In linguistics, ellipsis (from the Greek: ἔλλειψις, élleipsis, "omission") or elliptical construction refers to the omission from a clause of one or more words that would otherwise be required by the remaining elements. There are numerous distinct types of ellipsis acknowledged in theoretical syntax. This article provides an overview of them. Theoretical accounts of ellipsis can vary greatly depending upon whether a constituency-based or a dependency-based theory of syntactic structure is pursued.
Contents |
Varieties of ellipsis have long formed a central explicandum for linguistic theory, since elliptical phenomena seem to be able to shed light on basic questions of form-meaning correspondence: in particular, the usual mechanisms of grasping a meaning from a form seem to be bypassed or supplanted in the interpretation of elliptical structures, ones in which there is meaning without form. In generative linguistics, the term ellipsis has been applied to a range of phenomena in which a perceived interpretation is fuller than that which would be expected based solely on the presence of linguistic forms.
One trait that many types and instances of ellipsis have in common is that the appearance of ellipsis is optional. The occurrence of VP-ellipsis, for instance, is often optional, e.g. He will help, and she will (help), too. Whether or not the verb help is elided in this sentence is up to the speaker and to communicative aspects of the situational context in which the sentence is uttered. This optionality is a clear indication of ellipsis. At other times, however, ellipsis seems to be obligatory, for instance with cases of comparative deletion, e.g. *More girls were there today than girls were there yesterday. The second occurrence of girls must be omitted in this sentence (More girls were there today than were there yesterday). The obligatory occurrence of ellipsis complicates the analysis, since one can argue that obligatory cases are not really instances of ellipsis at all, but rather a null pro-form is involved. These aspects of the theory should be kept in mind when considering the various types and instances of ellipsis enumerated below.
There are numerous widely acknowledged types of ellipsis. Nine of them are mentioned and briefly illustrated below: 1) gapping, 2) stripping, 3) VP-ellipsis, 4) pseudogapping, 5) answer fragments, 6) sluicing, 7) N-ellipsis, 8) comparative deletion, and 9) null complement anaphora. One should note that there is no unanimity among experts that all nine of the mechanisms should indeed qualify as ellipsis. Most experts would agree, however, that most of the nine are in fact ellipsis. The discussion below takes their status as ellipsis largely for granted.
The example sentences below employ the convention whereby the elided material is indicated with subscripts and smaller font size.
Gapping occurs in coordinate structures. Redundant material that is present in the immediately preceding clause can be "gapped". This gapped material usually contains a finite verb. Canonical cases have a true "gap" insofar a remnant appears to the left and to the right of the elided material.
While canonical cases have medial gaps as in these two sentences, the gap need not be medial, and it can even be discontinuous, e.g.
While these two sentences again each have two remnants, the gapped material is no longer continuous. There are in a sense two gaps in each of the gapped clauses. Gapping has been thoroughly studied, and it is therefore reasonably well understood, although the theoretical analyses can vary significantly.
Stripping is also known as bare argument ellipsis. Many linguists take stripping to be a particular manifestation of gapping whereby just one remnant appears in the gapped clause instead of the two (or more) that occur in instances of gapping. The fact that stripping is limited to occurring in coordinate structures is the main reason why stripping is integrated into the analysis of gapping:
These examples illustrate that stripping is flexible insofar as the remnant in the stripped clause is not limited in function; it can, for instance, be a subject as in the first sentence or an object as in the second sentence. A particularly frequent type of stripping is not-stripping, e.g.
Not-stripping's status as a form of ellipsis can be debated, since the non-elliptical versions of these sentences are unacceptable. The key trait of ellipsis, namely, is that the both versions are supposed to be acceptable (the elliptical and non-elliptical version).
Verb phrase ellipsis (also VPE) is a particularly frequent form of ellipsis in English, absent from closely related languages but also used in Portuguese. VP-ellipsis elides a non-finite VP. The ellipsis must be introduced by an auxiliary verb or by the particle to.
An aspect of VP-ellipsis that is unlike gapping and stripping is that it can occur both forwards and backwards. That is, the ellipsis can both precede and follow its antecedent, e.g.
Of the various ellipsis mechanisms, VP-ellipsis has probably been studied the most and it is therefore relatively well understood.
Many linguists take pseudogapping to be a particular manifestation of VP-ellipsis (not of gapping). Like VP-ellipsis, pseudogapping is introduced by an auxiliary verb. Pseudogapping differs from VP-ellipsis, however, insofar as the elided VP is not entirely gone, but rather one (or more) remnants of the VP appear. This aspect of pseudogapping gives it the outward appearance of gapping. Pseudogapping occurs frequently in comparative and contrastive contexts:
Pseudogapping is more restricted in distribution than VP-ellipsis. For instance it can hardly occur backwards, i.e. the ellipsis can hardly precede its antecedent. Further examples:
Another noteworthy trait of pseudogapping (and one that supports the view that it is a type of VP-ellipsis) is that it absent from languages related to English.
