AUTHOR: Heck, Fabian
TITLE: On Pied-Piping
SUBTITLE: Wh-movement and Beyond
SERIES: Studies in Generative Grammar 98
PUBLISHER: Mouton de Gruyter
Joshua Bowles, Department of Linguistics, University of Utah
Let me begin by saying that I suspect Heck's monograph on Pied-piping (PiPing)
will become a widely cited resource for years to come. In other words, I would
be surprised and disappointed if it did not become an important work within
current research on PiPing and overall wh-movement. This is not to say that
everyone will (or should) agree with Heck's theory, only that his monograph
represents a substantial point of reference (or departure) for other generative
accounts of PiPing and gathers together a set of apparently disparate empirical
data under a small range of generalizations.
Heck's monograph is a revised and updated version of his 2004 dissertation from
the University of Tubingen; work done under the direction of Gereon Muller. It
should find a general readership within the broad spectrum of generative
approaches to linguistic theory. Specifically, those engaged with the various
theoretical outgrowths of the Minimalist Program and anyone interested in
Optimality Theoretic applications to syntax. The book is sectioned into six
chapters with each chapter containing many subsections.
Chapter 1 (pages 1-74) succinctly states Heck's argument for the systematicity
and regularity of PiPing in natural languages. Heck claims, contrary to most,
that PiPing is more common and systematic than has been previously recognized.
He uses this conjecture as the motivational force behind the need for a general
theory of PiPing; justification from empirical data comes in later chapters. The
bulk of the introduction is made up of brief sketches of other theoretical
accounts of PiPing. Heck reviews many of the ''[WH] feature percolation'' accounts
and finds them unsatisfying. He also reviews some non-percolation accounts,
including the dissertation work of Seth Cable, which was not available to Heck
during the writing of his own dissertation; see Cable (2008a,b) for brief
accounts. Needless to say, Heck does not find any past or present approach to
PiPing satisfying enough to extend it toward a general theory. He is
particularly concerned with showing that ''percolation cannot be reduced to Merge
or Move and must thus be stipulated as an additional operation of the grammar.
The conclusion is that it is conceptually attractive to eliminate wh-feature
percolation'' (page 7).
As a starting point for the framework of a general theory of PiPing, Heck adopts
the mechanisms and concepts of feature checking, probe-goal relations, and Phase
as defined in Chomsky (1995, 2000, 2001). It should be noted, as Heck is clear
to do, that he is concerned only with interrogative and relative clause
environments and does not look in any detail at the free relative,
wh-imperative, or wh-exclamative environments for PiPing.
Chapter 2 (75-181) proposes five generalizations of PiPing with the bulk of data
from Germanic languages (English, Danish, Dutch, German, Icelandic, Norwegian).
Other languages include Basque, Chamorro, Hungarian, Irish, Japanese, Mayan
(Chol, Tzotzil), Romance (French, Italian, Romanian), Slavic (Russian, Polish),
and others. Much of the interpretation of data is straightforward; stipulations
and conjectures are held to a minimum. I list here Heck's five generalizations.
1. Generalization on recursive pied-piping (76) If a wh-phrase A can pied-pipe a
constituent B, and if B is in a canonical position to pied-pipe C, then A can
also pied-pipe C.
2. Edge Generalization (88) If a wh-phrase A pied-pipes a constituent B, then A
has to be at the edge of B.
3. Repair Generalization (117) Pied-piping of B by a wh-phrase A is possible
only if (a) and (b) hold. (a). Movement of A out of B is blocked. (b).
Pied-piping of C by A out of B is blocked, where C is dominated by B and
4. Generalization on Massive Pied-piping (160) Massive pied-piping is only
possible if (a) and (b) hold. (a). The CP [complementizer phrase] whose
specifier is the target of (primary) wh-movement is not subordinated. (b). In
relative clauses, the pied-piper is selected by a D-element.
