In the comments on a recent question of mine about copulas and Chinese it has been stated:

... the concept of "copula" is a medieval invention of Latin grammarians, and not a modern linguistic term.

Is this a generally accepted view in linguistics? I seem to read about copulae pretty frequently and I don't think I'm reading more mediaeval sources than modern ones lately.

My understanding was that in theory copulas have specific functions which differentiate them from other verbs in languages such as English, and in fact I've been under the impression that cross-linguistically copulas exist which are not necessarily partilcularly verb-like at all.

But perhaps I'm wrong, I'm not a trained or professional linguist after all, just an interested amatuer. Perhaps I have been noticing only articles at similar levels to Wikipedia and perhaps in real linguistics sources the concept of copula has been obsolete for some time.

Can anybody clear this up? Is copula a live concept in modern linguistics, is it dead, or has it been superceded with one or more modern concepts?

  • 1
    It's traditional in analyzing European languages, which all have centuries of written Latin in their history, whether they're Romance languages or not. But it's a feature of some languages and not others, and its roles, in those languages where something called a "copula" occurs, are diverse.
    – jlawler
    Nov 23 '13 at 18:18
  • 1
    Well the OED is a generalist prescriptive work (of which I am a great admirer and have a copy stored at home) so it doesn't prescribe how modern linguists ought to use the term, nor would it necessarily cover any and all special senses that might be currently evolving just within linguistics. But again I'm not an expert, I'm just curious to know ... Nov 23 '13 at 18:30
  • 1
    Quite right. This doesn't settle the question; it just gives general background.
    – jlawler
    Nov 23 '13 at 18:46
  • 1
    First you analyse the language, then if you need to deploy the notion of a 'copula' you do that, but only if necessary. You don't start off by looking for a copula. So, to revisit what @jlawler said, copula is a feature of some languages, but not others, and its roles vary greatly from language to language. Nov 24 '13 at 0:11
  • 1
    I guess the upshot is that the notion of 'copula' is still available if useful in the analysis of a language. It certainly still turns up in modern descriptions. Nov 24 '13 at 8:30

It has been noted by Nancy Levin in her 1980 paper ‘Main Verb Ellipsis in Spoken English’ that copula be cannot be elided under pseudogapping ((2); an example of pseudogapping is given in (1)).

(1) She will enjoy your book more than she will enjoy mine
(2) *She had been happy more often than she had been miserable

We can check now whether be of the progressive behaves differently than what is called “copula be” in terms of pseudogapping, i.e. whether the first can be elided under pseudogapping.

(3) She has been enjoying your book more than she has been enjoying my book

It seems to me that (3) is grammatical, but not being a native speaker of English I can’t be sure. If (3) is indeed grammatical, then the contrast with (2) might show that we should distinguish between a copula be and a progressive be based on the syntactic behavior of the two verbs.

It might be argued that the difference between (2) and (3) is the result not of two different types of be but of the different elements left behind: an adjective in (2) (miserable) and a noun phrase in (3) (my book). But Ed Zoerner and Brian Agbayani argue in their 2002 paper ‘A Pseudogapping Asymmetry’ (http://www.ledonline.it/snippets/allegati/snippets5007.pdf) that adjectives can be the remnant of pseudogapping. Also, it seems to me that pseudogapping of copula be is impossible even when the remnant is a noun phrase (but NP-remnants are possible when progressive be is pseudogapped).

(4) a. *It has been John’s house for longer than it has been my house
     b. They have been living in my house for longer than they have been living in your house.

In addition, adjectives can be remnants when they are not complements of copula be:

(5) She is wearing blue and I am wearing red

Finally, apart from this possible syntactic argument, it has been assumed in the literature that the meaning of the be in (6a) is different than the meaning of the one in (6b).

