The cases when the governing or dependent part of a modification is not present in the surface structure of the sentence but when it is, however, present in the meaning of the sentence are regarded as cases of ellipsis.
Types of ellipsis. We distinguish several types of ellipses in the tectogrammatical annotation. On the most general level, the following types of ellipsis are distinguished:
In the case of textual ellipsis, the lexical content of the omitted expression is always clear from the context and is (unambiguously) recoverable. The given modification was omitted since it had already occurred elsewhere in the text; it does not have to be repeated for the sentence to be fully understood.
Grammatical ellipsis is a term used for such instances of elision in which the elided expressions do not corefer (there is no coreferred node), or such instances in which certain expressions are (obligatorily) non-expressed in the surface structure of the sentence although they are necessarily present (at the tectogrammatical level) for grammatical and semantic reasons.
The following sections describe individual cases of ellipsis, depending on what kind of position is omitted. These are:
The textual and grammatical ellipsis are distinguished for each type separately. A separate subsection is devoted to ellipsis (of dependent expressions) in paratactic structures, which forms a special case mainly due to the possibility of the coordinated expressions to have a single (shared) modifier (see Section 12.3, "Ellipsis and the principle of shared modification in paratactic structures").
Representing ellipsis in the tectogrammatical trees. There are the following ways to represent ellipsis:
by adding a new(ly established) node into the structure, in essentially two ways:
(newly established) node with a t-lemma substitute.
A new node is inserted into the tectogrammatical tree in the position of the omitted expression and one of the t-lemma substitutes is assigned to it (see Section 4, "T-lemma substitutes").
All the relevant attributes of the newly established node with the t-lemma substitute are filled-in (e.g.
This way of representing ellipsis is used for grammatical ellipsis of the governing element (see Section 18.104.22.168, "Grammatical ellipsis of the governing verb" a Section 22.214.171.124, "Grammatical ellipsis of the governing noun" and Section 12.1.3, "Ellipsis of the governing clause") and both the textual and grammatical ellipsis of a dependent element (see Section 12.2, "Ellipsis of the dependent element").
A new node, which is a copy of another node (representing an expression present in the surface form), is inserted into the tectogrammatical tree in the position on the omitted element. The original node does not necessarily have to be present in the same tectogrammatical tree, it is also possible to copy nodes from preceding (or following) trees.
The node is copied as a lexical unit represented especially by its t-lemma, its grammatemes and a valency frame. The values of the following attributes remain the same as in the original node (they do not change):
is_name_of_person and grammatemes
numertype. The values of the rest of the attributes of the newly established node need to be checked and changed if necessary. The attribute values can either change or remain the same as the values of the original node.
This way of representing ellipsis is used for textual ellipsis of the governing element (see Section 126.96.36.199, "Textual ellipsis of the governing verb" and Section 188.8.131.52, "Textual ellipsis of the governing noun" below). Exceptionally, the copied node can be used for representing ellipsis of dependent elements (see Section 12.3.1, "Textual ellipsis of a non-obligatory modification in paratactic structures")
!!! No subsequent check of the grammateme values of the copied nodes have been carried out in PDT. The values of all the grammatemes of the copied nodes, which may in fact differ from the values of the original node, remained the same in most cases. E.g. the
degcmp attribute of the copied comparative in comparative constructions is supposed to be different form the original node (see Section 4, "Constructions with the meaning of "comparison""); however, the data may not be quite consistent.
!!! Also the values of the attribute
a/aux.rf of the copied nodes have been checked only partially. All the values of the copied nodes remained the same (and we suppose this is right) in those cases in which the governing node of the copied node has the same t-lemma as with the original node. On the other hand, we have tested that none of the copied nodes has more than one reference to a conjunction or preposition in its
a/aux.rf attribute. Especially the references to auxiliary and modal verbs have remained unchecked. For more on the
a/aux.rf attribute see Section 1, "Relation between the tectogrammatical level and the lower levels".
A newly established node differs from all other nodes by the value
1 filled in the attribute
is_generated. See Table 6.7, "Values of the attribute
Table 6.7. Values of the attribute
the node represents an element that was expressed in the surface form of the sentence.
a newly established node that has no counterpart in the surface form of the sentence
If no value is filled in the attribute
is_generated, we assume the value
by using a shared modifier of paratactically connected elements.
Ellipsis is particularly common in paratactic structures. However, in compliance with the principle of shared modification (see Section 6.1.1, "Shared modifier of paratactically connected elements"), often, no new node has to be inserted into the tectogrammatical tree in place of the omitted modification.
Shared modifiers are used for representing grammatical ellipsis (of a dependent modification) with paratactically connected elements (see Section 12.3, "Ellipsis and the principle of shared modification in paratactic structures" for details).