Peter Culicover and Paul Postal (eds.): Parasitic Gaps MIT Press, 2001 ix + 447pp. Reviewed by Masao Ochi, Osaka University

1.

Introduction

Over the past few decades, we have witnessed a rapidly growing body of works on Parasitic Gaps (PGs) in the generative literature.

Typical examples of PG

are shown below.

(1)

a.

Which article did you t file without reading pg? 1

b.

Who did a picture of pg please t?

A rapid increase in the number of works on PG and related topics is a natural consequence of the theoretical interest and importance attached to PG.

First, it

is a curious syntactic phenomenon exhibiting a cluster of perplexing properties. A notable feature of this construction is that one filler is associated with multiple gaps.

In this sense, PG is similar to, but distinct in several crucial respects from,

the across-the-board (ATB) movement construction (see among others Munn’s chapter in the book under review).

PG also presents a challenge for any known

theory of locality, as PG is typically licensed in a configuration in which extraction is in general not permitted.

(2a) contains PG inside an adjunct clause.

However, as (2b) shows, the same filler-gap relation in general results in ungrammaticality.

The same point can be made on the basis of data like (3), in

which PG occurs within the subject.

It is in this sense that PG is “parasitic”: it

1

Throughout this article, t is used to refer to a real gap and pg is used to refer to a parasitic gap.

1

depends upon the presence of a legitimate, real gap (RG) with which it (apparently) shares the filler.

(2)

(3)

a.

Who did you scold t before interviewing pg?

b.

*Who did you scold John before interviewing t?

a.

Who did a picture of pg please t?

b.

*Who did a picture of t please Mary?

PG is also important from a broader theoretical perspective, as pointed out by Chomsky (1982).

One of the most important and forceful arguments for

the linguistic nativism is what is called the argument from the poverty of stimulus. 2

Since the input that children are exposed to is insufficient with

respect to acquiring certain (if not all) properties/constructions found in the adult grammar, human infants must be equipped with innate mental capacities which guide them through the course of language acquisition.

PG is of interest from

this standpoint as it is precisely the kind of a phenomenon for which direct evidence is scarce for the majority of infants, which means that any generalization about PG that adult speakers converge on must be, to a large extent, due to the innate principles of grammar. Several competing hypotheses have been available in the literature for the PG construction.

Some representative hypotheses include a null operator

movement analysis of Chomsky (1986), a null pronominal analysis of Chomsky (1982) and Cinque (1990), and their variants.

I should also add that the recent

literature contains some interesting approaches to PG, such as Nunes (2001) and 2

For a recent discussion of the poverty of stimulus argument, see a collection of articles in the Linguistic Review Volume 19, 1-2 (2002).

2

Nissenbaum (2000).

2.

Culicover and Postal’s Parasitic Gaps appeared as a much-anticipated

book against this background.

In my view, this volume will prove to be

valuable for anyone interested in the inner workings of the innate linguistic component of the human brain/mind.

Here, I would like to highlight a few

points that are especially noteworthy about this book.

First and foremost, it is

the first comprehensive book exclusively devoted to this intriguingly complex phenomenon. volume.

Some chapters are true classics while others are written for this

Together, they comprise a nice collection of competing and

interweaving ideas, new and old.

Second, the two editors each make unique and

important contributions that make this book even more special.

In his

contribution, Peter Culicover traced the history of PG studies and provided a comprehensive and thorough summary of the relevant literature, which has made this book accessible to those who are not familiar with the past literature.

Paul

Postal, who has written numerous important articles on English syntax, contributed three essays, thereby adding to this volume a wealth of interesting data about PG.

Third, all the chapters in this volume will serve as a rich source

of important data, which is particularly helpful for researchers working on PG and related topics. This book consists of four sections.

Section I features some old ideas.

Culicover (Chapter 1) goes over the past as well recent literature and comes up with six common consensus positions (CCPs), which one can safely regard as current views about PGs.

Although none of the CCPs has escaped

objections in one form or another, it is important to recognize such robust

3

generalizations about such a marginal construction, as Culicover correctly points out.

Engdahl’s seminal work, published in the early 80’s, is included as the

second chapter.

It is a true classic in the PG literature, where many of the

CCPs were first proposed.

If one has little time to spare and wants to pick

just one chapter in this volume (other than Culicover’s detailed introduction, which is 60+ page long), this would be the one.

