This book is a collection of original research articles on the representation and interpretation of indefinite and definite noun phrases, anaphoric pronouns, and closely related issues such as reference, scope and quantifier movement. A variety of frameworks for the formal analysis of discourse semantics are represented, including discourse representation theory, file change semantics, dynamic Logic, E-type theories, and choice function approaches, which was one of the main issues the Konstanz project were concerned with. All of these frameworks are couched in the tradition of Montague Grammar, even though they extend the classical formalism in different directions. The developments emerged from the modern discussion of problems for the representation of cross-sentential anaphoric relations involving pronouns, and the interpretation of definite and indefinite noun phrases. After a decade of controversy, these approaches are merging together, enriching each other's formal mechanisms rather than expressing opposition. Thus, the articles in this volume concentrate on fundamental semantic questions, and also extend the current formalisms in order to capture more data.

The roots of this book lie in the workshop "Reference and Anaphoric Relations", which was held at the University of Konstanz in June 1996. Five main issues were discussed: (i) the Stoic and scholastic treatment of reference and anaphora; (ii) quantification and scope, (iii) anaphoric reference; (iv) the representation of indefinite NPs with choice functions; and (v) the relation between linguistic expressions, their representations and their meanings. The workshop formed the final part of our project ``Interaktion von Wort- und Satzsemantik'' (``Interaction of word and sentence semantics''), which ran from 1992 to 1996. We gratefully acknowledge the financial support of the Deutsche Forschungsgemeinschaft and the University of Konstanz for both our project and the final conference.

The authors and editors have profited immensely from the presentations and comments by the other participants of the conference. We would like to thank the SLAP series editors, in particular Gennaro Chierchia, for their advice, and the external referee for the thorough review of the papers contained in this volume. Our special thanks go to Robert Schmaus, whose technical wizardry and tireless efforts produced the coherent layout, and to Nora Billermann, who did the painstaking job of updating the references and finding missing bibliographical information. Finally, we express our sincerest gratitude to all our colleagues who have so freely shared their time and friendship with us discussing the research of our project and the recent developments in the semantics of reference and anaphora. They not only contributed with their support, comments and new ideas to the success of the project but they also made this research worthwhile.

Historical aspects of anaphoric relations

The controversy about the nature of reference in general and the nature of anaphoric reference in particular goes back as far as Greek philosophy. URS EGLI guides us in his article Anaphora from Athens to Amsterdam through the ancient treatment on anaphora, which reveals important insights for the modern discussion. Egli presents sample texts that indicate that anaphora was a very prominent issue in the Stoic tradition dating back to the 3rd century B.C. The Stoics described quantifier raising and they recognized that a pronoun -- despite its name -- cannot always be replaced by the noun with which it is anaphorically linked, as is the case with proper names. They illustrated this with the so-called "nobody paradox", which arises if the pronoun he in (1) is replaced by its antecedent someone in (2):

Egli implements the Stoic view on anaphora and quantified sentences into a dynamic semantics, thus merging different traditions into a new and fruitful view.
(View ps / pdf)

Understanding the Semantics of "relativa grammaticalia":
Medieval Logicians on Anaphoric Pronouns
by REINHARD HÜLSEN

The modern discussion of the representation of cross sentential anaphoric pronouns can be traced back to the work of Geach (1962), who discussed medieval approaches to the study of anaphoric pronouns. Geach claimed that the scholastic tradition was not able to analyze more complex anaphoric constructions, like the celebrated donkey-sentences, which were a prominent example from the scholastic discussion. REINHARD HÜLSEN argues in his contribution Understanding the Semantics of "relativa grammaticalia": Medieval Logicians on Anaphoric Pronouns that Geach did not pay enough attention to the medieval discussion. For instance, the E-type theory of Evans had been anticipated by medieval logicians, such as Buridan. Hülsen first provides a sketch of the scholastic theory of suppositio or "kinds of reference" which was used during that time. He then explains different forms of suppositio, and finally discusses the problem of fitting anaphoric pronouns into this system. Like the Stoics, the medieval logicians dismissed the naive idea that a pronoun always refers to the same object as its antecedent expression. They discussed the possibility that pronouns refer in the same way, i.e. that they express the same suppositio as their antecedents. However, even this more sophisticated view was rejected by Buridan, who noted that man in (1) may have more ways of reference (suppositio) than the pronoun he. For the pronoun only refers to men who are debating and not to men in general, i.e. it refers to the "verifiers of the first clause".