(6) a. She is a surgeon
     b. She is drinking tea

It has been assumed in the literature (e.g. Barbara Partee’s 1986 paper ‘Ambiguous pseudoclefts with unambiguous be’) that the be in (6a) is a function taking as arguments a predicate-type argument and an individual-type argument. This has never been suggested for the be in (6b). If there are indeed two different meanings, then there should be two different lexical items. (See also P Elliott’s answer here: Interchangeable arguments with English copula)

When developing our syntactic parser for English, we at Contextors chose to distinguish between several kinds of be: copular be, progressive be, passive be, among others. Our motivation was mainly practical: we found that making these distinctions helped us manage our rule system more efficiently. In addition, it helped in developing further applications based on our parser, for example a tense recognizer.

  • Very interesting answer! As a native speaker, i agree with your assessment that (3) is grammatical whereas (2) is not. Based on data such as this, and the assumption that pseudogapping targets a fixed projection, researchers have drawn the conclusion that copula be sits in a position above the ellipsis, whereas progressive be sits in a position inside the ellipsis. I agree that it makes sense to distinguish between different kinds of be.
    – P Elliott
    Dec 2 '13 at 16:51
  • @ P Elliott, thanks for confirming the data and for mentioning the conclusion regarding the different targets of ellipsis. I might add here that the fact that copula 'be' can undergo VP-ellipsis but not pseudogapping suggests that pseudogapping and VP-ellipsis are not the same phenomenon. This conclusion is also suggested by the fact that VP ellipsis tolerates voice mismatches between the elided constituent and its antecedent whereas pseudogapping does not (Merchant 2008).
    – Shai Cohen
    Dec 2 '13 at 20:00
  • The explanation you mention but discount (in terms of what is left behind by pseudogapping) may be the better explanation. In the unacceptable examples of pseudogapping you produce, the remnant includes part of the matrix predicate, whereas in the acceptable examples, the remnant is an argument of the matrix predicate. The adjective "red" in (5) is analyzable as an argument, I think. It is similar to "a sweater" in "She is wearing a sweater". In short, the remnant of pseudogapping should be an argument of the predicate; it should not include part of the predicate. Dec 2 '13 at 22:24
  • @TimOsborne 1/2: I agree with you that the function of the remnant in the clause is the relevant factor (whether “function” is defined solely based on syntactic category and position or not). What I tried to discount was an account on which what matters is the syntactic category of the remnant, for example: possible if a VP, not possible if an adjective phrase. The grammaticality of (5) indeed shows that (2) is ungrammatical not because the remnant is an adjective but because it is has a certain function in the sentence.
    – Shai Cohen
    Dec 3 '13 at 9:04
  • @TimOsborne 2/2: The theoretical question now is how to explicate the different functions of the remnants in the grammatical and ungrammatical examples. Theories derived from the Government and Binding framework account for the difference by assuming that the remnants in the ungrammatical examples are syntactic complements of a copula be whose position in the tree is different from the position occupied by a progressive be (see P Elliott's comment above). But of course, as you seem to suggest, other syntactic frameworks may account for the difference in other ways.
    – Shai Cohen
    Dec 3 '13 at 9:05

You can find the terms copula and copular defined in most linguistics dictionaries, and they are widely used in modern syntax textbooks (e.g. Radford 2004, Carnie 2013), so the brief answer to your question is yes, the term copula is still quite current.

Where I think confusion arises about copulae in English concerns "be" in particular. Like other copulae (become, taste, feel, look, etc.), be takes a predicative expression as its complement, but unlike these other copulae, copula be behaves like an auxiliary verb:

It allows subject-auxiliary inversion: a. Fred is hungry. b. Is Fred hungry?

It takes not as a postdependent: a. Fred is hungry. b. Fred is not hungry.

It licenses VP-ellipsis: a. Susan is hungry, and Fred is hungry, too. b. Susan is hungry, and Fred is, too.

What this all means is that copula be has the status of an auxiliary verb. In other words, it is both a copula and an auxiliary verb. I think many linguistics texts fail to acknowledge this trait of copula be. They tacitly assume that copulae are not auxiliary verbs. While that is true of most copulae in English, it is not true of be, which is by far the most frequently occurring copula in the language.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.