EDITORS: Stepanov, Arthur; Fanselow, Gisbert; Vogel, Ralf TITLE: Minimality Effects in Syntax SERIES: Studies in Generative Grammar 70 PUBLISHER: Mouton de Gruyter YEAR: 2004
Rebecca Shields, Department of Linguistics, University of Wisconsin- Madison
Most of the twelve papers in the volume assume the framework of the Minimalist Program, but three make proposals within Optimality Theory (OT). Minimality effects are discussed in a range of environments, including A-movement (passivization, raising, unaccusatives), head movement, Object Shift, remnant movement, Stylistic Fronting, topicalization, and wh-movement (single and multiple). The papers would be of interest to anyone working on Relativized Minimality, intervention effects, and more generally the encoding of locality constraints in grammar.
The papers present novel data which are problematic for existing theories, and make interesting new connections between previously published facts. The focus of much of the volume is empirical, primarily apparent counterexamples to Chomsky's (1995) Minimal Link Condition (MLC). The authors suggest a variety of remedies: they tweak the MLC slightly, suggest separate principles that work together with the existing MLC, propose disposing of it and replacing it with something different entirely, or reanalyze the offending data so that no change to the theory is necessary. Some papers also extend the purview of Minimality by suggesting novel analyses of data which make use of the MLC or an OT variant. There is also much discussion of purely theoretical issues, such as the implications of Minimality effects for the derivation/representation debate, and the theoretical status of the MLC in the model (is it a primitive, or derivable from more principled aspects of grammar?).
SUMMARY OF EACH PAPER
Elena Anagnostopoulou. On clitics, feature movement and double object alternations.
Anagnostopoulou investigates A-movement (passive, unaccusative, and raising constructions) in Greek ditransitives. In Greek, goal/experiencer arguments may surface as dative PPs, genitive DPs, or cliticized genitives (with optional genitive DP doubling). But if the sentence involves A-movement of a theme or embedded subject, not all of these options remain possible. If the A-movement is biclausal, the goal/experiencer must be a cliticized genitive. If the A-movement is monoclausal, the goal/experiencer must surface either as a clitic or as a PP. Anagnostopoulou proposes an analysis in terms of Chomsky's (1995) notions Attract Closest and equidistance. Genitive DPs always block attraction by TP of a lower argument, because (following Marantz 1993) they are applicative arguments introduced by a light v, and are therefore closer to the attracting head than the theme or subject. PPs, on the other hand, are generated inside VP with the theme, so that the PP and clause-mate theme are equidistant to the attracting head. PPs therefore block attraction only of embedded subjects, but not clause-mate themes. (Note however that the appeal to equidistance is only necessary if it is NOT possible for the theme to c-command the PP underlyingly, an assumption which appears to be contradicted by the binding facts mentioned in the paper.) Anagnostopoulou proposes that clitics appear to be an exception to Attract Closest because they involve feature movement without pied- piping. The features of the goal/experiencer argument may raise first, obeying Attract Closest, in which case the goal/experiencer is spelled out as a clitic. The theme or embedded subject can subsequently raise without inducing a Minimality violation, on the assumption that the stranded material does not bear a feature relevant to Attract.
Željko Bošković. PF merger in stylistic fronting and object shift.
Bošković provides a morphophonological analysis for some well- known patterns in Scandinavian: the subject gap restriction on Stylistic Fronting in Icelandic, and V-topicalization repair in Scandinavian Object Shift configurations. The analysis is analogous to the Chomsky (1957) analysis of affix hopping. Bošković proposes that Stylistic Fronting targets a functional projection FP above IP headed by a phonologically null verbal affix F, which must be adjacent to the verb at PF, thus accounting for the null subject requirement. He shows that intervening adverbs do prevent the PF merger from taking place -- evidence against a late insertion analysis of adverbs. He suggests the same analysis for wh-questions in Bulgarian, which show a similar restriction. The Object Shift puzzle is from Holmberg (1999), who noticed that although Object Shift is not generally permitted when an auxiliary undergoes V-2 movement rather than the verb, if the V is topicalized Object Shift becomes possible. Bošković proposes extending Bobaljik's (1994) PF affixation analysis of Object Shift to these cases as well. On Bobaljik's analysis, Object Shift is ungrammatical only if it interferes with the PF adjacency required between a verb/verbal participle and the affix in I or head of ParticiplePhrase. Bošković shows that this straightforwardly accounts for the V-topicalization repair cases assuming multiple spell-outs. Although the shifted object intervenes between the V and its affix in the final spell-out, since the V moves to its topic position successive- cyclically, there will be an intermediate spell-out where it is adjacent to its affix and the two can phonologically merge.
