Perspectives on Prepositions [Reprint 2013 ed.] 9783110924787, 9783484304543

Table of contents :
Preface
Introduction
Part I: Syntactic-Semantic Perspectives on Prepositions
Prepositions, Features, and Projections
The Projections of Temporal Prepositions
Movement within a Spatial Phrase
The Effect of Prepositional Complements on the Choice of Synthetic or Analytic Comparatives
Processing Complexity and the Variable Use of Prepositions
Präpositionale Rektionsalternation unter dem Gesichtspunkt der Grammatikalisierung
Fortbewegungsverben und Ortswechsel im Französischen
Part II: Cognitive-Semantic Perspectives on Prepositions
Die Wegpräposition über
The Evolution of French Sur. Toward a Diachronic Cognitive Semantics
Force Dynamics in Static Prepositions: Dutch Aan, Op, and Tegen
Towards a Description of the Meaning of At
Why Similar To, but Different From?
Metonymy in Prepositions
Addresses of Contributors

Citation preview

Linguistische Arbeiten

454

Herausgegeben von Hans Altmann, Peter Blumenthal, Hans Jürgen Heringer, Ingo Plag, Heinz Vater und Richard Wiese

Perspectives on Prepositions Edited by Hubert Cuyckens and Günter Radden

Max Niemeyer Verlag Tübingen 2002

Die Deutsche Bibliothek - CIP-Einheitsaufnahme Perspectives on prepositions / ed. by Hubert Cuyckens and Günter RadJen. - Tübingen : Niemeyer, 2002 (Linguistische Arbeiten ; 454) ISBN 3-484-30454-5

ISSN 0344-6727

© M a x Niemeyer Verlag G m b H , Tübingen 2002 D a s Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung des Verlages unzulässig und strafbar. Das gilt insbesondere für Vervielfältigungen, Übersetzungen, Mikroverfilmungen und die Einspeicherung und Verarbeitung in elektronischen Systemen. Printed in Germany. Gedruckt auf alterungsbeständigem Papier. Druck: Weihert-Druck G m b H , Darmstadt Einband: Industriebuchbinderei Nadele, Nehren

Table of Contents

Preface

VII

Hubert Cuyckens & Günter Radden Introduction

IX

Part I: Syntactic-Semantic Perspectives on Prepositions

1

Gisa Rauh Prepositions, Features, and Projections

3

Dagmar Haumann The Projections of Temporal Prepositions

25

Niina Ning Zhang Movement within a Spatial Phrase

47

Britta Mondorf The Effect of Prepositional Complements on the Choice of Synthetic or Analytic Comparatives

65

Günter Rohdenburg Processing Complexity and the Variable Use of Prepositions

79

Claudio Dì Meóla Präpositionale Rektionsalternation unter dem Gesichtspunkt der Grammatikalisierung

101

Priska-Monika Hottenroth Fortbewegungsverben und Ortswechsel im Französischen

131

Part II: Cognitive-Semantic Perspectives on Prepositions

155

Birgitta Meex Die Wegpräposition über

157

Walter De Mulder & Anne Vanderheyden The Evolution of French Sur: Toward a Diachronic Cognitive Semantics

177

VI

Table of Contents

Maaike Beliën Force Dynamics in Static Prepositions: Dutch Aan, Op, and Tegen

195

Ignasi Navarro i Ferrando Towards a Description of the Meaning of At

211

Günter Radden & Elizabeth Matthis Why Similar To, but Different From?

231

Hubert Cuyckens Metonymy in Prepositions

257

Addresses of Contributors

267

Preface

The present volume evolved from a workshop on prepositions held at Hamburg University on June 26 and 27, 1998. Nine of the fifteen papers presented at the workshop are included in this volume; three more contributions were invited (Rauh, De Mulder & Vanderheyden, Navarro i Ferrando). The intention of the workshop was to bring together the multifaceted perspectives on prepositions that have been developed in modern linguistics. We would like to express our thanks to all the workshop participants in making this workshop a success. We are also grateful to Frau Elizabeth Himmler and the student assistants at the the Institut fur Anglistik und Amerikanistik for their organizational help. This workshop was organized while the first editor of this volume was a research fellow of the Alexander von Humboldt Stiftung at the Institut für Anglistik und Amerikanistik (University of Hamburg). The (financial) support of the Alexander von Humboldt Stiftung is hereby gratefully acknowledged. We would also like to thank all the authors for their contributions and for their speediness at every stage of the editorial process. A special word of thanks goes to the series editor, Prof. Dr. Heinz Vater, for his enthusiasm and support in seeing this project through. Finally, we would like to acknowledge the invaluable work of Bram Renmans in preparing the first formatted version.

Hubert Cuyckens & Günter

Rodden

Introduction

The papers collected in this volume offer different perspectives on prepositions. Some papers mainly discuss syntactic (and morphological) aspects of prepositions; other papers predominantly focus on cognitive aspects. All the papers are, however, concerned with the semantics of prepositions The book is, accordingly, divided into two main parts: Part I: "Syntactic-Semantic Perspectives on Prepositions"; Part II: "Cognitive-Semantic Perspectives on Prepositions."

Part I: Syntactic-Semantic Perspectives on Prepositions In the generative tradition, approaches to the syntax of prepositions tend to adopt the X-bar theory of phrase structure, according to which prepositions determine PPs as their maximal projections. In other words, prepositions function as heads of prepositional phrases and may, like the heads of other phrases, take complements and specifiers. For example, in right above your head, the determiner phrase your head is the complement of Ρ and the modifying adverb right is its specifier. Within this framework, prepositions share many syntactic properties with other word classes: they license, in particular, an argument structure potentially comprising external, internal, and referential arguments; like verbs, they constitute a case-assigning category, and they may be transitive or intransitive. The first three papers (Rauh, Haumann, Zhang) are set within this context of recent generative grammar. Rauh demonstrates that, in contrast to general assumptions in the literature, the potential internal structure of the projections of prepositions shows observable differences, and that these differences can be explained on the basis of the prepositions' different lexical properties. Haumann focuses on temporal prepositions, and discusses their internal and external syntax. Zhang provides a unified syntactic account of spatial phrases which incorporates the three semantic elements making up a spatial expression: the locative relation, the place value, and the reference entity. A second set of papers reports on results of research into determinants of grammatical variation in English. Mondorf examines the effect of different degrees of syntactic complexity manifested in the presence or absence of a prepositional complement on the choice of synthetic or analytic comparative forms. Rohdenburg looks at several prepositional variation phenomena that are sensitive to the so-called "Complexity Principle," which states that more explicit (lexico-)grammatical variants tend to be preferred in cognitively more complex environments. Di Meola's paper looks at the issue of the change from dative to genitive and from genitive to dative with prepositions in German. This case alternation, in Di Meola's view, can only be accounted for in terms of grammaticalization, a process which is characterized by

χ

Hubert Cuyckens & Günter

Rodden

progressively increasing morpho-phonological, semantic, and syntactic differentiation from the original form. The final paper in this section is situtated within the research tradition that investigates the relation between verb meaning and argument structure; in particular, its author, Hottenroth, focuses on the interaction between the semantic structure of verbs of motion and prepositional phrases denoting goal/source or place.

Individual contributions to Part I In her paper "Prepositions, Features, and Projections," Gisa Rauh argues against the view generally assumed in the literature that all prepositions head prepositional phrases with the same internal structure. Instead, she distinguishes between three different types of prepositional forms, and she demonstrates that any differences in the potential internal structure of their projections can be explained on the basis of the different lexical, i.e. grammatically relevant, properties of the prepositional types. In particular, Rauh distinguishes the following three types, which are exemplified in (l)-(3): (i) lexical prepositions, (ii) governed or case prepositions, and (iii) grammatical prepositions; the latter two categories are subsumed under the general heading "grammaticalized" prepositions. (1) (2) (3)

Mary looked at the vase on the shelf. Joe relied on Jim's promise. Bill bought the flowers at a good price.

The lexical properties a preposition can be specified for are: (i) theta-features (i.e. those features which are specified in an argument structure, namely external , internal , and referential arguments); (ii) operator-features; (iii) quantifier-features; and (iv) formal morpho-syntactic features. Lexical prepositions, such as on in (1), always have the theta features and and potentially also , they exhibit the case-assigning feature [assign OBJ], and they may be specified positively for measure phrases. Governed or case prepositions, such as on in (2), differ from lexical prepositions in that they are governed, or lexically selected, by their heads and, therefore, do not have thetafeatures of their own, nor do they license determiner-like operators or quantifier-like elements. Grammatical prepositions (cf. 3), finally, exhibit the theta features and and potentially the case feature [assign OBJ], Rauh concludes, then, that prepositions should not be defined in the lexicon on the basis of rather non-revealing descriptions such as [-N] and [-V], which assume that all prepositions are syntactically alike, but that prepositions fall into natural classes (such as illustrated in (l)-(3)). Each of these classes exhibits a different set of lexical, grammatically relevant, feature specifications and determines different projections. Dagmar Haumann's paper "The Projections of Temporal Prepositions" looks into the internal and external syntax of temporal prepositions within the framework of the Minimalist Program. Starting out from the assumption that homophonous elements such as before, which traditional grammar may categorize as prepositions, subordinating conjunctions, or adverbs, are in fact members of one and the same syntactic category, the author argues that the argument structure of temporal PPs contains one external argument, at most one internal argument, and a referential argument. Whereas the external and the internal

Introduction

XI

argument define syntactic argument positions within PP, the referential argument encodes the ontological category of time and reference properties of the temporal preposition. The core of the paper focuses on the set of functional projections within the extended projection of Ρ the presence of which is determined by the argument structure and the Case-properties of the individual temporal prepositions. Niina Ning Zhang's paper "Movement within a Spatial Phrase" is concerned with the issue of the internal semantic structure of spatial expressions. A spatial expression can be semantically decomposed into the three elements Locative Relation (LR), Place Value (PV), and Reference Entity (RE). The three elements are realized by separate forms in the English prepositional phrase from behind the table, where from expresses the LR of Source, behind expresses the PV of BACK SPACE, and the table expresses the RE element. Not all three elements need be lexicalized, however: for example, in (4) as well as in (5), the preposition behind expresses the PVs but the LRs are not lexicalized: in the former situation, the LR is understood to be that of a Route, in the latter that of a Goal. (4) (5)

Trevor went behind the curtain to the table. Trevor went behind the curtain and stayed there.

A central issue of this paper is that of accounting for the ordering of the three forms. In her cross-linguistic study, Zhang finds three basic orders: (i) LR - RE - PV (as in Chinese), (ii) LR - PV - RE (as in English), and (iii) RE - PV - LR (as in Japanese). Zhang derives these orders by raising operations from a split spatial phrase (SP) structure in which LR is the head of the higher SP and PV the head of a lower SP with RE as its complement. Britta Mondorf's paper "The Effect of Prepositional Complements on the Choice of Synthetic or Analytic Comparatives" examines the effect of the syntactic environment, and in particular the presence or absence of a prepositional adjective complement on the distribution between analytic or synthetic adjectival comparatives. There is general consent in the literature that trisyllabic words take the historically more recent analytic comparatives and superlatives and that monosyllables take synthetic variants, with disyllabic words being subject to variation. Mondorf s analysis of computerized corpora indicates that given the right syntactic environment, even monosyllables can strikingly often violate this rule. She argues that it is the complexity of the syntactic environment, and in particular the presence of a prepositional adjective complement, that calls for the analytic variant. In particular, it appears to be those adjectives that frequently take a closely bound prepositional complement, such as fond of, proud of, etc., that favor the analytic form. The patterns for adjectives that are followed by a than-clause are less consistent. Finally, she links up this correlation between the presence of prepositional complements and the use of the analytic comparative with Givón's Proximity Principle, which states that "functional operators will be placed closest, temporally or spatially at the code level, to the conceptual unit to which they are most relevant" (Givón 1991: 89). Günter Rohdenburg's paper "Processing Complexity and the Variable Use of Prepositions" explores four general kinds of prepositional variation: (i) the presence or absence of a given preposition, (ii) the distribution of competing prepositions, (iii) the choice between certain prepositions and the comparative particle than, and (iv) the rivalry between prepositions and so-called interpretative verbs. While Rohdenburg does not deny that each of the (lexico-)grammatical variants in (i)-(iv) may be largely controlled by semantic and

XII

Hubert Cuyckens & Günter Radden

stylistic tendencies, this paper presents a complementary view, in that it suggests that the more explicit grammatical or grammatico-lexical variants tend to be preferred in cognitively more complex environments (the so-called "Complexity Principle"). For each of the types of prepositional variation in (i)-(iv), Rohdenburg presents a detailed case study making use of a large corpus of British and American newspapers. The complexity factors giving rise to prepositional variation include (i) structural factors such as non-canonical orderings of major constituents, as produced, for instance, by relativization or interrogative clause formation, and (ii) semantic/conceptual factors such as the degree of semantic specificity of the superordinate predicates. For instance, the preposition in following have difficulty is far less omissible in a relativized sentence such as (6a), which is structurally and therefore cognitively more complex, than in (6b), which has canonical ordering: (6)

a. The difficulty they had (in) getting there, b. They had difficulty (in) getting there.

In a similar vein, in is lost much more frequently following the less specific verb have + difficulty than following the more specific verb experience + difficulty (which, being semantically more specific, is cognitively more complex). Claudio Di Meóla looks into the issue of case alternation with prepositions in German. In particular, his paper "Präpositionale Rektionsalternation unter dem Gesichtspunkt der Grammatikalisierung: Das Prinzip der 'maximalen Differenzierung'" examines the large number of morphologically transparent prepositions in German which may, without any change in meaning, be construed with a dative or a genitive NP. Originally, these casealternating prepositions were restricted to one case: prepositions such as trotz 'in spite of,' dank 'thanks to,' and entsprechend 'according to' governed the dative, while prepositions such as innerhalb 'within,' während 'during,' and statt 'instead of governed the genitive. Di Meóla demonstrates that the case shifts in these prepositions can only be accounted for in terms of a grammaticalization process, in which the later form (i.e. the preposition governing the new case) is morphologically, semantically, and syntactically maximally different from the original form (the principle of "maximal differentiation from the original structure"). This means that, after a preposition's semantic reanalysis or grammaticalization from content word to function word - whereby its original content meaning is often still transparent - its reanalysis can be further maximized (or made more visible) by using it with a different case form (and potentially change their word order with respect to the NP). The principle is iconically motivated in that a change in function is marked by a change in form. Large-scale corpus data reveal that number and complexity of the NP also play a role in the choice of case: singular and simple NPs tend to be construed in the dative while plural and complex NPs tend to be construed in the genitive. The case preferences with respect to number are also motivated by the principle of maximal differentiation: the case shifts tend to be strongest where the original case form is most marked. The genitive form is more marked in the singular and, therefore, tends to shift to the dative, while the dative form is more marked in the plural and, hence, tends to shift to the genitive. Priska-Monika Hottenroth's paper "Fortbewegungsverben und Ortswechsel im Französischen" contributes to the widely discussed issue of the syntax-semantics interface, and particularly to the question of the relation between verb meaning and argument structure. The study focuses on intransitive verbs of motion in French and their combination with

Introduction

XIII

prepositional phrases denoting the goal or the source of the movement. Generally, two major classes of verbs of motion are distinguished (see Talmy 1975 and others): (i) verbs like arriver 'arrive,' which take (static) spatial prepositional phrases to specify the goal, as in (7), and (ii) verbs like courir 'run,' and flotter 'float,' whose prepositional phrases do not denote a goal but function as PLACE-adjuncts, as in (8): (7) (8)

Jean est arrivé dans le parc. 'Jean arrived in the park.' La bouteille a flotté dans la grotte. 'The bottle floated in the cave.' (not: 'into the cave')

The contrasting behavior of the two verb classes is dependent on the presence of a sublexical predicate CHANGE (LOC(x, REGION (y))). This semantic component is present in verbs like arriver but missing in verbs like courir and flotter, which denote a continuous motion. It is one of the aims of the paper to explain apparent exceptions and to show that the CHANGE-component is also responsible for the fact that the arriver-class of verbs can take prepositional phrases specifying the source as in (9), while continuous motion verbs cannot (10). (9)

Jean sorti de l'université. 'Jean left the university.' ( 10) *Jean courait de l'université. ""Jean ran from the university.'

Part II: Cognitive-Semantic Perspectives on Prepositions

The semantics of prepositions became a major strand of research in cognitive linguistics in the early 1980s - in fact, studies on the polysemy of prepositions and particles, in particular the work of Brugman (1981) on over and Lindner (1981) on up and out, had a significant influence on the development of cognitive semantics. More than any other word class, prepositions and particles are multiply polysemous. Looked at from a cognitive perspective, their seemingly unrelated uses could be shown to form a complex "family-resemblance" network in which the different senses are meaningfully "chained" to one another. All the papers subsumed in this Part "Cognitive-Semantic Perspectives on Prepositions" are, in one way or other, concerned with the issue of prepositional polysemy. The central senses of the basic prepositions tend to belong to the domain of space, and as spatial expressions, prepositions reflect our physical experience of space. Primary among our spatial experiences is that of motion. In motional situations, the division of a scene into a moving trajector and a stable landmark is most conspicuous. The coding of the special situation of a trajector's motion "over" a path as a landmark in German über is analyzed by Meex. Abstract senses of prepositions tend to be derived from concrete, spatial senses by means of generalization or specialization of meaning or by métonymie or metaphoric transfer. The polysemy of prepositions is often claimed to be a reflection of diachronic

XIV

Hubert Cuyckens & Günter Rodden

evolution. De Mulder and Vanderheyden's study of the evolution of French sur 'on' corroborates the primacy and stability over time of the prototypical spatial senses of this preposition while changes of its extended senses defy any such systematic development. The prototype view of prepositions is not shared by all cognitivists. In her analysis of the three Dutch prepositions aan 'on,' op 'on,' and tegen 'against,' Belien argues that prepositional senses are constituted by a single schematic meaning, which she refers to as "Platonic" concept. The relevant component of meaning characterizing the prepositional concepts analyzed is that of force dynamics. In the same vein, Navarro i Ferrando rejects the traditional approach of solely considering the landmark's geometric configuration; prepositional meanings are, as illustrated for the preposition at, also defined by force dynamics and function. The two concluding papers are devoted to motivational aspects of prepositions. Radden and Matthis try to discover the cognitive motivation behind the use of the prepositions in similar to and different from and its variants to and than. Cuyckens examines changes of meaning in four English prepositions and concludes that metonymy is a more basic motivational factor triggering semantic extensions than metaphor.

Individual contributions to Part II In her paper "Die Wegpräposition über," Birgitta Meex applies the framework of cognitive grammar to the analysis of one of the central senses of the German preposition über, that of path. Its English equivalent, over, has been the subject of many cognitive-linguistic studies, which mainly focused on the semantics of the preposition. As a grammatical category, prepositions profile a relationship between between two entities, a trajector and a landmark. The paper is concerned with the sense of über in which the landmark defines a path for the motion of the trajector. Depending on aspectual notions of the motion event, two types of path need to be distinguished: imperfective motion as in He tramped over the field, which describes an incomplete event and involves a summary path covered by the trajector, and perfective motion as in He rowed over the Atlantic, which describes a completed event and involves the reaching of a goal or at least the crossing of the boundary of a landmark. The preposition über may also be used to describe a static situation as motional as in The linen was spanned over the street. This phenomenon of subjective motion needs to be distinguished from subjectifications as in The dwarfs live over the mountain, in which the speaker's reference point is construed subjectively. The conceptual difference between these situations is reflected in the case forms assigned by German über: the former situation is expressed by the accusative case, which is typically used for the notion of path, the latter situation is expressed by the dative case, which is typically used for static, unchanging situations. The aim of Walter De Mulder and Anne Vanderheyden's paper "The Evolution of French Sur: Toward a Diachronic Approach" is both descriptive and theoretical. Its descriptive purpose is to investigate the semantic evolution of the French preposition sur (whose earlier forms were sor, sus, and sur)·, theoretically, it wants to examine the relevance of diachronic prototype semantics - as presented in Geeraerts (1997) - to the study of the evolution of prepositions. In light of the view that the polysemy of current linguistic

Introduction

XV

expressions reflects their diachronic evolution, the authors look into the evolution of sur in order to find the cognitive links between its various meanings. In doing so, they start from the well-established idea in cognitive semantics that the basic meaning of prepositions such as sur is spatial, and that other meanings are derived from this spatial meaning by metaphorical and métonymie transfers. De Mulder and Vanderheyden's analysis reveals that most of the non-spatial meanings of sur were already present from Old French on, and that as such the diachronic evolution of sur can shed little light on its synchronic semantic structure. Still, their analysis corroborates a number of predictions and observations made by Geeraerts (1997): - it is sometimes difficult to precisely locate the origins of new meanings; - the boundaries of the different meanings are in constant evolution; - at least in the case of sur, the prototypical spatial meaning does not change. Furthermore, the authors show that the study of an items's diachronic evolution can help to establish which meanings are part of the synchronic networks and as such may partly solve Rice's (1996) question of how many distinct senses there are in a network. In her paper "Force Dynamics in Static Prepositions: Dutch Aan, Op, and Tegen, Maaike Beliën presents an analysis of the Dutch prepositions aan, op, and tegen which differs from current cognitive semantic analyses in two respects. Rather than treating prepositional meaning in terms of a family-resemblance network of interrelated senses, she proposes a more unified account that describes a preposition's meaning in terms of a single Platonic concept, i.e. a single maximally schematic concept that language users abstract from experience through their perceptual and cognitive apparatus and that may adapt under contextual pressure. In addition, unlike current analyses such as Cuyckens (1991), she accounts for the differences between aan, op, and tegen in terms of force-dynamic properties. As such, then, she characterizes the semantics of aan, op, and tegen in terms of Platonic concepts that each involve contact between two entities, but which differ crucially with respect to force dynamics. In his paper "Towards a Description of the Meaning of At," Ignasi Navarro i Ferrando sets himself the task of redefining the senses of at within the cognitive-linguistic paradigm. Traditionally, spatial prepositions are distinguished on the basis of the geometric shape of the landmark as a point, a surface or line, or an area or volume. The preposition at would, in this account, be described as referring to a landmark in the shape of a point. Navarro convincingly demonstrates the inadequacy of this approach. A prepositional concept involves many more dimensions which need to be considered for an adquate description. The conceptual schema of at is characterized by a specific kind of interaction of the trajector with the landmark. For example, the trajector has a functional front that determines the relationship with the landmark in a canonical way. Navarro describes this conceptual schema as "ENCOUNTER Schema." The ENCOUNTER Schema gives rise to three configurations or conceptual regions of at as well as the corresponding subsenses. The configuration senses are:

XVI

Hubert Cuyckens & Günter

Rodden

(i)

force-dynamic configuration senses, in which the trajector searches for contiguity as in to snatch at something or moves away from the landmark as in at my suggestion·, (ii) topological configuration senses where there is coincidence of trajector and landmark as for example with periods of time which are coincident with events as in The execution took place at dawn·, (iii) functional configuration senses, which highlight a canonical interaction as in ships at sea. The senses of at identified in this paper are shown to form a radial category. An answer to a seemingly trivial question is searched for in Günter Radden and Elizabeth Matthis's paper "Why Similar To, but Different FromT The use of spatial prepositions with similarity and difference is obviously motivated by conceptual metaphor: similarity is understood in terms of closeness as in This is close to the truth and difference is understood in terms of distance This is far from the truth. But why should dynamic prepositions be used to describe static static situations and why should closeness and similarity be seen as motion to a goal and distance and difference as motion away from a source? Crosslinguistic comparisons show that this distribution is not restricted to English but predominates as a general pattern. Radden and Matthis argue that this linguistic pattern points to an underlying folk model, in which close and similar things are seen as being attracted and distant and different things as being repulsed. English is unique among the languages studied in that different may not only be contraed with the Source preposition from, but also with the Goal preposition to and the Comparison preposition than. Usually, these prepositional alternatives are claimed to be no more than stylistic or geograpical variants. The authors claim, however, that each of the three prepositions is associated with its own schematic meaning: from evokes the repulsion schema, to the the attraction schema, and than the comparison schema. An empirical study based on questionnaries largely confirms these assumptions. In his article "Metonymy in Prepositions," Hubert Cuyckens examines semantic relations in four English prepositions. He shows that metonymy plays at least as important a role in motivating (synchronic) semantic extensions and (diachronic) semantic change as does metaphor. In particular, he demonstrates (i) that the metonymies CIRCUMSTANCE FOR CAUSE and GOAL FOR CAUSE underlie semantic extensions in the prepositions with and for, respectively, (ii) that SPATIAL GOAL FOR PURPOSE ultimately motivates the change from to as allative-dative marker to its use as a marker of the to-infinitive, and (iii) that in the change from by as a spatial preposition to its use as a passive marker such metonymies as PATH FOR MEANS OF TRANSIT and MEANS FOR CAUSE have played an important role. This analysis gives Cuyckens the opportunity to look more closely at the distinction between metaphor and metonymy. In line with recent studies on metonymy (Kövecses & Radden 1998; Radden 2000), he argues that when the semantic relation between two concepts can be motivated metaphorically as well as metonymically, the métonymie relation, which is based on contiguity within one domain, is more basic.