Hülsen shows in his interpretation that Buridan has anticipated one of Evans' most important arguments against the theory of Geach.
(View ps / pdf)

In their paper Meaning in Motion, JEROEN GROENENDIJK & MARTIN STOKHOF sketch the place of dynamic semantics within a broader picture of developments in philosophical and linguistic theories of meaning since the end of last century. They illustrate some basic concepts of dynamic semantics by means of a detailed analysis of anaphoric definite and indefinite descriptions, which are treated as contextually dependent quantificational expressions. They show how a dynamic view sheds new light on the contextual nature of interpretation, on the difference between monologue and dialogue, and on the interplay between direct and indirect information.
(View ps / pdf)

Quantification and Scope

Scope Matters
by DONKA FARKAS

Scope and particularly noun phrase scope has become a central topic again due to renewed interest in the syntax-semantic interface from both syntaciticians and semanticists. DONKA FARKAS treats the semantic reflexes of scope in her paper Scope Matters. She first reviews the main tasks that a theory of scope has to solve, and the ways in which they have been approached in the literature. In her view, the structural position of a variable underdetermines the possible readings. She presents a non-movement based theory of scope, and examines how it fares with respect to the desiderata established so far. In her own indexical theory of scope, she assumes that scope is not syntactically constrained at LF, but freely assigned semantically. Thus Farkas can account for data that are highly problematic for a movement account of scope.

(View ps / pdf)

The relations between scope, lexical composition and properties of generalized quantifiers are discussed by HENRIËTTE DE SWART, who explores alternatives to a lexical decomposition account of the German determiner kein "no" and its Dutch counterpart geen "no" in her paper Scope Ambiguities with Negative Quantifiers. The lexical decomposition was proposed to account for readings in which a scope-bearing operator such as an intensional verb or a universal quantifier intervenes between the negation and the existential quantifier as the two parts of the determiner. De Swart argues that the lexical decomposition is not only undesirable, but also empirically incorrect. As an alternative, she develops a higher-order interpretation of negative quantifiers in terms of quantification over properties. Her analysis is built on the observation that split readings are restricted to monotone descreasing NPs in predicative positions.
(View ps / pdf)

In Definiteness Effect: The Case of Russian, ELENA PADUCHEVA deals with the phenomena known as "definiteness effects"-- specifically, with the combinability restrictions that are at work in English there-sentences and their Russian equivalents. The distinction of strong vs. weak determiners was introduced in Milsark (1974), and formalized in Barwise & Cooper (1981) in order to give an account of these restrictions. Paducheva discusses a similar set of facts from a different perspective. It is claimed that a there is - construction, in its most common use, expresses existential quantification of its subject, and thus demands that the subject be a property denoting expression (i.e. a common noun) -- rather than referential or quantificational one. Explicit markers of definiteness or quantification in the subject NP -- either universal (as in *There is every tiger in the garden, *There are all solutions to this problem) or, sometimes, even existential -- contradict this requirement -- hence the explanation of ungrammaticality. Such determiners as most express double quantification (more precisely, quantification over one set and referentiality of the other), and are at place only in the context of a sentence with a very special topic-comment-structure; for example in such sentences as *There were most men in the garden, the topic-comment requirements of most contradict those of there is - construction.
(View ps / pdf)

STEPHEN NEALE shows in his puzzle Persistence, Polarity, and Plurality that the following three commonly accepted hypotheses are inconsistent: (i) A sentence of the form the As are Bs is true if and only if every A is B and there is more than one A. (ii) So-called "negative polarity" expressions like ever and any can occur only in "downward entailing" environments. (iii) Plural descriptions may contain negative polarity items. Since (i) defines the definite article as not downward entailing, it is predicted that the negative polarity item ever cannot appear in plural descriptions, contrary to assumption (iii). Neale elaborates this argument in detail by decomposing the meaning of the definite article into a complex quantifier phrase. He finally suggests different solutions to the puzzle.
(View ps / pdf)

Anaphoric reference

Anaphoric expressions can refer to antecedents in attitude contexts. This observation goes back to Geach's (1967) notorious Hob-Nob sentence (1), who dubbed it "intentional identity". The following two papers concentrate on the belief-belief variant of (1).

