') in Mandarin (44a). As a result, it is comprehensible for these two kinds of resultatives to have different compatibility with potential mode in the surface structure, since these two kinds of resultative semantics have different event structures within their underline forms.

hypothesis of the semantics of -dé- and the resultative categorization suggested by Kratzer (2004) to explain the two questions involving Mandarin V-part. constructions. First, the interpretation shift phenomenon has been explained with the inference of the bare NP that omits the standard degree ds for the dx to reach. The second phenomenon has been explained via two considerations – the rule applying restriction (i.e., A morphological rule cannot be applied in the syntactic domain and vice versa.) and the event structure difference between V-part. constructions. Consequentially, this proposal not only explained the questions raised in this paper, but also suggests that Mandarin, like German and English, has two types of resultatives. One is the eventual resultatives (ER) which merely represent the eventual result of some activity. The other is the consequential resultatives (CR) that indicate the resulting phase of one activity in consequence. Since these two types of RVPs are of different event structure, it is understandable for them to have different syntactic behaviors. Their syntactic structure and semantics will be revealed in the next section.