The evidence for the treatment of directed manner of motion verbs as underlyingly unaccusative hinges largely on the causative alternation data in English, as I discussed in Section 4.2.2. In that section, I pointed out the lack of consistency in the causativisation data, indicating that causativisation as conceived of by cannot be productively applied to the manner of motion verbs. I suggested there that causative uses of particular verbs of manner of motion are lexicalised. In this section, I would like to informally discuss the possibility of an alternate conception of causativisation which may account for the semi-productivity of the causativisation process with respect to (directed) manner of motion verbs and the requirement of the directional phrase in this context and doesn't require that these verbs have an unaccusative argument structure or that the causative form be lexically encoded for all of these verbs.
As introduced in Section 4.2.2, propose that causativisation occurs when an empty external argument position is filled by an external cause, hence requiring that any verb undergoing causativisation have an empty external argument position. For manner of motion verbs, their causative forms are generated via a two step process: first the manner of motion verbs are shifted to a directed manner of motion sense, which simultaneously gives them the required empty external argument position and adds a directional phrase, and second an external cause is inserted into the external argument position. Thus from John walked we generate Billy walked John to the store via John walked to the store, and *Billy walked John will be ruled out.
In lexical semantic work (e.g. Jackendoff 1990), causativisation has been treated as a process which embeds an event representation inside of a causation function (cause([CAUSER], [CAUSED-EVENT])). The corresponding change in syntactic form for the causativised verbs results from different linking from the CAUSE function to surface form than from a simple event representation. The causativisation of John walked to the store, represented as in res32a, thus generates the representation in res32b which maps to Billy walked John to the store.
This causativisation process could conceivably be applied to any caused event (subject of course to the condition that the causation is felicitous given the discourse or world knowledge); the difficulty then is to explain the differences with repect to causativisation between manner of motion verbs and these same verbs on a directed motion interpretation, noted by and shown in examples res20 and res22, repeated below for convenience. This data suggests that manner of motion verbs are felicitous in a causative construction only on their directed motion use.
[4.18]
The soldiers marched (to the tents). The general marched the soldiers to the tents. ??The general marched the soldiers.
[4.20]
The mouse ran (through the maze). We ran the mouse through the maze. *We ran the mouse.
A possible explanation for the necessity of an explicit path in the event representation embedded by the CAUSE function in order to generate a felicitous causitivised form (that is, to rule out *Billy walked John) is simply that there exists a constraint on the process of causativisation which requires explicit reference to the end state or location which is being caused. This constraint can be justified through consideration of the function of the causativisation process: its purpose is to express the cause of some change of state or location by something or someone. If the end state or location is not explicitly mentioned in the representation of the event which is caused, then what is caused remains unclear and the meaning of the causativised form is underspecified and therefore infelicitous. The constraint would in effect restrict caused events to be telic events.
This account treats causativisation as a semantic process rather than a syntactic one, with the appropriate syntactic changes being affected via the linking between semantics and syntax. It accounts for the directional phrase requirement for causativised forms of manner of motion verbs via a semantic constraint on causativisation rather than the underlying syntactic argument structure of the verb to be causativised. The inconsistency of the application of this process to verbs of manner of motion still needs to be investigated, but it is likely that the infelicitous examples (such as res6b, repeated here) are a result of further semantic constraints on causativisation.
[4.22b]
#John
limped/hobbled/ambled/meandered/swaggered/
sauntered/sashayed/wiggled
the child to the store.
The causativisation at issue here is very different from that expressed by the resultative construction. Although the surface forms are parallel, the semantics are very different. Compare resalt39a with resalt39b.
Billy rolled the ball down the incline.
Billy rolled
The ball rolled
Billy sneezed the tissue off the table.
Billy sneezed
The tissue sneezed
The caused event bears a different relation to the causer in each case. In the resultatives (e.g. resalt39b), the subject of the sentence is both causer and participant in the main event expressed by the verb, while in the `true' causatives (e.g. resalt39a) the subject is a causer but otherwise plays no role in the event expressed by the verb. In the case of manner of motion verbs, the causer is perceived as having control over both the caused motion and the manner of motion that in the `true' causative case, but only the caused motion itself, and not the type of that motion, in the resultative case. Consider resalt40. In resalt40a John explicitly causes the motion of the horse in a jumping manner, while resalt40b conveys only that the letter changed location, and what caused that change in location (John jumping) but not the manner in which the motion occurs (i.e. the letter does not jump as it moves to the post office).
John jumped the horse over the fence.
John caused the horse to go over the fence in a jumping manner
(means of causation unspecified)
John jumped the letter to the post office.
John caused the letter to be at the post office by John
jumping (manner of motion of the letter unspecified)
Thus we are dealing with distinct phenomena, which share certain similarities, but which have differing interpretations and which are subject to different semantic constraints. I suggest that differences in the felicity of particular verbs in instantiations of these constructions (e.g. the manner of motion verbs in the casuative form vs. in the Resultative construction) stem not from underlying differences in the unaccusativity of the verbs in the various instantiations, but from these different semantic constraints.