According to the received view in semantics, so-called unbound pronouns -- that is, pronouns not bound by a quantifier Q inside the smallest clause containing Q -- should either be treated as abbreviations for the antecedent clause or as variables bound by a dynamic existential quantifier. Following the discussions of Geach and Edelberg (1986, 1992), ROBERT VAN ROOY argues in his paper Anaphoric Relations across Attitude Contexts that the anaphoric dependencies in Hob-Nob sentences cannot be accounted for by treating pronouns simply as abbreviations for their antecedent clauses. This not only indicates that Hob-Nob sentences pose a serious problem for the proposal to analyse all pronouns as definite descriptions recoverable from the antecedent clause, but it also suggests that it won't be easy to account for Hob-Nob sentences for any approach that treats pronouns as abbreviations for indefinite descriptions, like standard dynamic semantics does. Van Rooy discusses various ways we might try to account for Hob-Nob sentences on the assumption that pronouns should be treated as variables bound by a dynamic existential quantifier, but finds none of them fully satisfactory. He concludes by suggesting that the problem might be solved when we take the notion of 'speaker's reference' seriously.
(View ps / pdf)

HARTLEY SLATER takes another direction in order to solve the problem of intentional identity. In his article The Grammar of the Attitudes, Slater suggests that a purely grammatical explanation is possible by using epsilon terms as representation for anaphoric pronouns. He gives a representation of (1) as Ta(there exists x)Sx & Tb C(epsilon x) Sx & (not(there exists x) Sx), in which the cross-referencing pronoun 'he' is captured by the epsilon term (epsilon x) Sx. He defends this kind of account in detail with respect to the twelve similar examples in Edelberg's (1995) paper.

Choice functions and the semantics of indefinites

ARNIM VON STECHOW compares the in situ approach of choice functions with the classical movement approach of quantifier raising at LF in his contribution Some Remarks on Choice Functions and LF-Movement. He discusses the use of choice functions for indefinites and for wh-phrases with a variety of data. For a long time, the standard assumption in Generative Grammar was that wh-phrases had to move at LF for semantic reasons, more precisely, for reasons of scope. Problems of wide scope indefinites and wh-phrases that cannot move for syntactic reasons have demanded a different semantic representation of these expressions. Reinhart (1992) and Kratzer (1998) applied the formal mechanism of choice functions to such cases. These in situ approaches to indefinites and wh-phrases receive support from current development in Generative Grammar. In recent work, Chomsky seems to hold the view that wh-movement serves the purpose of clause typing, i.e. a fronted wh-clause marks a construction as an interrogative construction. In other words, wh-movement is not motivated semantically but syntactically. The in situ interpretation by means of choice functions seems to provide a method to implement this idea semantically. Von Stechow then argues on the basis of more complex cases involving reconstruction and scrambling that the definition of the choice function must be modified to account for these cases. However, he also points to some data that cannot be described without movement. Although skeptical about the choice function approach, he tentatively concludes that it is an interesting alternative to the movement approach and that it might shed new light on quantification.
(View ps / pdf)

YOAD WINTER, in his article What Makes Choice Natural?, treats two aspects of the application of choice functions to wide scope indefinites. First, given that the semantics of indefinites involves functions, it still does not follow that these have to be choice functions. The common practice is to stipulate this restriction in order to get existential semantics right. However, he proposes a different way to derive this fact by referring to the conservativity, logicality and non-triviality universals of generalized quantifier theory. Winter then discusses the commonly assumed "specific" vs. "non-specific" contrast of indefinites. He classifies the choice function interpretation of indefinites as "definite" and "strong" in a precise sense, and treats it on a par with proper names, definites and other "referential" noun phrases. Finally, he explores the potential of such properties for describing the scope of indefinites in partitive constructions and there-sentences.
(View ps / pdf)

KLAUS VON HEUSINGER argues in his paper The Reference of Indefinites that indefinite NPs have a more complex referential nature than is usually supposed, and that this structure must be reflected in their semantic representation. He first presents three requirements toward a semantic representation of indefinites: (i) The descriptive material of indefinites, which serves to identify the referent, must be distinguished from the matrix predicate, which expresses the assertion of the sentence. (ii) The meaning of indefinites consists of the reference of the indefinite and its context change potential, which is necessary to link the indefinite to anaphoric expressions. (iii) Indefinites exhibit a fine-grained dependency structure. They are not only dependent on quantifiers but they are also dependent on other indefinites, which is shown by asymmetric readings of donkey sentences. Von Heusinger argues that these requirements are met if indefinites are represented by epsilon terms or choice functions. This allows one to keep the descriptive material separate from the assertive material of the sentence. He then presents a dynamic semantics in which indefinites introduce updates of choice functions in order to license anaphoric relations. Finally, he models the dependency structure of indefinites by Skolem functions that determine the particular choice function of the dependent term. This dynamic semantics with choice functions provides a feasible analysis of the so-called 'asymmetric readings' of donkey sentences.
(View ps / pdf)

