(1) Bill slept the afternoon (entirely) away.

A wide range of verbs is possible in the construction-provided they have an intransitive subcategorization. No postverbal NP may be present other than the time phrase, nor may the verb be one that requires a direct object. It can invert with the NP, especially if the NP is a little longer. The only possible particle in this construction, away, can also be modified, in which case it does not invert as (1). The NP is a free time expression, permitting a variety of determiners, and they are apparently VP complements (in direct object position), as if it usurps this position so that the verb itself cannot license an NP there.

The 'time-away construction' requires the rest of the sentence to express an atelic situation or a bounded but repeatable event. It requires its subject to be acting volitionally and the verb to denote an activity not a state. A more subtle aspect of the semantics of the construction is that the subject is in some sense understood as 'using' the time, or even better, 'using the time up'. A further subtlety is reflected in the verbs: there is an insinuation that the activity was heedless pleasure, or that the subject should have been doing something else, or both. A final point of the description of the semantics of the construction: in the context of a journey, say a crosscountry flight, the time expression can be replaced by a distance expression D which is read as 'the amount of time it took to travel/traverse D'.

(2) a. *The light flashed the morning away.
b. * Emma elbowed an afternoon away.

(3) a. The light flashed its way into our consciousness.
b. Emma elbowed her way into the room.

Next consider the status of the direct object. In the 'time'-away construction, it behaves like a fully referential phrase: it undergoes passive and other movements such as tough movement; it can be pronominalized and questioned. By contrast, his way in the way construction cannot undergo passive and tough movement; it cannot be pronomizalized, ellipted, or questioned. Finally, the way construction permits the time period to be identified by an event taking place in it; the 'time'-away construction does not. In the proper context, the 'time'-away construction allows the period of time to be identified by a stretch of space traveled during that period; the way construction does not.

There are three reasons to distinct the resultative and the 'time'-away constructions. First, the object in the resultative is a patient: it is acted upon by the subject and affected by the action. But a time period cannot be affected by what someone does during it. Hence the object of the 'time'-away construction fails the do to test and satisfies only the weaker do with test. Another reason is the status of away. There is no resemblance in the semantics of the away in the 'time'-away construction and the aspectual particle away, which lends some of its properties and syntactically looks much like the intransitive resultative: the aspectual particle away describes the state at the end of the event and is telic, while the away in the 'time'-away construction describes nothing of the state and is atelic. Finally, there is nothing about the resultative that explains the peculiar selectional restrictions on the verb in the 'time'-away construction and the restriction to volitional subjects.

The constructional approach is more convincing for me. It claims that a construction idioms has its own specific meaning, so when a verb occurring in this construction, we can automatically get the meaning from the construction not from the verb. We do not add more semantic meanings to the verb as a lexical rule approach, which violates the principle of economy. Adding senses is also a burden of lexicon to memorize more than one senses of a verb. On contrast, in constructional approach, each verb has one sense and the added senses could be derived from the construction.