Linguistic constructions usually treated as consisting of similar parts as well as comparative constructions which resemble them in some respects require a special representation for their semantic interpretation. Semantic representation of the constructions of this kind must have a place for a unifying object (situation, action, or predicate) having two subordinate objects ("predicate places") regarded as homogeneous. The unifying object may arise from the words denoting comparison or some other correlation between homogeneous objects (e.g., coming across each other) or from a coordinating conjunction. The interpretation task is to restore semantic structures in the homogeneous locations attached to the unifying object.
These constructions are usually represented in language with a complete first and a reduced second part. The text of this kind could be regarded as a result of a particular transformation applied to initially complete representation of the second part; but being aware that the reductions like that are widely spread and facing the necessity of performing text analysis, the constructions described above are considered as an independent and regular way of expressing the unit allocated in the second place of the unifying object rather than a reduced form of something else.
On the syntactic level we'll consider that the reduced construction occupies a continuous portion of the sentence and consists of one or more continuous syntactic groups each not being able to find its syntactic partner in the remaining part of the sentence. A corresponding semantic structure is built for each of the groups. Further for filling up the second place of the unifying object a copy of its first place filler is made and then each of the semantic structures of the second part "supersedes" a certain part in the copy. This process means putting the superseding object with all its "relatives" in the place of the object being superseded; some extra links and attributes of the latter may be preserved only if compatible with its new links and attributes. The superseding object must be similar to the object being superseded (in simplest cases they belong to the same semantic class), and if the object being superseded is the direct referent of a word then the superseding syntactic group must fit the place of the word in the syntactic structure (i.e., it must have the required syntactic form). Indeed there are cases of superseding elements of semantic structures which are not explicitly expressed, e.g., modality of assertion.
This can be illustrated with the following example where an ambiguous interpretation results from the choice of the superseding element: "Ракета летела быстрее, чем думал конструктор" ("The rocket flew faster than the designer thought"), an example due to S.Fitialov. One of the interpretation variants ("the rocket reached a certain point earlier than the designer completed thinking something over) results from superseding the action "летела" ("flew") and the actor "ракета" ("rocket") from the copy of the first sentence (note that a reference to the time points being compared is preserved). In the second variant ("the rocket flew faster in reality than it did in the designer's estimation") both the action and the actor are being preserved in the copy of the first sentence but the modality of assertion is being superseded.
A similar techniques of superseding must be probably used in semantic interpretation of substitution operating when a substituting word and a word being substituted have different direct referents.