Review of Derivation and Explanation in the Minimalist Program |
|
|
| Review: |
Date: Fri, 31 Oct 2003 21:23:50 +0200
From: Kleanthes Grohmann
Subject: Derivation and Explanation in the Minimalist Program
Epstein, Samuel David and T. Daniel Seely, eds. (2002) Derivation and
Explanation in the Minimalist Program, Blackwell Publishing.
Kleanthes K. Grohmann, University of Cyprus
BACKGROUND
The title of this volume hits a tune similar to one that sounds
familiar to many and really goes back to the pre-minimalist era:
derivation vs. representation. It is particularly fitting, however, for
the minimalist project as laid out in considerable detail in Chomsky
(1995) and much subsequent work by Chomsky (such as Chomsky 2001) as
well as many other scholars (see e.g. Epstein and Hornstein 1999b,
Baltin and Collins 2001, Hendrick 2003 for collections of recent
appraisals). The way Chomsky paints the picture, the computational
system of human language, C-HL, is derivational in nature, proceeding
bottom-up by successive application of the operation Merge with items
first taken from the lexicon (the numeration or lexical array), and
then with items already merged, aka Move. Under the copy theory of
movement, syntactic objects are copied and re-merged, accounting for
movement, followed by deletion of the lower copy (presumably for PF-
reasons; see Nunes 1995, Hornstein 2001 for detailed discussion). In
most "standard" approaches to movement, Move (qua Copy plus Merge) also
creates a chain, well known from GB, that contains the moved element
and its copy/trace.
This aspect of C-HL in particular is suspect for defenders of a
representational nature of language, casting doubt on the hybrid
character of the system to derive dependencies and also represent them
on the grounds of a very basic minimalist echo: economy. If one goal of
linguistic theory is to explain the workings of grammar in an
economical fashion -- meant both ways, an economic apparatus adhering
to Occam's razor as well as economizing nature of language itself - why
posit a two-part analysis if one is enough? The argument is that we
need some kind of representational system -- whether expressed pseudo-
derivationally as a chain or representationally as a, well, chain -- so
we should be able to dump one of them. In the case of all but one of
the contributors to the volume under review, the desire to derive the
outputs.
This is an extremely simplified exposition, of course. Among other
things, it should be noted that not all minimalist work assumes chains
as ingredients of C-HL (Hornstein 2001, Kiguchi 2002, Grohmann 2003),
that the chain-thing can be handled differently, perhaps truly
derivationally (Nunes 1995, Boeckx 2003), and that the representational
argument is not as simple as just presented (Brody 1995). But the point
of this book is very clear: IF (many) researchers agree that the system
is or should be derivational, what is the explanatory power of the
system? I believe this question makes even more sense if we see the
Minimalist Program (MP) as a more or less natural development of a
good, explanatory framework, Government-and Binding Theory (GB), a
conjecture not shared by everyone, but by a considerable part of the
Principles-and-Parameters population (made explicit in Hornstein,
Nunes, and Grohmann, in progress, for example). The rationale follows
well-known inductive procedure: If GB offered explanations and if GB
was representational, can a derivational MP provide explanations?
OVERVIEW
The editors hit the book off with their "Introduction: On the Quest for
Explanation" (pp. 1-18). Here they present the issue of derivation
and/or representation and resulting explanatory value in slightly more
elaborate ways than I did above; they also provide a brief chapter-by-
chapter overview. (First note to the reader: both my presentation above
and the overview below are totally independent from Epstein and Seely's
introduction.) The contributors offer eleven chapters in alphabetical
order. References are cited after endnotes individually at the end of
each chapter. Apart from Acknowledgements (x) and a List of
Contributors (xi), the book features an excellent one-for-all index,
containing subjects, names, and languages (305-317). (Second note:
given that the book appeared in 2002, it is shocking to see that many
references to work published after, say, early 2000 have not been
updated at all, something that could have been done trivially with a
more dedicated editorial eye; apart from a general discrepancy and -
continuity among the individual reference section, this is even more
striking since some authors have managed to update a given reference X,
while others cite X in its pre-published form, which suggests that
although the chapters were submitted considerably earlier than 2002, it
need not have been the final word.)
