Local Movement and Pseudocleft Connectivity1

John Moore
UC San Diego

1 Local Movement

Generative Grammar has, for a long time, attempted to unify the two intriguingly similar phenomena of anaphoric binding and local movement; these are illustrated in (1):

(1)a.The children bathed themselves in the river.anaphoric binding
 b.The children were sent home t.local movement

If one looks at the history of how these phenomena have been treated, there is a sense in which the pendulum has swung back and forth between transformational and non-transformational devices. During the Standard Theory period, both anaphoric binding and local movement were derived through bounded, cyclic transformations; however, by the Government and Binding Theory of the mid-80s, both were constrained by the static Principle A of the Binding Theory, and even the movement part of local movement was treated metaphorically by some (e.g. Rizzi 1986, among others). Since Lebeaux (1983) and Chomsky (1986), however, there has been a tradition of assimilating anaphoric binding to local movement. In particular, Lebeaux and Chomsky observe that some languages, most notably Romance languages, make use of reflexive pronominal clitics to indicate anaphoric binding, as in the Spanish example in (2):

(2)Los niños se bañaron t en el río.
 'The children bathed themselves in the river.'

On the assumption that these clitics undergo local movement by S-Structure, Chomsky proposes that anaphoric binding in languages like English, with no visible cliticization, also involves local movement of an anaphoric element, albeit at LF. In other words, Chomsky treats English reflexives as LF-clitics:

(3)The children [Infl selfi-bathedk] tk ti in the river.

In this manner, the locality of anaphoric binding reduces to the locality of local movement. Chomsky further proposes that the relationship between the reflexive clitic and its subject antecedent is one of government (or Spec-head), as the clitic is assumed to move to Infl (as proposed independently for Romance clitics by Kayne 1989 and Rosen 1990, among others). Hence, the representation in (3) follows both LF cliticization and verb-raising. A similar movement approach to reciprocal binding is proposed in Heim, Lasnik, and May (1991).

Another local dependancy that has been interpreted as local movement is quantifier float. Since Sportiche's (1988) treatment of French quantifier float, it has been widely assumed that floated quantifiers are actually stranded quantifiers in a base, VP-internal position:

(4)The children might [VP [all t] bathe in the river].

This movement analysis of quantifier float is not universally accepted; see McCloskey (2000) for discussion and further references.

In this squib I will compare the behavior of overt and covert clitic placement as they relate to connectivity effects found in pseudocleft constructions. We will see that there is a fundamental difference in the connectivity effects exhibited by the the two, processes. These results might provide evidence that non-clitic anaphoric binding, as found in English, should not be interpreted as covert movement. I conclude with a technical solution that saves the LF cliticization approach; this solution, however, requires that the movement analysis of quantifier float be abandoned. Hence, the rather inconclusive conclusion will be that either or both English anaphoric binding or/and quantifier float do not involve local movement.

2 Pseudoclefts and Connectivity

Pseudocleft constructions, as illustrated in (5), are traditionally divided into three parts: the wh-clause subject, the copula, and the focus constituent:

(5)What they did wassign the papers.
 wh-claude copulafocus

Because the focus is a VP in (5), the wh-phrase includes a form of the pro-VP do.2 A salient feature of pseudoclefts is the phenomenon of connectivity between the wh-clause and the focus (Akmajian 1970, Hankamer 1974, and Higgins 1979, among others). This refers to the fact that certain strictly local phenomena may be split between the wh-clause and the focus:3

(6)a.What he was doing was washing himself.reflexive binding
 b.What they were doing was washing each other.reciprocal binding
 c.What the little bastards did was all get into the tub at the same time. quantifier float
(Hankamer 1974:10–12)

There have been two general approaches to pseudoclefts: Hankamer proposes a Standard Theory transformation that derives these clefts from simple clauses; under this approach, the connectivity facts follow if, as cyclic rules, reflexivization, etc. apply before clefting (which he argues is post-cyclic). The other approach, proposed in Higgins (1979), base-generates pseudoclefts and resolves the connectivity effects in the interpretive component. One way to make this concrete would be to assume that the focus constituent reconstructs to the wh-clause at LF, creating a local domain between the connected elements.