Introduction

XVII

References

Brugman, Claudia. (1981): The Story of Over. - M.A. thesis. University of California at Berkeley. Cuyckens, Hubert (1991): The Semantics of Spatial Prepositions in Dutch: A Cognitive-Linguistic Exercise. - Unpublished Ph.D. dissertation. University of Antwerp (UIA). Geeraerts, Dirk (1997): Diachronic Prototype Semantics. - Oxford: Clarendon Press. Givón, Talmy (1991): Isomorphism in the Grammatical Code: Cognitive and Biological Considerations. - In: Studies in Language 15, 85-114. Kövecses, Zoltan & Günter Radden (1998): Metonymy: Developing a Cognitive Linguistic View. In: Cognitive Linguistics 9, 37-77. Lindner, Susan (1981): A Lexico-Semantic Analysis of Verb-Particle Constructions with Up and Out. - Ph.D. Dissertation. University of California at San Diego. Radden, Günter (2000): How Metonymie are Metaphors? - In: A. Barcelona (ed.): Metaphor and Metonymy at the Crossroads: A Cognitive Perspective, 93-108. Berlin: Mouton de Gruyter. Rice, Sally (1996): Prepositional Prototypes. - In: M. Pütz & R. Dirven (eds.): The Construction of Space in Language and Thought, 135-166. Berlin: Mouton de Gruyter. Talmy, Leonard (1975): Semantics and Syntax of Motion. - In: J. P. Kimball (ed.): Syntax and Semantics. Vol. 4, 181-238. New York: Academic Press.

Leuven and Hamburg, September 2001 Hubert Cuyckens and Günter Radden

Parti: Syntactic-Semantic Perspectives on Prepositions

Gisa Rauh Prepositions, Features, and Projections

1. Introduction*

This paper is concerned with an analysis of the projection properties of prepositional forms in examples like the following: (1) (2) (3)

a. b. a. b.

Mary put the vase [on the shelf]. The car stood [behind the bus], John relied [on Jim'spromise], They were angry [about the delay].

a. Bill bought the flowers [at a good price]. b. The child spoke [in a choked voice].

In general, the constituents in brackets are all represented as projections of P, i.e. as PPs. It is also claimed that the potential internal structuring of PPs conforms to the following scheme (see Radford 1988: 246f): (4)

Ρ

(Comp)

Spec(ifier) = measure phrases (e.g. two meters, Mod(ifier) = AP (e.g. far, deep), PP Comp(lement) = NP, CP, PP, 0

right)

What remains unaccounted for, however, is the fact that only the prepositional forms in examples like (1) provide the potential for a complete instantiation of (4). And what remains unexplained is what determines this potential. The present paper is concerned with these problems. In accordance with assumptions of the Principles and Parameters Theory (PPT)1 and the Minimalist Program (MP),2 it will be illustrated that the potential for lexical items to form heads of complex constructions is dependent on their lexical properties. These properties determine the licensing of constituents within maximal projections. With respect to prepositional forms like those in (1), (2), I dedicate my contribution in this volume to my valued colleague and friend Ekkehard König on the occasion of his 60th birthday. 1 2

See for example Chomsky (1985) and Speas (1990). See especially Chomsky (1993, 1995).

4

Gisa Rauh

and (3) it will be argued - and demonstrated - that observable differences concerning the potential internal structure of their projections can be explained on the basis of different lexical properties. The remaining part of this paper is structured as follows: Section 2 introduces lexical properties which license constituents in a structure. This is first demonstrated using examples from outside the problematic area of prepositional forms. Section 3 then illustrates the relationship between the licensing of constituents and the lexical properties introduced in Section 2, considering the prepositional forms of examples like those in (1). In various papers, I have called these "lexical" prepositions (e.g. Rauh 1995, 1996) and identified them as elements of a lexical category (Rauh 1997). Section 4 analyzes the licensing of constituents by the prepositional forms like those in (2) and (3) which together are considered as "grammaticalized prepositional forms." Finally, Section 5 discusses the results and comments on the differences which have been pointed out between the three types of prepositional forms.

2. Features and projections: On the licensing of constituents in a structure

For the licensing of constituents in a structure, the following types of lexical properties are relevant: (i) theta-features, (ii) operator-features, (iii) quantifier-features, and (iv) formal morpho-syntactic features.3 Of the set of features representing these feature types, only those will be considered which are relevant for prepositional forms and their projections. Theta-features are those features which are specified in a theta-grid or in an argument structure. They comprise external , internal , and referential arguments.4 Within projections of lexical heads, licenses specifiers which are interpreted as subjects; for example, John in (5), the licensing being represented by coindexation in (5c):5 (5)

a. snore·. b. John snores. c. VP DPi I John

V I snores

In this case, the type of licensing is theta-marking, which means that there is a thematic relation6 between the theta-marker and the theta-marked constituent and that the theta-

3 4

5

6

These features are discussed in Rauh (2000a, Section 4.1 ) in more detail. The modes of licensing by theta-features characterized here are based on suggestions presented by Higginbotham (1985). The analysis in (5c) which locates the subject internal to VP in underlying structure is motivated, for example, by Koopman & Sportiche (1991). Examples of thematic relations are "Agent," "Theme," "Location," etc.

Prepositions, Features, and Projections

5

marked constituent as a constant replaces the variable in the argument structure of the thetamarker. Via theta-identification licenses adjuncts as modifiers: (6)

a. tall: b. tall boy c.

NP

AP

NPj

tall

I boy

Internal arguments, i.e. complements, are licensed by , again via theta-marking: (7)

a. hit b. the ball hit the fence c.

VP DP¡7'

Δ

the ball

V' V

DP,

hit

the fence

Via theta-identification, on the one hand licenses adjuncts. In this case, the feature of the head of a construction is identified with the feature of the adjunct: (8)

a. boy b. boy behind the curtain c. NP NP

PP

boy behind the curtain

Semantically this identification or "unification" yields coordinate predicates providing the following interpretation for boy behind the curtain: (9)

boy' (R) & behind-the-curtain' (R)

in this case is a specification of and determines the sort of referent denoted by nouns, namely individuals, i.e. objects in the widest sense.

7

DP as the maximal projection of D(eterminer) reanalyzes traditional NPs in the light of insights concerning, for example, structural properties of pronouns like we, these, those, and others which may be used either intransitively or transitively, as in we/these!those linguists, and which are therefore analyzed as intransitive or transitive representatives of D. More important for a reanalysis of NPs and DPs, however, is the fact that elements of the category D are operators in the sense described below.

6

Gisa

Rauh

On the other hand, licenses operators, which are themselves specified by operatorfeatures. In the context of nouns, operators are lexically represented by determiners: (10)

a. a:

[ - d e f i n i t e ) ] ()

this: b. a/this

[+def,+deic(tic),+prox(imal)] () 8 boy

c.

DP D

NP

I

I

a [-def]i this [+def, +deic, +prox],

Ν I boy

The licensing in this case is achieved via theta-binding. Semantically this means that out of the set of individuals satisfying the predicate expressed by the NP, the determiners select referents according to their specification. Within the PPT and the MP framework, determiners are syntactically described as elements of the functional category D in the extended projection of N, not inside the NP, as is shown in (10c). In contexts other than N, the analysis of operators is similar. Quantifier-features are those features which license quantifiers in the context of lexical items. For nouns, Löbel (1989, 1990) demonstrates that quantifiers are dependent on the inherent features [±count]. These, then, are the features which license quantifiers as functional elements in extended projections of N, given that the quantifiers as well as the nouns exhibit these features and that there is agreement between the two: (11)

a.

many

[+count]

much [-count] b.

many [+count] books [+count] *much [-count] books [+count]

In the framework of the MP, agreement of these features is checked via head-to-head movement: (11)

c.

many books¡ [+count] [+count]

If the features do not agree or if one of the partners lacks this feature, then a potential quantifier is not licensed.

8

In the context o f [+deic], the feature [+def] is redundant. It is therefore not represented in the examples to follow.

Prepositions,

Features,

and

7

Projections

Amongst the formal morpho-syntactic features there are case-features. 9 They are of two types, namely those that specify case-bearers and those that specify case-assigners: (12)

a. him hit

[OBJ] [assign OBj(ective CASE)] 1 0

Since only NPs or DPs bear case, the contextual feature [assign OBJ] licenses the syntactic category of complements which as such are licensed by : (12)

b. hit

[assign OBJ] him [OBJ]

In the MP, the licensing of NPs or DPs is described via spec(ifier)-head agreement. To check agreement - in this case of the two case-features - the NP or DP is moved to the specifier position of a superordinate agreement phrase Agr0P'> at LF, i.e. invisibly to the phonetic interpretation of the phrase: (12)

c.

Agr 0 P

hit [assign OBJ]

him, [OBJ]

If the case-features do not agree, the constituent is not licensed. It is important to note that there is an immediate dependency relation between the licensing of constituents and lexical properties: the presence of the lexical properties is a prerequisite for the licensing of constituents in a projection. Therefore, the lexical prop-erties of lexical items can be derived from their empirically attested projection potential. On the other hand, the presence of these properties explains this potential. In what follows, this dependency will be exploited to describe systematic differences between prepositional forms like those presented in (1), (2), and (3). To avoid confusion, it should be noted beforehand that, due to the modes of licensing discussed here, possible projections of prepositions will crucially differ from (4), in that the elements under Spec, which are located within PP in (4), will, in the analysis presented here, appear outside of PP in the projection extended by functional categories.

9

10

11

The formal morpho-syntactic features include also gender, number, and tense features (cf. Chomsky 1995). The notation [assign OBJ] is here adapted to the notation applied in the MP by Chomsky (1995). In previous papers, I represented this property as ">+S:obj , [assign OBJ], and [±ext]. The subscript "sca(lar)" of S indicates that scalar PPs denote scalar spaces. Scalar prepositions differ from spatial and temporal prepositions in that they always license DP-complements and can never be used intransitively.21 The latter restriction is due to the fact that they are identical in form with spatial prepositions and that the scalar meaning is interpretable only on the basis of the complement. Like temporal prepositions, scalar prepositions license either R or Ext in their extended projection, but not both. The reasons are the same: the linear conceptualization of scales does not allow a double specification. With respect to their phonetic form, scalar prepositions are a subset of spatial prepositions. Apart from a few forms like after, before, since, and until, this is also true for temporal prepositions such as at, on, and in. It is therefore worth considering whether these forms are equipped with a specification for "locative," "temporal," and "scalar" or whether they are lexically underspecified and receive a specification only through a given context. 22 By indexing S as "locative," "temporal," or "scalar," the first of these solutions was chosen here. The following arguments support this decision: the licensing of constituents is based on a mutual relationship of two partners. Spatial, temporal, and scalar prepositions license amongst other constituents - different types of measure phrases: spatial prepositions two meters or far, temporal prepositions two hours or immediately, and scalar prepositions two degrees or well, for example. This can be explained by a specification of measure phrases as [+loc], [+temp], and [+sca] and the requirement of agreement between these features and the subscripts of the referential argument . If there is no agreement, licensing is not possible. 20

21

22

A more detailed analysis of temporal prepositions with some different suggestions is provided by Haumann (1997; this volume). Prepositional forms with scalar meanings belong to various grammatical categories, as Wege (1997) points out. Here only those forms are considered which belong to a lexical category. For a treatment of underspecification as a model of lexical representation the reader is referred to Pinkal (1996), for example.

Prepositions,

Features, and

15

Projections

In principle, underspecification can only be assumed if a preposition may unrestrictedly be used in all contexts or if restrictions derive from elsewhere. The first case is found only with very few prepositions, for example, at, from, and to. The second might be considered for scalar prepositions because the way scales are conceptualized means that only selected forms of the inventory of spatial prepositions can be used in this domain, for example above and below but not behind. However, not every spatial preposition which is conceptually adequate may in fact be used scalarly, as the example of beneath demonstrates.23 Facts like these provide arguments against lexical underspecification. Another argument against underspecification may be seen in the fact that apparently non-spatial uses of prepositions are closely connected to spatial ones, the reinterpretation being determined by a conceptualization which suits the new semantic domain. In the representation suggested here, this relationship is expressed explicitly by the different subscripts of . Therefore, prepositional forms with different semantic - and syntactic properties will here be associated with different sets of features. The examples below illustrate this: (36)

(37)

(38) (39)

on,: on2'. on}: below,: below2: belowy behind behind2: after,·, after 2. aftery.









[assign [assign [assign [assign

OBJ] OBJ] ING] OBJ]

[assign OBJ] [assign OBJ] [assign OBJ] [assign +FINITE] [assign ING]

[-ext] [-ext] [-ext] [+ext] [+ext] [+ext] [+ext] [+ext] [+ext] [+ext] [+ext]

Spatial, temporal, and scalar prepositions are all referred to as "lexical" prepositions because they exhibit properties characteristic of lexical categories as opposed to functional categories.24 A major characteristic is that "lexical" prepositions have theta-features. All of the prepositions considered "lexical" contain, in addition to an external argument , a referential argument in their feature specification which determines spaces, i.e. spatial, temporal, or scalar spaces, as referents of lexical PPs. In addition, many forms have an internal argument : scalar prepositions always have it whereas spatial and temporal prepositions may also be intransitive. It is above all theta-features which distinguish lexical prepositions from those forms which are here referred to as "grammaticalized"

23 24

See Wege (1997) for further discussion. The differences betweeen lexical and functional categories have been described in detail by Abney (1987), Fukui & Speas (1986), and Grimshaw (1991). A survey is provided in Rauh (1997). In addition to theta-features, which characterize elements of lexical categories but not those of functional categories, the major differences are the following: Elements of lexical categories have descriptive meaning, they are grammatically autonomous, they form open sets, and as a class they exhibit variation concerning the categories of their complements. Elements of functional categories, on the other hand, do not have descriptive meaning, are grammatically dependent, form closed sets, and always have the same category as their complement.

16

G isa Rauh

prepositional forms and whose features and projections are the subject of the following section.

4. Grammaticalized prepositional forms and their features and projections

Examples of grammaticalized prepositional forms are given in (2) and (3), repeated below for convenience: (2) (3)

a. b. a. b.

John relied on Jim 's promise. They were angry about the delay. Bill bought the flowers at a good price. The child spoke in a choked voice.

The label "grammaticalized" forms was chosen to express that in most cases these forms are identical to spatial prepositions but that they have reduced sets of features in comparison to those of the spatial prepositions. This relationship is comparable to processes of grammaticalization first characterized by Meillet (1912) and more recently by Lehmann (1985, 1986/87). Two types of preposition have to be distinguished which I have called "governed" or "case" prepositions and "grammatical" prepositions in previous papers (e.g. Rauh 1996, 1997). "Governed" or "case" prepositions in construction with NPs or DPs always form complements of lexical heads. They are lexically selected or - in traditional terms - "governed" by these heads in the same way that morphological cases are. The two labels express the two aspects of this property. Governed prepositions are lexically selected by elements of the categories V, N, A, and Ρ in the same way. (2) provides examples of selection by verbs and adjectives, and (40) provides examples of selection by nouns and prepositions or adverbs:25 (40)

a. The lawyer had no influence over his client, b. The horse raced out o / t h e barn.

Projections of governed prepositions differ from those of lexical prepositions in that neither determiner-like operators (41) nor quantifier-like elements (42) nor adjuncts (43) are licensed: (41) (42) (43)

25

a. b. a. b. a. b.

The village is [somewhere/right over the hill]. *The lawyer had no influence [somewhere/right over his client], The village is [all the way over the hill]. *The lawyer had no influence [all the way over his client]. The village is [over the hill in a valley]. *The lawyer had no influence [over his client in his worries].

It should be noted that a subset of traditional adverbs is here considered as prepositions, following Jackendoff (1973) and Emonds (1976). For the relevance or irrelevance of categories or category names like these, see Rauh (1999, 2000a).

Prepositions,

Features, and

Projections

17

Since these constituents are licensed by referential arguments, this leads to the conclusion that no such feature is included in their feature specification. This conclusion is supported by the fact that projections of governed prepositions - as opposed to projections of lexical prepositions - do not denote preposition-specific referents, i.e. spaces of any kind. Linguistically, this is reflected by the the fact that the sequence [P+DP] cannot be replaced by an appropriate proform, which is possible in the case of lexical prepositions, as demonstrated by the examples in (44) and (45): (44)

(45)

a. The village is over the hill and Bill lives there. b. Bill promised to come at five o 'clock and came then. c. The temperature rose above freezing point and stayed there. "John relied on Jim'spromise and Mary relied there as well.

In addition, governed prepositions differ from lexical prepositions in that their relation to their complements is not thematic, i.e. not determined by theta-marking. A thematic relation exists, however, between the governing lexical head and the complement of the governed preposition. This explains the awkwardness of the following examples: (46)

a. ?The lawyer had no influence over his door. b. ?John relied on his table.

The combinations influence and door as well as relied and table are hardly interpretable, whereas the combinations over and door as well as on and table would not cause any problem. The interpretability of the examples in (2) and (40) is consequently determined by the thematic relation between the lexical head and the complement of the governed preposition and not by one between this preposition and its complement. This leads to the conclusion that governed prepositions do not have the feature , i.e. it is not the prepositions that license their complements. Since governed prepositions are selected and are always part of the complement of the selecting head and never part of adjuncts or predicates, there is no evidence for the feature either. It follows, therefore, that governed prepositions have no theta-features at all. Neither do they have quantifier-features. Their being lexically dependent on governing lexical heads, the fact that their complements are always DPs, and the fact that there is a thematic relation between their governing lexical heads and their complements all support their analysis as case forms, in particular as inherent cases in the sense of Chomsky (1981), which I have suggested elsewhere (e.g. Rauh 1991, 1993, 1996, 1997). It should be noted also that governed prepositions form a closed set. 26 Finally, further support for this analysis is provided by the following examples, showing that sequences of governed prepositions and DPs behave like DPs and not like lexical PPs: (47)

(48)

26

a. b. c. a. b. c.

John¡ shaved [ D P himself,/*him¡]. John, put the book [ PP near *himself¡/him¡]. John¡ relied [ PrcgP on himself¡/*him¡]. Mary saw [ D P the kids¡] several times each,. *Mary sat [ PP near the kids¡] several times each¡. Mary thought [ PregP about the kids¡] several times each.

The following forms have been identified as governed prepositions: about, after, against, between, by, for, from, in, into, of, on, over, to, towards, under, up, with, without.

at,

18

Gisa Rauh

Governed prepositions differ from other case forms, however, in that they are not suffixes and in that they themselves contain the feature [assign OBJ], which is evident in the following examples: (49)

a. John relied on him. b. The lawyer had no influence over her.

This illustrates that although governed prepositions have the grammatical function of case affixes, the grammaticalization of lexical prepositions to inherent cases has left some prepositional properties unchanged. As a result of the properties discussed above, governed prepositions have no potential to license lexical projections of their own. They rather represent functional K(ase)-heads in extended projections of N. As such they represent lexically selected forms of inherent cases with the prepositional feature [assign OBJ]: (50)

VP V

KP

Ι

rely Κ

I [assign on [assign OBJ]] on [assign OBJ]

DP

Jim's promise

[OBJ]

In addition to the feature [assign OBJ], governed prepositions have inherited the phonetic form, the position, and certain semantic features from lexical prepositions. The inheritance of semantic features is especially evident in those contexts where alternatives are possible: (51)

a. to talk to someone about

b. to fight for something against

Governed prepositions are richer and semantically more differentiated than case affixes, which may be the reason why they have entered the case system. 27 The prepositional character of governed prepositions is also evident in constructions which include preposition stranding, as in (52): (52)

Who¡ did he rely on t¡?

A comparable construction involving case affixes is impossible. This shows that governed prepositions are double-faced, in part already case forms, in part still prepositions. The second type of grammaticalized prepositional forms exhibits characteristics quite different from those of governed prepositions. The forms referred to as "grammatical" prepositions do not represent parts of complements of lexical heads and are not selected by the latter. On the contrary, they are always part of constructions which function as adjuncts or predicates. Following the discussion in Section 2, this means that the theta-feature , which licenses adjuncts and predicates, is included in their feature specification.

27

For a further discussion of governed prepositions as cases, see Rauh (1993, esp. Section 4).

Prepositions,

Features, and

19

Projections

Grammatical prepositions share with governed prepositions the inability to license determiner-like operators or quantifier-like elements in their projections: (53)

a. *Bill bought the flowers [somewhere at a good price], b. *Bill bought the flowers [close at a good price].

They cannot license adjuncts either: (54)

*Bill bought the flowers [at a good price near three

dollars].

This indicates that they are like governed prepositions in not having a referential argument in their feature specification. This is confirmed by the fact that the sequence [P+DP] cannot be replaced by a corresponding proform: (55)

*Bill bought the flowers at a good price and John bought some there as well.

Semantically this means that projections of grammatical prepositions do not refer to spaces conceptualized in any way comparable to spatial, temporal, or scalar spaces. If traditional conjunctions like although or because are not considered as prepositional forms, then grammatical prepositions in general license DP- or NP-complements and seldom sentence-like complements, as in (56): (56)

With the bus drivers striking/on strike, we have to walk.

It follows that the theta-feature is included in their feature specification as well as the case-feature [assign OBJ], which is however not present if sentence-like complements are licensed. (57) provides examples of maximal projections of grammatical prepositions: (57)

a.

Agr 0 P

b.

PP V¡ngP¡/Sc¡ with

at

[assign OBJ]

the bus drivers striking on strike

a good price

[OBJ]

Grammatical prepositions exhibit a strong tendency to form fixed phrases together with their complements, as the examples in (58) illustrate: (58)

a. b. c. d.

Bill was in a hurry. He was under pressure for weeks. Work is now in progress. The guard on duty became ill.

These complements are not referential. It is also debatable whether there are any case requirements to fulfill. Since the phrases are fixed and no pronominal substitution can be made for the NP or DP, an empirical test is impossible. A structure like the following can be assumed for these examples:

Gisa Rauh

20 (59)

PP Ρ

DP¡/NP¡

Simple N-complements are assumed to be lexically incorporated into P: PP

(60) Ρ Ρ

NP N¡

Ν t,

This assumption explains not only why the constructions in question are absolutely invariable, but also why they behave like intransitive prepositions and can be made transitive under certain circumstances, as the examples in (61) illustrate: (61)

a. He said nothing in addition {to what he had said before), b. John left instead (of Mary).

The representation of in+stead as a single graphical unit in (61b) illustrates that lexical incorporation has already taken place.

5. Summary and conclusion

A comparison of the feature specifications and the projections o f lexical, governed and grammatical prepositions explains why the last two are referred to as "grammaticalized" forms: in both cases, the sets o f features - and the projections - are substantially reduced compared to the set o f features of lexical prepositions. The feature set o f lexical prepositions always includes a referential argument , which identifies spaces as the referents o f their projections. Neither governed nor grammatical prepositions exhibit this feature. A s a consequence, they do not project their own referring projections. What is referring in these [P+DP]-sequences is only the DP and even this is not always the case with grammatical prepositions, as has been illustrated. Compared to transitive lexical prepositions, grammatical prepositions may lack the internal argument as well as . Governed prepositions always lack these features. In addition, they lack the external argument , since they always mark the complement o f a governing lexical head. Grammatical prepositions, on the other hand, do exhibit this feature since their projections are licensed as adjuncts or predicates. Apart from grammatical prepositions in fixed phrases, all prepositions with NPor DP-complements exhibit the case-assigning - or case-requiring - property [assign OBJ]. For governed prepositions, the effect o f grammaticalization is a semantification o f cases. For grammatical prepositions, it is the "predicatizing" of nouns with the additional function of adding partial semantic prepositional content. "Predicatizing" here means that constitu-

21

Prepositions, Features, and Projections

ents are made available for licensing as adjuncts or predicates, which is achieved by adding to the feature structure of the projections of nouns. The following examples of so-called "telegraphic style" illustrate that, in fact, grammaticalized prepositions are "more grammatical" than lexical prepositions, since the meanings of expressions are much easier to derive if a grammaticalized preposition is dropped (62, 63) than if this happens to a lexical preposition (64): (62) (63) (64)

a. b. a. b. a. b.

John rely Jim promise. They angry delay. Bill buy flower good price. Child speak choked voice. Work four o'clock. Car stand bus.

-> at?, before?, after?, until? —> behind?, near?, in front of?

This fits in well with a general characterization of telegraphic style which claims that grammatical or functional elements may be dropped - or learned later as far as language acquisition is concerned - because they are less relevant for the semantic interpretation of the expression as a whole than content or lexical elements. It has thus been demonstrated that a uniform representation of [P+DP] sequences, as in (1), (2), or (3), as PPs is inadequate. What is inadequate, to begin with, is the uniform representation of the prepositional forms as members of the category P. It is to be expected that the category to which elements are assigned determines their grammatical behavior, that it predicts which constructions they may form part of, which projections they license. The assignment of the prepositional forms in (1), (2), and (3) to Ρ allows no such predictions, since the prepositional forms, as has been demonstrated, form part of different constructions rather than the same one. It has also been demonstrated that this is due to their differing lexical, grammatically relevant properties, here identified as theta-features, operator-features, quantifier-features, and case-features. As a consequence, linguistic items in the lexicon should not be represented by grammatically rather non-revealing categories like Ν, A, V, or Ρ or their specifications on the basis of [±N] and [±V] but by individual sets of feature specifications which predict the grammatical behavior of each given item. This has been suggested in Rauh (1998a). 28 On the basis of these sets of features, various grammatical categories can be defined as natural classes. The definition is based on single features or sets of features which represent an intersection of the feature specifications of their members. In this sense, lexical, governed, and grammatical prepositions may be defined as three different categories. For lexical prepositions, the features , , and [±ext] are defining, for governed prepositions it is [assign OBJ], and for grammatical prepositions together with the optionally lexicalized combination of and [assign OBJ]. Additional categories could be the following:

28

It should be noted that according to this suggestion, nodes in phrase-structure trees are not labeled with categories such as Ρ, V, N, D, K, Q, etc. and their projections but rather with feature specifications. It is obvious that this kind of representation is much more specific and avoids constantly recurring problems of categories and categorization.

22

G isa Rauh a. b. c. d. e.

spatial prepositions: temporal prepositions: scalar prepositions: transitive prepositions: intransitive prepositions:









[±ext] [±ext]
[assign OBJ] [assign OBJ] [±ext]

The feature representations list the features which all the members o f the category exhibit. Individual members m a y and in general will exhibit additional features. Thus, for example, spatial prepositions which license DP-complements in addition exhibit the features and [assign OBJ], In this kind o f categorization, individual lexical items may and in general will belong to several categories. Naturally, this means that the defining features o f a single category will not predict all the grammatical properties o f a given item, including projection properties, as would be expected o f P. These will rather be predicted by all the categories to which the item belongs or, o f course, b y its individual set o f features. A n d it has b e e n demonstrated that these sets o f features as well as the projections they determine are different for the prepositional forms in examples like those in (1), (2), and (3) and also for others. Therefore, these forms are not simply Ps and their projections not simply PPs, as is generally assumed in the literature.

