Ken Wexler (MIT)

Developing Phases, Verbal, Resultant State and Target State Passives, Inverse Copulas, Clefts

Location: Wang Center, Lecture Hall 2

This paper presents a biolinguistic solution to a long-standing puzzle in grammatical development, namely the question of why comprehension of cleft sentences like (1) is so late (not good until about age 7), along with experimental tests of the theory.

(1) it’s John that/who Mary pushed

No theory until recently has managed to fit the developmental patterns of such sentences. In this talk we’ll present a solution to the problem and link to many other problems of syntactic development, with experiments to test the prediction. The solution will come for free; no new assumptions about syntax or development will have to be made. The solution will involve several independent and well-attested pieces of syntax, including minimalist considerations of edges and phases and well-attested developmental principles. Following a standard procedure in attempting to solve a problem in linguistic development, we’ll start from an analysis, namely an analysis of clefts like (1) (Percus’). Namely (1) has an underlying form like (2).

(2)[the x that Mary pushed] is John

This analysis has the advantage of explaining the existence and uniqueness (maximality) presuppositions on the focus NP (John) in (1), since the definite determiner the in (2) has such properties. Percus proposes that (1) is derived from (2) by extraposing the relative clauses in (1) and spelling out the x as it.

If forms like (2) develop late, then clefts like (1) should develop late. Note that (2) is a “Specificational” sentence, or “inverse copula” in the sense of Moro’s classic work on copulas with DP predicates. The predicate appears before the copula and the subject after the copula, unlike the usual (non-inverse) copula, where the order is reversed. The problem now devolves into why (2) should be late in development. Moro shows that inverse copulas have quite different properties than non-inverse copulas. He proposes that copula sentences arise from small clauses selected by the copula. The first DP in the small clauses is the subject, the 2nd the predicate. Moro argues that an inverse copula arises from raising the 2nd (predicate) DP, while a non-inverse copula arises from raising the 1st (subject) DP. (2), then, involves predicate raising.

Invoking Bakers’ arguments that small clauses involve hierarchical structure, unlike Moro’s original proposal, we’ll take the subject DP to be higher than the predicate DP. Assuming that the 2nd (predicate) DP is the complement of a phasal (perhaps light verb v) head, we can see that this phasal head must be defective or non-phasal in order for the predicate (in its complement) to raise. There is extensive evidence that children until about age 7 think all heads of certain types are in fact phasal (Wexler’s Universal Phase Requirement). (I’ll very briefly review the extensive experimental evidence for UPR, especially passive development). Thus all inverse copulas, including forms like (2) will not develop as grammatical until about age 7. This immediately implies that cleft sentences like (1) will not develop as grammatical until about age 7, so that children will fail comprehension tests on cleft sentences, as they are well-known to do.

The immediate and surprising prediction of this for-free analysis is that children should not comprehend simple inverse copulas sentences like (3) until about age 7.

(3) The person who Mary pushed is John

They should actually fail even on identifying whether Mary pushed John or John pushed Mary. I will present experiments that shows this is exactly the case, children fail dramatically on (3) while doing very well on non-inverse DP is DP copulas like (4).

(4) John is the person who Mary pushed.

At the same time, we’ll see that relative clause constructions, which have quite similar surface structure to the copula sentences above, present no trouble at all to children of the relevant ages.

This is a stunning experimental result, one that never would have been predicted or even thought of without a serious theory of the internal representations and how they develop in children. DP is DP sentences, they seem very simple and should develop quite early. In fact, they do, if they are forms like (4), subject is predicate sentences. But just turn them around and make them predicate is subject sentences, like (3) and they are seriously non-understood until children are about 7.

They biological facts seem to be that children prefer (at least certain) categories to be phases, and they have to (biologically) mature or grow out of this stage. The genetic system underlies this growth. This is the type of solution that Biolinguistics looks for, as we attempt to integrate the theory of linguistics with the theory of biology.