5. Intervention Generalization (173) Within a massively pied-piped constituent
no overt specifier may c-command the pied-piper.
Chapter 3 (183-210) extends Heck's theoretical vision for a general theory of
PiPing. In reference to the empirically grounded generalizations in the previous
chapter, Heck proposes what he calls ''Derivational Optimization'' (186, also see
Heck and Muller 2000, 2006a,b). Derivational Optimization is a hybrid generative
approach incorporating (i) standard Minimalist Program operations Merge, Move,
and Agree (feature-checking and probe-goal relations) as well as the concept of
Phase (Heck assumes Determiner Phrase, and possibly participial Verb Phrase
(224) are phases, along with the standard Complementizer Phrase and ''little v''
verb Phrase): all defined in Chomsky (1995, 2000, 2001); and (ii) the Optimality
Theoretic concepts of Optimality (that a syntactic object with the best
''candidate profile'' of the candidate set S and the ranked constraints R is
grammatical) and Constraint Profile (the best profile is the one that violates
the least number of high ranked constraints C in the set of ranked constraints
R) (185). This particular application of optimality theory to syntax emphasizes
the non-standard (in optimality theory) nature of locality and serialization of
the operations Gen(erate) and H(armony)-Eval(uation); that is, evaluation of
output candidates is standardly assumed to apply to all completed derivations in
a parallel and global fashion. This gives standard optimality its
representational structure. However, if evaluation is applied in a piece by
piece fashion, then this opens the door to a local and cyclic input-output
approach to derivations. It is worth quoting Heck in full about this hybrid system:
''[The] minimalist assumption that syntax unfolds in a derivational fashion on
the one hand and the assumption from optimality theory that the determination of
grammaticality involves input/output optimization on the other hand, consists of
interspersing them in a cyclic interleaving fashion: each phrasal cycle of
structure-building operations is followed by a cycle of input/output
optimization of different variants of the phrase in question. The output of
optimization then serves as part of the input for the next structure-building
cycle etc., until the numeration is empty'' (209).
Heck defines many more important theoretical terms in Chapter 3 (repeating some
from earlier pages), including Agree (191), Active Probe (191), Accessibility
(194), Edge domain (195), and Domain (195). The last three concepts are defined
in a recursive fashion, allowing goals to become accessible to their matching
probes at the edge of phase domains. Heck also introduces and defines various
optimal constraints, including FEATURE CONDITION (189), LAST RESORT (189), and
STRICT CYCLE CONDITION (193). Many of these optimality constraints are concepts
from derivational theories now adapted to work in an optimality format. The most
important optimality constraints to Heck's system appear to be LOCAL AGREE and
6. LOCAL AGREE (191) For every active probe B, there is a different matching
goal C in Z such that no XP [phrase boundary] dominates C but not B.
7. PHASE BALANCE (202) For every single probe B in the numeration there is a
different accessible matching goal C in the current phase Z.
These two constraints (coupled with the PHASE EDGE CONDITION given in Chapter 4
and defined below) typically force an unmatched goal to move to the phase edge
if it does not establish an agreement relation with a matching probe in the
current phase. By this, Heck must adopt non-feature-driven movement, but only to
intermediate positions. Movement to the final position is feature-driven by
agreement with the probe. It should be clear that Heck must adopt a multiple
Spell-Out model. Generally, Spell-Out must occur for each phase domain at every
cycle. Only material at the edge of phases (i.e., in the specifier position) is
accessible to the next derivational cycle; this material having been subject to
the optimality constraints on well-formed output representations, which are then
recycled as input to the next derivational cycle. Heck's hybrid system builds on
the notion of phase (and the idea that a derivation can in essence ''bootstrap''
an eligible goal as long as there exists a possible matching probe in the
numeration, i.e., an active probe) by developing specific constraints that place
PiPed material at the edge of each phase cycle.