References

Abney, Steven P. (1987): JTie English Noun Phrase in Its Sentential Aspects. - Unpublished Ph.D. dissertation. Cambridge, MA: MIT. Chomsky, Noam (1981): Lectures on Government and Binding. - Dordrecht: Foris Publications. - ( 1986): Knowledge of Language: Its Nature, Origin, and Use. - New York, NY: Praeger. - (1993): A Minimalist Program for Linguistic Theory. - In: K. Hale & S. J. Keyser (eds.): The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger, 1-52. Cambridge, MA: The MIT Press. - (1995): The Minimalist Program. - Cambridge, MA: The MIT Press. Emonds, Joseph E. (1976): A Transformational Approach to English Syntax: Root, Structure Preserving and Local Transformations. - New York, NY: Academic Press. - ( 1987): The Invisible Category Principle. - In: Linguistic Inquiry 18, 613-632. Fukui, Naoki & Margaret Speas (1986): Specifiers and Projection. - In: MIT Working Papers in Linguistics 8, 128-172. Cambridge, MA: The MIT Press . Grimshaw, Jane ( 1991 ): Extended Projection. - Ms. Waltham, MA: Brandeis University. Haider, Hubert (1988): Die Struktur der deutschen Nominalphrase. - In: Zeitschrift für Sprachwissenschaft 7, 32-59. Haumann, Dagmar ( 1997): The Syntax of Subordination. - Tübingen: Niemeyer. - (this volume). The Projections of Temporal Prepositions. Jackendoff, Ray S. (1973): The Base Rules for Prepositional Phrases. - In: S. Anderson & P. Kiparsky (eds.): A Festschrift for Morris Halle, 345-366. New York, NY: Holt, Rinehart & Winston. Higginbotham, James (1985): On Semantics. - In: Linguistic Inquiry 16, 547-593. Koopman, Hilda & Dominique Sportiche (1991): The Position of Subjects. - In: Lingua 85, 211-258.

Prepositions, Features, and Projections

23

Lehmann, Christian (1985): Grammaticalization: Synchronic Variation and Diachronic Change. - In: Lingua e Stile 20, 303-318. - (1986/87): Grammaticalization and Linguistic Typology. - In: General Linguistics 26/27, 3-22. Löbel, Elisabeth (1989): Q as a Functional Category. - In: C. Bhatt, E. Löbel & C. Schmidt (eds.): Syntactic Phrase Structure Phenomena in Noun Phrases and Sentences, 133-158. Amsterdam: John Benjamins. - (1990): D und Q als funktionale Kategorien in der Nominalphrase. - In: Linguistische Berichte 127, 232-264. Meillet, Antoine (1912): L'évolution des formes grammaticales. - In: A. Meillet (ed.) (1921): Linguistique historique et linguistique générale. Tome I, 130-148. Paris: Klincksieck. Pinkal, Manfred (1996): Wie die Semantik arbeitet: Ein unterspezifiziertes Modell. - In: G. Harras & E. Löbel (eds.): Wenn die Semantik arbeitet, 57-88. Tübingen: Niemeyer. Radford, Andrew (1988): Transformational Grammar: A First Course. - Cambridge: Cambridge University Press. Rauh, Gisa (1991): Prepositional Forms in the Lexicon. - In: G. Rauh (ed.): Approaches to Prepositions, 169-223. Tübingen: Narr. - (1993): Kasus und Präpositionen im Englischen. - In: Indogermanische Forschungen 98, 252292. - (1995): Präpositionen und Rollen. - In: Sprachwissenschaft 20, 123-167. - (1996): Zur Struktur von Präpositionalphrasen im Englischen. - In: Zeitschrift für Sprachwissenschaft 15, 178-230. - (1997): Englische Präpositionen zwischen lexikalischen und funktionalen Kategorien. - In: E. Löbel & G. Rauh (eds.): Lexikalische Kategorien und Merkmale, 125-167. Tübingen: Niemeyer. - (1999): Adverb oder Präposition? Von der Notwendigkeit einer Abgrenzung von Wortarten und grammatischen Kategorien und der Gefahr einer terminologischen Falle. - In: E. Eggers, J. Becker, J. Udolph & D. Weber (eds.): Florilegium Linguisticum: Festschrift für Wolfgang P. Schmid zum 70. Geburtstag, 367-392. Frankfurt a.M.: Lang. - (2000a): Don't Call it "X"! or: Why X Does Not Represent Grammatical Categories. - In: H. Janßen (ed.): Verbal Projections, 7-37. Tübingen: Niemeyer. - (2000b): Determinantien und Lokaladverbien. Projektionsrelevante Analogien. - In: J. Bayer & C. Römer (eds.): Von der Philologie zur Grammatiktheorie, 139-160. Tübingen: Niemeyer. Speas, Margaret (1990): Phrase Structure in Natural Language. - Dordrecht: Kluwer. Wege, Barbara (1997): Skalare Präpositionen - grammatikalisierte Varianten? - In: D. Haumann & S. Schierholz (eds.): Lexikalische und grammatische Eigenschaften präpositionaler Elemente, 19-48. Tübingen: Niemeyer. Zwarts, Joost (1992): X-Syntax - X-Semantics: On the Interpretation of Functional and Lexical Heads. - Ph.D. dissertation. Utrecht University (published in the UiL OTS Dissertation Series).

Dagmar Haumann

The Projections of Temporal Prepositions

1. Introduction*

The aim of this paper is to give an account of both the internal and external syntax o f temporal prepositional projections as illustrated by the italicized examples in (1): (1)

a. b. c. d. e. f. g. h.

We couldn't get hold of him until the end of the conference. We planned to meet two hours after the conference was over. I haven't see him before. I haven't seen him since the conference. We talked about this during dinner. They met three weeks ago. We talked about this while the kids were at home. They were shocked when they saw what had happened.

The general background o f the analysis presented is within the context o f recent generative grammar, or, to be more precise, the Minimalist Program as developed in Chomsky (1993, 1995). The focus of the proposed analysis is on licensing, i.e. on the conditions under which the presence, the distribution, and the internal structure o f temporal PPs are sanctioned within a given complex structure. The following, interrelated assumptions are at the heart o f the analysis presented: first, prepositions, like verbs and nouns, have an extended projection, 1 i.e. they are dominated by a set o f specific functional projections (FP) as indicated in (2); 2 second, the amount o f functional architecture dominating temporal prepositions is determined by the argument structure and the Case-properties of the individual temporal prepositions. 3

1

2

3

This paper is dedicated to Hans Ulrich Boas on the occasion of his 60th birthday. The term "extended projection" is due to Grimshaw (1991); see Haider (1988) and van Riemsdijk (1990) for related concepts. It should be noted, though, that Grimshaw assumes that prepositions do not have an extended projection but figure within the extended projections of verbs and nouns. The most detailed analysis of PPs and the functional structure that dominates PP presented is that by Koopman (1993); see also van Riemsdijk (1990), Zwarts (1992), Rooryck (1993), Haumann (1997, 1999), Rauh (1997a, 1997b, 1998, this volume), and Wege (1997) for functional projections dominating PP. The second assumption reflects Haider's (1993: 50) contention that there are as many functional heads as there are projectible features. That is to say, the relation that exists between functional projections and grammatical features is isomorphic.

Dagmar Haumann

26 (2)

FP

spec

F'

\ FP

F

PP

In the following, the focus will be on the argument structure and the Case-properties of temporal prepositions. I will try to show that these two ingredients determine both the amount and the shape of the functional architecture dominating temporal prepositions. The focus will be on feature checking and licensing within each of the functional layers which dominate temporal prepositions. I shall start in Section 2 with the description of the internal structure of temporal PPs. The analysis presented basically incorporates Jespersen's (1924) and, more recently, Jackendoff s (1973, 1977) and Emonds' (1976) contention that homophonous elements such as before in (3) should be assigned to only one category, namely prepositions, and not, as in traditional grammar, to three, i.e. prepositions (3a), subordinating conjunctions (3b), or adverbs (3 c). (3)

a. I didn't see him before the conference. b. The dizzy turns started long before the conference started.

c. I haven't seen him before. Rather than adopting this traditional classification, Jackendoff (1973, 1977) and Emonds (1976) assume that prepositions, like verbs, come in three flavors, namely as transitive prepositions, taking either a nominal or a sentential complement as in (3a) and (3b), respectively, and as intransitive prepositions as in (3c). The focus in this section will be on the internal argument and the Case-assigning and non-Case-assigning properties of temporal prepositions. In Section 3, I turn to what, in the generative tradition, or more specifically after Williams' (1980, 1981) papers on argument structure, has been referred to as the external argument of a lexical head. Section 4 focuses on the distribution and licensing of nonargumental elements within the prepositional projections. The elements under consideration are elements that Jackendoff (1973, 1977) and Emonds (1976) refer to as "modifiers or specifiers of P" such as two hours in (lb) or long as in (3b). Section 5, finally, is concerned with the linearization of elements contained within the projections of temporal prepositions. A short summary is presented in Section 6.

The Projections of Temporal Prepositions

27

2. The internal argument

Temporal prepositions have an argument structure that contains one external and at most one internal argument. In addition to these, their argument structure contains a referential argument which encodes the ontological category "time" as well as the reference properties of the temporal preposition. The provisional argument structure of temporal prepositions is illustrated in (4), where the integer designates the external argument, whilst designates the internal and the referential argument encoding "time": (4)

a. < 1, 2, t > 4 b. < l , t >

If the argument structure contains an internal argument as indicated in (4a), the temporal preposition can take a nominal complement as in (5a, b) or a sentential one as in (5c, d). If the argument structure, however, does not contain an internal argument as is shown in (4b), the temporal prepositions can only occur intransitively as in (5e, f): (5)

a. b. c. d. e. f.

The dizzy turns started after the conference. I haven't seen him since his divorce. You will have to wait until Santa Clause comes to town. We all knew what they were up to long before Tom told us. 1 haven't seen him before. I haven't seen him since.

Whereas before, after, since, and until, as can be seen in (6), may take sentential and nominal internal arguments, the preposition during and the post-position ago in (7) only allow a nominal internal argument, and finally, when and while, as can be seen in (8), are restricted to taking a sentential internal argument. (6)

4

a. b. c. d. e. f. g. h.

I didn't see him before the conference. I didn't see him before the conference started. The dizzy turns started after the conference. The dizzy turns started after the conference was over. I haven't seen him since his divorce. I haven't seen him since he got divorced. You will have to wait until December. You will have to wait until Santa Clause comes to town.

The assumption that the argument structure of Ps is associated with a referential argument is not beyond controversy. See Haumann (1990, 1993, 1997), Zwarts (1992), and Rauh (1993, 1997a, 1997b, 1998, this volume) for analyses that include a referential argument in the argument structure of spatial and temporal prepositions, and see Bierwisch (1988), Wunderlich (1992), and Steinitz (1996, 1997) for analyses that do not. See especially Rauh (1993,· 1997a, 1997b, 1998, this volume), and Wege (1997) for an argument structure-based distinction between lexical and nonlexical or functional prepositions; see also Zwarts (1997) for a discussion of the lexical and functional properties of prepositions.

28

Dagmar Haumann (7)

(8)

a. b. c. d. a. b. c. d.

I fell asleep during the movie. *I fell asleep during the movie was on. They met three weeks ago. *They met holiday had started ago. I was delighted when the party was over. I slept while dinner was being prepared. *I was delighted when the party. *I slept while dinner.

The feature the temporal prepositions in (6), (7), and (8) have in common is that their argument structures contain an internal argument that is lexicalized by either a nominal or a sentential constituent. During and ago in (7) and when and while in (8) differ from the temporal prepositions in (6) in that they do not allow categorial variation of their structural complement, i.e. the internal argument must be nominal with during and ago and sentential with when and while.

2.1. Nominal internal arguments It is generally assumed that if the internal argument of a preposition is a nominal constituent, as in (6a, c, e, g) and (7), it must bear Case. In addition to having an argument structure that contains an internal argument, after, before, since, until, during, and ago must be specified for Case-assigning properties. As can be seen from the contrast between the grammatical strings in (6a, c, e, g) and (7) on the one hand, and the ungrammatical ones in (8c, d) on the other hand, when and while are not endowed with this Case-assigning property and are thus unable to take nominal internal arguments. To distinguish between Case-assigning and non-Case-assigning prepositions, I propose introducing the binary feature [±CAS], where [+CAS] lexically specifies Case-assigners as in (9), and [-CAS] non-Case-assigners as in (10). 5 (9)

{< 1, 2, t >, [+CAS]}

(10)

{< 1, 2, t >, [-CAS]}

( = after, before, since, until as in (6a, c, e, g) and during and ago in (7a, c)) ( = when and while in (8c, d) and after, before, since, until as in (6b, d, f, h))

As has already been mentioned in the introductory Section, Case features are morphosyntactic features and trigger the presence of a functional projection. Checking, as is assumed in the Minimalist Program, takes place prior to Logical Form (LF), or at LF. The point in the derivation at which checking takes place is basically determined by the relative strength of the features involved. If the features are strong, the effects of movement are visible; i.e. there is word-order variation. However, if the features are weak, there are nonobservable effects of movement. Specifying the temporal prepositions in (6a, c, e, g) and (7) as Case-assigners leads us to assume a functional projection dominating PP, where Case is checked under specifier-head

5

See also Rauh (1997a: 160ff.), who proposes encoding Case-assigning properties as [±K].

The Projections

of Temporal

29

Prepositions

agreement. In analogy to structural Case-checking in extended verbal projections, 6 1 assume that Ρ is dominated by an object agreement phrase, which I refer to as P-ArgoP as in (11). Specifier-head agreement within this functional projection is achieved by two related movements: the prepositional head moves and adjoins to P-Agr0, and the internal nominal argument, i.e. DP, moves into the specifer position of P-AgroP. These movements are illustrated in (12). (11)

P-AgroP spec

P-Agr 0 '

Ρ (12)

DP

P-AgroP spec

P-Agr 0 ' P-Agro

a. pre-LF: three weeksj b. pre-LF: LF: the movies¡

PP

ago, during¡

As has already been mentioned, these movements can take place either overtly, i.e. prior to LF, and thus have visible effect as in the case of ago, or the aforementioned movements take place covertly, i.e. at LF, and thus do not have visible effects as is the case with after, before, since, until, and during. This analysis is similar to the analyses presented by Koopman (1993) and van Riemsdijk (1990) in that so-called post-positions are conceived as underlying prepositions.7 The postpositional order is arrived at by overt movement of ago and its internal argument, as indicated in (12a). With during in (12b), movement is postponed until LF. As can be seen from the representations in (12a) and (12b), the movements that are involved are identical and so are the Logical Forms; (12a) and (12b) differ only with respect to "when" movement takes place. 6 7

See, for example, Sportiche (1990: 74), Zwart (1994), and Chomsky (1995: 122, 149, passim). See also Rooryck (1993) for the assumption that the postpositional order is derived by movement. It should be noted, though, that Koopman's (1993), van Riemsdijk's (1990) and Rooryck's (1993) analyses are not fully compatible. While Koopman's (1993) analysis incorporates Kayne's (1994) "leftward movement only" postulate, van Riemsdijk's (1990) and Rooryck's (1993) analyses are based on the assumption that movement within the functional structure dominating PP is rightward movement.

30

Dagmar Haumann

2.2. Sentential internal arguments As can be seen from the contrasts between during and ago in (7) and when and while in (8), nominal and sentential internal arguments do not share the same distribution. Sentential internal arguments do not bear Case. Hence, temporal prepositions that govern these elements cannot be classified as Case-assigners. This has already been stated in (10) above. When and while in (8c, d) and after, before, since, and until as in (6b, d, f, h) are specified as transitive, wow-Case-assigning prepositions. As neither after, before, since, and until in the relevant contexts nor when and while in general assign Case to their structural complements, it cannot be argued that they are associated with a morphosyntactic case feature. If there is no such Case feature, then no Case-related functional projection can be triggered. In other words, there is no need for a functional projection dominating PP that would be relevant with respect to the internal argument of P. Thus, the structure for temporal prepositions taking sentential internal arguments (which I take to be subject agreement phrases, i.e. AgrsPs and not CPs8) is given in (13): (13)

PP spec

PP Ρ

AgrsP

In (12), we have seen that nominal internal arguments - at some time in the derivation move into the specifier position of the functional projection P-AgroP to have their objective case features checked. It cannot, however, be argued that sentential internal arguments undergo this kind of movement. This well-known contrast between nominal and sentential internal arguments can be observed in subordinate clauses in so-called SOV languages such as German and Dutch. In these languages, nominal internal arguments and sentential ones clearly assume different surface positions. As can be seen in (14a), nominal internal arguments precede the verb, and sentential internal arguments, as in (14b), follow it; as can be seen from the ungrammatical string in (14c), the sentential internal argument does not precede the verb. (14)

a. ... dai Jan zijn rijbewijs haalt ... that John his driver's license gets b. ... dat Jan wil [dat hij zijn rijbewijs haalt] ... that John wants that he his driver's license gets c. *... dat Jan [dat hij zijn rijbewijs haalt] wil ... that John that he his driver's license gets wants

(examples from Zwart 1994: 4f.)

Zwart (1994), following Kayne (1994), argues that the SOV word order in (14a) is derivative: in the non-derived structure, the nominal internal argument follows the lexical verb, that is to say that (14a) and (14b) have the same underlying structure. The surface asymmetry is related to the fact that nominal constituents, unlike sentential ones, must bear Case. In order to have its Case checked, the nominal constituent must move to specAgroP as 8

Cf. Haumann (1997: 212ff.) for the categorial status of the sentential complement of temporal prepositions.

31

The Projections of Temporal Prepositions

indicated in (15). As sentential elements are not associated with Case features, they must not move to specAgr0P, as can be seen from the ungrammatical string in (14c). (15)

a. b.

zijn rijbewijSi dat hij zijn rijbewijs haalt

The main point I wish to make with respect to temporal prepositions taking sentential arguments is that these elements are not associated with a Case feature, and therefore cannot be dominated by Case-related functional projection. Thus, the two types of transitive temporal prepositions differ in structure. Only those elements that are specified as Case-assigners, as in (9) above, trigger the presence of the functional projection P-AgrJP. Since non-Case-assigning prepositions, as in (10), are not associated with Case, they cannot license a dominating functional projection. Intransitive prepositions, to which I turn now, have an argument structure that lacks an internal argument. And, clearly, as there is no internal argument, there is no structural position Case could be assigned to.

2.3. Intransitive temporal prepositions The argument structure of intransitive prepositions such as before and since in (5e, f), is given in (16). As can be seen from the representation, there is an external and a referential argument but no internal argument and hence no potential bearer of objective Case: (16) < 1, t > ( = before and since in (5e, f))

Alternatively, it could be argued that the intransitive prepositions are in fact transitive with the internal argument realized as little pro, i.e. as a non-overt nominal element, as indicated in (17). Under such an analysis, the internal argument position occupied by pro would need to be Case marked because of the nominal nature of pro. This kind of analysis has been proposed by Koopman (1993), Sportiche (1990), and Rooryck (1993). (17) < 1, pro, t > Since the argument structure in (16) does not contain an internal argument, it cannot be specified for Case; hence there is no Case-related functional projection dominating PP. The structure headed by an intransitive temporal preposition is shown in (18):

Dagmar Haumann

32

(18)

PP spec

PP

If we assume that intransitive temporal prepositions are associated with an argument structure, as in(17), then the structure of intransitive prepositions would be identical to that of transitive temporal prepositions taking nominal internal arguments, as in (12) above and as repeated in (19). Here, we would have to assume that pro, presumably prior to LF, moves to the specifier of P-Agr()P to have its Case checked. As there is no independent or empirical evidence for this structure, I propose subscribing to the simpler one shown in (18).9 (19)

P-AgroP spec

pre-LF:

proj

To conclude this section, I want to briefly summarize the main points. Temporal prepositions are either transitive or intransitive. If they are transitive, their argument structure contains an internal argument. If they are intransitive, then there is no internal argument. Transitive temporal prepositions come in two flavors, namely as Case-assigners and nonCase-assigners, with only the former taking a nominal internal argument and thus being dominated by the functional projection P-AgroP. The three types of temporal prepositions, together with their respective extended projections, are summarized in (20):10

9

Although the pro-analysis of "intransitive" Ps would allow for a reduction of three to t w o types o f Ps, namely Case-assigning and non-Case-assigning ones, it is far from clear h o w pro is prevented from functioning as the internal argument of, for example, during, ago, and until: (i)

W e talked about this during They met pro ago.

pro.

W e couldn't get hold of him until 10

pro.

Rauh (1998: 215ff.), in her discussion o f temporal prepositions, identifies four possible structures o f temporal PPs. The structures she gives for intransitive temporal prepositions and transitive ones with a nominal or a sentential argument are identical to those given in (20) above. In addition to these, she discusses transitive temporal prepositions that take a prepositional internal argument such as since in (i): (i)

He had known her since before

the war

(Rauh 1998: 215).

The Projections of Temporal (20)

33

Prepositions

a. transitive with nominal internal argument I didn't see him before the conference. The dizzy turns started after the conference. I haven't seen him since his divorce. You will have to wait until December. I fell asleep during the movie. They met three weeks ago.

{< 1, 2, t >, [+CAS]} P-Agr„P spec

P-Agro' P-AgTo

PP

spec

P' DP

b. transitive with sentential internal argument , [-CAS]}

spec

PP Agr s P

I didn't see him before the conference started. The dizzy turns started after the conference was over. I haven't seen him since he got divorced. You will have to wait until Santa Clause comes to us. I was delighted when the party was over. I slept while dinner was being prepared.

c. intransitive I haven't seen him before. I haven't seen him since. spec

PP I ρ

3. T h e external a r g u m e n t

In m y analysis, I follow Williams (1994) and others in assuming that the argument structure o f prepositions contains an external argument by virtue o f which they function as one-place predicates." That is to say that temporal prepositions locate entities, usually events, in time.

11

(...continued) The type of temporal preposition that takes a prepositional internal argument, i.e. some extended projection of P, is roughly identical to that given in (20b) above. Rauh (1998: 216) points out that prepositional and sentential internal arguments are idiosyncratic choices of governing Ps, i.e. their categorial format is not sanctioned by a morphosyntactic feature such as [+CAS] in connection with nominal internal arguments, but by being categorially selected and strictly subcategorized by the governing head. See also Higginbotham (1985, 1987), Bierwisch (1988), Haumann (1990, 1992, 1993, 1997), Zwarts (1992), Wunderlich (1992), Rauh (1993, 1997a, 1997b, 1998, this volume), Zimmermann (1993), and Steinitz (1996, 1997).

Dagmar

34

Haumann

Syntactically, these one-place predicates are typically conceived as adjuncts in the extended verbal projection. 1 2 A s is s h o w n in (21), PPs are adjoined to TP, the projection o f tense, and/or to VP. 1 3 (21)

A s has already been mentioned, the projections o f temporal prepositions are conceived semantically as one-place predicates because temporal prepositions are associated with argument structures that contain an external argument. A s is indicated in (22), the external argument o f the temporal preposition before must be coindexed with s o m e element outside the minimal phrase that contains Ρ and its internal argument (if there is one): (22)

12

13

X,

[ before

[noon/the conference started]]

It should be noted that there is an alternative standard analysis of adjuncts, the so-called "specifier analysis of adjuncts." According to this analysis, with Kayne (1994) and Cinque (1999) as its main proponents, adjuncts, i.e. the XPs in (i) below, are conceived as specifiers within functional projections (FP): (i) [ F P n X P [ F n [FJ [... [ F P 2 X P [F'2 M [FPI XP [F I [F,] [YP]]]]]]] Under this analysis, classical adjuncts (cf. Kayne (1994: lOOf.,137 note 31) for adnominal APs, and Cinque (1994, 1999) and Alexiadou (1997) for AdvPs) are reanalyzed as unique specifiers within functional projections. In these configurations, both the functional head and the AdvP that occupies its specifier positions can be licensed under specifier-head agreement, provided these heads and specifiers have features to share. In Haumann (1997, 1999) I have argued that this specifier analysis is not available for adverbial PPs, such as the projections of the temporal prepositions under consideration, since these are oneplace predicates and as such must have subjects in order to be interpretable. Cf. Chomsky (1986: 3If.) and Manzini (1992: 51) for adjunction to TP, and Cinque (1990: 27) for adjunction to VP. See also Rizzi (1990: 50) for adjunction to TP and/or VP within the framework of Relativized Minimality, where the licensing of PP is brought about under head-government in conjunction with Relativized Minimality, as defined in (i) and (ii), respectively: (i) X head-governs Y iff a. X e {Α, Ν, Ρ, V, Agr, T} b. X m-commands Y c. no barrier intervenes d. Relativized Minimality is respected (cf. Rizzi 1990: 6) (ii)X [head]-governs Y only if there is no Ζ such that a. Ζ is a typical [head]-govemor for Y b. Ζ c-commands Y and does not c-command X. (cf. Rizzi 1990: 7); [DH])

The Projections

of Temporal

35

Prepositions

Higginbotham (1985, 1987), Bierwisch (1988), Haumann (1990, 1992, 1993), Zwarts (1992), Wunderlich (1991), Rauh (1993, 1997b, 1998), Zimmermann (1993), and Steinitz (1996, 1997) assume, among others, that these modifying phrases, such as the temporal PP in (22), are licensed by the mechanism described in (23): (23)

A lexical head L is modified by a phrase FP iff: 14 (i) L governs FP, and (ii) the external argument of the semantic head of FP is coindexed with the referential argument of L. (adapted from Zwarts 1992: 63)

According to (23), the external argument of a semantic head, such as the temporal preposition before in (24), must be coindexed with the referential event argument of the verb, which is encoded as "e": 15 (24)

The dizzy turn started before noon/the conference started VP VP spec

V'

spec

I ν started

The advantage of this analysis (as opposed to those sketched in notes 12 and 13) lies in the integration of thematic properties as well as the relations that exist between argument structures. This analysis, however, fails to syntactically reflect the subject predicate relation that exists between the event denoted by the main clause and the projection of the temporal preposition, the semantic predicate. In the following I shall present an analysis of temporal prepositions that syntactically reflects the predicatehood of PP.16 That is to say, I propose extending the internal subject hypothesis17 to PPs, as indicated in (25): (25)

PP subject( Ρ

14

15

16 17

DP/Agr s P

This is a modified version of Higginbotham's (1985: 563ff.) original definition of "Thetaidentification." See, among others, Higginbotham (1983: 397ff., 416ff.; 1985: 556, 560ff.), Pollock (1989: 392), Zwarts (1992: 58ff.), Steinitz (1992: 38), Wunderlich (1992: 1), and Bierwisch (1988: 42ff.). I assume that this analysis carries over to spatial prepositions without modification. The internal subject hypothesis was originally proposed for VP (cf. Koopman & Sportiche (1985, 1991), Kuroda (1988), Burton & Grimshaw (1992), and McNally (1992)).