Gisbert Fanselow. The MLC and derivational economy.
Fanselow looks at anti-superiority effects in multiple wh-questions in a range of languages. He notes that while the MLC appears to hold strictly for head-movement, in the case of wh-operator movement there are many counterexamples. He proposes that the MLC is not able to block movement which affects the semantic scope of operators. Specifically, he claims that the MLC only applies when the movement involved results in a distinct Logical Form (LF). On this view, the MLC is a global economy constraint. Fanselow claims that obeying the MLC is not correlated with either nestedness effects or wh-islands cross-linguistically, so these effects must therefore be produced by some principle other than the MLC. He suggests that other possible counterexamples to his proposal stem from the interaction of the MLC and various pragmatic constraints.
Susann Fischer. Stylistic Fronting: A contribution to information structure.
The topic of this paper is Stylistic Fronting (SF) in Old Catalan. Fischer shows that the null-subject requirement is not respected in this language, unlike in other languages that show SF (cf. Bošković above). Several previous proposals for SF in Germanic are presented, critiqued, and shown to make incorrect predictions for Old Catalan. Fischer then presents her own account in terms of the types of features involved in motivating the fronting. Specifically, she proposes that in Old Catalan SF is motivated by the need to check off a strong V feature on SigmaP, a functional category above IP, while in Germanic languages the movement is triggered by the Extended Projection Principle (EPP) feature on IP. The constituent in SigmaP is interpreted as semantically emphasized. Thus, in Old Catalan SF is not simply the grammatical reflex of a projection attracting the closest head because it needs a specifier; rather, it stems from the speaker's intention to express emphasis, and has significance for information structure. Fischer also points out that some previously published data call into question the Attract Closest account for Icelandic, suggesting that her analysis may be extendable to some Germanic languages as well.
Hubert Haider. The superiority conspiracy: Four constraints and a processing effect.
Haider takes a cross-linguistic look at multiple wh-questions, and shows that there is considerably more variation than is predicted by the MLC. Superiority is sometimes respected, but not always; furthermore, the contexts where it is respected vary from language to language. He argues that the Superiority/MLC effects are in fact an epiphenomenon, and the patterns we observe are due to the conspiracy of four grammatical constraints, plus a processing restriction. The four constraints are: 1) obligatory operator - an in situ wh in the specifier of a functional projection must be an operator binding a variable, 2) semantic type - the moved wh and the in situ wh cannot both range over higher-order types, 3) domain-mapping - operators must c-command their semantic domain, and 4) minimal binding - an in situ wh-element must be licensed by a c-commanding wh-element in a minimal domain. Haider shows how the observed variation (he focuses primarily on English, German, and Dutch) follows from his constraints plus the head-parameter setting for a given language.
John Hale and Géraldine Legendre. Minimal links, remnant movement, and (non-) derivational grammar.
This paper takes another look at some previously reported facts about German remnant movement. Several researchers have noticed that the two movements involved (raising the NP out of the VP, and subsequent raising of the VP containing the NP trace) must be of different types. So if the NP is moved out of the VP via scrambling, the VP remnant can undergo wh-movement or topicalization, but not scrambling. Müller (1998) analyzes this as a type of MLC effect, and claims that it argues in favor of a derivational view of grammar. Hale and Legendre point out that the same effect can be captured representationally in terms of link minimization on surface structures. They go on to propose just such an account in an OT framework, and they show how variation in ranking of their violable constraints can account for certain differences between German and Japanese.
Winfried Lechner. Extending and reducing the MLC.