The major claim of Kiss’s

chapter (Chapter 3) is that the parasitic gap chain must share the antecedent with the real gap chain, from which it follows that the two chains need to transmit to its antecedent such syntactic features as Case (Case-matching condition), gender, number, and definiteness (which is easily detectable in Hungarian).

Kiss’s Case matching condition inspired reactions from

authors such as Nakajima and Horvath.

Some of the replies from Kiss ar e

included in this revised version of the article. Section II, entitled “what is a parasitic gap?”, starts with Engdahl’s other essay (chapter 4), which investigates a correlation between PG and pro-forms in several languages.

In response to Cinque’s (1990) and

Postal’s (1994) arguments for a correlation between the possibility of PG and the availability of pronouns in given contexts, Engdahl claims, on the basis of a range of Swedish data, that the correct characterization should b e stated in terms of pro-forms rather than pronouns.

Ouhalla’s chapter

(chapter 5) is likewise built on Cinque’ and Postal’s idea that PG is a null pronominal.

Ouhalla considers examples from Moroccan Arabic (MA) that

utilize an overt pronoun instead of a null gap. By showing that PG in languages like English and its overt counterpart (weak pronoun) in MA exhibit a parallel distribution, the author concludes that they should be

4

given a unified account.

The apparent differences between the two should be

attributed to the phonological content of the pronominals in the two languages, Ouhalla

argues.

Levine

et

al.

(Chapter

6)

provides

a

number

of

counterexamples to the prevalent views (largely due to Cinque and Postal) that (i) PGs are restricted to nominal categories, and (ii) PGs cannot occur in anti-pronominal contexts.

They discuss and challenge such views, concluding

that there is nothing unique about PGs.

A surprising aspect of Postal’s chapter

(Chapter 7) is his novel observation that a subset of the anti-pronominal contexts (contexts precluding the occurrence of weak definite pronouns) is relevant not only for PG but also for RG. Section III is concerned with “what is not a parasitic gap?”. two essays.

It contains

Postal (Chapter 8) goes to great length to show that the examples

in which PG is apparently licensed by rightward movement (such as heavy NP shift and right node raising) are not genuine instances of PG. discussion of PG in parenthetical sentences.

Of interest is his

In order to reconcile conflicting

evidence, Postal argues that such examples should be analyzed as involving both rightward movement and leftward extraction. related to Postal’s chapter in an obvious sense.

Kathol’s article (Chapter 9) is It is argued that multiple gap

constructions in German that seemingly correspond to English PG constructions should be best analyzed as instances of Pseudo-P gaps.

The discussion in this

chapter raises important questions about PG from a cross-linguistic perspective: Why is it that some languages allow (genuine) PG while others allow only pseudo PG? Section IV, “Restrictions on Parasitic Gaps”, contains four chapters. Tellier (Chapter 10) examines PG constructions in French and English,

5

presenting four aspects in which the two languages behave in a non-uniform fashion.

To be specific, PGs are disallowed in relatives, tensed adjunct clauses,

and tough-constructions in French.

Further, the two languages differ with

respect to the choice of the determiner when parasitic gaps occur within DPs. Adopting the null operator movement analysis of Chomsky (1986), Tellier presents several interesting arguments for the view that the null operator must be licensed via spec-head agreement with C or D.

Such an analysis has important

implications for the overall theory of PG, as I will discuss later.

Munn (Chapter

11) focuses on one of the fundamental issues about PG constructions, namely, a correlation between PG and ATB.

Munn’s claim is that PG construction

necessarily involves resumptive pronouns whereas ATB employs this strategy only optionally.

Munn demonstrates this point forcefully by showing a number

of similarities and dissimilarities between the two constructions.

Kennedy’s

short chapter’s (Chapter 12) focus is on Kim and Lyle’s (1996) interesting observation that VP-ellipsis apparently ameliorates island violations induced by alleged movement from the locus of PG.

As an alternative to their analysis,

Kennedy argues that such examples do not involve PG but a bound pronoun inside the elided VP by appealing to Fiengo and May’s (1994) notion of reconstruction and vehicle change.

In the final chapter of this volume, Postal

takes up this issue further, demonstrating that there are in fact genuine instances of PG in the ellipsis site.

3.