36

Dagmar

Haumann

In (25), both the internal and the external arguments are structurally realized within the domain of the head. The structure in (25) corresponds to the structure that, since Stowell's (1981, 1983) original proposal, has widely been assumed for small clauses as in (26): (26) XP

X'

DPi Subj

X

a. b. c. d.

She considered We wanted We considered Consider

[them [him [them [me

liars. out of the car. nice. gone!

]

nominal small clause prepositional small clause adjectival small clause verbal/participial small clause

The generalized internal subject hypothesis, i.e. a small clauses analysis across categories, makes it possible to treat semantic predicates, such as the projections of temporal prepositions, on a par with syntactic predicates, i.e. elements that take a structural subject. One of the consequences of this analysis concerns licensing: it is no longer the temporal "PP-adjunct," i.e. the one place-predicate, that has to be licensed in a complex structure but its subject, i.e. the entity of which it is predicated. 18 In constructions such as (25) and (26), both thematic arguments, i.e. the internal and the external argument, are structurally realized and licensed within PP. As temporal prepositions are associated with at most two arguments, only two structural positions within PP can be licensed, i.e. the complement position and the specifier position: (27)

PP specifier

P' complement

Modifying elements such as two hours or long cannot be realized and licensed within PP, as the structural positions within PP, as shown in (27), are reserved for the external and the internal argument. The following section focuses on the distribution and licensing of these modifying elements within the extended projection of Ρ and on the remaining ingredient of argument structure, the referential argument.

18

Under the analysis presented, the analyses sketched in Notes 12 and 13 can be discarded.

The Projections of Temporal

37

Prepositions

4. Non-arguments within the extended projections of temporal prepositions

As has already been pointed out in Section 2, the referential argument of a temporal preposition encodes temporal semantics and allows for modifying elements such as right, two hours, long, etc. as in (28) to occur in the projections of temporal prepositions. (28)

a. The dizzy turns started right after dinner. b. The dizzy turns started two hours after he had been administered the poison. c. The dizzy turns started long before noon.

Right, two hours and long are clearly not arguments, i.e. they are hosted outside PP, but as is also argued by Koopman (1993) - within the extended projection of P. 19 I assume that PP and P-Arg()P - if the latter is required for Case-related reasons - are dominated by a special degree phrase as in (29): (29)

P-DegP spec

P-Deg' P-Deg

PP

(or P-Agr0P)

Furthermore, I assume that modifiers such as right, two hours, long, etc. are base-generated specifiers in a Kayne- and Cinque-style analysis. 20 That is to say that they are analyzed as base-generated specifiers of P-DegP as indicated in (30): (30)

P-DegP spec

P-Deg'

P-Deg

P-Agr 0 P spec

if and only if Ρ takes nominal internal argument (DP), as in (b) P-Agr0'

P-Agr 0

PP spec

a.

b.

right two hours

P' Ρ

AgrsP/DP

when after

Tom came in the conference

These elements enter a checking relation with the head of P-DegP. This checking relation is based on semantic features, not on morphosyntactic features as has been the case with Case-checking in P-Agr 0 P.

19

20

Modifiers such as right, two hours and long cannot occupy the complement or the specifier position within PP (cf. (27)). Cf. note (12) for discussion. But see Koopman (1993), who treats right, as opposed to modifiers such as two hours, as the head, not as the specifier of a degree phrase.

Dagmar Haumann

38

The set of semantic features relevant to specifier-head agreement within P-DegP derive f r o m the temporal relations individual temporal prepositions specify. 2 1 These relations are relations of simultaneity [±simult], posteriority [+post] and anteriority [-post]. Furthermore, temporal prepositions must be differentiated with respect to the relation they enter with the matrix clause: temporal prepositional projections are terminative with respect to the duration of the matrix event {since, until, and while),22 or they are not {after, before, and when). The (non-)terminative function of temporal prepositional projections can be encoded as [±perf(ective)]. 2 3 On the basis of the features proposed, temporal prepositions are characterized as in (31) and (32): (31) a. b. c. (32) a. b. c.

after before when since until while





{[-simult], {[-simult], {[+simult], {[-simult], {[-simult], {[+simult],

[-post], [-perf]} [+post], [-perf]} [-perf]} [-post], [+perf]} [+post], [+perf]} [+perf]}

From the set of the identified features, [±perf] and [±simult] can be observed to play a crucial role with respect to the admissibility of modifiers. As can be seen f r o m the data in (33) and (34), only after, before, and when, i.e. temporal prepositions that are specified as non-perfective (cf. 31), allow modifying measure phrases. Since, until, and while, all of which are specified as perfective (cf. 32), cannot take modifying measure phrases at all. 2 4 · 2 5 (33) a. The dizzy turns started right!long!two hours before noon. b. The dizzy turns started right!long!two hours after the conference started. c. He kissed her right! long! two hours when Tom came in.

21 22

23

24

25

Cf. Haumann (1997: 218ff.) for a detailed discussion. Since, until and while delimit the duration of the matrix event by either specifying its endpoint (since) or its beginning (until), or by specifying it as temporally congruent (while). I am endebted to Thilo Tappe (p.c.) for suggesting [±perf(ective)] as a handy feature capturing (non-)terminative function of temporal prepositional projections. The example in (34c) is adapted from Emonds (1976: 174). It is problematic for the analysis presented: while, although specified as a perfective temporal preposition, may support right (at least for some native speakers). Haumann (1997: 225f.), based on a suggestion made by Friederike Pfingsten (p.c.), proposes to assume two lexical variants of while, one of which is [+perf] and thus does not allow right. The second variant of while, which is specified as [-perf], supports right. This second variant is interpreted as being synonymous with when: (i) while < 1, 2, t > {[+simult], [+perf]} -> no modification due to [+perf]-specification; (ii) while < 1, 2, t > {[+simult], [-perf]} -> modification by right due to {[+simult], [-perf]} -specification The fact that since, until and while do not take modifiers, however, should not lead to the assumption that the argument structures of these elements lack a referential argument (cf. Haumann 1997, 220ff.). Rauh (1998: 223) conjectures that the non-admissibility of modifiers with since and until could be due to their lacking a referential argument or to specific semantic properties of temporal prepositions, e.g. reference to one-dimensional space.

The Projections of Temporal

39

Prepositions *

*

*

(34) a. I haven't talked to him ri^htl long/ two hours since his divorce. b. I couldn't get hold of hrn right/ long/ two hours until the conference was over. c. He kissed her bright/*long/ two hours while her boyfriend was looking. As can be seen from the contrast between (33a, b) and (33c), only after and before allow the full range of modifying elements, i.e. right-like elements and measure phrases. With when, the only admissible modifier is right. This difference can be attributed to the fact that when, as opposed to after and before, is a preposition of simultaneity, i.e. the events related by when have to be temporally congruent. Thus - though quite redundantly - the only admissible modifier for when is right. The advantage of analyzing right, two hours, long, etc. as specifiers within a special functional projection, as indicated in (30) above, lies in the unique relations, i.e. relations of specifier-head agreement that are established between the elements that are hosted in specP-DegP and the semantic features of temporal prepositions (cf. (31) and (32)) hosted in the head ofP-DegP. 2 6 At the same time, however, the analysis of right, long and two hours as specP-DegPelements gives rise to a linearization problem.

5. Linearization

Under the analysis presented, modifying phrases that are base-generated in specP-DegP, e.g. right, two hours, long, etc., precede the matrix clause which is the structural subject of PP, i.e. it is located in specPP. 27 The linearization problem that arises under the present analysis is illustrated by the examples given in (35): (35) a. * Right/long/two hours [PP the dizzy turns started before noon], b. * Right/long/two hours [PP the dizzy turns started after the conference started]. c. * Right [pp he kissed her when Tom came in]. Comparing the ungrammatical strings in (35) with the grammatical ones in (33), we are led to assume that the subject of Ρ leaves specPP and moves overtly into the specifier position of some higher functional projection, which I refer to as ΔΡ, as indicated in (36). While the ultimate reason for the movement of a nominal constituent, namely Case-requirement, is perfectly clear, there is no such obvious reason why sentential elements as in (36) should move at all.

26

27

The semantic features of temporal prepositions that are listed in (31) and (32) above are represented twice: on the individual temporal prepositions and at their checking site, the head of PDegP. That is to say that, at some point in the derivation, Ρ must adjoin to P-Deg to check its semantic features. Cf. (25) above.

40

Dagmar Haumann (36)

a. the dizzy turns started ¡ b. the dizzy turns started ¡ c. he kissed her ,

long two hours right

[ DP noon] Ugrsp the conference started] Ugrsp Tom came in]

As the nominal subjects of PP in (37), i.e. them and that party,29 are assigned objective case by the governing verb want in (37a) and the governing preposition with in (37b), they have to move into appropriate specifier positions for reasons of feature checking. (37)

a. I want [them]¡ [p.Degp right t¡ on the table]. b. With [that party]¡ [ P . DcgP three years t¡ ago], they will not throw another one within the next couple of years.

The nominal constituent them, as it is governed by the verb want, moves into the specifier of Agr 0 P within the extended projection of want, as is illustrated in (38a), 30 and the nominal constituent the party, as it is governed by the preposition with, moves into the specifier of P-Agr 0 P, as is shown in (38b). The representations given in (38) are representations of the LFs of the strings under consideration. Sentential constituents that function as internal arguments, such as dat hij zijn rijbewijs haalt 'that he his drivers license gets' in (14b) above, were argued not to undergo any syntactic movement at all. As they do not bear Case, they do not have to meet Case-requirements. 31 The fact that the sentential subject of Ρ has to leave its base position, as shown in (36), cannot be derived from the morphosyntactic requirement that Case be checked.

28

29

30

31

N.B.: If P, like before, takes a nominal internal argument such as noon, P-Agr 0 P (cf. (11), (20a) above) is projected, i.e. the extended projection of Ρ contains the functional projection P-Agr 0 P which sandwiches between PP and P-DegP, as indicated by the bold face dots in (36). Nominal subjects, like the nominal internal argument in (12) and (14a) above, have Case-features to check. Structure (38a) corresponds to the analysis of Exceptional Case marking constructions and predicative adjectives sketched in Chomsky (1993: 6ff.). Above it was argued that Agr0P and P-Agr0P are projected if and only if V or Ρ assign Case.

41

The Projections of Temporal Prepositions (38)

\

Agr0P/P-AgroP spec

Agr0'/P-Agro' Agro/P-Agro

VP/PP

spec

V'/P' V/P

spec

Ρ a. themk b. that partyk

t

want] witht

ti t|

right

oHj

the table¡

tj t^

tj

agOj

three yearS[ tj t^

tj

Similarly, as can be seen in (39), the sentential subjects of verbs may not stay in spec VP but must move overtly to some higher specifier position which - in the case of verbal predicates - must be one that precedes the modal will. Again, movement of the sentential subjects (as opposed to nominal subjects) is not triggered by Case-requirements: (39) a. *Will [to have not been invited to the party] upset Bill. b. Will [that Jill has not been invited to the party] upset Bill. c. [To have not been invited to the party]¡ will t¡ upset Bill. d. [That Jill has not been invited to the party]¡ will t¡ upset Bill. The fact that sentential subjects cannot stay in their base positions, e.g. specPP (35) or spec VP (39a, b), but must move to some higher specifier position, as illustrated in (36) and (39c, d), seems a bit of a puzzle. The solution of this puzzle is based on the assumption that sentence type information must be visible on the highest sentential level for both interpretation and selection. 33 32

33

In Haumann (1997: 228, note 70) it is assumed that the extended projection of Ρ contains a subject agreement phrase, P-AgrsP, if the subject of PP is nominal. This additional phrase is assumed to be sandwiched between the lexical head governing the extended projection of Ρ and P-DegP. SpecPAgrsP is the pre-LF host of the nominal external argument of PP. Cf. Grimshaw (1979) for arguments asserting that the selection of a complement makes reference not only to its syntactic format, but also to semantic type, e.g. (non-)interrogative.

42

Dagmar

Haumann

The sentential external argument of a temporal preposition can, for example, be realized as an interrogative clause, i.e. CP, as shown in (40): (40)

a. Who did he call right before the party started? b. Why won't you have your hair cut before your birthday?

Complex sentences, such as those in (40), in turn, can function as internal arguments of lexical heads, such as ask and wonder in (41), which select interrogative complements: (41)

(42)

a. She asked who he called right before the party started. b. I wonder why you won't have your hair cut before your birthday. V'

Ρ

a. asked [[who]j she called tj] b. wonder [[why]j you won't have your hair cut tj]

right

t¡

before

[Agrsp the party started]

t¡

before

[DP your birthday]

Selection of sentential complements, as is well-known, is closely tied to sentence type information.35 Under the analysis presented, the category of complex sentences in (41) is ΔΡ as in (36) above, with the specifier position of ΔΡ hosting the matrix clause, the CP that is marked as [+wh]. As is indicated by the arrows in (42), CP[+wh] enters a specifier-head agreement relation with the head of ΔΡ. 36 · 37 The ultimate reason for the movement of CP[+wh] into specAP lies in the checking of sentence type features: as ΔΡ is sentential, its head must be specified for sentence type infor34

35 36 37

If the internal argument of Ρ is nominal, such as your birthday in (42b), the extended projection Ρ contains P-Agr 0 P (cf. notes (28), (31)). Cf. note (33). I should like to express my thanks to Gisa Rauh for reminding me of feature checking. In (42), the index [+wh] appears on the governing verb to indicate that this verb selects an interrogative complement; namely ΔΡ. The selection relation that exists between V and ΔΡ is indicated by the dotted arrow. ΔΡ is an interrogative constituent, i.e. AP[+whJ, by virtue of its specifier hosting an interrogative CP, i.e. CP[+Whj , and by virtue of its head, Δ, agreeing with its specifier. Since specCP hosts a [+wh]-constituent and since CP and C agree in the feature [+wh], CP is an interrogative constituent.

The Projections

of Temporal

Prepositions

43

mation. Given the configuration in (42), sentence type information is available exactly where it is needed, namely within the highest extended projection of P, i.e. in ΔΡ.

6. Summary

Temporal prepositions were argued to be dominated by at most three layers of functional structure: P-Agr 0 P, P-DegP, and ΔΡ. Whereas P-AgroP is present if and only if a temporal preposition qualifies as a structural Case-assigner, such as during and ago in (7) or after, before, since, until in (6a, c, e, g), P-DegP and ΔΡ, by contrast, figure within the projection of every preposition. The function of P-DegP within the extended projection of Ρ was argued to be essential for the licensing of measure and degree elements which are conceived as base-generated specifiers within P-DegP (cf. (29) and (30) above). And finally, ΔΡ has been argued to be a functional projection dominating PP and P-DegP with specAP being the landing site of the external argument originating in specPP (cf. 36). Evidence for this projection derives from both linearization facts and from the requirement that sentence type information be available within the highest functional projection (cf. (41) and (42) above).

References

Alexiadou, Artemis (1997): Adverb Placement: A Case Study in Antisymmetric Syntax. - Amsterdam: John Benjamins. Burton, Strang & Jane Grimshaw (1992): Coordination and VP-internal Subject. - In: Linguistic Inquiry 2 3 , 3 0 5 - 3 1 3 . Chomsky, Noam (1986): Barriers. - Cambridge, MA: The MIT Press. - (1993): A Minimalist Program for Linguistic Theory. - In: K. Hale & S. Keyser (eds.): The View from Building 20: Essays in Honor of Sylvain Bromberger, 1-52. Cambridge, MA: The MIT Press. - (1995): The Minimalist Program. - Cambridge, MA: The MIT Press. Cinque, Guglielmo (1990): Types of A '-Dependencies. - Cambridge, MA: The MIT Press. - (1994): Functional Structure in the Romance Sentence and DP. - Course material presented at Girona International Summer School in Linguistics, 1994. - (1999): Adverbs and Functional Heads: A Cross-Linguistic Perspective. - New York, NY: Oxford University Press. Emonds, Joseph (1976): A Transformational Approach to English Syntax: Root, Structure Preserving and Local Transformations. - New York, NY: Academic Press. Grimshaw, Jane (1979): Complement Selection and the Lexicon. - In: Linguistic Inquiry 10, 2 7 9 326. - ( 1991 ): Extended Projection. - Ms. Waltham, MA: Brandeis University.

44

Dagmar

Haumann

Haider, Hubert (1988): Matching Projections. - In: A. Cardinaletti, G. Cinque & G. Giusti (eds.): Constituent Structure: Papers from the 1987 GLOW Conference, 101-21. Dordrecht: Foris Publications. - ( 1993): Deutsche Syntax - Generativ. - Tübingen: Narr. Haumann, Dagmar (1990): Θ-Rollen Englischer Lokalpräpositionen: Syntaktische und Semantische Aspekte. - Unpublished M.A. thesis. Wuppertal: Bergische Universität-GH Wuppertal. - (1993): Referential Arguments. - Wuppertal: Bergische Universität-GH Wuppertal (= Arbeiten des Sonderforschungsbereichs 282. Theorie des Lexikons 44). - (1997): The Syntax of Subordination. - Tübingen: Niemeyer (= Linguistische Arbeiten 373). - (1999): Adverbial Subordination: Main Clauses as Specifiers. - In: F. W. Neumann & S. Schülting (eds.): Anglistentag: Proceedings of the Conference of the German Association of University Teachers of English. Vol. XIX, 77-97. Trier: Wissenschaftlicher Verlag Trier. Higginbotham, James (1985): On Semantics. - In: Linguistic Inquiry 16, 547-93. - ( 1987): Elucidations of Meaning. - Cambridge, MA: MIT (Lexicon Project Working Paper 19). Jackendoff, Ray (1973): The Base Rules for Prepositional Phrases. - In: S. Anderson & P. Kiparsky (eds.): A Festschrift for Morris Halle, 345-56. New York, NY: Holt, Rinehart & Winston. - (1977): X-Syntax: A Study of Phrase Structure. - Cambridge, MA: The MIT Press. Jespersen, Otto (1909-1949): A Modern English Grammar on Historical Principles: In Seven Parts. London: George Allen & Unwin. - ( 1924/1992): The Philosophy of Grammar. - Chicago, IL: The University of Chicago Press. Kayne, Richard (1994): The Antisymmetry of Syntax. Cambridge, MA: The MIT Press. Koopman, Hilda (1993): The Structure of Dutch PPs. - Ms. University of California at Los Angeles. Koopman, Hilda & Dominique Sportiche (1985): Θ-Theory and Extraction. - In: GLOW Newsletter 14, 57-58. - (1991): The Position of Subjects. - In: Lingua 85,211-258. Kuroda, Shige-Yuki (1988): Whether We Agree or Not: A Comparative Syntax of English and Japanese. - In: W. Poser (ed.): Japanese Syntax, 103-43. Stanford: Center for the Study of Language and Information. Manzini, Rita (1992): Locality: A Theory and Some of Its Empirical Consequences. - Cambridge, MA: The MIT Press. McNally, Louise (1992): VP Coordination and the VP-internal Subject Hypothesis. - In: Linguistic Inquiry 23, 336-341. Rauh, Gisa (1993): Grammatische Kategorien. - Wuppertal: Bergische Universität-GH Wuppertal (= Arbeiten des Sonderforschungsbereichs 282. Theorie des Lexikons 39). - (1997a): Englische Präpositionen zwischen lexikalischen und funktionalen Kategorien. - In: Löbel, E. & G. Rauh (eds.): Lexikalische Kategorien und Merkmale, 125-167. Tübingen: Niemeyer (= Linguistische Arbeiten 366). - (1997b): Lokale Präpositionen und referentielle Argumente. - In: Linguistische Berichte 171, 415— 442. - (1998): Zur Struktur von Präpositionalphrasen im Englischen. - In: Zeitschrift für Sprachwissenschaft 15, 178-230. - (this volume): Prepositions, Features, and Projections. Rizzi, Luigi (1990): Relativized Minimality. - Cambridge, MA: The MIT Press. Rooryck, Johan (1993): Prepositions and Minimalist Case-Marking. - Ms. Leiden University. Sportiche, Dominique (1990): Movement, Agreement and Case. - Ms. University of California at Los Angeles. Steinitz, Renate (1996): Valenznotwendige Präpositionalphrasen: Weder Argumentnoch Adjunktposition. - Ms. Berlin: Zentrum für Allgemeine Sprachwissenschaft (ZAS).

The Projections of Temporal -

Prepositions

45

(1997): Lexikalische Kategorisierung: Ein Vorschlag zur Revision. - In: E. Löbel, & G. Rauh (eds.): Lexikalische Kategorien und Merkmale, 1-26. Tübingen: Niemeyer (= Linguistische Arbeiten 366). Stowell, Timothy (1981): Origins of Phrase Structure. - Unpublished Ph.D. dissertation. Cambridge, MA: MIT. - (1983): Subjects across Categories. - In: The Linguistic Review 2, 285-312. Van Riemsdijk, Henk (1990): Functional Prepositions. - In: H. Pinkster & I. Genee (eds.): Unity in Diversity: Papers Presented to Simon C. Dik on his 50th Birthday, 229-241. Dordrecht: Foris Publications. Wege, Barbara (1997): Skalare Präpositionen - grammatikalisierte Varianten? - In: D. Haumann & S. Schierholz (eds.): Lexikalische und Grammatische Eigenschaften Präpositionaler Elemente, 1945. Tübingen: Niemeyer (= Linguistische Arbeiten 371). Williams, Edwin (1980): Predication. - In: Linguistic Inquiry 11, 203-238. - (1981): Argument Structure and Morphology. - In: The Linguistic Review 1,81-114. - (1994): Thematic Structure in Syntax. - Cambridge, MA: The MIT Press. Wunderlich, Dieter (1991): How Do Prepositional Phrases Fit into Compositional Syntax and Semantics? - In: Linguistics 29, 591-621. Zimmermann, Ilse (1993): Zur Syntax und Semantik der Satzeinbettung. - In: I. Rosengren (ed.): Satz und Illokution. Vol.2, 231 -251. Tübingen: Niemeyer (= Linguistische Arbeiten 279). Zwart, Jan-Wouter (1994): Dutch is Head Initial. - Ms. University of Groningen. Zwarts, Joost (1992): X-Syntax - X-Semantics. On the Interpretation of Functional and Lexical Heads. - Ph.D. dissertation. Utrecht University (published in the UiL OTS Dissertation Series). - (1997): Lexical and Functional Properties of Prepositions. - In: Haumann, D. & S. Schierholz (eds.): Lexikalische und Grammatische Eigenschaften Präpositionaler Elemente, 1-18. Tübingen: Niemeyer (= Linguistische Arbeiten 371).

Niina Ning Zhang

Movement within a Spatial Phrase

1. Introduction 1

This paper presents a unified syntactic account of the different morphological and syntactic representations of such spatial expressions as English from behind the table and Chinese cong zhuozi houmian 'from table behind.' Based on facts from various languages, I will arrive at the following theoretical conclusions: 1. A spatial expression can be semantically analyzed into three elements: LOCATIVE RELATION (LR), PLACE VALUE (PV), and REFERENCE ENTITY (RE). The morphological realization of each of these elements can be found in various languages. 2. A spatial expression has a split spatial phrase (SP) structure. The higher SP is headed by an LR element, and the lower SP is headed by a PV element, as shown in (1): (1)

3. The three elements undergo merging and moving operations in the computational system deriving various orders of a spatial expression across languages. 4. Each of the two SPs can be either prepositional, postpositional or nominal, even within a single language. In Section 2 , 1 will present my semantic decomposition analysis of spatial phrases. Section 3 will show how the syntactic analytic elements which encode the three semantic elements are ordered differently in various languages. In Section 4, then, I will put forward my

1

1 wish to thank Hubert Cuyckens, Dieter Gasde, Brigitta Haftka, Daniel Hole, Günter Radden, Tonjes Veenstra, Chris Wilder, members of Focus-Syntax project group of ZAS, and the participants of the Workshop on Prepositions at Hamburg University (June 26-27, 1998) for their constructive comments and discussion. I would also like to thank Päivi Koskinen and Chris Wilder for their patience to discuss data with me of their native languages (Finnish and English, respectively). All remaining errors are mine. The term spatial expression in this paper is not a technical term. It is used to refer to any linguistic element encoding spatial information, regardless of the syntactic category of the element in the language.

48

Ν Una Ning

Zhang

proposal for split spatial phrases and illustrate how the various orders listed in Section 3 are derived. I will then present four pieces of evidence for my proposal in Section 5. Finally, I will conclude my paper in Section 6 with a brief summary.