Chapters 4 (211-296) and 5 (295-373) build on the theoretical background of
Chapter 3 in order to derive the empirical generalizations in Chapter 2 from his
hybrid system. Crucial is Heck's definition of the constraint PHASE EDGE CONDITION.
8. PHASE EDGE CONDITION (218) A wh-phrase A can be spelled out at the edge of a
phase P if and only if an Agree relation between the head of P and A has been
This constraint basically places a restriction on how far a goal can be
''bootstrapped'' and when it must be spelled-out. It says that once a matching
probe is introduced into the derivation (from its active state in the
numeration), an agreement relation with the goal can (must?) be established.
This goal can (must?) now be spelled out and no longer needs to be
non-feature-driven to phase edges.
The two PHASE constraints are key players in Heck's hybrid derivational
optimization; both are interpreted as inviolable, but need not be. (And in fact
are not in all cases; see Heck's discussion of the possible re-ranking of the
standard optimality theory EMPTY OUTPUT CONSTRAINT to account for variation of
inviolable constraints, namely the PHASE EDGE CONDITION, related to apparent
clausal PiPing in Basque (237).) It is this cyclic movement of wh-material to
phase edges, and subsequent cyclic movement of PiPed material to phase edges,
that is defined through the PHASE and AGREE constraints and constitutes Heck's
general alternative to PiPing; compare to fairly standard wh-percolation
accounts or other alternatives such as Cable (2008a,b); see also Hagstrom (2008).
The remainder of Chapters 4 and 5 consist of a detailed look at (optimality)
derivations and the application of Heck's constraints and theoretical concepts
to empirical data (mostly taken from Chapter 2). Chapter 4 is restricted to the
first three generalizations (1-3), termed non-massive, and Chapter 5 to the last
two (4-5), termed massive. This grouping is more than convenient and implies the
possibility of some further restrictions, reductions, or explanations within the
(non)/massive groups. Finally, CHAPTER 6 (375-377) is a very brief conclusion,
summarizing the basic arguments and their rationale.
Overall, Heck is honest and clear, aware that no theory is perfect. He never
appears to operate under the illusion that his theory can explain everything;
and even points out data that is problematic for which he can offer no
substantive solution at the moment.
I will keep critical discussion to a minimum, focusing only on the general issue
of Heck's hybrid approach to syntax (i.e., derivational optimization). There is
a lot (perhaps too much) that can be said about the debate between derivational
versus representational approaches to syntax; or Minimalism versus Optimality.
Note here that a Principles and Parameters system is probably equivalent to an
Optimality system that contains all and only inviolable constraints (stress
''probably''); see Legendre (2003) and brief comments in Kager (1999). The two
approaches are, at the very least, conceptually linked and need not always be
viewed as antagonistic to each other.
The crucial difference is that while (most) minimalist approaches consider
natural language syntax to be built bottom-up in a derivational, step-by-step
process, optimality syntax assumes well-formedness conditions on output
representations that are ranked relative to universal competing constraints that
are surface-violable; the optimal (winning) candidate being the grammatical one.
However, output representations can be built by derivations (cf. Ackema and de
Hoop 2006, and also interesting comments in Hale 1999). Within Minimalism it is
widely agreed that at least two representational levels exist, Logical Form and
Phonological Form. Furthermore, empirical data suggest that some aspects of
syntactic derivations may be representational, resulting in well-formedness
conditions imposed internal to those derivations (see Hornstein et al. 2005 for
the ''weakly derivational'' approach and Lasnik 2001 for discussion).
Well-formedness conditions internal to derivations can lead to a hybrid system,
and although such hybridity need not incorporate the representationalism of
Optimality Theory, one might wonder ''Why not?'' as many syntacticians, including
leading minimalist ones, already have (remember that Minimalism is not a theory
itself, but a research program intended to incite competition between theories
under the umbrella of general assumptions about the nature of I(nternal)-language).