Something like the LF reconstruction approach is required if we adopt the LF-cliticization analysis of anaphoric binding described above. This is because the chain consisting of the clitic and its trace does not exist until LF; LF reconstruction subsequently resolves the locality of the chain:4

(7)a.What he was doing was [washing himself].S-Structure
 b.What he [self-was] doing was [washing t]LF cliticization
 c.What he [self-was] [washing t]LF reconstruction

3 Connectivity in Spanish Pseudoclefts

With respect to the restricted class of pseudoclefts under consideration, Spanish behaves very much like English, both in terms of the form of the clefts and connectivity effects:5

(8) Lo que hicieron fuefirmar los papeles.
  wh-clause copulafocus
  'What they did was sign the papers.'
(9)a.Lo que hicieron fue bañarse. reflexive binding
  'What they did was bathe themselves.' 
 b.Lo que hicieron fue lavarse el uno al otro.reciprocal binding
  'What they did was wash each other.' 
 c.Lo que los niños hicieron fue meterse todos en el baño.quantifier float
  'What the kids did was all get into the bath.' 

Thus, the LF reconstruction account proposed for English should directly carry over to Spanish. However, Spanish, unlike English, has overt reflexive clitics; these are commonly claimed to undergo overt movement. Thus, if the reflexive connectivity effect is resolved through LF reconstruction of a clitic-trace chain for English, then we would expect Spanish to exhibit overt clitic connectivity effects. That is, we should expect clitics in the wh-clause to correspond to arguments in the focus. As the data in (10) show, this is not possible—neither in the case of reflexive/reciprocal clitics, nor other object clitics:6

(10)a.*Lo que los niños se hicieron fue bañar. reflexive clitic
   'What the kids did was bathe themselves.' 
 b.*Lo que los niños se hicieron fue lavar el uno al otro. reciprocal clitic
   'What the kids did was wash each other.' 
 c.*Lo que lo hizieron fue comprar para su hermano.direct object clitic
   'What they did was buy it for their brother.' 
 d.*Lo que les hizieron fue pegar. indirect object clitic
   'What they did was hit them.' 
 e.*Lo que se quisieron hacer fue bañar. reflexive clitic climbing
   'What they wanted to do was bathe themselves.' 
 f.*Lo que lo trataron de hacer fue comprar para su hermano. direct object clitic climbing
   'What they tried to do was buy it for their brother.' 

Given the LF reconstruction account, there is nothing that would rule out these examples. The lack of locality and c-command between the clitics in the wh-clause and their traces in the focus is restored through reconstruction. Hence, we see a clear contrast in connectivity effects between overt clitics and the putative LF clitics—this may be interpreted as an argument against the latter.

4 An Alternative and Tentative Conclusions

The LF cliticization approach may be saved through an intrinsic ordering of LF reconstruction and LF cliticization. What the Spanish anti-connectivity facts show is that cliticization is strictly local. Nevertheless, it should be possible for overt clitics to undergo further local LF movement after reconstruction:

(11)a.Lo que los niños hicieron fue [bañarse t].S-Structure—local cliticization to verb
 b.Lo que los niños [bañarse t] LF reconstruction
 c.Lo que los niños [Infl se] bañar-t tLF cliticization to Infl

This derivation puts the clitic into Infl by LF through short, local movements. Note that this type of derivation is essentially guaranteed in a derivational framework that constrains local movement through an economy condition such as shortest move (Chomsky 1993).

If (11) is a possible derivation in Spanish, then the same type of derivation should be possible in English, excpet that both clitic movements would take place covertly—one before and one after reconstruction:

(12)a.What the kids did was [bathe themselves].S-Structure
 b.What the kids did was [self-bathe t] LF cliticization to verb
 c.What the kids [self-bathe t]LF reconstruction
 d.What the kids [Infl self] [t-bathe t]LF cliticization to Infl

This approach unifies the treatment of reflexive connectivity in Spanish and English, modulo the difference between overt and covert reflexive clitics. The ungrammatical examples in (10) are ruled out because movement has not been local at some point in the derivation—the clitic has moved too far overtly, before reconstruction created an extended local domain.

However, this analysis runs into problems with a movement account of quantifier float, as this movement necessarily involves overt movement from the focus to wh-clause:

(13)[What the little bastards did] was [[all t] get into the tub at the same time].

LF reconstruction comes too late to allow us to view this in strictly derivational, short moves.