2. Semantic decomposition of a spatial phrase

2.1. Three semantic elements Generally speaking, a spatial expression can express the Position, the Goal, the Source, or the Route of an entity or a state or event. I will group these four notions under the general term "Locative Relations."2 For instance, the English on in (2a) denotes a Position relation, from in (2b) a Source relation, to in (2c) a Goal relation, and via in (2d) a Route relation. (2)

a. There is a book on the table. b. He came from Berlin. c. She went to Berlin. d. The train g o e s to Berlin via Hannover.

Thus, the first semantic category of a spatial expression is "Locative Relation" (LR), which is a generalization capturing the notions Position, Goal, Source, and Route. The relation between the abstract LR and the Goal, for example, is similar to that between the general category "Tense" and past tense. The four locative relations exclude each other. For instance, if the LR in (2a) is Position, it cannot be Goal or Source. Accordingly, one spatial expression in a specific clause can denote only one kind of concrete locative relation. Secondly, a spatial expression must contain a "Reference Entity" (RE). For example, in on the table, the table is the reference entity of the spatial expression. It contrasts with other entities such as a chair or a house. Finally, in many cases, a spatial expression encompasses a "Place Value" (PV). For example, on the table expresses the place value SURFACE with respect to the reference entity the table. Similarly, under the table lexicalizes the place value SPACE BELOW with respect to the table. Different place values contrast with each other, e.g. SURFACE is in contrast to SPACE BELOW. 3

2

Source, Goal, and Route can be grouped into a more general dynamic category Path, and as such contrast with the static category Position. Other terms for Position and Path used in the literature are: Locative and Directional (cf. Bennett 1975; Somers 1987), Place and Path (Jackendoff 1990), and non-motional and motional (Gruber 1976). For a finer semantic analysis o f these locative relations, see Svorou (1994) and Kracht (1998), among many others.

3

For a detailed Place Value analysis o f English locative prepositions, see Bennett (1975: 92) and Lindstromberg (1998), among many others. For an interesting discussion o f realization o f PV in various languages, see Heine, Claudi & Hünnemeyer (1991). The authors o f these studies use different terms for what I refer to in this paper as Place Value.

Movement within a Spatial

49

Phrase

2.2. Morphological realizations of the three semantic elements LR, PV, and RE The morphological realization of the three semantic elements LR, PV, and RE is different across languages and may even be different within one language.

2.2.1. Analytical forms denoting LR, PV, and RE PV and LR, as well as RE, are each encoded in a separate morpheme or word in Chinese, Japanese, Indonesian, Saramaccan, and in some English and German expressions. In (3), the noun xiangzi 'chest' is an RE element, the prepositions zai 'at,' cong 'from,' and wang 'to' denote the LRs of Position, Source, and Goal, respectively, and the post-nominal particles shang 'surface,' xia 'under,' and li 'in' express the PVs of SURFACE, LOWER, and INTERIOR, respectively. Similarly, in (4), the noun teeburu 'table' is an RE element, the postposition ni 'at, to' denotes the LR of Position or Goal, kara 'from' denotes the LR of Source, and the post-nominal ué 'surface' and sità 'under' express the PVs of SURFACE and LOWER, respectively. (3)

(4)

Chinese a. zai xiangzi shang at chest surface 'on the chest'

b. cong xiangzi xia from chest under 'from under the chest'

c. wang xiangzi li to chest in 'into the chest'

Japanese 4 a. teeburu-no ué ni table-GEN surface at 'on the table'

b. teeburu-no sità kara table-GEN under from 'from under the table'

c. teeburu-no ué ni table-GEN surface to 'onto the table'

In Indonesian, the LR Position is realized as the word di, the LR Source is realized as the word dari, and the LR Goal is realized as the word ke or kepada. Among the various PV elements, dalam encodes INSIDE, atas encodes SURFACE, bawah encodes UNDER, and depart encodes FRONT. Thus in (5a), the three words di 'at,' atas 'top,' and medja 'table' encode the LR Position, the PV SURFACE, and the RE table, respectively. In (5b), the three words ke 'to,' dalam 'inside,' and kelas-nja 'class-his' encode the LR Goal, the PV INTERIOR, and the RE his class, respectively. (5)

4

a. Djaket jang saja djahit untuk Hasan ter-letak di atas media. Coat COMP I sew for Hasan ACCID-lie at top table 'The coat that I sewed for Hasan is lying on the table.' (Chung 1976: 62) b. Setiap anak mem-bawa buku ke dalam kelas-nja. Each child TRANS-bring book to inside class-his 'Each child brought a book to his class.' (Chung 1976: 42)

The following abbreviations are used in the glosses in examples throughout this paper: ii 2nd person; 3SG 3rd person singular; 3PL 3rd person plural; ACC accusative; APPL applicative; ASP aspect; COMP complementizer; HON honorific; DAT dative; FM focus marker; GEN genitive; HAB habitual; IND indicative; MOD modifier; NOM nominative; PL plural; PRS present; Q question; RES resultative; SB subject.

50

ΝUna Ning Zhang

In English phrases such as into the lake, onto the table, and from behind the table, the forms to and from express the LR of Goal and Source respectively, while in in into, on in onto, and behind in from behind the table express the P V of INSIDE, SURFACE, and BACK 5 SPACE, respectively. Similarly, German von hinter dem Tisch 'from behind the table' also encodes LR, PV, and RE by separate forms: von 'from' is an LR-Source element, hinter 'behind' is a P V - S P A C E BEHIND element, and the nominal dem Tisch 'the:DAT table' is an RE element.

2.2.2. Portmanteau morphs denoting PV and LR PV and LR are expressed by a single portmanteau morph in Finnish and Chichewa. In Finnish, PV and LR are encoded by a case marker on nomináis or pre- or postpositions (Päivi Koskinen, p.c.). In Chichewa, PV and LR are encoded by a noun-class prefix (Bresnan & Kanerva 1989). The data in (6) are Finnish examples. The adessive case marker -Ila in (6a) encodes the LR of Position and the PV of OUTSIDE, while the illative case marker -on in (6b) encodes the LR of Goal and the PV of INSIDE. A locative preposition or postposition can also encode a more specific PV. For instance, in (6c), the adessive case marker -Ila indicates the LR of Position and the general PV of OUTSIDE, and the preposition lähe 'near' expresses the more specific feature of the PV, PROXIMITY. In (6d), however, the postposition -alia 'under' encodes three pieces of information: the LR of Position, the general PV of CLOSENESS, and the more specific feature of the PV, SPACE BELOW. 6 (6)

a. pöydä-llä

b. talo-on

table-ADESSiVE

house-iLLATiVE

'on the table'

'into the h o u s e '

c. lähe-llä

koti-a

d. sängy-n

alla

near-ADESSiVE home-PARHTivE

bed-GEN under-ADESSiVE

'near h o m e '

'under the b e d '

2.2.3. Zero form o f P V In Chinese, a proper noun RE, e.g. Zhongguo Yinhang 'China Bank,' or an RE comprising both a proper noun and a common noun, e.g. Beijing cheng 'Beijing city,' can take various PVs, as illustrated in (7). However, if the RE is a bare proper name denoting a Place, such as Bolin 'Berlin' in (8) and (9), the PV of INSIDE always takes a zero form in Chinese.7

5

6 7

The English PV-LR-RE order can be derived from incorporation. The order of elements of an incorporation word can be different from that of the corresponding non-incorporation form (Baker 1988). Thus from behind the table has the order LR-PV-RE, while onto the table has the order PV-LR -RE. See the Appendix for more Finnish data. In English, the PV of INSIDE takes the form of inside rather than in when the LR is Source: from inside/*in, regardless of whether the RE is a place name or not.

Movement

(7)

within a Spatial

a. zai {Zhongguo at

Yinhang/

China

b. cong

Bank

{Zhongguo

Bank

c. xiang (Zhongguo

(8)

China

a. zaiBolin At Berlin b. cong

Bolin

/ nanmian /

nearby / south

limian}

/ inside

nearby/ south

/Beijingcity

limian} /inside

/nanmian} /nanmian}

nearby/south /nanmian}

nearby/south (*li/*limian)

a', za hezi

inside

Bolin

from Berlin c. xiang Bolin to

{fujin

{fujin / nanmian /

{fujin

a. zaiBolin b. cong

cheng}

/ Beijing city

limian}

/ inside

Yinhang / Beijing cheng}

{fujin

Berlin

at Berlin

/ nanmian /

nearby/ south

nearby/south

c. xiang Bolin

(9)

(fujin

Bank

{fujin

from Berlin to

Beijing cheng} / Beijing city

Yinhang/Beijing

from China to

51

Phrase

Berlin

(*li/*limian)

b'. cong

inside (*li/*limian)

li/limian

at box inside hezi

li/limian

from box inside c'. xiang hezi

inside

to

li/limian

box inside

2.3. Non-lexicalized LR, PV, and RE An LR may not be lexicalized. In such a case, the exact LR is understood from the context. This is shown by the English prepositional phrase behind the curtain. The preposition behind expresses the PV SPACE BEHIND, but the related LR may vary. The examples in (10) show that behind the curtain can denote the LR of Route (10a), Goal (10b), and Position (lOc-d). Similarly, the preposition in expresses the PV INTERIOR consistently, but the related LR can be either Position, as shown by the sentence-initial in in (10e), or Goal, as shown by the other in in (10e). (10)

a. Trevor went behind the curtain to the table.

(Route)

b. Trevor went behind the curtain and stayed there.

(Goal)

c. Trevor trembled behind the curtain.

(Position)

d. The man behind the curtain cried.

(Position)

e. In the mall, he put a letter in the mailbox.

(Position, Goal)

In Section 5.4,1 will assume that in cases such as (10), the syntactic category for LR is not lexicalized, and thus the exact LR is expressed by the syntactic and semantic context. In particular, if these spatial phrases are integrated into an atelic predicate, as in (10c) or (lOd), or are external adverbials, as in (10e), they denote a Position LR; otherwise, they may denote a Non-Positional LR, such as Goal, Source, and Route, integrating with the verb meaning (cf. Talmy 1991). The linguistic context which helps to express the LR can also be morphological. While such German prepositions as von 'from,' zu 'to,' nach 'to,' gegen 'towards' are lexicalized LR markers, many other prepositions are not (some of them, for instance, are overt PV realizations; e.g. the preposition auf 'on' or the postposition gegenüber 'across from, opposite'). LRs which are not lexicalized are expressed by both the verb meaning and the case

Νiina Ning Zhang

52

marker of the RE nominal. In (11), for instance, the dative case of the determiner expresses the LR of Position, and the accusative case of the determiner expresses the LR of Goal. (11) German:

auf dem Tisch surface the:DAT table 'on the table'

auf

den

Tisch

surface the:ACC table 'onto the table'

Similarly, it is possible that a PV is not lexicalized. In such a case, no specific place value is implied. This is shown by the contrast in the following pair: (12) a. Trevor walked to the supermarket, but he didn't enter it. b. Trevor walked into the supermarket, but he didn't enter it. In (12a), there is no PV element in the first conjunct; as such, the negation of the entering event in the second conjunct, which implies the negation of the PV INTERIOR, does not cause a semantic conflict in the sentence. In contrast, in (10b), the morpheme in in the word into expresses the PV INTERIOR, which is in contradiction with the negation of the PV meaning of INTERIOR in the second conjunct. Thus the sentence is unacceptable. Sometimes, a non-lexicalized PV is understood from our world knowledge. In (13), the preposition at lexicalizes a LR. The PV of (13a) is understood as PROXIMITY while that of (13b) is understood as INTERIOR. (13) a. He sat at the table, b. He was at his office. The absence of RE is shown in the following Chinese example: (14) Tazai shangmian xie le zi. I at surface write ASP character Ί wrote characters on [a certain object].' The RE of the spatial phrase zai shangmian 'at surface' in (14) is implied in the discourse context, e.g. a piece of paper or a board. I will assume that the RE here is a context-bound pro-form.

2.4. LR forms with context-sensitive interpretations The English preposition to can express either Goal, as in (15a), where the PP is integrated with the motion verb ride, or Position, as in (15b), where the PP is integrated with the nominal the buildings:8 (15) a. Each morning, we rode our bike to school. b. The buildings to your left are over 400 years old. (examples from Cuyckens 1998)

8

The English prepositions to and from mean 'with respect to' rather than Goal or Source when the PP they head is an adjectival complement.

Movement within a Spatial Phrase

53

The English preposition at can express either Position, as in (16a), or Goal, as in (16b): (16) a. He was at his office. b. He {looked/shot/threw a stone/rushed} at her. Similarly, in Chichewa, noun class prefix 17 ku- can denote three LRs: Position, as in (17a), Goal, as in (17b), and Source, as in (17c): (17) a. Mú-ma-thamáng-á

ku-ti?

II:HON:SB-PRS:HAB-run-IND

17-Q

'Where do you usually run?' b. Mu-ma-thamang-ír-ά

ku-ti?

II:HON:SB-PRS:HAB-run-APPL-IND 1 7 - q

'Where do you usually run to?' c. Mu-ma-chokér-ά ku-ti? II:HON:SB-PRS:HAB-COme-IND

17-Q

'Where do you come from?' (examples from Bresnan & Kanerva 1989: 39)

3. Basic orders o f the three elements

Different languages may employ different basic orders of the analytic forms denoting LR, RE, and PV. (18) illustrates the basic order of the three elements in Indonesian, German, English, Chinese, and Japanese. (18) a. Indonesian German English

9

di atas medja at surface table 'on the table' von hinter dem Tisch from behind the:DAT table 'from behind the table'

Chinese9

zai zhuozi shang at table surface 'on the table'

Japanese

teeburu-no ué ni table-GEN surface at 'on the table'

The order of LR, RE, and PV is also found in Saramaccan (T. Veenstra, personal communication).

54

Ν Una Ning Zhang

4. The syntactic structure of spatial expressions

4.1. Proposal: Split spatial phrases An important assumption of this paper is that a phrase projection in all languages is in the order of Specifier-Head-Complement (Kayne 1994). For spatial expressions, I propose a structure which contains two layers of maximal projections: SpatialP-High (SPH) and SpatialP-Low (SPL). The former is headed by a syntactic element (word or morpheme) which denotes LR, while the latter is headed by a syntactic element which denotes PV. As shown in (1), repeated here as (19), the complement of SH is SPL, and the complement of SL is a nominal (DP, for instance) which denotes RE. I will discuss the syntactic categories of both SPs in Section 5.5. (19)

SL PV

DP RE

4.2. Deriving various orders Cross-linguistic order variations of spatial expressions can be derived by movement, as shown in (20). In (20a), the three elements, LR, PV, and RE, all remain in situ, and thus the order of LR-PV-RE is derived. This order is found in Indonesian, English expressions such as from behind the table, and German expressions such as von hinter dem Tisch 'from behind the table.' In (20b), the RE DP undergoes raising to the Spec position of SPL, thus giving the order LR-RE-PV. This order is found in Chinese and Saramaccan. In (20c), the RE DP first moves to the Spec position of SPL, as in (20b), and then the whole SPL undergoes raising to the Spec position of SPH- In this way, the order of RE-PV-LR is derived. This order is found in typical head-final languages such as Japanese. (20)

a.

SP H

LR-PV-RE Indonesian di atas média at surface table 'on the table' German: von hinter dem Tisch English: from behind the table

55

Movement within a Spatial Phrase LR-RE-PV

b.

Chinese

ζai zhuoz

shang

at table surface 'on the table'

SL PV

DP RE

5. Evidence for the proposal

5.1. Pure LR words as Xo elements There is no dispute over the prepositional status of the English words at, from, and to. Each of these LR prepositions heads a prepositional phrase (PP). In Chinese, the four pure LR words, zai 'at,' cong 'from,' wang 'to,' and xiang 'to,' take the word or phrase which encodes RE and PV as their complement. This can be seen from the fact that no focus marker can occur between LR and the RE-PV elements (cf. 21c), which follows from the general requirement that focus markers cannot occur between a head and its complement in Chinese (cf. 21a-b). (21) a. Tamai-le

(*shi) zheiben

he buy-ASP FM this 'He bought this book.'

b. Genju

(*shi)zheige

shu. book

yuanze

according:to FM this principle 'according to this principle'

c. Ta cong (*shi)jia

li lai.

he from FM home in come 'He came from home.'

56

ΝUna Ning

Zhang

My conclusion is that there exists a maximal projection headed by a pure LR word. In English, this projection is a PP; in Chinese, the words cong 'from' and wang 'to' are typical prepositions, while zai 'at' and xiang 'to' can be used as either verbs or prepositions. Regardless of their syntactic categories, these LR words can head a phrase.

5.2. Pure PV forms as X o elements In this paper, I have pointed out that all locative particles encode PV only in Chinese. As argued by Liu (1997), some locative particles are nouns and others are phrasal suffixes (enclitics). If a locative particle is a noun, the relevant maximal projection is an NP. Accordingly, the SPL is an NP. If a locative particle is a phrasal suffix, it can also head a phrase. Like the English possessive marker 's (cf. 22a), a phrasal suffix locative in Chinese can target a phrase as well as a word. In (22b), the locative li 'in' targets the whole DP naben Lao Hu xie de shu 'the book which was written by Lao Hu.' (22)

a. [the Queen of E n g l a n d ' s hat b. zai [[DPnaben [cp Lao Hu xie e¡ de] shu¡] li] at that Lao Hu write COMP book in 'in the book which was written by Lao Hu'

In recent literature (reviewed by Manzini 1998), a phrasal affix, or clitic, has been analyzed as a functional head. I thus conclude that there exists a second phrasal projection, which is headed by a pure PV element.

5.3. Head position parameters and the complement raising of SP Why can the complement of S L or SH undergo raising, as I proposed in (20b) and (20c), respectively? I will show that this kind of complement raising is required by the consistency of the complement position of the category in the given language. In English, PPs are head-initial. If the proposed two SPs are PPs, as in from behind the table, no complement-raising occurs (see 20a). In Japanese, PPs and NPs are both headfinal. If the proposed SPH is a PP and SPL is an NP, the complement of both heads needs raising (see 20c). German PPs can be either head initial, as shown in (23a) and (24a), or head-final, as shown in (23b) and (24b): (23)

a. gemäß seinem Wunsch according:to his wish b. den Stimmungen gemäß the moods according:to

(24)

a. wegen

schlechten

Wetters

due:to bad weather b. der großen Kälte wegen the extreme coldness due:to

Movement within a Spatial

57

Phrase

Accordingly, on the one hand, data like (20a) [PP von [PP hinter den Tisch]] show that both SPs are PPs. On the other hand, a complement of an S head raises and derives a head-final order, as shown below. (25)

a. gegenüber [der Post] opposite the:DAT post-office b. [dem Gast]¡ gegenüber t¡ the:DAT guest opposite

[head-initial] [head-final]

Chinese PPs and VPs are head-initial, while Chinese NPs are head-final (see, e.g., Li 1990: 6). A phrase headed by a pure LR element, such as zai 'at,' cong 'from,' wang 'to,' and xiang 'to,' is generally a PP, and it is always head-initial. On the other hand, in an expression consisting of an RE nominal and a locative particle, the locative particle is always at the end. In the following, I will show that such an expression is a nominal. If it is a nominal, the locative particle, which heads the whole nominal, is expected to occur at the end, since nomináis in Chinese are head-final. The form RE-PV in Chinese is indeed a nominal. Li (1990: 4) and McCawley (1992: 230) present some evidence to show that the form of RE-PV can appear in certain positions where NPs are allowed but PPs are excluded. For instance, the form of RE-PV in Chinese can occur to the right of the causative marker ba, which requires the element to its right to be a nominal. In contrast, a PP can never occur to the right of ba. In the following data, jia 'home' expresses RE, li 'in' expresses PV, and thus jia-li 'home-in' is a RE-PV form. The preposition zai 'at' expresses LR, and thus the PP zai jai-li 'at home in' is an LR-RE-PV form. The contrast in grammaticality between (26a) and (26b) suggests that, unlike the form zai jia-li (LR-RE-PV), the form jai-li (RE-PV) is a nominal. (26)

a. Haizi-men ba jia-li nong-de hen haokan. kid-PL Β a home-in make-RES very pretty ' T h e children made inside of the house pretty.' b. * Haizi-men ba zai jia-li nong-de hen haokan. kid-PL β a at home-in make-RES very pretty

In addition, the RE-PV form jia-li in (27a), but not the PP zai jia-li in (27b), can occur as the subject of a ¿>e/-passive: (27)

a. Jia-li bei haizi-men nong-de luanqibazao. home-in by kid-PL make-RES mess 'The inside of the house was made a mess by the children.' b. * Zai jia-li bei haizi-men nong-de luanqibazao. At home-in by kid-PL make-RES mess

The claim that RE-PV is a nominal is further supported by the following two pieces of evidence. First, in Chinese the complement of a preposition must be a nominal, and the form RE-PV in Chinese can be a complement of a preposition (not necessarily a locative preposition). A PP, by contrast, can never occur as a complement of a preposition. In (28), for instance, gei 'for' is a beneficiary preposition, which allows its complement to be the REPV form suo-li 'institute-in,' but not the PP zai suo-li 'at institute-in.'

58

Νlina Ning

(28)

a. Ta gei

suo-li

mai-le

Zhang

yitaidiannao.

he for institute-in buy-ASP one computer 'He has bought a computer for the institute.' b. * Ta gei

zaisuo-li

he for at

mai-le

institute-in

yitai

diannao.

buy-ASP one

computer

Second, if the verb kanqilai 'seem' is a raising verb in Chinese, as it is treated in the literature (Qu 1994), then the fact that the RE-PV form can occur as the subject of this verb shows that RE-PV can undergo NP-raising, like a regular nominal. The occurrence of a PP in the subject position of the verb kanqilai, however, makes the sentence unacceptable. The following data show that the RE-PV form jie-shang 'street-on' can be the subject of the raising verb kanqilai, while the PP zai jie-shang 'at street-on' cannot. (29)

a. Jie-shang street-on

kanqilai

hen

seem

very lively

renao.

'The street seems to be lively.' b. *Zai jie-shang at

street-on

kanqilai

hen

seem

very lively

renao.

The above evidence indicates that the RE-PV forms in Chinese are nomináis. In other words, the proposed SPL in Chinese is a nominal phrase. Since Chinese nomináis must be head-final, the raising of an RE element to the left of a PV element, which is assumed to be the head of SPL (cf. 20b), is expected. One might wonder what drives the complement of the head of an SP to undergo raising. The RE element raising data considered in this paper cover Japanese, Chinese, Finnish, and German (25b). In Japanese, a raised RE element has a genitive case marker no (see 4). If, in Finnish, a raised RE word is independent of a LR-PV form, in other words, if a LR-PV form is not a suffix of an RE element, a genitive case marker shows up on the RE form. The contrast between the presence of a genitive case marker with a raised RE element and the absence of a genitive case marker with an unraised RE element can be seen in (30). In (30a), the phrase-final RE koti 'home' does not have a genitive case marker; instead, it has a partitive case marker. In (30b) and (30c), however, the phrase-initial koti does take a genitive case marker. (30)

a. lähe-llä

koti-a

near-ADDESSiVE

home-PARTinvE

'near h o m e ' b. kodi-n

(LR = Position) lähe-lle

home-GEN near-ALLATiVE 'to near h o m e ' c. kodi-n

(LR = Goal)

lähe-ltä

home-GEN near-ABLATIVE 'from near h o m e '

(LR = Source)

In Chinese, PV forms can be either monosyllabic or disyllabic. In the latter case, de, which functions like a genitive case marker, can occur with a RE nominal. In (31), the PV element

Movement within a Spatial

is shangmian word zhuozi (31)

59

Phrase

'surface,' which is disyllabic, and the word de is optional between the RE 'table' and the P V element. 1 0 · 1 1

za hong zhuozi (de) shangmian at red table MOD surface 'on the red table'

Based on the above fact, I claim that complement raising in SPs m a y be genitive case related. 1 2

5.4. Semantic evidence A s w e have seen in Section 2.3, if a word which exclusively denotes an LR, such as from in English and cong 'from' in Chinese, occurs in a spatial expression, a specific LR is expressed; otherwise, no specific LR is implied. I repeat the example discussed in Section 2.3 as ( 3 2 ) here. (32)

a. Trevor went behind the curtain to the table. (Route) b. Trevor went behind the curtain and stayed there. (Goal) c. Trevor trembled behind the curtain. (Position)

In the three sentences in (32), there is no syntactic element to encode an LR exclusively. The spatial expression behind the curtain encodes P V and RE only, without an LR. In cases like these, the relevant LR is understood from the context. It can be a Route as in (32a), a Goal as in (32b), or a Position as in (32c). In the split spatial phrase structure proposed here, S H is the base-position for an element encoding LR, while S L is the base-position for an element encoding PV. The expression from behind the curtain has the structure o f

10

It has been suggested to me that the mere mention of de-insertion between RE and PV already justifies the analysis that a RE-PV form is a nominal in Chinese, presumably making all the other tests illustrated in (26-29) into strawman arguments. However, not all R E - P V forms can contain de, e.g. zhuozi (*de) shang 'table surface.' If an element cannot contain de, this does not mean that the element cannot be a nominal, so one still needs further tests to show the nominal properties of the R E - P V form. Thus de-insertion is not an ideal test for the nominal properties of an element. My own tests show that, in Chinese, RE-PV is always a nominal, regardless of whether it allows de-insertion or not.

11

I have no explanation for the lack of a genitive case marker for the complement of a German spatial postposition. If RE raising is a genitive case related movement, how does this analysis fit in with other possessive cases in Chinese. Does it follow that Akiu-de che 'Akiu's car' is derived from che Akiu-de 'car Akiu's'? In this paper, I restrict the discussion to inalianable possessive relations, leaving the study of expressions such as Akiu-de che 'Akiu's car' for future research. In fact, we can find possessor raising in Chinese independently of spatial expressions. For instance, in (i) below, the possessor Wangmian is raised out of a nominal and surfaces as the subject of the sentence, and no resumptive pronoun is allowed in its base position:

12

(i)

Wangmian si le (*ta/*tade) Wangmian die ASP (he/his) 'Wangmian's father died.'

fuqin. father

60

Niina Ning

Zhang

(33a). Without a context, behind the curtain has the structure of (33b), where SH is empty and the relevant LR is semantically unspecified. The two-layered structure thus captures the distinc-tive semantic features between LR expressions and other spatial expressions. In other words, the semantic differences between forms like behind the curtain and forms like from behind the curtain need to be represented syntactically. The split spatial phrase structure analysis gives the two types of expressions different syntactic representations, and thus covers their semantic contrast. (33) a.