Chapter 1 by Michael Brody is "On the Status of Representations and
Derivations" (19-41). Here he presents aspects of his own framework,
"elegant syntax" (a.k.a. "mirror theory" or "perfect syntax;" see
especially Brody 2000), pertaining to the representational-derivational
issue. Brody's discussion revolves around issues of "duplication" (in
the sense of the above-mentioned possible redundancy),
"restrictiveness" (that duplicated systems are by nature less
restrictive), and "explanation" (can a derivational approach be truly
explanatory?). Slightly simplifying here again, he also addresses weak
and strong versions of both the derivational- and representational view
of grammar.
Chris Collins is still busy with "Eliminating Labels" (42-64) in
chapter 2, which has floated around as a (rightly so) widely cited
manuscript for a few years now and seen the glimpse of publication
daylight in 2001. The main proposal is a redefinition of Merge not in
terms of projection (where a generalized Merge(X, Y) = {X, {X, Y}}),
but a pure set (Merge(X, Y) = {X, Y}). Not being able to talk about
intermediate projection levels (X-bar) is surely a welcome result from
a bare-phrase-structure perspective, but Collins also manages to
eliminate any notion of a label, including maximal projections. He
applied his system to areas that make traditionally heavy use of
mention to labels, namely X'-Theory, selection issues, the Minimal Link
Condition, and the PF-interface. His quest for explanation in a D-
approach is the derivation of well-established generalizations in these
domains without resorting to labels.
Chapter 3 by the editors, Samuel David Epstein and T. Daniel Seely,
suggests to understand "Rule Applications as Cycles in a Level-Free
Syntax" (65-89). So, just as Collins got rid of all labels, Epstein and
Seely get rid of all levels. Noting difficulties with the timing of
Spell Out (in the more recent minimalist system of Chomsky 2001), they
propose a radical departure from either the original single application
(Chomsky 1995) or the multiple/cyclic application (originally due to
Uriagereka 1999, but only more widely appreciated with Chomsky 2001 and
his related work in a phase-system): Spell Out applies after each
application of an operation ("transformational rule"). One interesting
side-aspect of this truly derivational system (getting rid of one thorn
in any theoretician's eyes, the Extended Projection Principle, for
example, on principled grounds) is the serious doubts it casts on the
assumption that cyclicity is packed into "phases" (see also Uriagereka
1999, Grohmann 2003).
John Frampton and Sam Gutmann develop a "Crash-Proof Syntax" (90-105)
in chapter 4. This is the fourth largely conceptual chapter in a row,
followed directly by one more. They explore a pretty simple question:
since MP is concerned with optimal design specifications that C-HL must
meet at the (PF and LF) interfaces, couldn't there be a way that only
considers derivations that conform to these design specifications and
excludes crashing derivations in the first place? Why do only those
derivations "win" (over others) that don't "crash" (fail to meet design
specifications at the interface)? Frampton and Gutmann propose one such
system that furthermore does away with any notion of "comparison of
derivation" and the like; again, in an economizing setting, this makes
a lot of sense (see Uriagereka 1998, Epstein and Hornstein 1999a,
Hornstein 2001, Grohmann 2003 on the relevance of economy and the form
of grammar).
Norbert Hornstein and Juan Uriagereka suggest the possibility of
"Reprojections" (106-132) in grammar in chapter 5. After a hiatus of
three chapters, Hornstein and Uriagereka mention the representation vs.
derivation question again (see BACKGROUND above) and remark on the
difficulty of deciding on one or the other, or possibly a combination
thereof, with honesty. They set the stage for their contribution with
an important suggestion as a means of decision: "The best kind of
argument for derivations involves the need for representations whose
information is required piecemeal - being vital at one point and
unwarranted at another" (106). This chapter deals with one such case of
"loss of information" (where a representationally enriched system might
really face difficulties): reprojection. Re-what? Well, not going with
Collins' definition of Merge above (see also Hornstein and Uriagereka's
revealing note 2 on p. 129), the label L of merging two objects X and Y
is either X or Y. Based on how projectionality is derived in X'-Theory,
or more recently bare phrase structure (namely, basically in a what-we-
need-is-what-we-get kind of way), it really is either X or way, but
never both. So a VP carries the label of V merged with D just because
(simplifying again). Hornstein and Uriagereka positively evaluate the
idea that a label may change in the course of the -- well yes,
derivation, not representation. They look at the syntax of binary
quantification and come to the conclusion that "binary quantifiers are
allowed to covertly reproject after meeting their syntactic
requirements" (128).