In the remainder of this review article, I would like to call attention to

locality issues arising from PG and consider their implications for a theory of grammar, drawing ideas from some of the chapters in this book.

6

In doing so, I

will make crucial reference to some of the CCPs that Culicover summarized in Chapter 1. Let us first consider CCP2, which states that PG must be licensed in overt syntax.

A strong argument for CCP2 is due to Kayne (1983).

As shown

in (4), PG typically occurs inside an island, and yet relevant sentences become degraded when PG occurs inside an island which is in turn embedded in another island.

If we assume with Huang (1982) that Subjacency constrains only overt

movement, such facts show that whatever occupies the PG-site must be licensed via overt movement. 3

(4)

a.

the man that Peter scolded t after John said that Mary invited pg

b.

*the man that Peter scolded t after John invited Mary without talking to pg

What is the landing site of this movement? becomes relevant here.

Tellier’s (Chapter 10) analysis

Based on the comparison of French and English data,

Tellier argues that PG must be licensed by the complementizer or the determiner that bears agreement features.

This point is important, for it motivates a

movement step from the PG site to the edge of the adjunct/subject domain. Now, let us turn to CCP5, which states that PG is in a chain with the antecedent of a true gap.

Proposing that CCP5 holds in the form of a

3

There is another sense in which PG must be licensed in overt syntax. RG, whose presence is required for licensing PG, must be created in overt syntax, as discussed by Engdahl (Chapter 2), among others. The contrast in (i) illustrates this point. (i)

a. b.

I wonder which articles you file d t without reading pg. *I wonder who file d which articles without rea ding pg.

In short, both PG and RG must be created/licensed in overt syntax.

7

cliticization process at LF, Ouhalla (Chapter 5) further argues that PG is a null resumptive pronoun, which (by definition) must be licensed via direct A-bar binding.

The key term here is “direct A-bar binding,” which is related to the

anti-c-command condition (CCP4), illustrated by data such as those in (5).

(5)

a.

*Who t remembered talking to pg?

b.

*Which articles t got filed by John without him reading pg?

Ouhalla’s idea is that such examples are ruled out because PG fails to be directly A-bar bound, due to the presence of a closer A-trace.

On the assumption that

PG as well as RG must establish a dependency relation with the same filler, Ouhalla’s insight can be reinterpreted as demanding that neither gap should be closer to the filler than the other, where closeness is calculated in terms of c-command.

According to this view, CCP4 is a consequence of the equidistance

requirement imposed on multiple gaps, where closeness is defined from the viewpoint of the target (Attract/Agree of Chomsky 1995, 2000).

This approach

may be a viable alternative to CCP6 that states that CCP4 is due to the Condition C of the binding theory. 4 Discussions in the preceding paragraphs seem to yield an interesting puzzle.

On the one hand, PG is sensitive to syntactic islands insofar as the

4

CCP6 and the Ouhalla-based approach sketched here overlap to a large extent in their empirical coverage, but not always so. For instance, consider the double object construction. Binding facts in (i) indicate that the two internal arguments c-command each other (see Burzio 1986). An analysis that defines closeness (to a target) in terms of asymmetric c-command would presumably predict that such arguments are equally close. Nevertheless, (ii) is bad, which would be predicted under CCP6. (i) (ii)

a. John showed Bob and Tom to each other’s friends. b. John showed each other’s friends to Bob and Tom. *Which person did you show e to e?

8

movement within the adjunct/subject domain is concerned.

On the other hand,

it is apparently legitimate to extract PG out of the adjunct/subject island.

One

would like to know how to reconcile these two observations 5 Several answers come to mind.

For example, Richards (1998) offers a

promising approach to this puzzle in terms of the Principle of Minimal Compliance. 6

Alternatively, one could entertain the idea that the PG can escape

an island precisely because of the first step of movement, that is, movement to the edge of the island.

For concreteness, let us adopt the analysis of

subject/adjunct islands by Takahashi (1994) and Saito and Fukui (1998) and interpret it in a derivational manner.

Let us reconsider (2) and (4), repeated

below as (6) and (7), respectively.

(6)

(7)

a.

Who did you scold t before interviewing pg?

b.

*Who did you scold John before interviewing t?

a.

the man that Peter scolded t after John said that Mary invited pg

b.

