Rezension von: Freidin, Robert (ed.) (1991): Principles and Parameters in Comparative Grammar. Cambridge, MA:
MIT Press
It is not the principal aim of this volume to present a collection of articles on principles and parameters. Although the editor has chosen a programmatic title, the book covers a wide variety of topics. Studies include papers on virtually every subtheory of the PPT and therefore embody a representative documenta- tion of ongoing investigation within Generative Grammar. The book consists of fourteen articles which were presented at the First Princeton Workshop on Comparative Grammar in 1986. Some of them have already been published in similar form (Belletti/Rizzi, Travis, Aoun/Li, for instance); others are, to my knowledge, new. This means that the book leans towards being documentative rather than suggesting innovative directions in syntactic research.
The first two papers deal with Binding Theory. In his article On the Necessity of Binding Conditions, Howard Lasnik provides arguments in favour of the existence of (a reformulated version of) Binding Principle C. He argues that Thai doesn't show certain Condition C effects: An R-expression can c-command another coreferential R-expression. In Vietnamese, an R-expression has to be free only within its governing category. In both languages, however, pronouns are generally banned from binding R-expressions. This calls for a reformulation of Principle C requiring R-expressions to be pronoun free. The remainder of the paper examines the distribution of anaphoric epithets. Epithets are nominal expressions which further qualify an N P (like, for instance, John, the nut, ...).
Beside being R-expressions, they show pronominal behavior in that even in Thai they may not be bound in their governing category. Additionally, if they are binders, they cannot c-command names. Anaphoric epithets are therefore characterized as pronominal R-expressions. This extension of the nominal inventory needs an additional feature in order to categorize the distribution of the nominal expressions. A feature [ ± referential] is added to the anaphoric/
pronominal matrix, increasing the number of possible types from four to eight.
Pronominale differ from epithets in that the latter are [ + referential]. R- expressions are [—pronominal] but epithets [-I-pronominal]. The binding conditions follow from a referential hierarchy: A less referential expression may not bind a more referential one where pronouns are less referential than epithets and epithets are less referential than names.
Zeitschrift für Sprachwissenschaft 13,1 (1994), 129-133
©Vandenhoeck & Ruprecht, 1995 ISSN 0721-9067
Wayne Harbert discusses an appropriate formulation of Binding Domains. In Binding, SUBJECT, and Accessibility he proposes that the notion of accessible SUBJECT would bp better defined in terms of potential binders. This reformulation accounts for a familiar asymmetry found with object anaphors in complement clauses. They can be coindexed with the matrix subject in infinite but not in finite sentences, which shows that only a finite AGR coindexed with a nominal subject may serve as a binder. The anaphor is therefore required to be bound within the domain of that AGR. In languages which lack AGR, or in infinitive clauses, there is no accessible SUBJECT and the anaphor can be bound by a clause-external binder. Although the proposal seems convincing, it is sometimes hard to follow the line of argument. The pros and cons for AGR being a SUBJECT in the Chomskian sense are manifold and it is only at the end of the paper that the author gives his own view on the subject.
C.-T. James Huang's article Remarks on the Status of the Null Object presents several arguments against analyzing Chinese null objects as empty pronouns.
The arguments are considered as evidence ex negativo in favor of his own claim that Chinese null objects are variables bound by an empty operator in SpecCP.
One argument against a pro analysis shows that, with respect to binding, null objects behave as if they were zero counterparts of anaphoric epithets. However, they do not share the binding properties usually attributed to empty pronouns.
Another argument stems from VP-ellipsis. As in English, VP-ellipsis construc- tions exhibit the strict/sloppy ambiguity in Chinese. Huang assumes with Sag (1976) that the empty VP is a lambda expression at LF. Now, the sloppy reading can be achieved only if the pronoun within the empty VP is analyzed as a variable bound by whatever the lambda expression is a predicate of. pro is excluded from such a position because it can never get a variable interpretation.
Finally, adnominal emphatic ziji (himself ) can emphasize a subject pro, but it cannot occur with an empty object. Again, this is evidence against the construal of the empty object as a null pronominal.