Chapter 6 is provided by Richard S. Kayne who looks at "Pronouns and
Their Antecedents" (133-166). In line with recent research on a
movement-heavy approach to traditional construal relations (citing
Hornstein 2001, among others, but not, however, work that led to e.g.
Boeckx 2003 or Grohmann 2003, which is equally relevant to Kayne's
interests -- and has surely been around prior to deadlines pertaining
to this volume), such as Conditions A and B of Binding Theory or
control structures, Kayne develops Condition C derivationally as well
(see also Zwart's contribution below). Ultimately, he argues for a "Big
DP"-approach under which the antecedent of a given coreference relation
moves out of a structure that contains the coreferent element as well.
The explanation coming out of this chapter is a rejection of
representational Binding Theory in favour of a derivational account.
"Scrambling, Case, and Interpretability" (167-183) are the issues
Hisatsugu Kitahara is concerned with in chapter 7. He proposes the
first serious analysis of scrambling within the phase-based Probe-Goal
system of Chomsky (2001), appealing to the distinction between the
(new) operations Agree and Match. His approach is "strongly
derivational" (like so many others in this volume) in which LF
interprets Case immediately upon its licensing (previously checking,
now valuing) and deletes it. Kitahara is now on home free, solving
long-known binding-theoretic properties/-blems without invoking
mechanisms that induce a violation of (the) Inclusiveness (Condition).
Chapter 8, by James McCloskey, investigates "Resumption, Successive
Cyclicity, and the Locality of Operations" (184-226). As can be
expected from this scholar, the paper is empirically rich and well
presented, and it is theoretically demanding and interesting. I would
even go so far and say that this chapter is a truly beautiful piece of
work which makes heads and tails out of intricate patterns (empirical)
with an elegant explanation (theoretical). The data under discussion
are instances of non-local A'-dependencies in Irish, though the
proposal is of course more far-reaching -- it's just that English
doesn't show such complex patterns. He looks at A'-relations (concerned
with both the A'-dependencies themselves and instances of A'-binding)
that could be rendered into English as something like, to use
McCloskey's example (184), 'He's the guy that they said they thought
they wanted to hire __.' What we see in Irish is a three-way distinct
complementizer depending on which kind of A'-dependency it's involved
in (the famous 'aL', 'aN', and 'go'; see McCloskey 1990). McCloskey
couches his analysis within recent extensions of MP (e.g. Chomsky 2001)
and explores in considerable detail the role of (and issues involving)
intermediate movement steps.
Norvin Richards, in chapter 9, argues for "Very Local A'-Movement in a
Root-First Derivation" (227-248). Working in a root-first approach to
the derivation (one implementation of a top-down, rather than bottom-
up, system), Richards explores and corroborates an interesting
prediction of an earlier paper of his: A'-movement, the traditional
least local type of movement, should in fact be very local (under
certain circumstances), much more in line with A-movement. In the
course of the discussion, Richards demonstrates that, at least in this
system, Move should also be preferred over Merge (contra Chomsky 1995
and much research that upholds something like the "Merge over Move
Preference" -- see also Castillo, Drury, and Grohmann 1999, and
especially Drury 1998 in a more Richards-like system, or rather one
that takes Phillips 1996 as its stepping stone). Along the way,
Richards argues in favour of dropping the need for numerations from the
theory of grammar, develops the operation Sinking, and discusses tough-
movement, vacuous movement, contained relative clauses in Japanese,
tense islands, scrambling in Persian, and extraction in Tagalog as
instances of very local A'-movement. The interesting aspect of this
chapter is that one can find explanation in derivation regardless of
whether it proceeds bottom-up or top-down.