SL PV behind

DP RE the curtain

SL PV behind

DP RE the curtain

b.

Summarizing, I have listed four pieces of evidence to argue for a syntactic structure which has two layers of maximal projections for a spatial expression. First, a pure LR element can head a maximal projection. Second, a pure PV element can also head a maximal projection. Third, RE element raising corresponds to head position parameters. Finally, the semantic differences between LR and PV expressions need to be represented syntactically.

5.5. The syntactic categories of SP If a spatial expression has two layers of maximal projection, as I have argued, a relevant issue to consider is what the categorial features are for each of the heads of the two SPs. If the two PPs of English phrases like from behind the table correspond to the two SPs, we can claim that both SPs in English can be PPs. I have argued in Section 5.3 that in Chinese, SPL is a nominal phrase. It seems that SPL can also be nominal in English. In (34), the SPL is the DP the inner part of the box or the bottom of the vase. (34) from/to [dp the inner part [pp of the box]] / [dp the bottom [pp of the vase]]

Movement within a Spatial

61

Phrase

In fact, not only SPL, but also SPH can be nominal. Koskinen (1998) argues for the nominal status of Finnish spatial expressions such as those underlined in (35). (35)

a. Sofia on talo-ssa. Sofia:NOM be:3sG house-INESSIVE 'Sophia is in the house' b. Sofia mene-e talo-on. Sofia:NOM go-3sG house-ILLATIVE 'Sofia goes into the house' c. Sofia tule-e talo-sta. Sofia:NOM come-3SG house-ELATLVE 'Sophia comes out of the house'

She argues that, unlike adverbs, these locative expressions move like DPs (into topic position, where only DPs go, and not adjectives, participles, verbs, manner adverbs, etc.), and that, like DPs, they bear case and nominal number marking. Thus, if spatial expressions like those in (35) are nomináis, we can assume that in additon to SPL, SP h can also be nominal. My conclusion is that both the higher and the lower heads of a spatial phrase can be either adpositional or nominal. Recall that English SPL can be either a nominal or a PP. These facts show that the categorial variations of a SP occur both in a single language and across languages.

6. Conclusions

A spatial phrase can be semantically analyzed into three elements LOCATIVE RELATION, PLACE VALUE, and REFERENCE ENTITY. Each of these elements is realized morphologically in various languages. Syntactically, a spatial expression has a split spatial phrase structure. The higher projection is headed by an LR element, and the lower projection is headed by a PV element. Furthermore, the three elements undergo merging and moving operations in the computational system deriving various orders of a spatial expression across languages. Finally, each of the two projections can be either prepositional, postpositional or nominal, even within a single language.

Appendix: Morphological realizations of the three semantic elements in Finnish

In Finnish, LR and PV are realized in portmanteau morphs:

PLACE VALUE

INSIDE OUTSIDE

LR: Position

LR: Source

LR: Goal

inessive case adessive case

elative case ablative case

illative case allative case

62

Ν Una Ning

Type-I: Nominal spatial phrase (36)

a. Kirja on pöydä-llä. b00k:N0M be:3sG table-ADESSlVE 'The book is on the table.' b. Sofia laitta-a kirja-n pöydä-lle. Sophia:NOMput-3sG book-ACC table-ALLAUVE 'Sophia is putting the book onto the table.' c. Sofia ott-I kirja-n pöydä-ltä. Sophia:NOMtake-PAST:3sG book-ACC table-ABLATIVE 'Sophia took the book from the surface of the table.'

(37)

a. Sofia on talo-ssa. Sopia:NOM be:3sG house-INESSLVE 'Sophia is in the house.' b. Sofia mene-e talo-on. S0phia:N0Mg0-3sG house-iLLATiVE 'Sofia is going into the house.' c. Sofia tule-e talo-sta. Sophia:NOM come-3sG house-ELATiVE 'Sophia is coming out of the house.'

Type-II: Postpositional spatial phrases (38)

a. Kengä-t o-vat sängy-n a-lla. shoe-PL:NOM be-3PL bed-GEN under-ADESSivE 'The shoes are under the bed.' b. Kissa juoks-I sängy-n a-lle. cat:NOM run-PAST:3sG bed-GEN under-ALLAUVE 'The cat ran to under the bed.' c. Kissa juoks-I sängy-n a-lta. cat:NOM run-PAST:3sG bed-GEN under-ABLATIVE 'The cat ran from under the bed.'

(39)

Lintu lentä-ä kato-n pää-llä. bird.NOM fly-3sG roof-GEN over-ADESSIVE 'The bird is flying over the roof.'

(40)

a. kodi-n lähe-lle home-GENETiVE near-ALLATiVE 'to near h o m e ' b. kodi-n lähe-ltä home-GENiTiVE near-ABLATiVE 'from near h o m e '

Type-Ill: Prepositional spatial phrases (41)

lähe-llä koti-a near-ADESSivE home-PARTiTiVE 'near h o m e '

Zhang

Movement within a Spatial Phrase

63

References Baker, Mark (1988): Incorporation: A Theory of Grammatical Function Changing. - Chicago, IL: The University of Chicago Press. Bennett David C. (1975): Spatial and Temporal Uses of English Prepositions. - London: Longman. Bresnan, Joan & Jonni M. Kanerva (1989): Locative Inversion in Chichewa: A Case Study of Factorization in Grammar. - In: Linguistic Inquiry 20, 1-50. Chung, Sandra (1976): An Object-Creating Rule in Bahasa Indonesia. - In: Linguistic Inquiry 7, 4187. Cuyckens, Hubert (1998): The Preposition To: Prepositional and Infinitival Uses. - Paper presented at the Workshop on Prepositions, Universität Hamburg, June 26-27, 1998. Gruber, Jeffrey S. (1976): Lexical Structures in Syntax and Semantics. - Amsterdam: North-Holland. Heine, Bernd, Ulrike Claudi & Friederike Hunnemeyer (1991): Grammaticalization: A Conceptual Framework. - Chicago, IL: The University of Chicago Press. Jackendoff, Ray (1990): Semantic Structure. - Cambridge, MA: MIT Press. Kayne, Richard (1994): The Antisymmetry of Syntax. - Cambridge, MA: MIT Press. Koskinen, Päivi (1998): Features and Categories: Non-Finite Constructions in Finnish. - Ph.D. dissertation. University of Toronto. Kracht, Marcus (1998): On the Semantics of Locatives. - Ms. Free University of Berlin. Li, Audrey (1990): Order and Constituency in Mandarin Chinese. - Kluwer Academic Publishers: London. Lindstromberg, Seth (1998): English Prepositions Explained. - Amsterdam: John Benjamins. Liu, Feng-hsi (1997): A Clitic Analysis of Locative Particles. - In: Journal of Chinese Linguistics 26, 48-70. Manzini, Rita (1998): Syntactic Approaches to Cliticization. - In: Glot International 3, 3-7. McCawley, James (1992): Justifying Part-of-Speech Assignments in Mandarin Chinese. - In: Journal of Chinese Linguistics 20, 211-246. Qu, Yanfeng (1994): Object Noun Phrase Dislocation in Mandarin Chinese. - Ph.D. dissertation. Vancouver, BC: University of British Columbia. Somers, H. L. (1987): Valency and Case in Computational Linguistics. - Edinburgh: Edinburgh University Press. Svorou, Soteria (1994): The Grammar of Space. - Amsterdam: Philadelphia: John Benjamins. Talmy, Leonard (1991): Path to Realization: A Typology of Event Conflation. - In: Proceedings of the Annual Meeting of the Berkeley Linguistics Society 17, 480-519.

Britta Mondorf The Effect of Prepositional Complements on the Choice of Synthetic or Analytic Comparatives

1. Introduction 1

In the various English grammar books, rules are formulated that govern the choice between synthetic comparatives, e.g. proud - prouder - proudest, and analytic comparatives, e.g. important - more important - most important. These rules reflect the generally held wisdom that the factors conditioning the choice between the synthetic and analytic variants are predominantly morphological, such as the number of syllables and the nature of the word ending (e.g. syllabic - / , final -y). This paper examines the effect of a new factor determining the choice between these two comparative forms, namely that of the syntactic environment, and in particular the presence or absence of a prepositional adjective complement. While there is general consent in the literature that trisyllabic words take the historically more recent analytic comparatives and superlatives and that monosyllables take synthetic variants, with disyllabic words being subject to variation, a first analysis of computerized corpora indicates that given the right syntactic environment, even monosyllables can strikingly often violate this rule. It will be argued that it is the complexity of the syntactic environment, and in particular the presence of a prepositional adjective complement that calls for the analytic variant.

1.1. Monosyllabic adjectives The question of what factors condition the choice between the historically earlier synthetic form, e.g. prouder, and the more recent analytic variant, e.g. more proud, has already been commented on by Pound (1901: 18), who argued that during the fifteenth century, "the form of comparison is governed by no fixed principle, such as length, ending, accent, or the source of the word [but that] the two methods are used quite indiscriminately, according to the author's choice." In a similar vein, Jespersen (1949: 347) has observed that the "rules given in ordinary grammars are often too dogmatic." The treatment of monosyllabic adjectives in the literature today seems to mirror that found in the traditional grammar books. König (1994: 540) states that the choice between the synthetic vs. analytic comparative [...] depends primarily on the syllable structure of the adjective. Monosyllabic adjectives take the inflectional form: big, bigger, biggest·, adjectives with three syllables take the analytic form: intelligent more intelligent, most intelligent. Disyllabic adjectives accept both strategies. 1

This paper is based on work supported by the Deutsche Forschungsgemeinschaft (DFG, RO 2271/1-1).

66

Britta Mondorf

Likewise, Bauer (1994: 51) holds that "generally speaking, monosyllabic adjectives [...] add the affix -er for the comparative." Empirical evidence comes from Kytö & Romaine (1997), who have observed the synthetic -er variant o f 1,248 monosyllabic adjectives in 99% o f the cases. The present paper shows that the rule that monosyllabic adjectives require the synthetic comparative appears not to apply to a wide range of monosyllabic adjectives.

1.2. The data Text samples exceeding a total of 500 million words form the basis o f the present study. The samples are taken from a selection o f British newspapers and the complete British National Corpus. 2 CORPORA

NUMBER OF W O R D S

British National Corpus The Guardian 1990-1994 + The Observer 1994 The Daily Mail 1993-1994 The Daily Telegraph + The Sunday Telegraph 1991-1994 The Times & Sunday Times 1990-1994

10,010,608 140,593,452 38,294,699 127,906,346 192,191,5 26 599,092,031

Table I: The data Even though the British National Corpus (BNC) is tagged for the syntactic function of each item, this unedited automatic tagging produces severe problems for empirical analyses. For instance, the search command apter with the additional requirement that the search string should be the comparative form o f an adjective yields the following 12 occurrences. CMU(352) CRE(2628) F9E(1237) F9E(1238) F9E(1240) F9E(1242) F9E(1244) G1B(536) GVB(2406) HTF(1037) HTF(1037) K97(9153)

[...] ecological perspectives ( [...], and they'd put Gentle in an [···]> we can consider the views of [...] [...] [...] But [...] [...] petals and drooping leaves, [...] And I could not have an [...] more elaborate; for example, as [...] governmental stages" ( [...] success over Crusader (Adrian

Apter 1982, Wolfendale 1987a [...] apter mood for this study. Apter (1988) and Mouzelis [...] Apter argues that although the [...] Apter argues that both perspectives [...] Apter sees such violence as fun [...] Apter's attempt to synthesise [...] apter tributes to a 300-year-old [...] apter pupil. Apter suggests, 'proceeding [...] Apter 1965, p. 55). Apter, Stuart Thornton).

Table 2: Search results for apter in the British National Corpus - untagged

2

version

The BNC comprises written as well as spoken samples. This is not considered to be of major relevance here since the proportion of spoken to written English data for this study amounts to approximately 1:500 million words, i.e. 0,2%.

Prepositional Complements and Synthetic or Analytic Comparatives

67

Out of these twelve occurrences of apter, nine refer to some Mr. Apter. On closer inspection the tagging reveals that even proper names are explicitly labeled AJC "comparative form of adjective." CMU(352) CRE(2628) F9 E( 123 7) F9E(1238) F9E( 1240) F9E( 1242) F9 E( 1244) G1 B(536) GVB(2406) HTF( 1037) HTF( 103 7) K97(9153)

perspectives (>>-phrases: 9x)

I in

II 0

III total

IV percentage of in

86 20 55 11

31 4 26

117 24 81 12

73.5% 83.3% 67.9% 91.7%

1

Table 5: The variable use of the preposition in with -ing constructions dependent on (the verb) face + (the object) noun problem in The Daily Telegraph and The Sunday Telegraph for 1991-1994

2.2. Gerundial constructions involving spend + financial amount term With very few exceptions, the object phrases associated with gerundial constructions dependent on the verb spend + object can be assigned to three semantic types: time expres-

Processing

83

Complexity

sions, as shown in (4a); nouns of effort like energy, force, or effort itself, as in (4b); and various monetary or financial amount terms, as in (4c). (4)

a. She spent several weeks ( ? in) preparing her talk. b. They spent a great deal of effort (in) organizing the conference. c. He spent 20,000 (pounds)/a large amount of money (on) refurbishing the flat.

In all three types, the more explicit option including the preposition has been receding dramatically over the last 100 years or so, with British English clearly lagging behind American English (cf. Rohdenburg 1996: 168-170). Even in British English, type (4a) has virtually lost its preposition in straightforward active uses. By contrast, types (4b) and (4c) have preserved their prepositional variants very much better in both British and American English. The three classes of noun are often coordinated with each other. This is facilitated by the fact that all of them are in theory compatible with either (optional) in or (optional) on. In general, however, types (4a) and (4b) have always preferred in to on. As is suggested by the data in Table 6, type (4c) seems to have switched over from (optional) in to (optional) on in the course of the twentieth century; in this way, it has become clearly differentiated from both (4a) and (4b).

1. all examples (pre-1925/post-1925) 2. active uses

3. passive uses

I in

II on (upon. \)

III 0

IV total

12 (11/1)

13 (3/10)

2 (0/2)

27 (14/13)

9 (1842,1847,1870, 1898, 1900, 1912, 1918, 1920, 1922))

9 (1915,1922,1930,1933, 1938, 1963 (upon), 1964, 1971, 1991)

1 (1987)

3

4

(1901,1918,1974)

(1898,1946,1961,1965)

1

1

9

8

(1972)

Table 6: The variable use of prepositions with -ing constructions dependent on (the verb) spend + object involving exclusively monetary or financial amount terms in the Changing Times (1785-1992)

With respect to type (4c), the complexity principle would lead us to expect that the decline of the prepositional variant should be slowed down in the passive and that any remaining instances of the less specific preposition in might be attracted to the active rather than the passive. Both expectations are confirmed by the analyses presented in Tables 7 and 8, despite the fact that the decrease of the prepositional options is clearly more advanced in The Daily Mail, a tabloid, than in The Times, a so-called quality paper. (Incidentally, this contrast supports a general stylistic principle advanced by Haiman (1983), which might be captured by the phrase "More form, more formal.")

84

Günter

I prepositions (a) (b) (c) on in total A all examples

Rohdenburg

II 0

III total

IV percentage of on/in

68

3

71

173

244

29.1%

3

Β active uses 1. all examples 2. non-canonical orderings involving fronted or relativized objects 3. canonical orderings (a) all examples (b) the type 5,000 pounds (c) other amount terms

36 5

—

39 5

141 5

180 10

21.7% 50.0%

31 16 15

3 2 1

34 18 16

136 113 23

170 131 39

20.0% 13.7% 41.1%

C passive uses 1. all examples 2. the type 5,000pounds 3. other amount terms

32 14 18

32 14 18

32 24 8

64 38 26

50.0% 36.8% 69.2%

-

—

Table 7: The variable use of prepositions with -ing constructions dependent on (the verb) spend + object involving exclusively monetary or financial amount terms in the 3rd and 4th quarters of The Daily Mail & The Sunday M a i l / o r 1994

I prepositions (b) (c) (a) in total on

II 0

III total

IV percentage of on/in

A all examples

50

5

55

63

118

46.6%

Β active uses 1. all examples 2. the type 5,000pounds 3. other amount terms

31 20 11

5 3 2

36 23 13

54 44 10

90 67 23

40.0% 34.3% 56.5%

C passive uses 1. all examples 2. the type 5,000 pounds 3. other amount terms

19 10 9

-

19 10 9

9 6 3

28 16 12

67.9% 62.5% 75.0%

-

Table 8: The variable use of prepositions with -ing constructions dependent on (the verb) spend + object involving exclusively monetary or financial amount terms in the 1st quarter o / T h e Times Genitiv anzunehmen, wobei der ursprüngliche Kasus in unterschiedlichem Maße weiterexistieren kann.

3. Rektionswechsel in Verbindung mit Stellungswechsel

Bisher haben wir ausschließlich Präpositionen betrachtet, bei denen ein Genitiv in Prästellung mit einem Dativ in Prästellung alterniert. Es gibt ferner Präpositionen, bei denen Genitiv und Dativ in Prästellung alternieren, zugleich jedoch der ursprüngliche Kasus (Genitiv oder Dativ) bei Poststellung erhalten bleibt. Auch hier können wir zwei Untergruppen je nach Ursprungsrektion unterscheiden.

3.1. Genitiv (Poststellung) > Dativ (Prästellung) Im Deutschen liegt eine einzige Präposition vor, die ursprünglich den Genitiv in Poststellung erforderte und nun auch den Dativ in Prästellung regieren kann: 11

12

So in der einschlägigen Literatur und in fast allen Grammatiken. Eine Ausnahme stellen lediglich Zifonun u.a. (1997: 2082) dar: Sie fuhren für inmitten einen Dativbeleg an, den sie jedoch als Verletzung der allgemeinen Genitivregel einstufen. Diese verbreitete These wird u.a. noch von Willems (1997) in seinem Werk zu kognitiven Kasustheorien vertreten.

Präpositionale Rektionsaliernation

109

wegen Vom morphologischen Gesichtspunkt ist wegen undurchsichtig, denn die ursprüngliche Übereinstimmung mit dem Dativ Plural des Substantivs Weg ist synchron nicht mehr zu erkennen. Sehen wir uns jeweils ein Korpusbeispiel an, bei dem wegen in Poststellung mit Genitiv, in Prästellung mit Genitiv und in Prästellung mit Dativ vorkommt: (30) a. Heike immer wieder vor dem Verzehr brandenburgischer Fische gewarnt, denn was hier als Delikatesse galt, mußte in Westberlin von den Fischern der hochbelasteten Gewässer wegen noch immer als Sondermüll entsorgt werden, doch das störte ihn wenig. (Bosetzky, 145) b. Allein in den letzten zwölf Monaten nahm die öffentliche Schuld um 190 Milliarden Mark zu, zum größten Teil wegen der Zinszahlungen für die Gesamtschuld. (FAZ 6.1.1994, 2) c. »Wie komme ich zu der Ehre?« »Wegen deinen Beinen«, sagt Czemi. »Deiner Beine!«, sage ich. (Borowiak, 46)

3.2. Dativ (Poststellung) > Genitiv (Prästellung) Folgende Dativ-Postpositionen nehmen in Prästellung zusätzlich einen Genitiv an:13 entlang, entgegen, gegenüber, nahe, ähnlich, gleich, gemäß-, entsprechend. Alle diese Bildungen sind morphologisch transparent, denn entlang, entgegen und gegenüber sehen aus wie ein Adverb, nahe usw. wie ein Adjektiv, entsprechend wie ein Verb (Partizip). Was die Rektion betrifft, so wird lediglich für entlang der Genitiv standardsprachlich als Variante akzeptiert. Für sämtliche anderen Präpositionen ist die Möglichkeit einer Genitivrektion in der einschlägigen Literatur bisher nicht zur Kenntnis genommen worden. Betrachten wir zunächst einige Korpusbeispiele mit entlang und gemäß (jeweils Dativ in Poststellung, Dativ in Prästellung und Genitiv in Prästellung): (31) a. Ein wenig weiter, den Gleisen entlang, zu Fuß zu erreichen, kommt der unterirdische Karstfluß Timavo ans Licht [...]. (Handke, 118) b. Am Ende der schnurgeraden Pappelallee, entlang dem Forstmeister-Kanal, liegt ein Erholungszentrum mit Restaurant, Schwimmbad, Minigolfanlage und Campingplatz. (Kohut u.a., 169) c. Entlang des Malecón befinden sich noch einige kleine Badebuchten, die jedoch meist nicht sehr sauber sind. (Griebel & Reithmeir, 80)

13

Vgl. Di Meola (1999); für eine detaillierte Analyse der verschiedenen Rektionsmöglichkeiten von entlang - Dativ, Genitiv und Akkusativ - siehe auch Di Meola (1998).

Claudio Di Meóla

110 (32)

a. Diesen Grundsätzen gemäß müssen Eigentumsbindungen stets verhältnismäßig (Hesse, 184) b. Letztere sind gemäß dem Organprinzip Verwaltungsstellen in einer zentralistischen wirtschaft, während öffentliche Unternehmen dezentrale Wirtschaftssubjekte in Marktwirtschaft sind. (Jens, 119-120) c. Auch im Magen- und Darmbereich können Barotraumen entstehen, da auch hier vorhanden sind, die sich gemäß der Druckgesetze verhalten. (Naglschmid, 79)

sein. Planeiner Gase

Für die übrigen Präpositionen seien lediglich B e l e g e mit der "neuen" Genitivrektion angeführt: (33)

Entgegen verhaltener Eingangserwartungen habe die Domotex mit nochmals verbesserten Ergebnissen abgeschlossen, berichtet die Deutsche Messe AG. ( F A Z 13.1.1994, 12) (34) »Der DFB muß sich darüber im klaren sein, daß er die WM dann nicht bekommt, wenn er solche Forderungen unterstützt«, sagte Burger am Sonntag gegenüber des Deutschlandfunks. (FAZ 10.1.1994,22) (35) Dann schwenkten wir durch die Grands Boulevards nahe des Flughafens und vorher vorbei am Bahnhof, in dem täglich zwölf Züge ankamen und abfuhren, dann zurück zur KoutoubiaMoschee [...]. (Herburger, 176) (36) Wird der Schnitt von oben nach unten, ähnlich des Bleistiftspitzens, vollzogen, entstehen Wölbungen auf der Schnittfläche, und die Veredlungen wachsen schlecht oder überhaupt nicht zusammen. (Zettl, 70) (37) Bei gegebener Menge der Produktionsfaktoren ist der Output somit gleich der kleineren der in der Klammer enthaltenen Verhältniszahlen. (Fuhrmann, 96) (38) Insgesamt ergab sich für Deutschland mit 4,0% ein Preisanstieg des Bruttoinlandsprodukts entsprechend der Erwartungen. (Jahreswirtschaftsbericht, 55)

3.3. D i e A b f o l g e v o n Rektions- und Stellungswechsel Bei den unter 3.1 und 3.2 betrachteten Präpositionen verändert sich i m Z u g e der Grammatikalisierung auf syntaktischer Ebene sowohl die Rektion als auch die Stellung. W i e bereits gesagt, ist in Poststellung ausschließlich der ursprüngliche Kasus zu finden - Genitiv und Dativ alternieren nur in Prästellung. Daraus können wir einen Rückschluß auf die relative Chronologie der beiden Differenzierungsprozesse ziehen: SCHEMA 1 : ABFOLGE DER DIFFERENZIERUNGSPROZESSE

a. Poststellung > Prästellung b. Genitiv > Dativ bzw. Dativ > Genitiv Zuerst findet der Stellungswechsel, dann der Kasuswechsel statt. M a n kann sich nun fragen, aus w e l c h e m Grunde die umgekehrte A b f o l g e nicht belegt ist. Mit anderen Worten: Warum kommt in Poststellung kein strukturfremder Genitiv bzw. Dativ auf? D i e Erklärung liegt wahrscheinlich darin, daß ein Stellungswechsel als Differenzierungsprozeß auffalliger ist als ein Kasuswechsel - der ja oftmals durch verschiedene Formen des Synkretismus verdeckt bleibt - und somit Priorität genießt.

Präpositionale

Rektionsalternation

111

4. Eine statistische Analyse

Die Rektionsalternation zwischen Dativ und Genitiv soll nun statistisch dokumentiert werden. Im folgenden werde ich die beiden großen Gruppen - Genitiv-Präpositionen, die zum Dativ übergehen (vgl. 2.1 und 3.1) und Dativ-Präpositionen, die zum Genitiv übergehen (vgl. 2.2 und 3.2) - auf drei Aspekte hin untersuchen: (i) Frequenz der jeweils "neuen" Rektion; (ii) Zusammenhang zwischen Rektion und Numerus der NP; (iii) Zusammenhang zwischen Rektion und Struktur der NP.