*the man that Peter scolded t after John invited Mary without talking to pg

According to Takahashi (1994) and Saito and Fukui (1998), the ungrammaticality of (6b) is due to the conflicting requirements imposed on the movement involved. On the one hand, a principle of the grammar pertinent to movement demands that a moving item goes through every potential landing site, which in the cases under 5

Cinque (1990) offers a possible answer to this puzzle. See Levine et al. (Chapter 6) for a critical evaluation of Cinque’s approach. 6 Due to lack of space, I cannot discuss Richards’ work in any detail. Let me just note that the principle he proposes, though general enough, does not explain why both RG and PG must be created in overt syntax, nor does it derive the anti-c-command condition without additional assumptions.

9

consideration would include the specifier position of before-clause.

On the

other hand, a movement step into the specifier of an adjunct phrase/clause violates another principle of grammar. 7

(6b) is then ruled out as it has no

derivation in which both of the principles are observed.

The same logic applies

to (7b) and excludes it: The null element base-generated in the PG-site cannot reach the specifier of after-clause without violating either of the principles introduced above.

In contrast to the (b)-examples, (6a) and (7a) are ruled in

because there is a legitimate derivation in which neither principle is violated. This is the derivation in which the movement to the spec/edge of the adjunct clause takes place before the latter acquires an adjunct status (see Saito and Fukui (1998) for much relevant discussion). 8

4.

Conclusion

Upon going over each chapter of the volume under review, the reader may be overwhelmed by a huge amount of intricate data, discussions that crosscut competing theoretical frameworks, and conflicting grammatical judgments on some of the core data.

One may even get the impression that no unified analysis

of PG is feasible in the near future.

As the editors note in the preface, however,

this book also shows us how much progress we have made in elucidating the nature of PG in the last twenty years or so.

There is no doubt in my mind that a

progress in research in this area will be (or has already been) facilitated by the 7

See Takahashi (1994) and Fukui and Saito (1998) for discussion of this principle. Whatever the answer is, we need to consider an implication of this two-step movement approach for the entire theory of movement. In particular, we need to ask why the null element can move further after moving to the edge of the adjunct/subject domain, in light of the well-known generalization tha t a n e le ment moved to an operator position is frozen for a furthe r move ment (see (i)). 8

(i)

*Who do you wonder bought what?

10

publication of this book.

References

Burzio, L. (1986) Italian Syntax: A Government and Binding Approach. Dordrecht: Reidel. Chomsky, N. (1982) Some Concepts and Consequences of the Theory of Government and Binding. Cambridge, MA: MIT Press. Chomsky, N. (1986) Barriers. Cambridge, MA: MIT Press. Chomsky, N. (1995) The Minimalist Program. Cambridge, MA: MIT Press. Chomsky, N. (2000) “Minimalist inquires: The framework,” in Step by Step: Essays on Minimalism in Honor of Howard Lasnik, ed. R. Martin, D. Michaels, & J. Uriagereka, 89-155. Cambridge, Mass.: MIT Press. Cinque, G. (1990) Types of A-bar Dependencies. Cambridge, MA: MIT Press. Fiengo, R. and R. May (1994) Indices and Identity. Cambridge, MA: MIT Press. Huang, C.-T. J. (1982) Logical Relations in Chinese and the Theory of Grammar. Ph.D. dissertation, MIT. Kayne, R. (1983) “Connectedness,” Linguistic Inquiry 14, 223-250. Kim, S and J. Lyle (1996) “Parasitic gaps, multiple questions, and VP ellipsis,” in The Proceedings of the Fourteenth West Coast Conference on Formal Linguistics, 287-301. Calif.: CSLI Publications. Saito, M. and N. Fukui (1998) “Order in phrase structure and movement, ” Linguistic Inquiry 29, 439-474.

11

Nunes, J. (2001) “Sideward movement,” Linguistic Inquiry 32, 303-344. Nissenbaum, J. (2000) Investigations of Covert Phrasal Movement. Ph.D. dissertation, MIT. Postal, P. (1994) “Contrasting extraction types,” Journal of Linguistics 30, 159-186. Richards, N. (1998) “The principle of minimal compliance,” Linguistic Inquiry 29, 599-630. Takahashi,

D. (1994)

Minimality of Movement.

University of Connecticut.

12

Ph.D.

dissertation,