Esther Torrego provides "Arguments for a Derivational Approach to
Syntactic Relations Based on Clitics" (249-268) in chapter 10. She
investigates subject-to-subject raising across experiencers in Romance
and addresses three syntactic problems associated with it, finding a
derivational solution derived from the Minimal Link Condition: Why can
subjects only raise across cliticized, but not fully nominal,
experiencers in French and Italian? Why can't they raise even across a
cliticized experiencer in Spanish? Why can French have an expletive
subject preceding an experiencer only when the experiencer is
cliticized (also in Spanish, with a null expletive subject)? (I might
have got these questions wrong; I'm not an expert on these issues at
all, Torrego's exposition including glosses is not always the clearest,
and the "problems" mentioned in the beginning are never returned to or
addressed as such again; then just take these questions as some of the
questions Torrego aims to account for.) A careful derivational
understanding of the Minimal Link Condition explains these cases, while
the condition as laid out representationally (possibly as stated
originally in Chomsky 1995, at least at first glance), fails to do so.
In the final chapter of the book, Jan-Wouter Zwart considers "Issues
Relating to a Derivational Theory of Binding" (269-304). It is similar
to Kayne's contribution in spirit, but adopts the strictly derivational
approach to C-HL of Epstein et al. (1998). In a way, though, it can be
thought of as a companion piece to Kayne's chapter, since Zwart deals
with those relations that Kayne is not concerned with, namely strictly
local anaphor dependencies. In a nutshell, Zwart analyses locally bound
anaphors as "the marked Spell Out of a generic variable referential
element ("PRONOUN")" (294) with the coreferentiality property acquired
in the course of the derivation (making this type of Spell Out
different from the notion of Copy Spell Out argued for in both
Hornstein 2001 and Grohmann 2003, again without mention of either
work). The big idea to take home here is that coreferentiality is not
the result of an interpretive mechanism, or an interpretive procedure
itself, but "a function of the operation Merge" (295) - in other words,
it receives a truly derivational explanation.
CLOSURE
I already threw in my general impressions of the book and some
shortcomings (mainly on the editorial/literature side: shortcomings in
updating references, lack of coherence in citing references, and
absence of some relevant references). I also added my own salt to some
of the chapters above. Therefore, I really want to close with only one
comment. Now that we have seen such a fine collection of arguments for
explanation in derivation, maybe the next collection could re-address
the question of representation vs. derivation in a new light (beyond
Lasnik 2001 and more generally the, in this respect, disappointing Part
I of Baltin and Collins' 2001 rather sensationally titled "Derivation
versus Representation"). It has become clear that this question doesn't
have a simple answer; and in the present volume, Hornstein and
Uriagereka are particularly honest in this respect: "if reprojection
should prove to occur grammatically, it would argue that grammars are
(at least in part) derivational systems" (106). Maybe it's not even
desirable to have one. But it would be interesting if we collectively
could address this question one day without rhetoric or prejudice.
Epstein and Seely's volume strongly suggests that (pardon my
Shakespeare) not all is rotten in the State of Derivation.
In sum, despite the negative aspects I managed to throw in here and
there, I think this is an excellent addition to any theoretically
minded linguist's library (though minimalistically tempted
syntacticians may get most out of it). I think it's a healthy sign if
some questions are asked over and over again. Good thing about the
contributors to this book is that they don't ask too many old
questions. They just take some aspects of the theory and run with it,
(by and large) each for him- and herself. The result is an intriguing
potpourri of some of the hottest researchers around today. Definitely
worth one's time.
REFERENCES
Baltin, Mark R. and Chris Collins, eds. (2001) The Handbook of
Contemporary Syntactic Theory. Oxford: Blackwell.
Boeckx, Cedric (2003) Islands and Chains. Amsterdam: John Benjamins.
Brody, Michael (1995) Lexico-Logical Form. Cambridge, Mass.: MIT Press.
Brody, Michael (2000) Mirror Theory: Syntactic Representation in
Perfect Syntax. Linguistic Inquiry 31, 29-56.
Castillo, Juan Carlos, John Drury, and Kleanthes K. Grohmann (1999)
Merge over Move and the Extended Projection Principle. University of
Maryland Working Papers in Linguistics 8, 63-103.