4.1. Frequenz der jeweils "neuen" Rektion Sehen wir uns als erstes die zahlenmäßige Verteilung von Dativ und Genitiv bei den Präpositionen mit Rektionsalternation an. Es sei vorausgeschickt, daß ich nur diejenigen Belege berücksichtige, die dem Genitiv bzw. Dativ zweifelsfrei zugeschrieben werden können. Nicht betrachtet werden somit mehrdeutige Formen wie z.B. innerhalb der Stadt oder gegenüber der Kirche sowie Erweiterungen mit von (innerhalb von, gegenüber von usw.). Bei Präpositionen mit unterschiedlichen Stellungsvarianten werden ausschließlich Prästellungsbelege einbezogen, denn nur in Prästellung alternieren die beiden Kasus. Wenden wir uns zunächst der Gruppe von Präpositionen zu, die neben einer ursprünglichen Genitivrektion einen Dativ annehmen können. Es handelt sich um die Präpositionen innerhalb, bezüglich, hinsichtlich, einschließlich, zuzüglich, abzüglich, südlich, westlich, voll, voller, während, mangels, mittels, zwecks, kraftlKraft, Mittelmitte, statt, anstatt (2.1) sowie um die ursprüngliche Postposition wegen (3.1). Insgesamt weist die Gruppe der Genitiv-Präpositionen 2690 relevante Belege auf: 2284 Genitiv-Formen (ca. 85%) und 406 Dativ-Formen (ca. 15%). Als Genitive werden gerechnet: 2236 PPs, die ausschließlich Genitiv sein können (wie z.B. statt des Kontrasts, statt der Kontraste) sowie 48 PPs, die außer einer Genitivinterpretation eine Nominativbzw. Akkusativinterpretation zulassen - nicht jedoch Dativ sein können (Beispiele wie statt Kontraste). Als Dative werden gerechnet: 105 PPs, die ausschließlich Dativ sein können (z.B. wegen dem Regen)·, 300 PPs, die außer einer Dativinterpretation eine Nominativ- bzw. Akkusativinterpretation zulassen, nicht jedoch Genitiv sein können (wegen Regen usw.); 1 Beispiel von Dativ-/Akkusativsynkretismus (wegen uns). Für die einzelnen Präpositionen ergibt sich die in Tabelle 1 aufgefiihre Verteilung (geordnet nach dem prozentmäßigen Anteil der strukturfremden Dativbelege). Es gilt hier der offensichtliche Zusammenhang: Je höher der Dativ-Anteil, desto fortgeschrittener die Grammatikalisierung bezüglich der Rektion.

112

Claudio Di Meóla

innerhalb während Mitte/mitte westlich hinsichtlich kraft/Kraft südlich bezüglich wegen abzüglich mangels einschließlich mittels statt anstatt zwecks voll voller zuzüglich

gesamt

Genitiv

Dativ

gesamt

Anteil Dativ

340 445 69 16 100 59 21 52 798 9 28 118 29 119 6 2 31 40 2

2 10 3 1 7 5 2 6 147 2 7 34 9 44 3 3 48 68 5

342 455 72 17 107 64 23 58 945 11 35 152 38 163 9 5 79 108 7

1% 2% 4% 6% 7% 8% 9% 10% 16% 18% 20% 22% 24% 27% 33% 60% 61% 63% 71%

2284

406

2690

15%

Tabelle 1: Genitiv-Präpositionen

mit

Dativrektion

Untersuchen wir nun die Präpositionen, die neben einer ursprünglichen Dativrektion einen Genitiv aufweisen können: die Präpositionen dank, trotz, inmitten, binnen, samt, mitsamt (2.2) sowie die ursprünglichen Postpositionen entlang, entgegen, gegenüber, nahe, ähnlich, gleich, gemäß, entsprechend (3.2). Insgesamt haben sich im Korpus 1963 relevante Belege gefunden: 1248 Dativ-Formen (ca. 64%) und 715 Genitiv-Formen (ca. 36%). Als Dative werden auch hier gerechnet: 1058 PPs mit ausschließlicher Dativinterpretation, 188 PPs mit Dativ-/Nominativ-/Akkusativsynkretismus, 2 PPs mit Dativ-/Akkusativsynkretis-mus. Als Genitive werden gerechnet: 712 PPs mit ausschließlicher Genitivinterpretation, 3 mit Genitiv-/Nominativ-/Akkusativsynkretismus. Aufgeschlüsselt nach den einzelnen Präpositionen ergibt sich das in Tabelle 2 aufgeführtes Bild (wobei wie-derum die Korrelation gilt: Je höher der Anteil des "neuen" Kasus, desto fortgeschrittener die rektionsmäßige Grammatikalisierung). Vergleicht man beide Gruppen von Präpositionen, so fällt zunächst auf, daß die GenitivPräpositionen einen niedrigen bis hohen Prozentsatz (1-71%) von Dativbelegen aufweisen können (im Schnitt 15%), während die Dativ-Präpositionen niedrige (1-25%), aber auch sehr hohe Prozentsätze (75-97%) einer Genitivrektion zeigen (im Schnitt 36%).

Präpositionale

113

Rektionsalternation

Dativ

gegenüber gleich samt entgegen mitsamt ähnlich nahe entsprechend gemäß binnen entlang dank trotz inmitten gesamt

725 65 64 55 33 25 57 79

Genitiv

gesamt

1 1

726 66

3

67 58 35 27 62

3 2 2

Anteil Genitiv

>

Prästellung Poststellung Dativ Genitiv

Die syntaktischen Entwicklungen (a) Poststellung > Prästellung und (c) Genitiv > Dativ gehen mit der Prototypisierung konform, nicht jedoch die Entwicklungen (b) Prästellung > Poststellung und (d) Dativ > Genitiv. Es gibt also zwei Konfliktfälle: (i)

(ii)

Soll eine Präposition in Prästellung nach dem Prinzip der Differenzierung zur Poststellung übergehen oder nach dem Prinzip der Prototypisierung die Prästellung beibehalten? Soll eine Dativ-Präposition nach dem Prinzip der Differenzierung zum Genitiv übergehen oder nach dem Prinzip der Prototypisierung den Dativ beibehalten?

22

Absehen muß man hier offensichtlich von den Präpositionen, die die Form eines Substantivs haben. Hier kann der Genitiv auch adnominal sein.

23

Zu den Anwendungsmöglichkeiten der Prototypentheorie auf grammatische Kategorien allgemein siehe vor allem Lakoff (1987) und Taylor (1995).

124

Claudio Di Meóla

Der Übergang von Prä- zu Poststellung (Fall b) ist nicht belegt, der Übergang von Dativ zu Genitiv (Fall d) ist sehr wohl belegt. Man sieht also, daß in Konfliktfallen bezüglich der Stellung das Prinzip der Prototypisierung die Oberhand behält, bezüglich der Rektion hingegen das Prinzip der Differenzierung. Dieses Hierarchisierungsmuster ist durchaus motiviert. Die Stellung ist nämlich als prototypisches Charakteristikum von Präpositionen offensichtlicherweise markanter als die von Synkretismen oftmals verdeckte Rektion. Man könnte also eine generelle Hypothese fur Grammatikalisierungsprozesse aufstellen: Bezüglich markanten Charakteristiken einer grammatischen Kategorie überwiegt das Prinzip der Prototypisierung, bezüglich weniger auffalligen Charakteristiken überwiegt das Prinzip der Differenzierung.

7. Zusammenfassung und Ausblick

Wir haben gesehen, daß zahlreiche Präpositionen der deutschen Gegenwartssprache eine Kasusalternation Dativ/Genitiv aufweisen. Es handelt sich um nichtprototypische Präpositionen, die sprachgeschichtlich noch relativ jung sind und den Weg zur "charakteristischen" Präposition erst teilweise beschritten haben. Sie stimmen fast immer formal mit Inhaltswörtern überein (Adverbien wie z.B. gegenüber, Adjektiven wie nahe, Verben wie dem Partizip entsprechend, Substantiven wie Dank) und erweisen sich somit als morphologisch vollkommen transparent. Oftmals können sie - wie die soeben genannten Beispiele - in ihrer ursprünglichen Funktion als Inhaltswort weiterverwendet werden. Obwohl Alternationen des Typs Dativ/Genitiv in den meisten Fällen vom normativen Standpunkt aus nicht akzeptiert werden, sind sie dennoch belegt. Statistiken auf der Grundlage eines reichhaltigen, breitgefacherten Korpus der deutschen Gegenwartssprache haben gezeigt, daß selbst in gehobener und kontrollierter Schriftsprache das Auftreten einer normabweichenden Rektion keine Ausnahme ist: Der Anteil eines standardsprachlich "falschen" Dativs bei Genitiv-Präpositionen liegt - je nach Einzelfall - zwischen 1% und 71%, der Anteil eines "falschen" Genitivs bei Dativ-Präpositionen zwischen 1% und 25%. Die vom präskriptiven Standpunkt aus jeweils kritisierte Kasuswahl ist jedoch keinesfalls eine zufallige Entgleisung, zurückzufuhren auf die Unfähigkeit eines Sprechers, die standardsprachliche Norm zu erfüllen. Der Kasus erklärt sich einzig und allein aus der Dynamik von Grammatikalisierungsprozessen. Die synchron zu beobachtenden Kasusalternationen sind eine Folge von grammatikalisierungsbedingten Kasuswechseln. Bezüglich der Grammatikalisierung von Präpositionen habe ich ein Prinzip ikonischer Natur angenommen, das ich "Prinzip der maximalen Differenzierung" genannt habe: Um die durch Reanalyse bereits stattgefundene semantische Entwicklung von Inhalts- zu Funktionswort sichtbar werden zu lassen, findet eine progressive Abkehr von der ursprünglichen syntaktischen Umgebung der jeweiligen Form statt. Was die Rektion betrifft, so gehen Genitiv-Präpositionen zum Dativ übergehen, Dativ-Präpositionen zum Genitiv. Eine Kasusalternation liegt somit immer dann vor, wenn der präpositionale Grammatikalisierungsprozeß noch im Gange ist. Es ist demzufolge evident, daß Kasusalternation bei

Präpositionale

Rektionsalternation

125

Präpositionen nicht die Ausnahme, sondern die Regel darstellt. Lediglich Präpositionen höchsten bzw. niedrigsten Grammatikalisierungsgrades weisen keinerlei Alternation auf. Die Tatsache nun, daß zahlreiche Dativ-Präpositionen eine Genitivrektion annehmen, verdient besondere Beachtung. Zum einen ist deutlich geworden, daß der präpositionale Genitiv äußerst produktiv ist und keineswegs vom Aussterben bedroht, wie immer wieder behauptet wird. Zum anderen ist das Aufkommen einer Genitivrektion bei DativPräpositionen unter dem Gesichtspunkt der Prototypisierung interessant. Prototypische Präpositionen regieren den Dativ (oder Akkusativ), nicht den Genitiv. Das Grammatikalisierungsprinzip der "maximalen Differenzierung" kontrastiert in diesem Fall also mit der erwartbaren Annäherung an den Prototyp der betreffenden grammatischen Kategorie. Es hat sich ferner gezeigt, daß das zahlenmäßige Verhältnis zwischen Dativ und Genitiv mit Numerus und Struktur der jeweils regierten NP zusammenhängt. Diese beiden Faktoren können sich auf einen eventuellen Rektionswechsel fördernd, aber auch hemmend auswirken. Ist die NP singularisch und/oder besteht sie aus einem einfachen Nomen, scheint allgemein eine Präferenz für den Dativ vorzuliegen: Dativ-Präpositionen bewahren eher ihren ursprünglichen Kasus, Genitiv-Präpositionen gehen eher zum Dativ über. Ist die NP hingegen pluralisch und/oder enthält sie ein Adjektiv bzw. Determinane, scheint eine Präferenz für den Genitiv vorzuliegen: Genitiv-Präpositionen bewahren eher ihren ursprünglichen Kasus, Dativ-Präpositionen gehen eher zum Genitiv über. Für den Faktor Numerus bietet sich eine Erklärung im Rahmen der "maximalen Differenzierung" an: Die Tendenz zum Kasuswechsel ist jeweils in dem Numerus am stärksten, in dem der Ausgangskasus am markantesten ist. Demgegenüber ist den Gründen für das Verteilungsmuster bezüglich der Struktur der NP noch im einzelnen nachzugehen. Es ist des weiteren deutlich geworden, daß die adäquate Beschreibung synchroner Alternationen nicht ohne diachrone Erwägungen auskommen kann. Ein Vergleich mit literarischen Texten aus den Jahren 1750-1920 hat die vorliegende Analyse der Gegenwartssprache bekräftigt. Bei Präpositionen, die gegenwärtig eine Kasusalternation Dativ/Genitiv aufweisen, ist der "alte", strukturkonforme Kasus in früheren Sprachstufen prozentmäßig häufiger belegt als heutzutage, der "neue", strukturverletzende Kasus hingegen prozentmäßig geringer (oder überhaupt nicht). Die diachronische Grammatikalisierungsentwicklung geht also erwartungsgemäß in Richtung einer progressiven Verbreitung der "neuen", höhergradig grammatikalisierten Variante auf Kosten der "alten", niedriggradig grammatikalisierten Variante. Schließlich ist noch ein wichtiger Problembereich der Grammatikalisierungsforschung allgemein zur Sprache gekommen. Auch für den präpositionalen Rektionswechsel hat sich nämlich gezeigt, daß das Grammatikalisierungsmodell lediglich eine begrenzte Prädiktabilität bezüglich Sprachwandelprozessen erlaubt: Es läßt sich nicht vorhersagen, ob ein Rektionswechsel überhaupt in Gang kommt, ob der Übergang partiell oder vollständig sein wird und wie schnell er sich vollziehen wird. Es sind nämlich stets mehrere - potentiell widersprüchliche - Faktoren am Werke. Für die präpositionale Grammatikalisierung sind es zumindest die Prinzipien der Differenzierung, Prototypisierung und funktionalen Auffälligkeit. Das Prinzip der Differenzierung (Abkehr von der Ursprungsstruktur) ist ausgangspunktorientiert, die Prinzipien der Prototypisierung (Annäherung an den Prototyp) und funktionalen Auffälligkeit (Erkennbarkeit der syntaktischen Funktion) sind endpunktorientiert. Der Übergang Genitiv > Dativ wird den Prinzipien der Differenzierung und Prototypisierung gerecht, kontrastiert jedoch

126

Claudio Di Meóla

mit d e m Prinzip der Auffälligkeit. Der Übergang Dativ > Genitiv wird den Prinzipien der Differenzierung und Auffälligkeit gerecht, kontrastiert j e d o c h mit d e m Prinzip der Prototypisierung. W e l c h e s Prinzip j e w e i l s überwiegt, ist nur sehr schwer zu prognostizieren, w e n n nicht andere präpositionale Charakteristiken in die Untersuchung miteinbezogen werden (beispielsweise Stellung). Soviel steht j e d o c h fest: B e z ü g l i c h der drei obengenannten Prinzipien der Sprachoptimierung ist ein Kasuswechsel stets "besser" als eine Kasusbeibehaltung. Während der Wechsel j e w e i l s nur ein Prinzip verletzt, gerät die Beibehaltung mit j e w e i l s z w e i Prinzipien in Konflikt: Für die Dativ-Präpositionen sind es die Differenzierung und die Auffälligkeit, für die Genitiv-Präpositionen die Differenzierung und die Prototypisierung.

Literatur

1. Zitierte Quellen Becker, Hermann J. u.a. (eds.) (1994): Fusionsreport '93. - Heidelberg: Decker's. Bild-Zeitung, September 1998. Borowiak, Simone (1992): Frau Rettich, die Czerni und ich. - Frankfurt a.M.: Eichborn (Taschenbuchausgabe O.J.; München: Goldmann). Bosetzky, Horst (1993): Blut will der Dämon. - Reinbek: Rowohlt. Drawert, Kurt (1992). Spiegelland: Ein deutscher Monolog. - Frankfurt a.M.: Suhrkamp. Frankfurter Allgemeine Zeitung, Januar 1994. Freudenberger, Axel (1991): Sizilien. - Köln: Hayit. Fuhrmann, Wilfried (1991 3 ): MakroÖkonomik: Zur Theorie interdependenter Märkte. - München: Oldenbourg. Grass, Günter (1995): Ein weites Feld. - Göttingen: Steidl. Griebel, Michael & Claudia Reithmeir (1991 2 ): Senegal mit Gambia und Mauretanien. - Köln: Hayit. Grube, Tina (1995): Männer sind wie Schokolade. - Frankfurt a.M.: Fischer. Handke, Peter (1990): Versuch über die Jukebox: Erzählung. - Frankfurt a.M.: Suhrkamp. Heger, Anna-Maria (1994): Hunde: Rassen - Erziehung - Pflege. - Köln: Buch und Zeit Verlagsgesellschaft. Herburger, Günter (1994): Traum und Bahn. - München: Luchterhand. Hesse, Konrad (1993 19 ): Grundzüge des Verfassungsrechts der Bundesrepublik Deutschland. Heidelberg: Müller. Hilbig, Wolfgang (1993): "Ich ": Roman. - Frankfurt a.M.: Fischer. Jahreswirtschaftsbericht 1994 der Bundesregierung. 1994. - Stuttgart: Metzler-Poeschel. Jens, Uwe (ed.) (1991): Der Umbau: Von der Kommandowirtschaft zur Öko-sozialen Marktwirtschaft. - Baden-Baden: Nomos. Keller, Claudia (1991): Der Flop. - Frankfurt a.M.: Fischer. Kirsch, Sarah (1994): Das simple Leben. Stuttgart: - Deutsche Verlags-Anstalt. Köbler, Gerhard (1991): Deutsches Privatrecht der Gegenwart. - München: Beck. Kohut, Jutta u.a. (1993): Österreich. - Berlin: Reise- und Verkehrsverlag. Kronenberg, Susanne (1994): Freizeitpferde: Rassen - Eignung - Traditionen. - Stuttgart: FranckhKosmos. Lind, Hera (1992): Frau zu sein bedarf es wenig. - Frankfurt a.M.: Fischer.

Präpositionale

Rektionsalternation

127

Maier, Walter (1990 2 ): Staats- und Verfassungsrecht. - Achim: Fleischer. Malik, Fredmund & Daniel Stelter (1990): Krisengefahren in der Weltwirtschaft: Überlebensstrategie für das Unternehmen. - Stuttgart: Schäffer. Mühlbauer, Winni (1993): So einfach ist Laufen: Das Programm fur den leichten Einstieg. München: Winfried-Muehlbauer-Verlag (Taschenbuchausgabe 1997; Reinbek: Rowohlt). Naglschmid, Friedrich (1990): Fit mit Sporttauchen. - Niedernhausen/Ts.: Falken. Schneider, Peter (1992): Paarungen: Roman. - Berlin: Rowohlt. Siebert, Horst (1994): Geht den Deutschen die Arbeit aus? Neue Wege zu mehr Beschäftigung. München: Bertelsmann. Wemer, Olaf (1995 2 ): Fälle zum Erbrecht. - München: Beck. Wolter, Jürgen (ed.) (1995): Zur Theorie und Systematik des Strafprozeßrechts. - Neuwied/ Kriftel/Berlin: Luchterhand. Zettl, Erich (1993): Der richtige Schnitt von Obst- und Ziergehölzen, Rosen und Hecken. - Niedernhausen/Ts.: Falken. Zinn, Dorit (1995): Mit fünfzig küssen Männer anders. - Frankfurt a.M.: Fischer.

2. Sonstige Literatur Abraham, Werner (1990): Postposition, Präpositionaladverb oder Präfix? Zum Status der sog. "trennbaren Verbpräfixe" im Deutschen. - In: A. Katny (Hg.): Studien zum Deutschen aus kontrastiver Sicht, 7-34. Frankfurt a.M.: Lang. - (ed.) (1993): Grammatikalisierung und Reanalyse: Konfrontation. - Berlin: Mouton de Gruyter (Folia Linguistica Histórica 13). Ágel, Vilmos (1992): Die deutschen Genitivpräpositionen: Prinzipien ihrer Verwendung. - In: S. R. Anschütz (ed.): Texte, Sätze, Wörter und Moneme: Festschrift für Klaus Heger zum 65. Geburtstag, 17-33. Heidelberg: Orientverlag. Bartels, Gerhard & Birgit Tarnow (1993): Von 'à'bis 'zwischen': Das Beziehungswort der deutschen Gegenwartssprache. - Frankfurt a.M.: Lang. Bebermeyer, Renate (1994): 'Trotz', 'laut' und die (schweizer)deutsche Grammatik. - In: Sprachspiegel 50, 33-35. Behagel, Otto (1924): Deutsche Syntax: Eine geschichtliche Darstellung. Bd. II: Die Wortklassen und Wortformen, B. Adverbium, C. Verbum. - Heidelberg: Winter. Berger, Dieter, Günther Drosdowski & Otmar Käge (eds.) (1985): Duden: Richtiges und gutes Deutsch - Wörterbuch der sprachlichen Zweifelsfälle. - Mannheim: Dudenverlag. Bertram, Mathias u.a. (eds.) (1997): Deutsche Literatur von Lessing bis Kafka. - Berlin: Directmedia. Compes, Isabel, Silvia Kutscher & Carmen Rudorf (1993): Pfade der Grammatikalisierung: Ein systematischer Überblick. - Köln: Institut für Sprachwissenschaft der Universität zu Köln (Arbeitspapier Neue Folge Nr. 17). Di Meola, Claudio (1997): Grammatikalisierungsprozesse am Beispiel subordinativer Konzessivkonnektive. - In: Papiere zur Linguistik 57, 183-203. - (1998): Semantisch relevante und semantisch irrelevante Kasusaltemation am Beispiel von 'entlang'. - In: Zeitschrift für Sprachwissenschaft 17, 204-235. - (1999): 'Entgegen', 'nahe', 'entsprechend' und 'gemäß': Dativpräpositionen mit Genitivrektion. In: Zeitschrift für germanistische Linguistik 27, 344-351. - (2000a): Die Grammatikalisierung deutscher Präpositionen. - Tübingen: Stauffenburg (= Studien zur deutschen Grammatik 62). - (2000b): Deutsche Präpositionen im Überblick: Form, Stellung und Rektion. - In: Pandaemonium Germanicum 4, 321-368.

128 -

Claudio Di Meóla

(2000c): Le preposizioni in tedesco: il problema dell'alternanza genitivo/dativo e la sua didatticizzazione. In: Studi italiani di linguistica teorica e applicata 29, 423-450. - (2001): Vom Inhalts- zum Funktionswort: Grammatikalisierungspfade deutscher Adpositionen. In: Sprachwissenschaft 26, 59-83. Diewald, Gabriele (1997): Grammatikalisierung: Eine Einfiihrung in Sein und Werden grammatischer Formen. - Tübingen: Niemeyer (= Germanistische Arbeitshefte 36). Drosdowski, Günther u.a. (eds.) (1995 5 [1959]): Duden: Grammatik der deutschen Gegenwartssprache. - Mannheim: Dudenverlag. Durreil, Martin (1993): The use of 'entlang' in modern standard German. In: J.L. Flood u.a. (eds.): Das unsichtbare Band der Sprache: Studies in German Language and Linguistic History in Memory of Leslie Seiffert, 521-538. Stuttgart: Heinz. Engel, Ulrich (1996 3 [1988]): Deutsche Grammatik: Vollständige Darstellung der deutschen Gegenwartssprache. - Heidelberg: Groos. Fries, Norbert (1988): Präpositionen und Präpositionalphrasen im Deutschen und Neugriechischen: Aspekte einer kontrastiven Analyse Deutsch-Neugriechisch. - Tübingen: Niemeyer (= Linguistische Arbeiten 208). Gallmann, Peter& Horst Sitta (1996): Duden: Die Neuregelung der deutschen Rechtschreibung Regeln, Kommentar und Verzeichnis wichtiger Neuschreibungen. - Mannheim: Dudenverlag. Gelhaus, Hermann, Roger Frey & Otfried Heyne (1972): Vorstudien zu einer kontrastiven Beschreibung der schweizerdeutschen Schriftsprache der Gegenwart: Die Rektion der Präpositionen 'trotz', 'während' und 'wegen '. - Frankfurt a.M.: Lang. Giacalone Ramat, Anna & Paul J. Hopper (eds.) (1998): The Limits of Grammaticalization. Amsterdam: John Benjamins. Haspelmath, Martin (1998): Does Grammaticalization Need Reanalysis? - In: Studies in Language 22,315-351. - ( 1999): Why Is Grammaticalization Irreversible? - In: Linguistics 37, 1043-1068. Heine, Bernd, Ulrike Claudi & Friederike Hünnemeyer (1991): Grammaticalization: A Conceptual Framework. - Chicago, IL: The University of Chicago Press. Heibig, Gerhard & Joachim Buscha (1986® [1972]): Deutsche Grammatik: Ein Handbuch für den Ausländerunterricht. - Leipzig: Enzyklopädie. Hentschel, Elke, Harald Weydt(1994 2 [1990]): Handbuch der deutschen Grammatik. - Berlin: Mouton de Gruyter. Hopper, Paul. J., Elizabeth C. Traugott (1993): Grammaticalization. - Cambridge: Cambridge University Press. Klaus, Cäcilia (1999): Grammatik der Präpositionen: Studien zur Grammatikographie. Frankfurt a.M.: Lang. König, Ekkehard & Johan van der Auwera (1988): Clause Integration in German and Dutch Conditionals, Concessive Conditionals, and Concessives. - In: J. Haiman & S. A. Thompson (eds.): Clause Combining in Grammar and Discourse, 101-133. Amsterdam: John Benjamins. König, Ekkehard & Bernd Kortmann (1991): On the Reanalysis of Verbs as Prepositions. - In: G. Rauh (ed.): Approaches to Prepositions, 109-125. Tübingen: Narr. Kortmann, Bernd & Ekkehard König (1992): Categorial Reanalysis: The Case of Deverbal Prepositions. - In: Linguistics 30, 671-697. Lakoff, George (1987): Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. - Chicago, IL: The University of Chicago Press. Lehmann, Christian (1995 2 [1982]): Thoughts on Grammaticalization. - München: Lincom Europa. - (1995b):. Synsemantika. - In: J. Jacobs u.a. (eds.) 1995: Syntax: Ein internationales Handbuch zeitgenössischer Forschung, 1251-1266. Berlin: Mouton de Gruyter. Lessau, Donald A. (1994): A Dictionary of Grammaticalization, 3 Bde. - Bochum: Brockmeyer (Bochum-Essener Beiträge zur Sprachwandelforschung 21).