Representation and interpretation

The relation between the concept of reference and that of inference is investigated by JAROSLAV PEREGRIN in his article Reference and Inference: The Case of Anaphora. He argues that reference is parasitic on inference, and not vice versa, as is commonly assumed. Peregrin divides his argument into two parts. In the first part he provides an overview of the development of logical tools which have been employed in the course of the analysis of referring expressions, such as definite and (specific) indefinite singular terms. He contrasts the Russellian Theory of Definite Descriptions with Hilbert's epsilon-calculus, and introduces the dynamic semantics with choice functions of Peregrin & von Heusinger (1995). He argues that the Russellian and Hilbertian ideas of singular terms only yield a satisfactory result if reconstructed in a dynamic framework. In the second part he turns to the discussion of the nature of such explications, and especially considers the question of whether a representational view of language is necessary. Peregrin shows that anaphoric pronouns are merely indicators (markers, slots) for singular terms already used. These indicators are necessary for inferences between sentences that are linked by anaphoric relations. In this way, he derives anaphoric relations from inferential relations. As a byproduct of his argument, Peregrin denies that the semantic representation can be taken as directly related to a mental reality: It must not be understood as depicting some cognitive machinery of the participants in the linguistic community.
(View ps / pdf)

PAUL DEKKER reflects on the relation between interpretation and representation in his paper Coreference and Representationalism. Research on the semantics of inter-sentential anaphoric relationships has led to a debate about the necessity of an additional level of representation of meaning. These relations are not established directly between the linguistic terms or the objects of the worlds, but between mediating entities, like discourse referents, file cards, or pegs, at a representational level. The additional level is necessary for the interpretation of discourses, in particular for the resolution of anaphoric pronouns. Dekker raises the question of whether the interpretation, i.e. the meaning, also needs such an additional level. He designs a dynamic epistemic predicate logic that offers a unified treatment of anaphoric and demonstrative pronouns. The different ways of representing discourse referents known from the literature cause different interpretational strategies, which Dekker characterizes as strong, midweakly and weakly representational. He argues that his own system is as weakly representational as possible. Thus, this investigation into anaphoric relations and the notion of subjects shows how the interpretation of anaphoric terms promotes new levels of representation of linguistic meaning.
(View ps / pdf)

In his contribution Underspecified Semantics, REINHARD MUSKENS investigates the interaction between representation and interpretation with respect to underspecified structures. Ambiguities in natural language can multiply so fast that no person or machine can be expected to process a text of even moderate length by enumerating all possible disambiguations. A sentence containing n scope bearing elements which are freely permutable will have n! readings, if there are no other, say lexical or syntactic, sources of ambiguity, and a series of m such sentences would lead to (n!)m possibilities. Muskens proposes not to generate and test many possible interpretations, but to first generate one `underspecified' representation which in a sense represents all its specifications, and then use whatever information is available to further specify the result. Muskens shows that it is possible to use a standard logic for underspecified representations, provided that we use this logic not only to reason about the content of any given statement, but also about its form. One central hypothesis in the paper is that the relation between an underspecified representation and its full representations is not so much the relation between one structure and a set of other structures but is in fact the relation between a description (a set of logical sentences) and its models.
(View ps / pdf)

Preface	i
Picture of the participants
Klaus von Heusinger & Urs Egli/ Introduction: Reference and the Semantics of Anaphora (ps / pdf)	1-13
Historical Aspects of Anaphoric Relations
Urs Egli / Anaphora from Athens to Amsterdam	17-29
Reinhard Hülsen / Understanding the Semantics of "relativa grammaticalia": Medieval Logicians on Anaphoric Pronouns	31-46
Jeroen Groenendijk & Martin Stokhof / Meaning in Motion	47-76
Quantification and Scope
Donka Farkas / Scope Matters	79-108
Henriette de Swart / Scope Ambiguities with Negative Quantifiers	109-132
Elena Paducheva / Definiteness Effect: The Case of Russian	133-146
Stephen Neale / Persistence, Polarity, and Plurality	147-153
Anaphoric Reference
Robert van Rooy / Anaphoric Relations Across Attitude Contexts	157-181
Hartley Slater / The Grammar of the Attitudes	183-190
Choice Functions and the Semantics of Indefinites
Arnim von Stechow / Some Remarks on Choice Functions and LF-Movements	193-228
Yoad Winter / What Makes Choice Natural?	229-245
Klaus von Heusinger / The Reference of Indefinites	247-265
Representation and Interpretation
Jaroslav Peregrin / Reference and Inference: The Case of Anaphora	269-286
Paul Dekker / Coreference and Representationalism	287-310
Reinhard Muskens / Underspecified Semantics	311-338
Index of Subjects	339-344
Index of Names	345

von Heusinger, Klaus & Egli, Urs 2000.

Reference and Anaphoric Relations. Dordrecht: Kluwer.

Studies in Linguistics and Philosophy 72

Preface

Historical aspects of anaphoric relations

Quantification and Scope

Anaphoric reference

Choice functions and the semantics of indefinites

Representation and interpretation