Is Isn't Be

The main claim of this paper is that the uninflected copula (be) and the inflected copula (is, am, etc.) in English are syntactically different entities: the uninflected copula heads a VP projection, while the inflected copula heads IP. That is, the inflected copula is not simply the inflected instantiation of a single underlying copula, raised from a V position. Evidence for the existence of a V-type copula and an Infl-type copula comes from a consideration of a semantic alternation in standard English make-complements, African American English (AAE) habitual be, Hebrew predicatives and omission of the copula in child English. By associating V but not Infl with the projection of an Event argument (Davidson, 1967), we can account for the eventive or active interpretation of complements with be and the lack of this interpretation of predicates with is.

Small clause complements of make may or may not contain an uninflected copula, as in (1).

(1) a. Max made Rodney polite.

b. Max made Rodney be polite. As discussed in Rothstein (1999) (1b) means that Max forced Rodney to act in a polite way, while (1a) means that Max coached or tutored Rodney and thereby made him into a polite person. Rothstein argues that the reason for this difference in meaning is that (1b) contains a verbal predicate, and all verbal predicates project an Event argument. (1a) contains an adjectival (nonverbal) predicate, and as such does not project an Event argument. It is the presence of the Event argument in (1b) that yields the eventive/stage-level reading.

Moreover, the predicate in (2b) has a somewhat less inherent or individual-level feel to it than the predicate in (2a) (Rothstein, 1999).

(2) a. Mark considers Tim very clever.

b. Mark considers Tim to be very clever. This semantic effect of the presence of be in these constructions supports the view that be is a verb and that it projects an Event argument (Rothstein, 1999; Sch"utze, 2000). The problem is that in contrast to clauses with be, no such interpretation is forced when the predicate occurs with the inflected copula.

(3) a. Rodney is polite.

b. Tim is clever. Thus, whatever property the uninflected copula has that induces an active/eventive meaning of the predicate, the inflected copula does not share this property. I agree with Rothstein that the relevant property of uninflected be is that it projects an Event argument, and it does so because it is a verb. The suggestion being made here is that the inflected copula does not project an Event argument because it is not a verb. Rather, it is generated in Infl as the pronunciation of finiteness features.

In AAE the be vs. is contrast plays out in main clauses. There is a clear semantic difference between main clauses with an uninflected (invariant) be and those with the inflected copula, which may be either null or overt.

(4) a. Sean ('s/is/0) tired.

b. Sean be tired.

(4a) means that Sean is tired at the time of utterance; (4b) means that Sean is habitually tired. Green (2000) argues that in clauses with invariant be, the predicate projects an Event argument which is then bound by a Habitual operator (hab). That invariant be occupies a V position (not Infl) can be shown by its position with respect to negation (it is lower than the inflected/null copula), and the fact that a tag question applied to a clause with this copula contains the auxiliary do, not be, as with main verbs (e. g. Sean be tired/like cake, don't/*ain't/*isn't he?). Since invariant be is a V it projects an Event argument, and in this it is like the standard English be of embedded clauses, illustrated above.

Hebrew also contains two syntactically distinct copulas. The copula in past and future tense predicatives (h. y. y) is a V, as can be shown by its tense inflection and its position with respect to negation (e. g. Rapoport, 1987) (in both respects it is like other verbs). In present tense predicatives the copular element is not a V (by the same criteria that show the past/future copula to be a V); it is the spell-out of agreement features in Infl.

Further support comes from child standard English (around age 2), in which the nonfinite version of is is arguably a null copula (0), not be. Children's main clauses may be nonfinite, unlike in adult English. In addition to the finite/nonfinite alternation in clauses containing main verbs (e. g. lady have/has a dress on), children's predicative constructions can be either finite (one finds an inflected copula) or nonfinite (there is no copula). Thus, I argue here that a null copula, not an uninflected copula, is the nonfinite form of the inflected copula: children almost never produce an overt uninflected copula in main clauses (less than 1%) (Becker, 2000).

One of the consequences of this argument is that it allows us to dispense with the stipulation (Chomsky, 1957, and much other work) that be and have are the only verbs in English that raise to I. That is, it allows us to say simply that no verbs in English raise from V to I (see Lasnik (2000) for discussion), because is is not a V to begin with
(although something more must be said about auxiliary have). Another welcome consequence is that we have a straightforward account of the ellipsis asymmetry noted by Lasnik (1999) between main verbs and the copula.

(5) a. John runs faster than Bill will [run]

b. * John is taller than Bill will [be] Ellipsis is possible in (5a) because the main verb is deleted under identity in the second clause; it is not possible in (5b) because there is no identity: be is not the same thing as is.

Another piece of data we can account for is the particular meaning of the so-called "active be" construction (John is being polite=John is acting in a polite way (Partee,
1977)). Here the added verbal copula (being) yields an eventive interpretation.

Apparent counterexamples are discussed. For example in free adjuncts containing stage-level predicates the presence of being affects the interpretation, but not by making it more eventive (Stump, 1985). Nevertheless, these data are unproblematic: in free adjuncts without being (e. g. Clean-shaven, John might impress the dean), the predicate is interpreted modally; no eventuality of being clean-shaven is actually asserted. In free adjuncts with being (e. g. Being clean-shaven, John might impress the dean), an eventuality is asserted to hold at the time of utterance, consistent with there being an Event argument projected by the predicate.

References

Becker, Misha. 2000. The Development of the Copula in Child English: The lightness of be. Doctoral  Dissertation, UCLA.

Chomsky, Noam. 1957. Syntactic structures. The Hague: Mouton. Davidson, Donald. 1967. The logical form of action sentences. In Essays on actions and events, 105-148.

Oxford: Oxford University Press.

Green, Lisa. 2000. Aspectual be-type constructions and coercion in African American English. Natural

Language Semantics .

Lasnik, Howard. 1999. Minimalist analysis. Malden, MA: Blackwell. Lasnik, Howard. 2000. Syntactic structures revisited . Cambridge, MA: MIT Press. Partee, Barbara. 1977. John is easy to please. In Linguistic structures processing, ed. A. Zampolli. Amsterdam: Holland.

Rapoport, Tova. 1987. Copular, nominal and small clauses: A study of Israeli Hebrew. Doctoral Dissertation,

MIT.

Rothstein, Susan. 1999. Fine-grained structure in the eventuality domain: The semantics of predicative

adjective phrases and be. Natural Language Semantics 7:347-420.

Sch"utze, Carson. 2000. Semantically empty lexical heads as last resorts. In Semi-lexical categories: On the content of function words and the function of content words, ed. Norbert Corver and Henk van Riemsdijk. Berlin: Mouton de Gruyter.