There appear to be three solutions to this conundrum:

 (i)

Abandon the idea that cliticization involves movement. Rather, base-generate clitics in their S-Structure postion and impose a locality via some mechanism other than Shortest Move. I have argued for this treatment of clitics elsewhere (Moore 1994, 1996); I proposed that their locality be imposed by proper head government, which is not salvageable via reconstruction. Under this approach, quantifier float could move non-locally, with its locality restored through reconstruction. This solution also entails abandoning the abstract clitic analysis of English reflexives.

 (ii)

Maintain the movement account of clitics, but abandon the movement account of quantifier float (again, see McCloskey's 2000 discussion of this proposal). The locality of quantifier float would have to be achieved in some way other than Shortest Move, and whatever that mechanism is, it would have to be recoverable under reconstruction.

 (iii)

Abandon both the movement analysis of cliticization and the movement analysis of quantifier float.

What seems to be excluded is a theory that analyzes both cliticization and quantifier float as local movement.

References

Akmajian, A. (1970) "On Deriving Cleft Sentences from Pseudocleft Sentences", Linguistic Inquiry 1, 149–168.

Chomsky, N. (1986) Knowledge of Language, Praeger, New York.

Chomsky, N. (1993) The Minimalist Program, The MIT Press, Cambridge, MA.

Goldsmith, J. (1981) "On Spanish Pseudoclefts", Glossa 15, 3–15.

Guitart, J. (1989) "On Spanish Cleft Sentences", in C. Kirschner and J. Decesaris, eds., Studies in Romance Linguistics 17, John Benjamins, Amsterdam, 129–137.

Hankamer, J. (1974) "On the Non-cyclic Nature of WH-clefting", CLS 10, 221–233.

Heim I., H. Lasnik, and R. May (1991) "Reprocity and Plurality", Linguistic Inquiry 22, 63–103.

Higgins, F. (1979) The Pseudo-Cleft Construction in English, Garland, New York.

Kayne, R. (1989) "Null Subjects and Clitic Climbing", in O. Jaeggli and K. Safir, eds., The Null Subject Parameter, Kluwer, Dordrecht, 239–262.

Lebeaux, D. (1983) "A Distributional Difference between Reciprocals and Reflexives", Linguistic Inquiry 14, 723–730.

Moore, J. (1994) "Romance Cliticization and Relativized Minimality", Linguistic Inquiry 25, 335–343.

Moore, J. (1996) Reduced Constructions in Spanish, Garland, New York.

Rizzi, L. (1978) "A Restructuring Rule in Italian Syntax", in S. Keyser, ed., Recent Transformational Studies in European Languages, The MIT Press, Cambridge, MA, 113–158.

Rizzi, L. (1986) "On Chain Formation", in H. Borer, ed., Syntax and Semantics, volume 19, The Syntax of Pronominal Clitics, Academic Press, New York, 65–95.

Rosen, S. (1990) Argument Structure and Complex Predicates, Garland, New York.

Sportiche, D. (1988) "A Theory of Floating Quantifiers and its Corollaries for Constituent Structure", Linguistic Inquiry 19, 425–450.

Strozer, J. (1981) "An Alternative to Restructuring in Romance Syntax", in H. Contreras and K. Zagona, eds., Papers in Romance Linguistics, vol. 3, supp 2, 177–184

Endnotes

1 This squib is dedicated to Jorge Hankamer on the occasion of his sixtieth birthday. What made me think about pseudoclefts was Jorge's 1974 paper. Thanks to Ana Ardid Gumiel and Eric Bakovic for Spanish judgments; all errors and shortcomings are my responsibility. [Back]

2 Other constituents may be clefted; I will limit discussion to VP clefts. [Back]

3 An additional connectivity effect involves negative polarity items. [Back]

4 Alternatively, LF reconstruction could precede LF cliticization; a version of this approach is considered below. [Back]

5 See Goldsmith (1981) and Guitart (1989) for discussions of various ways that Spanish pseudoclefts differ from English ones. [Back]

6 In some cases the corresponding English clefts are ungrammatical because the focus is not a constituent. In Spanish, however, the focus is a constituent under the assumption that the clitic attaches to Infl and the verb remains in the VP. Even if these assumptions are abandoned, the clitic climbing example in (10e–f) cannot be due to non-constituent clefting. All the examples in (10) become grammatical if the clitics are left on the verbs in the focus constituent. [Back]