Edwin Williams supports a different approach to empty categories, whose existence he simply negates. In his article The Argument-Bound Empty Catego- ries he claims that Θ-roles but not NPs refer. Thus, implicit arguments which pose a problem for a traditional formulation of the Θ-Theory are now analyzed as unassigned Θ-roles. It follows that they are visible as antecedents for control relations and for the Binding Theory. The pro drop characteristic of some languages is seen as a parametrization of INFL: INFL may satisfy the external Θ-role itself - no empty pronominal is needed.
Ken Safir, Evaluative Predicates and the Representation of Implicit Arguments, investigates complementation of evaluative predicates such as merit or deserve.
He argues that the object of gerundive and deverbal nominal complements is implicit since the complements themselves express the action the subject will undergo. The representation of the implicit argument cannot be accounted for by a movement analysis. Instead of a syntactic account, Safir advocates
pragmatic coconstrual of the patient matrix subject with the implicit argument of the gerundive/deverbal nominal complement.
Another paper on argument structure is Adriana Belletti and Luigi Rizzi's Notes on Psych-Verbs, Θ-Theory, and Binding. This p a p e r is a reprint of Belletti/Rizzi (1988). Its main idea is that psych-verbs fall into two main classes.
In the temere class, the experiencer is base generated in subject position, whereas the preoccupare ¡piacere class resembles a double object construction at d-struc- ture, with a non-thematic subject position and raising-to-subject of the theme object. Evidence for this analysis comes from i) causative constructions (they exclude embedded derived subjects and are also impossible with preoccupare verbs), ii) passive constructions (nonthematic subjects cannot passivize), iii) binding (an experiencer in object position can bind an anaphor contained within the subject), and iv) long distance anaphors (the experiencer can bind an anaphor within the theme but not vice versa).
In their article The Interaction of Operators, Joseph Aoun and Yen-hui Audrey Li formulate two principles which account for scope ambiguities found in multiple operator constructions in English and Chinese. Ambiguities arise if two quantifiers take scope over each other or a part of the chain of the other (Scope Principle). If each quantifier has scope over its own variable only (Minimal Binding Requirement), the sentence is not ambiguous. These princip- les also account for the interaction of wh-operators and quantifiers, once we assume that variables bound by wh-operators are subject to binding principle C, whereas variables bound by a quantifier are not.
Tim Stowell, Small Clause Restructuring, argues that small clauses are clausal constituents at d-structure which undergo restructuring at LF. A small clause analysis is supported by the fact that the subject and the predicate of the small clause form a constituent. Furthermore, no main clause constituents may be interpolated between the predicate and the small clause suggesting a case driven adjacency requirement. Finally, small clauses are governing categories wrt. the Binding Theory, that is, an anaphor must be bound, whereas a pronoun must be free within a small clause. At LF, the small clause is restructured forming a complex predicate. Thus, clitic movement out of a small clause is shown to be possible only after restructuring. Additionally, a quantifier cannot adjoin to the small clause after restructuring, which accounts for the lack of narrow scope readings with quantified expressions in small clauses. As expected, small clause subjects behave like objects at LF, since they can host traces of empty operators.
Knut Tarald Taraldsen in A Directionality Parameter for Subject-Object Linking accounts for the word order regularities in Scandinavian causative constructions. Taraldsen stipulates a parameter in the spirit of Burzio's Generalization: In so-called "suject driven" languages a canonical subject is licensed if the unergative verb case marks an object. In "object driven"
languages an object is licensed if the verb has a canonical subject. Assuming two additional requirements, the composition of categorically non-distinct projec-
tions, and the prohibition of intersecting governing domains, not only the word order facts of causatives follow from this parameter but also the variation found within particle constructions, clitic-participle agreement and auxiliary selection.
In their paper The Derived Constituent Structure of the West Germanic Verb-Raising Construction, Anthony Kroch and Beatrice Santorini defend a syntactic treatment of verb-raising in the West Germanic languages. They argue that neither clause pruning nor verb cluster formation are valid mechanisms in order to correctly represent the data. Instead, they show that the formalism of Tree Adjoining Grammar is able to account for the variation found between German and Dutch, for instance. Following Zaenen (1979), verb-raising is basically analyzed as infinitive extraposition, each sentence being adjoined to the root node of the innermost infinite sentence. In Dutch, the infinite verbs are extraposed before tree adjunction takes place. This accounts for the different serialization of verbs in German and Dutch.