Given these comments, Heck's hybrid system should not surprise anyone, nor need
it offend purist sensibilities. The concepts he employs appear to be
interpretable in either framework. I would not be surprised to see future
re-interpretations of his hybrid system into strictly derivational or
representational theories (e.g., in Heck 2009 the optimality theory has been
somewhat smoothed over). As Heck claims, central tenets to the Minimalist
Program and Optimality Theory are not ''incompatible but rather orthogonal to
each other'' (209).
Stepping aside theoretical and methodological issues, Heck has clearly done a
tremendous job in gathering together a typologically diverse set of languages
and providing a small range of generalizations that account for the apparent
variety of PiPing. Overall, the book is written well. There are, however,
numerous typographical errors, but they are not so bad that they get in the way
of appreciating the analysis.
Ackema, Peter, and Helen de Hoop. 2006. Current issues in optimality theoretic
syntax. _Linguistics_ 44:873–887.
Cable, Seth. 2008a. Q-particles and the nature of WH-fronting. In
_Quantification: Universals and Variation_, ed. Lisa Matthewson, 105–178. North
Holland Linguistic Series.
Cable, Seth. 2008b. Towards the elimination of pied-piping: Evidence from
Tlingit. Ms., University of Massachusetts, Amherst.
Chomsky, Noam. 1995. _The minimalist program_. Cambridge, MA.: MIT Press
Chomsky, Noam. 2000. Minimalist inquiries: The framework. In _Step by step:
Essays on Minimalist Syntax in honor of Howard Lasnik_, ed. Roger Martin, David
Michaels, and Juan Uriagereka, 89–155. Cambridge, MA: MIT Press.
Chomsky, Noam. 2001. Derivation by phase. In _Ken Hale: a life in language_, ed.
Michael Kenstowicz, 1–52. Cambridge, MA: MIT Press.
Hagstrom, Paul. 2008. Concerning Cable (2008) and the (non)existence of
pied-piping. Ms., Boston University.
Hale, John. 1999. Representational versus derivational approaches to syntax:
Optimality theory and chains.
http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.37.1827, Ms., Johns
Heck, Fabian. 2009. On certain properties of pied-piping. _Linguistic Inquiry_
Heck, Fabian, and Gereon Muller. 2000. Successive cyclicity, long-distance
superiority, and local optimization. In _WCCFL 19: Proceedings of the 19th West
Coast Conference on Formal Linguistics_, ed. Roger Billerey and Brook Danielle
Lillehaugen, 218–231. Sommerville, MA: Cascadilla Press.
Heck, Fabian, and Gereon Muller. 2006a. Derivational optimization of
wh-movement. Ms., Univeristy of Tubingen.
Heck, Fabian, and Gereon Muller. 2006b. Extremely local optimization. In
_Proceedings of WECOL_ 34, ed. Erin Bainbridge and Brian Agbayani, 170–182.
University of Fresno, CA.
Hornstein, Norbert, Jairo Nunes, and Kleanthes K. Grohmann. 2005. _Understanding
minimalism_. Cambridge: Cambridge University Press.
Kager, Rene. 1999. _Optimality theory_. Cambridge: Cambridge University Press.
Lasnik, Howard. 2001. Derivation and representation in modern transformational
syntax. In _The handbook of contemporary syntactic theory_, ed. Mark Baltin and
Chris Collins, 67–88. Blackwell.
Legendre, Geraldine. 2003. Optimality theory in syntax. In _International
Encyclopedia of Linguistics_, ed. William Frawley. Oxford: Oxford University
ABOUT THE REVIEWER
Joshua Bowles Received his MA from the University of Utah (December 2008) and
will be pursuing a PhD at another university soon. His research interests
include theoretical syntax and semantics (agreement, phrase structure,
evidentiality, and questions) as well as biolinguistics and the foundations of
generative grammar. At the moment he is an adjunct instructor in the English
Department at Utah Valley University and is working on a book length manuscript
which surveys the various approaches to linguistic research for a general audience.