This paper is primarily concerned with the theoretical status of constraints on the operations Move and Merge in the grammar. Lechner views the conditions on these two operations as having more overlap than is generally assumed. First, he provides a redefinition of the MLC which eliminates the need for Chomsky's 1995 explicit Merge over Move economy condition. This is achieved essentially by widening the domain of Attract to include items in the Numeration as well as items already Merged into the tree, so that Merge and Move compete directly within the MLC. Lechner shows how his redefinition makes some good predictions regarding Case freezing (A-movement out of finite clauses) and Superraising. A second proposal in the paper is that part of the MLC can be derived from Kayne's 1994 Linear Correspondence Axiom (LCA). Lechner achieves this by positing that the LCA imposes an ordering on several sets of nodes per derivation: the set of all terminals, as in Kayne's original version, and additionally the subset(s) of these nodes that contain a feature relevant to a particular type of movement. In such a subset (but crucially NOT in the superset), a moved expression and its copy count as a single item for the purposes of the ordering algorithm, and any distinct item which asymmetrically c-commands only the copy will therefore result in a contradiction for the LCA. Lechner further claims that the LCA is computed over phases, rather than over entire sentences, so that an ordering is required only for nodes within the same phase. The phase implementation predicts that although movement across a phase-mate intervener is ruled out, movement across an intervener in a higher phase should be allowed, contrary to fact. Lechner suggests that the latter cases constitute a residue of the MLC which is not derivable from the LCA.
Hanjung Lee. Minimality in a lexicalist Optimality Theory.
Lee discusses some previously reported facts about word order freezing. In Hindi, although variable word order is in general possible, the object may not precede the subject if both bear the same morphological case marker. Lee claims that this data is problematic for an MLC account, and offers an alternative in an OT Lexical-Functional Grammar framework. The data is explained by the interaction of violable constituent alignment constraints in the bidirectional model of Smolensky (1996). Two optimal candidate sets are computed, from the perspectives of production on one hand and comprehension on the other, and the overall winning candidate is chosen from the intersection of these sets. In this model recoverability from surface structures may play a direct role in grammar, if the comprehension optimization yields a narrower candidate set than the production optimization. Lee discusses further cases where recoverability constrains syntax in this way from Chamorro and Tzotzil.
Gereon Müller. Phrase impenetrability and wh-intervention.
Müller proposes abandoning the MLC and replacing it with a PHRASE Impenetrability Condition -- a strengthened version of Chomsky's (2000) PHASE Impenetrability Condition (PIC). The arguments in favor of such an approach are conceptual: Müller's condition is purely derivational, while a phase evaluation of the MLC is strongly representational; Müller's condition eliminates redundancies in the MLC and PIC; and Müller's condition is symmetrical, in that the probe must be in the head or specifier of the phrase currently undergoing Merge, and the goal must be in the head or specifier of the next phrase, while the MLC contains an asymmetry -- the domain of the probe is the head and specifier of the phrase undergoing Merge, but the domain of the goal is larger: the current phase. Actually, his explanation of the Minimality effects also relies crucially on a condition Müller calls Phrase Balance, which allows certain requirements of the derivation to be satisfied by elements in the Numeration as well as by elements already Merged. In this respect his approach is similar to Lechner's broadening of the domain of Attract mentioned above. Müller further shows how his approach, but not the MLC, can be extended to certain cases of wh-intervention by non-c-commanding wh-phrases in English and German.
Geoffrey Poole and Noel Burton-Roberts. MLC violations: Implications for the syntax/phonology interface.
The topic of this paper is apparent MLC violations in Stylistic Fronting in Icelandic and Long Head Movement in Breton. Poole & Burton- Roberts claim that these types of movement are in some sense phonological rather than syntactic phenomena, which explains why they are insensitive to the MLC, a constraint on syntactic movement. Their analysis leads the authors to support their Representational (as opposed to Realizational) theory of the syntax-phonology interface. The movement operations in question, although they do not appear to be purely syntactic, do not appear to be purely phonological either --- for example, they make reference to syntactic category and constituency. Poole & Burton-Roberts claim that these phenomena can be explained by the "representational conventions" of a given language, which determine the mapping between syntactic and phonetic representations, and are thus neither strictly syntactic nor strictly phonological.
Arthur Stepanov. Ergativity, Case and the Minimal Link Condition.
Stepanov's puzzle is Nominative Case assignment to the absolutive argument of transitive verbs in Hindi. Assuming Nominative Case is assigned by T, the fact that it can be assigned to the absolutive object across an interfering ergative subject appears problematic for the MLC. Stepanov argues that Case assignment in this configuration is possible because ergative-absolutive verbs in Hindi are actually unaccusative, with the ergative argument receiving inherent Case. Such NPs, he claims, are merged counter-cyclically, in this case at a point in the derivation following the establishment of a structural Case dependency between T and the absolutive object (via Agree). At the point when the Nominative Case dependency is established, then, the MLC is respected. Stepanov gives examples from English raising to show that inherently Case marked NPs do not induce Minimality effects for structural Case related dependencies in general. Additionally, the counter-cyclic merge analysis helps to explain some scope freezing facts for ergative subjects as well.