Chomsky, Noam (1995) The Minimalist Program. Cambridge, Mass.: MIT
Press.
Chomsky, Noam (2001) Derivation by Phase. In Michael Kenstowicz (ed),
2001. Ken Hale. A Life in Language. Cambridge, Mass.: MIT Press, 1-52.
Drury, John Edward (1998) The Promise of Derivations: Atomic Merge &
Multiple Spell-Out. Groninger Arbeiten zur germanistischen Linguistik
42, 61-108.
Epstein, Samuel David, Erich M. Groat, Ruriko Kawashima & Hisatsugu
Kitahara (1998) A Derivational Approach to Syntactic Relations. Oxford:
Oxford University Press.
Epstein, Samuel David & Norbert Hornstein (1999a) Introduction. In
Epstein & Hornstein (1999b), ix-xviii.
Epstein, Samuel David & Norbert Hornstein, eds. (1999b) Working
Minimalism. Cambridge, Mass.: MIT Press.
Grohmann, Kleanthes K. (2003) Prolific Domains: On the Anti-Locality of
Movement Dependencies. Amsterdam: John Benjamins. [November 2003]
Hendrick, Randall, ed. (2003) Minimalist Syntax. Oxford: Blackwell.
Hornstein, Norbert (2001) Move! A Minimalist Theory of Construal.
Oxford: Blackwell.
Hornstein, Norbert, Jairo Nunes, and Kleanthes K. Grohmann (in
progress) Understanding Minimalism: An Introduction to Minimalist
Syntax. Cambridge: Cambridge University Press.
Kiguchi, Hirohisa (2002) Syntax Unchained. Doctoral dissertation,
University of Maryland, College Park.
Lasnik, Howard (2001) Derivation vs. Representation in Modern
Transformational Syntax. In Baltin & Collins (2001), 197-217.
McCloskey, James (1990) Resumptive Pronouns, A'-Binding and Levels of
Representation in Irish. In Randall Hendrick, ed. Syntax and Semantics
23: Syntax of the Modern Celtic Languages. New York: Academic Press,
199-248.
Nunes, Jairo (1995) The Copy Theory of Movement and Linearization of
Chains in the Minimalist Program. Doctoral dissertation, University of
Maryland, College Park.
Phillips, Colin (1996) Order and Structure. Doctoral dissertation, MIT.
Uriagereka, Juan (1998) Rhyme and Reason: An Introduction to Minimalist
Syntax. Cambridge, Mass.: MIT Press.
Uriagereka, Juan (1999) Multiple Spell Out. In Epstein & Hornstein
(1999b), 251-282.
|
| |
ABOUT THE REVIEWER:
ABOUT THE REVIEWER
I'm interested in syntactic theory (esp. within Principles-and-
Parameters approaches) and comparative syntax (esp. Germanic, Romance,
Slavic, Greek) and have worked on different topics, such as wh-
constructions, left dislocation and resumption, cliticization, and
reverse locality effects, what I call "anti-locality." You find
personal and professional information about me on my homepage at
http://www.punksinscience.org/kleanthes. For more on PUNKS IN SCIENCE,
a project I'm involved in with Jeff Parrott from Georgetown University,
please go to http://www.punksinscience.org, laugh, and get in touch
with us. (The manifesto is pretty cool though, and we ARE serious!) I'm
also involved with the Ask-A-Linguist service offered by LINGUIST List,
where I am a member of the expert panel. Two previous reviews of mine
appeared on LINGUIST List: Three Types of Extraction, by Paul Postal
, and On the Nature of
the Syntax-Phonology Interface: Cliticization and Related Phenomena, by
Zeljko Boskovic .
|
|
| Versions: |
| Format: |
Hardback |
| ISBN: |
0631227326 |
| ISBN-13: |
N/A
|
| Pages: |
336 |
| Prices: |
U.S. $
77.95
U.K. £
60.00
|
| Format: |
Paperback |
| ISBN: |
0631227334 |
| ISBN-13: |
N/A
|
| Pages: |
336 |
| Prices: |
U.S. $
38.95
U.K. £
21.99
|
|
| |
|
|