In Parameters of Phrase Structure and Verb-Second Phenomena, Lisa deMena Travis summarizes some ideas of her dissertation (Travis 1984). She claims that the Germanic languages have a uniform clause structure. Subject initial sentences are analyzed as IPs. The V-2 characteristic of some of these languages follows from a parameter which disallows IP adjunction for [— wh] elements.
Topicalization must therefore be movement to the specifier of CP, and the ECP forces verb-movement into C. If adjunction to IP is allowed, this position serves for fronting and no verb-movement is triggered. Travis finally accounts for the complementary distribution of finite verbs and complementizers postulating another parameter: In German and Swedish, for instance, C can identify the content of I. Thus, the ECP is fulfilled and the verb is banned from raising to I. If C is too weak, as in English and Icelandic, the verb has to raise to I in order to satisfy the ECP.
Norbert Hornstein and David Lightfoot, On the Nature of Lexical Govern- ment, dissociate lexical government from antecedent government. While lexical government serves mainly for the recoverability of deletion and is suggested to be a Ρ F requirement, antecedent government is made superfluous, being a subcase of anaphor binding: Both antecedent government and Binding Principle A are defined as coindexing with a local c-commanding element. This account, which is based on the work of Aoun (1985, 1986) and Aoun/Hornstein/
Lightfoot/Weinberg (1987), is compared to Lasnik/Saito (1984) and shown to be superior in that it does not need to stipulate different levels for trace-binding - Generalized Binding applies at LF uniformly.
In their paper Lexical Case Phenomena, the editor Robert Freidin and Rex Sprouse present their insights on lexical case. They show that lexical case is resistent to several syntactic processes. In Russian, case splitting, which is common with structural case, does not occur with lexical case. If lexical case verbs are passivized in German, the lexical case is not absorbed and it appears on the fronted object. This is different from Icelandic, where subjects can be
lexically case marked. Here, a lexically case marked passivized Ν Ρ can appear as the passive subject.
The last paper of the collection is N o a m Chomsky's Some Notes on Economy of Derivation and Representation. The article has been circulating since its preprint appeared in the M I T Working Papers Series in 1989. At the publication date of the present volume, Chomsky has partially out-dated his own paper with his newest theory of the Minimalist A p p r o a c h . T h e paper is deliberately placed at the end of the book. Restricting the way parametric variation ought to be represented in Generative G r a m m a r , it dwells upon the foundations of various contributions. Once we accept that parametrization affects functional categories only, the properties of these categories trigger a set of syntactic processes. These processes are restricted by economic considerations: Derivations have to proceed with least possible effort. However, a formal elaboration of what exactly counts as derivationally or representationally economic is only partially given. The notion is sharpened in C h o m s k y (1992) where principles like G R E E D and P R O C R A S T I N A T E more generally define minimality of steps and symbols.
Literaturnach weis
Aoun, J. (1985): A Grammar of Anaphora. Cambridge: MIT Press.
Aoun, J. (1986): Generalized Binding. Dordrecht: Foris.
Aoun, J./N. Hornstein/A. Weinberg (1987): Two Types of Locality. In: Linguistic Inquiry 18, 537-577.
Belletti, A./L. Rizzi (1988): Psych-verbs and Theta-Theory. In: Natural Language and Linguistic Theory 6, 291-377.
Chomsky, N. (1992): A Minimalist Programfor Linguistic Theory. MIT Occasional Papers in Linguistics 1.
Lasnik, H./M. Saito (1984): On the Nature of Proper Government. In: Linguistic Inquiry 15, 235-298.
Sag, I. (1976): Deletion and Logical Form. Doctoral Dissertation, MIT.
Travis, L. (1984): Parameters and Effects of Word Order Variation. Doctoral Dissertation, MIT.
Zaenen, A. (1979): Infinitival Complements in Dutch. Papers from the Fifteenth Regional Meeting of the CLS. Chicago Linguistic Society, University of Chicago.