Ralf Vogel. Correspondence in OT syntax and Minimal Link effects.
Vogel provides an OT analysis of Minimality effects in topicalization and wh-movement. In his model, violable constraints govern the possible correspondences between three levels of representation: semantic, syntactic, and phonological. The mappings are evaluated bidirectionally: like Lee above, Vogel discusses how his model handles the factor of recoverability in word order freezing. Vogel derives the differences between English and German (German allows Superiority violations in multiple questions, while English does not) by ranking the relevant Faithfulness constraints above Markedness in German, with English displaying the opposite ranking. Since Superiority violations are marked, English will never allow them, even if a violation is specified in the input. German, however, will tolerate markedness violations in order to remain faithful to the input, thus allowing marked Superiority violations to surface.
Most of the papers are clearly written and well-argued; all of them present new and interesting data, analyses, and/or argumentation points. Anyone interested in locality effects in syntax will find here new puzzles to ponder and new approaches to consider.
A few things I found lacking:
It would have been interesting to see the papers dealing with anti- Superiority effects in wh-movement compare their proposals with the featural movement analysis in Pesetsky (2000). Is there any way to distinguish between these proposals?
Also, in presenting wh-movement paradigms that are known to be sensitive to D-linking, some authors used wh-words that are ambiguous between D-linked and non-D-linked interpretations, such as 'who' and 'what.' In order to control for D-linking, it is necessary to compare unambiguously D-linked wh-phrases such as 'which person' with unambiguously non-D-linked wh-phrases such as 'who the hell,' and/or to present informants with explicit contexts. If this factor is not controlled for, it makes it difficult to know how to interpret the judgment patterns reported.
One other small but noticeable problem is that some papers contain an exceedingly large number of typos and grammatical errors, which in some cases compromise clarity. Such a nice and expensive publication surely deserved better proofreading.
Nevertheless the book is well worth reading. Particularly interesting in my view is the variety of environments where Minimality effects show up, and the variety of solutions adopted for similar problems across environments. This work naturally invites questions of further synthesis:
Are the various approaches presented here empirically distinguishable, or are they notational variants?
For those approaches which are empirically distinguishable, what is the scope of their explanatory power? Where do they overlap, and where do they complement each other? Can some be subsumed under others, or do we in fact need some or all of them to achieve maximum coverage?
For those approaches which are notational variants, can they be distinguished given some metric of simplicity or elegance? And do these metrics give the same results when applied to Minimality effects in various environments, or do some approaches seem better suited to some environments than others?
More generally, is Minimality a unified phenomenon, encoded "in one spot" in the grammar, or is it encoded redundantly (and perhaps somewhat differently) in various components of grammar?
Bobaljik, Jonathan. 1994. What does adjacency do? MITWPL 22:1-32.
Chomsky, Noam. 1957. Syntactic structures. The Hague: Mouton.
Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press.
Chomsky, Noam. 2000. Minimalist inquiries: The framework. In R. Martin, D. Michaels, and J. Uriagereka, eds., Step by step 89-155. Cambridge, MA: MIT Press.
Holmberg, Anders. 1999. The true nature of Holmberg's generalization. Studia Linguistica 53:1-39.
Kayne, Richard. 1994. The antisymmetry of syntax. Cambridge, MA: MIT Press.
Marantz, Alec. 1993. Implications of asymmetries in double object constructions. In S. A. Mchombo, ed., Theoretical aspects of Bantu grammar 113-150. Stanford: CSLI.
Müller, Gereon. 1998. Incomplete category fronting: A derivational approach to remnant movment in German. Dordrecht: Kluwer.
Pesetsky, David. 2000. Phrasal movement and its kin. Cambridge, MA: MIT Press.
Smolensky, Paul. 1996. On the comprehension/production dilemma in child language. Linguistic Inquiry 27:720-731.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Rebecca Shields is a Ph.D. student in the Department of Linguistics at the University of Wisconsin-Madison. Her research focuses on Relativized Minimality, intervention effects, functional categories, and adjuncts.