In linguistics, verb phrase ellipsis (VP-ellipsis or VPE) is an elliptical construction in which a non-finite verb phrase has been left out (elided). VP-ellipsis is a well studied kind of ellipsis,[1] although it appears to occur mainly in English.[2]
Contents |
The first thing to acknowledge about VP-ellipsis is that the elided VP must be a non-finite VP; it cannot be a finite VP. Further, the ellipsis must be introduced by an auxiliary verb (be, can, do, don't, could, have, may, might, should, will, won't, would) or by the infinitive particle to. The elided material of VP-ellipsis is indicated using subscripts and a smaller font:
Attempts at VP-ellipsis that lack an auxiliary verb fail:
Apparent exceptions to this aspect of VP-ellipsis may be instances of null complement anaphora, e.g. ?Bill tried to leave, and Jim tried, too.
The examples just produced have the antecedent to the ellipsis preceding the ellipsis, and ellipsis is operating across coordinated clauses, whereby these clauses (since they are coordinated) are equi-level. The following examples illustrate in this area that VP-ellipsis can operate forwards or backwards as well as downwards or upwards:
VP-ellipsis operates forwards when the antecedent to the ellipsis precedes the ellipsis. It operates downwards when the antecedent appears in a clause that is superordinate to the clause containing the ellipsis. Three of the four combinations are acceptable. VP-ellipsis is, however, impossible when it operates both backwards and upwards.
An aspect of VP-ellipsis that has been the cause of much theorizing occurs when the ellipsis appears to be contained inside its antecedent.[3] The phenomenon is called antecedent-contained ellipsis (=antecedent-contained deletion). Canonical cases of antecedent-contained ellipsis occur when the ellipsis appears inside a quantified object NP, e.g.
The apparent antecedent to the ellipsis is the VP in italics. The fact that this antecedent VP contains the ellipsis itself means that an infinite regress occurs. An infinite regress is, however, an impossibility, since it would mean that the ellipsis could never acquire full semantic content.
One means of addressing antecedent-contained ellipsis that is pursued in some phrase structure grammars is to assume quantifier raising (QR). Quantifier raising raises the quantified NP to a position where it is no longer contained inside its antecedent VP. An alternative explanation (i.e. an alternative to the QR analysis) pursued in dependency grammars is to assume that the basic unit of syntax is not the constituent, but rather the catena.[4] A catena-based analysis of sentence structure sees the antecedent to the ellipsis as a non-constituent catena. What this means is that no movement mechanism (e.g. no QR) is needed to account for the fact that an infinite regress is an impossibility. The antecedent to the ellipsis is a catena that does not in any sense contain the ellipsis.
A phenomenon that overlaps to an extent with antecedent-contained ellipsis is argument-contained ellipsis.[5] Canonical cases of argument-contained ellipsis have the VP-ellipsis appearing inside the subject argument (as opposed to inside the object argument). The 'forwards/backwards' examples above involve argument-contained ellipsis, and the following examples illustrate the phenomenon further:
The second sentence in each pair is an instance of argument contained ellipsis. The VP-ellipsis appears inside an argument of its antecedent predicate. Note that these cases do not involve antecedent contained ellipsis because the ellipsis appears in the subject argument, which is indisputably outside of the antecedent VP.
The mysterious aspect of argument contained ellipsis concerns the environments in which it is and is not possible, e.g.[6]
Comparing these two sentences with those immediately above, a fundamental question arises: When is argument contained ellipsis possible, and when is it not possible? The situation seems even more convoluted when one examines the following contrast:
Argument contained ellipsis is possible in the first pair of sentences, but not in the second. These data remain mysterious to the present day, since a satisfactory theoretical account of the contrast is lacking.