The answer fragments of question-answer pairs involve ellipsis. The question focuses an unknown piece of information, often using a question word (e.g. who, what, when, etc). The corresponding answer provides the missing information and in so doing, the redundant information that appeared in the question is elided, e.g.
The fragment answers in these two sentences are verb arguments (subject and object NPs). The fragment can also correspond to an adjunct, e.g.
The ellipsis of answer fragments occurs in most if not all languages. It is a very frequent type of ellipsis that is omnipresent in everyday communication between speakers.
Sluicing usually elides everything from a direct or indirect question except the question word. It is a frequent type of ellipsis that appears to occur in most if not all languages. It can operate both forwards and backwards like VP-ellipsis, but unlike gapping, stripping, answer fragments, and pseudogapping, e.g.
The sluicing illustrated with these two sentences has occurred in indirect questions. Sluicing in direct questions is illustrated with the following two examples:
Sluicing has been studied intensely in the past decade and can be viewed as a relatively well understood ellipsis mechanism, although the theoretical analysis of certain aspects of sluicing remains controversial.
N(noun)-ellipsis occurs when the noun and perhaps accompanying modifiers are omitted from a noun phrase. N-ellipsis occurs with a limited set of modifiers in English (cardinal and ordinal numbers and possessive determiners), whereas it is much freer in other languages. The following examples illustrate N-ellipsis with cardinal and ordinal numbers:
And the following two sentences illustrate N-ellipsis with possessive determiners:
The fact that hers (as opposed to her) must appear in the second sentence could be interpreted to mean that the modifier that introduces the ellipsis is actually not an adjective or determiner, but rather it is a pronoun. Based on this observation, one could argue that N-ellipsis is in fact not actually a type of ellipsis, but rather the modifier serves as a pronoun of a sort, which means nothing has been elided.
Comparative deletion occurs in comparative clauses introduced by than in English. The expression in the comparative clause is elided that corresponds to the expression focused by a comparative morph such as more or -er in the antecedent clause, e.g.
Comparative deletion is different from many of the other optional ellipsis mechanisms insofar as it is obligatory. The non-elliptical versions of these sentences are unacceptable.
Null complement anaphora elides a complete complement, whereby the the elided complement is a finite clause, infinitive phrase, or prepositional phrase. The verbal predicates that can license null complement anaphora form a limited set (e.g. know, approve, refuse, decide). Interestingly, the elided complement cannot be a noun phrase.
Of the various ellipsis mechanisms, null complement anaphora is the least studied. In this regard, its status as ellipsis is a point of debate, since its behavior is not consistent with the behavior of many of the other ellipsis mechanisms.
Further instances of ellipsis that do not (in a clear way) qualify as any of the ellipsis types listed above:
With such data in mind, it is apparent that more work on ellipsis needs to be done before any sort of complete inventory of ellipsis mechanisms can be stipulated.
Theoretical accounts of ellipsis struggle. One reason why they struggle is that the elided material of many instances of ellipsis (e.g. the subscripted material above) often does not qualify as a constituent, the constituent being the fundamental unit of syntactic analysis associated with phrase structure grammars.[2] What this means is that formal accounts of ellipsis must seek some way of accounting for the fact that many of the ellipsis mechanisms enumerated above can elide word combinations that do not qualify as any recognizable unit of (phrase structure) syntax.
One widespread approach to the challenge is to assume movement (or some notion akin to movement).[3] What happens is that remnants are moved out of a greater constituent first so that the greater constituent can then be elided in full. By assuming movement first and ellipsis second, a theory of syntax can be maintained that continues to build on the constituent as the fundamental unit of syntactic analysis.
Another, more recent approach states that the challenges posed by ellipsis to phrase structure theories of syntax are due precisely to the phrase structure component of the grammar. In other words, the difficulties facing phrase structure theories are due the theoretical prerequisite that syntactic structure be analyzed in terms of the constituents associated with constituency grammars (= phrase structure grammars). If the theory departs from phrase structures and acknowledges the dependency structures of dependency grammars[4] instead, the ability to acknowledge a different sort of syntactic unit as fundamental opens the door to a much more parsimonious theory of ellipsis. This unit is the catena.[5] The assumption is now that ellipsis mechanisms are eliding catenae, whereby many of these catenae fail to qualify constituents. In this manner, the need to posit movement to "rectify" much of the ellipsis data disappears.