Präpositionale

Rektionsalternation

129

Lindqvist, Christer (1994): Zur Entstehung von Präpositionen im Deutschen und Schwedischen. Tübingen: Niemeyer (= Linguistische Arbeiten 311). Müller, Wolfgang (1990): Mittels neuen Kleinbusses oder mittels neuem Kleinbus? - In: Muttersprache 100, 53-59. Pagliuca, William (ed.) (1994): Perspectives on Grammaticalization. - Amsterdam: John Benjamins. Paul, Hermann (1920): Deutsche Grammatik. Band IV. - Halle a.S.: Niemeyer. Petig, William E. (1997): Genitive Prepositions Used with the Dative in Spoken German. - In: Unterrichtspraxis 30, 36-39. Ramat, Paolo (1992): Thoughts on Degrammaticalization. - In: Linguistics 30: 549-560. Rentsch, Hans U. (1986): Trotz/dank des, trotz/dank dem - was stimmt? - In: Sprachspiegel 42, 178— 179. Sauter, Roger (1998): Der Genitivschwund im verbalen Bereich. - In: M. Vuillaume (ed.): Die Kasus im Deutschen. Form und Inhalt, 181-191. Tübingen: Stauffenburg (= Eurogermanistik 13). Schröder, Jochen (1986): Lexikon deutscher Präpositionen. - Leipzig: Enzyklopädie. Sommerfeldt, Karl-Ernst, Günter Starke (1998 3 [1988]: Einführung in die Grammatik der deutschen Gegenwartssprache. - Tübingen: Niemeyer. Taylor, John R. (1995 2 [1989]): Linguistic Categorization. Prototypes in Linguistic Theory. - Oxford: Oxford University Press. Traugott, Elizabeth C. & Bernd Heine (eds.) (1991): Approaches to Grammaticalization, 2 Vols. Amsterdam: John Benjamins. Weinrich, Harald (1993): Textgrammatik der deutschen Sprache. - Mannheim: Dudenverlag. Willems, Klaas (1997): Kasus, grammatische Bedeutung und kognitive Linguistik: Ein Beitrag zur allgemeinen Sprachwissenschaft. - Tübingen: Narr (= Tübinger Beiträge zur Linguistik 427). Wittich, Ursula u.a. (1975): Präpositionen in der deutschen Gegenwartssprache. - In: Sprachpflege 24, 16-21. Zifonun, Gisela, Ludger Hoffman, Bruno Strecker, et al. (1997): Grammatik der deutschen Sprache. Berlin: Mouton de Gruyter.

Priska-Monika

Hottenroth

Fortbewegung und Ortswechsel im Französischen: Die semantische Struktur intransitiver Fortbewegungsverben und ihre Kombinatorik mit lokalen und direktionalen Präpositionen

1. E i n l e i t u n g

Das Ziel der Arbeit ist die Untersuchung der sprachlichen Möglichkeiten des Französischen, einen Ortswechsel - im Sinne der Bewegung eines Individuums zu einem Ziel hin oder von einem Ausgangsort fort - auszudrücken, wie etwa in den folgenden Beispielen: (1)

a. Jean arrivera à la gare. 'Jean wird am Bahnhof ankommen.' b. Il est entré dans la maison. 'Er ist in das Haus hineingegangen.' c. Ce train part de Paris à huit heures du matin. 'Dieser Zug fährt um acht Uhr morgens aus Paris ab.' d. Jean est sorti en ville acheter son journal. 'Jean ist in die Stadt (hinaus-) gegangen, um seine Zeitung zu kaufen.' e. Il vient de sortir de la maison. 'Er ist gerade aus dem Haus gegangen.' f. Il a couru jusqu 'à la gare. 'Er ist bis zum Bahnhof gerannt.' g. L'enfant court à l'école.1 'Das Kind rennt zur Schule.'

Da entweder das Erreichen des Ziels bzw. Endpunkts der B e w e g u n g oder das Verlassen des Ausgangsortes impliziert ist, liegt hier nicht nur eine Fortbewegung im Sinne einer

1

Die angeführten Beispiele stammen überwiegend aus den gängigen Lexika oder sind aus anderen linguistischen Arbeiten zum Thema übernommen worden, insbesondere aus älteren, rein deskriptiv orientierten Arbeiten wie z.B. Bergh (1948), Doli (1967), Hajdù (1969), Welti (1969) und Stöcklin (1974), die umfangreiches Beispielmaerial enthalten, sowie auch aus Sablayrolles (1993) Sablayrolles & Asher (1995), Schlyter (1981) u.a. Alle angeführten Beispiele wurden zusätzlich von muttersprachlichen Informanten überprüft. Den französischen Beispielen wurden zur besseren Verständlichkeit der Argumentation auch für Leser, die mit dem Französischen weniger vertraut sind, die deutschen Übersetzungen hinzugefügt. Es ist jedoch zu beachten, daß diese häufig nur approximativ sein können, da Ortswechselprädikate im Deutschen nach einem anderen Lexikalisierungsmuster Zustandekommen und die darin auftretenden Bewegungsverben und lokalen Präpositionen typischerweise eine andere semantische Struktur haben als im Französischen.

132

Priska-Monika

Hottenroth

kontinuierlichen Ortsveränderung vor, sondern ein Ortswechsel im Sinne eines Übergangs von einer Region in eine andere. 2 Für den Ausdruck der Fortbewegung und des Ortswechsels stehen im Französischen sowohl transitive als auch intransitive Bewegungsverben zur Verfügung. Im folgenden sollen nur Konstruktionen mit intransitiven Verben betrachtet werden, d.h. es geht um Verben, die die Eigenbewegung eines Individuums kodieren und mit Präpositionalphrasen kombiniert werden können, die sich auf das Ziel oder den Ausgangsort der Bewegung beziehen,wie z.B. arriver, entrer, partir, sortir u.ä., aber auch courir, marcher etc. 3 Bei den meisten Bewegungsverben sind diese Präpositionalphrasen fakultativ, nur bei relativ wenigen obligatorisch, wie es z.B. bei aller, se rendre, parvenir (in lokaler Bedeutung) und einigen wenigen anderen der Fall zu sein scheint. Die letzteren sollen hier ausgeklammert werden, da sie eigene Probleme bieten. Gegenstand der Untersuchung ist die semantische Struktur der verschiedenen an Ortswechselprädikaten beteiligten Bewegungsverben sowie ihre Kombinatorik mit lokalen oder direktionalen Präpositionalphrasen, die das Ziel oder den Ausgangsort realisieren. Die zentrale Frage ist, ob sich die Restriktionen, die zwischen bestimmten Fortbewegungsverben und bestimmten Präpositionen bestehen, aus der Bedeutungsstruktur der Verben erklären lassen, und welche Bedeutungskomponenten es sind, die die Kombination mit statischen oder direktionalen Präpositionen in den Ziel- oder Ausgangsortangaben determinieren. Präpositionalphrasen, die sich auf den Zielort oder den Ausgangsort der Bewegung beziehen, haben bei Fortbewegungsverben wie arriver, partir u.ä. den Status von Argumenten. Es gibt jedoch auch Konstruktionen, in denen sie einen Zwischenstatus zwischen Argumenten und Adjunkten innezuhaben scheinen. Dies scheint vor allem bei Verben wie courir, marcher u.ä. der Fall zu sein. In jedem Fall enthalten die Verben Anforderungen an diese Präpositionalphrasen, und umgekehrt enthalten auch die Präpositionalphrasen Anforderungen an die Verben, mit denen sie kombiniert werden können. Im folgenden wird von der Grundannahme ausgegangen, daß die Verbbedeutung in sublexikalische Prädikat-Argumentstrukturen zerlegbar ist und daß durch diese die Argumentpositionen bereitgestellt werden, die beim Verb realisiert werden. Sie determinieren somit, welche Argumente beim Verb realisiert werden können und welche nicht, und sie legen deren thematische Rollen fest. 4 Somit muß aus der semantischen Struktur erklärbar sein, ob z.B. das Ziel oder der Ausgangsort oder beide realisiert werden können, und auch, ob die entsprechenden Präpositionalphrasen statisch lokalisierende oder direktionale Präpositionalphrasen sind. D.h. es sollen aus der semantischen Struktur der Verben z.B. die folgenden unterschiedlichen

2

Obwohl prinzipiell ein Ortswechsel auch durch Verben oder Verbalphrasen ausgedrückt werden kann, die eine Bewegung durch einen Ort oder Objekt (ganz) hindurch, über ihn (ganz) hinweg, um ihn ganz herum u.ä. spezifizieren, werden wir diese hier nicht mitbehandeln. Sie bleiben einer späteren Untersuchung vorbehalten.

3

Verben, die sowohl transitiv als auch intransitiv sind, werden in die Untersuchung miteinbezogen. Bzgl. dieser Grundannahmen und auch der Terminologie schließe ich mich weitgehend an die folgenden Arbeiten zum Deutschen an: Kaufmann (1995a, 1995b), Maienborn (1990, 1994, 1996), Rapp (1997), Wunderlich (1991), Wunderlich & Kaufmann (1990) und Wunderlich & Herweg (1991).

4

Fortbewegungsverben

und Ortswechsel im

Französischen

133

kombinatorischen Eigenschaften bzw. Restriktionen gegenüber Präpositionalphrasen erklärbar sein: - Bei Verben wie arriver, entrer, rentrer u.ä. sowie auch bei partir, sortir etc. kann die Präpositionalphrase, die sich auf das Ziel der Bewegung bezieht, außer mit der Präposition à auch mit dans, en, sur, sous, devant, derrière, chez etc. gebildet werden: (2)

a. Jean est arrivé sur la colline/dans le jardin/sous les arbres/devant la maison/chez le boulanger etc. 'Jean ist auf dem Hügel/im Garten/unter den Bäumen/vor dem Haus/beim Bäcker angekommen.' b. Il est entré dans la cuisine. 'Er ist in die Küche hineingegangen.' c. Il est sorti en ville acheter son journal. 'Er ist in die Stadt (hinaus-)gegangen, um eine Zeitung zu kaufen.' d. Il s'évada dans les champs. 'Er ist in die Felder geflohen.'

- Bei Verben wie courir, marcher, nager u.ä. können sich Präpositionalphrasen mit denselben Präpositionen dagegen im allgemeinen nicht auf das Ziel beziehen, sondern nur auf den Ort, an dem die Bewegung stattfindet. Sie sind dann Adjunkte, nicht Argumente des Verbs. Es ergibt sich kein Ortswechselprädikat. (3)

a. Les enfants courent dans le jardin. 'Die Kinder rennen im Garten umher.' b. Paul marche dans sa chambre (i.e. Paul marche de long en large dans sa chambre.) 'Paul läuft im Zimmer umher (d.h. Paul läuft im Zimmer hin und her).' c. La souris court sous l'armoire. 'Die Maus rennt unter dem Schrank umher.' d. Paul a nagé sous le pont. 'Paul ist unter der Brücke geschwommen.' e. Il a marche sur la terrasse. 'Er ist auf der Terrasse herumgelaufen.'

- Verben wie z.B. partir, sortir u.v.ä. aber auch arriver, venir, rentrer u.ä. können sich auch nur mit einer Präpositionalphrase verbinden, die sich auf den Ausgangsort der Bewegung bezieht: (4)

a. Ce train part de Paris à huit heures du matin. 'Dieser Zug fährt um acht Uhr morgens von Paris ab.' b. Le ballon est sorti du terrain de football. 'Der Ball ist vom Fussballfeld geflogen.' c. Il vient de chez soi. 'Er kommt von zu Hause.' d. Il est arrivé de Paris vers sept heures. 'Er ist gegen sieben Uhr aus Paris angekommen.' e. Paul rentre de l'université. 'Paul kommt von der Uni zurück.'

134

Priska-Monika

Hottenroth

- Verben wie courir, marcher, nager u.ä. lassen dies dagegen nicht zu: (5)

a. * Paul marche de l'université. '*Paul läuft von der Uni.' b. *L 'enfant a couru de l'école. 'Das Kind ist aus der Schule gerannt.'

Ortswechselprädikate können jedoch auch mit Fortbewegungsverben in Kombination mit direktionalem à oder jusqu'à gebildet werden. Hier zeigen sich jedoch interessante Beschränkungen. Die kombinatorischen Eigenschaften der Fortbewegungsverben in Ortswechselprädikaten bzgl. ihrer präpositionalen Komplemente korrelieren mit bestimmten zeitstrukturellen Eigenschaften (Aktionsarteigenschaften) der Verben wie Telizität und Atelizität. Es wird gezeigt werden, daß für die Kombinatorik mit jusqu'à nicht nur Eigenschaften wie die Telizität bzw. Atelizität relevant sind, sondern darüber hinaus Eigenschaften, die (in Vendlers 1967 Terminologie) accomplishments von achievements unterscheiden.

2. Das romanische und das germanische "Lexikalisierungsmuster"

Bei Fortbewegungsverben lassen sich zunächst zwei Klassen unterscheiden: (i) (ii)

Verben wie courir, marcher, nager, voler u.ä., die sich auf eine kontinuierliche Fortbewegung beziehen; Verben wie arriver, partir, entrer, sortir u.ä., die das Erreichen des Ziels oder das Verlassen des Ausgangsortes implizieren.

Talmy (1975, 1985) hat auf einen grundsätzlichen typologischen Unterschied zwischen den romanischen und den germanischen Sprachen aufmerksam gemacht, der die Struktur von Ortswechselprädikaten betrifft: In den germanischen Sprachen werden zum Ausdruck zielgerichter Bewegung typischerweise kontinuierliche Fortbewegungsverben verwendet, d.h. Verben, die nur den Prozeß der Fortbewegung und meist eine Spezifizierung des Bewegungsmodus enthalten. Das Ziel oder der Ausgangsort wird durch "Satelliten", d.h. direktionale Präpositionalphrasen, Partikel oder Präfixe u.a. ausgedrückt. Erst durch sie wird das Prädikat zu einem Ortswechselprädikat; siehe z.B. im Englischen (6) und ebenso im Deutschen (7): (6)

(7)

a. b. c. a. b. c.

He ran to the store. She limped out of the house. He rode away. Er rannte zum Geschäft. Sie hinkte aus dem Haus. Er ritt fort.

In den romanischen Sprachen werden dagegen für die Bewegung zu einem Ziel hin oder von einem Ausgangsort fort im allgemeinen Verben verwendet, die das Erreichen des Zielortes oder das Verlassen des Ausgangsortes, d.h. das Resultat der Bewegung in ihrer

Fortbewegungsverben

und Ortswechsel

im

Französischen

135

Bedeutung schon enthalten, wie es im Französischen bei den Verben arriver, partir, entrer, sortir u.ä. der Fall ist. Diese Verben enthalten damit selbst schon den Ortswechsel. Präpositionalphrasen (oder Adverbien) spezifizieren den Zielort oder den Ausgangsort nur zusätzlich. Die Art der Bewegung bleibt in den so gebildeten Ortswechselprädikaten in den romanischen Sprachen meist unspezifiziert. Sie kann, wenn sie ausgedrückt werden soll, nur durch hinzugesetzte, dem Basisverb subordinierte Gerundialkonstruktionen oder durch Präpositionalphrasen spezifiert werden. Das beinhaltet dann eine besondere Betonung des Bewegungsmodus, die in den englischen und deutschen Entsprechungen nicht gegeben ist (siehe dazu Slobin 1997). Vergi, im Französischen: (8)

a. Il arriva au magasin (en courant). Wörtl.: 'Er kam (laufend) im Geschäft an.' = 'Er lief ins Geschäft.' b. Elle est sorti de la maison (en clopinant). Wörtl.: 'Sie ist (hinkend) aus dem Haus gegangen.' = 'Sie hinkte aus dem Haus.' c. Il est parti (au galop). Wörtl.: 'Er ist (im Galopp) losgeritten.' = 'Er galoppierte los.'

Schon Bergh (1948), Green (1973) u.a. haben in diesem Zusammenhang darauf aufmerksam gemacht, daß den englischen Sätzen (9) nicht die französischen Sätze (10) entsprechen: (9)

(10)

a. b. c. a. b. c.

The bottle floated into the cave. He ran into the house. The mouse crawled under the table. La bouteille a flotté dans la grotte. Il a couru dans la maison. La souris a rampé sous la table.5

In den französischen Sätzen wird keine Bewegung zu einem Ziel ausgedrückt; die Präpositionalphrasen können nur die gesamte Situation lokalisieren, d.h. den Ort angeben, an dem die Bewegung stattfindet. Sie sind Adjunkte, nicht Argumente der Verben. Sie entsprechen somit den deutschen Sätzen (11): (11)

a. Die Flasche trieb in der Höhle. b. Er rannte im Haus umher. c. Die Maus kroch unter dem Tisch herum.

Dies erklärt sich jedoch zumindest nicht allein aus der Tatsache, daß die romanischen Sprachen das Lexikalisierungsmuster aus zielimplizierendem Verb und Präpositionalphrase bevorzugen, sondern zunächst einmal daraus, daß Ortswechselprädikate auch in den germanischen Sprachen mit kontinuierlichen Fortbewegungsverben nur gebildet werden können, indem eine direktionale Präpositionalphrase (bzw. ein direktionales Adverb, eine Partikel oder ein Präfix) hinzutritt. Die französischen Präpositionen dans, sur, sous etc. sind jedoch 5

Diese Eigenschaft der romanischen Sprachen korreliert mit dem Phänomen, daß sie generell resultative Konstruktionen aus einem Prozeßverb und z.B. einem Adjektiv wie in den englischen Sätzen He hammered the metal flat und He wiped the table clean nicht zulassen, sondern auch hierfür Verben, die den Resultatzustand beinhalten, d.h. Verben des Zustandswechsels gewählt werden müssen wie z.B. aplatir, nettoyer etc.

136

Priska-Monika Hottenroth

statisch lokalisierende Präpositionen. Das Französische besitzt keine Entsprechungen zu den englischen Präpositionen into, onto etc. bzw. den deutschen direktionalen Varianten in, auf, unter u.a. (mit dem Akkusativ). Aber es besitzt direktionale Präpositionen, z.B. à (das sowohl eine statische als auch eine direktionale Variante hat) und jusqu'à. Beide können sich durchaus mit kontinuierlichen Fortbewegungsverben zu Ortswechselprädikaten verbinden, worauf in einem späteren Abschnitt zurückzukommen sein wird. 6

3. Die semantische Struktur der zielimplizierenden Verben (GOAL-Verben) und die Angabe des Ziels

Verben, die nicht nur eine Fortbewegung, sondern auch das Erreichen des Ziels implizieren, sind im Französischen: arriver, venir, entrer, rentrer, retourner, revenir, accourir, affluer, refluer, pénétrer, s'introduire, s'infiltrer, s'enfoncer, s'engouffrer, s'enfouir, s'enfiler, s'engager (z.B. dans une rue), immigrer, plonger, s'immerger, couler (in der Bedeutung 'versinken'), sombrer, s'embarquer, atterrir, amerrir, emménager u.v.a.7 Auffallend ist, daß eine ganze Reihe dieser Verben Präfixverben sind. Obwohl ihre Bildungsmuster überwiegend nicht mehr produktiv zu sein scheinen, sind sie doch transparent, und es zeigt sich, daß der Verbstamm den Bewegungsmodus, das Präfix den Bezug

6

7

Talmy selbst hat darauf aufmerksam gemacht, daß es sich bei den Lexikalisierungsmustern eher um Tendenzen als um feste Regelmäßigkeiten handelt. Zudem ist die Tendenz zum romanischen Lexikalisierungsmuster in den verschiedenen romanischen Sprachen unterschiedlich ausgeprägt. Am deutlichsten scheint sie sich im Spanischen durchzusetzen. Die meisten Abweichungen davon zeigt das Italienische. Vom Französischen wird oft behauptet, daß es eine Zwischenstellung einnimmt. Bei einigen Verben wie émerger, surgir, déboucher, débarquer ist nicht ganz klar, ob sie eher einen Bezug zu einem Ausgangsort oder zu einem Zielort enthalten; ihre Präfixe weisen auf einen Bezug zu einem Ausgangsort; dieser bleibt aber in den meisten Kontexten unspezifiziert, wohingegen der Zielort als Ort des Sprechers oder im Kontext erwähnter Ort rekonstruiert wird. Das würde dafür sprechen, sie zu den zielimplizierenden Verben zu zählen. Vergi.: Tout à coup, il débouche d'une petite rue. 'Plötzlich taucht er aus einer Gasse auf.' Ils ont débarqué à l'improviste (chez nous). 'Sie sind unerwartet (bei uns zu Hause) eingetroffen.' Mais, au loin, un homme surgit. 'Aber in der Ferne erschien ein Mann.' Il émergea brusquement (de l'ombre). 'Er tauchte unversehens aus dem Schatten auf.' Zu Verben wie monter, descendre u.ä., die eine zielimplizierende Variante haben, siehe Punkt 6.

Fortbewegungsverben

und Ortswechsel im Französischen

137

zum Zielort enthält.8 Damit weichen diese Verben genaugenommen bereits vom typisch romanischen Lexikalisierungsmuster ab. Daß die genannten Verben in ihrer Bedeutung bereits einen Bezug zum Zielort enthalten, zeigt sich daran, daß das Ziel - auch ohne daß es durch eine Präpositionalphrase realisiert wird - mitverstanden wird und automatisch aus dem Kontext, d.h. deiktisch oder anaphorisch rekonstruiert wird (siehe dazu Maienborn 1990: 57ff.). Vergi.: (12)

a. A ce moment, il arriva quatre personnes de plus. 'In diesem Augenblick kamen vier weitere Personen an.' b. Entrez donc! 'Treten Sie doch ein!' c. Je viens tout de suite. 'Ich komme sofort.' d. Il est venu nous voir. Wörtl.: 'Er ist gekommen, um uns zu sehen.' = 'Er ist uns besuchen gekommen. 1 e. Il η 'est pas encore revenu. 'Er ist noch nicht zurückgekommen.' f. Dès l'ouverture du magasin, la foule affluait. 'Kurz nach der Eröffnung des Geschäfts strömte die Menge herbei.'

Im Fall von arriver, accourir, affluer u.ä. ist der im Verb implizierte Zielort relativ unspezifiziert. Die meisten Verben dieser Klasse enthalten jedoch Spezifizierungen des Zielorts; z.B. ist es bei venir der Ort des Sprechers oder dessen Nahbereich, bei retourner, revenir der Ausgangsort einer vorherigen Bewegung, bei entrer, s'introduire u.v.a. ein Innenraum oder eine Binnenregion etc. Einige Verben wie z.B. plonger, s'embarquer, atterrir, amerrir u.ä. enthalten einen so weitgehend spezifizierten Zielort, daß eine zusätzliche kontextuelle Spezifizierung nicht notwendig erfolgen muß. Vergi.: (13)

a. Le sous-marin plonge. 'Das U-Boot taucht unter.' b. L'avion atterrit à dix heures. 'Das Flugzeug landet um zehn Uhr.'

Zielimplizierende Verben sind, da ihnen der Endpunkt der Bewegung inhärent ist, telisch. Das unterscheidet sie von den kontinuierlichen Fortbewegungsverben, die einen Prozeß bezeichnen, ohne daß irgendein Bezug auf dessen Begrenzungen in der Verbbedeutung enthalten ist, und die daher atelisch sind. Da diese Verben das Erreichen des Zielorts implizieren und der Zielort - auch ohne hinzutretende Präpositionalphrase - mitverstanden und aus dem Kontext rekonstruiert wird, ist anzunehmen, daß er als Argument in der semantischen Struktur dieser Verben verankert ist.

8

Das Präfix ist als inkorporierte Präposition des Zielarguments anzusehen; siehe dazu Stiebeis

(1991).

138

Priska-Monika Hottenroth

3.1. Ziel-Verben und statische Präpositionalphrasen Im allgemeinen verbinden sich diese Verben aber mit einer Präpositionalphrase, die sich auf das Ziel der Bewegung bezieht. 9 Da das Ziel bereits in der Verbbedeutung angelegt ist, realisiert diese Präpositionalphrase dieses Argument und spezifiziert es. Die Frage ist nun, warum das Ziel-Argument bei diesen Verben nicht durch eine direktionale, sondern durch eine statische Präpositionalphrase realisiert wird, also durch Präpositionalphrasen, die mit dans, sur, sous, devant, derrière, chez u.ä. gebildet sind. Siehe z.B.: (14) a. Jean arrivera dans le jardin/sur la colline/sous les arbres/devant la maison/chez le boulanger. 'Jean wird im Garten/auf dem Hügel/unter den Bäumen/vor dem Haus/beim Bäcker ankommen.' b. Il est rentré chez lui. 'Er ist wieder nach Hause gegangen.' c. Il entre dans sa chambre. 'Er geht in das Zimmer hinein.' d. Toute la population est accourue sur la place centrale. 'Alle Leute strömten auf den zentralen Platz.' e. Le voleur s 'est introduit dans la maison par la fenêtre. 'Der Dieb ist durch das Fenster ins Haus hereingekommen.' f. Nous nous enfonçons dans la neige. 'Wir sinken in den Schnee ein.' g. Il s'enfouit sous ses draps. 'Er vergräbt sich unter seinen Decken.' h. Nous avons embarqué sur le "France" pour faire une croisière. 'Wir sind an Bord der "France" gegangen, um eine Kreuzfahrt zu machen.' Allerdings kann auch die Präposition à im Zielargument erscheinen. (15) a. Jean arrive à la gare. 'Jean kommt am Bahnhof an.' b. Nous sommes entrés à la cuisine. 'Wir sind in die Küche gegangen.' c. Il est rentré à la maison. 'Er ist wieder nach Hause gegangen.' Die Präposition à besitzt jedoch eine statische und eine direktionale Variante, so daß entsprechend zu den obigen Beispielen (14) - anzunehmen ist, daß hier die statische Variante vorliegt. Die Tatsache also, daß Verben wie arriver, entrer etc. das Zielargument durch eine statisch lokalisierende und nicht durch eine direktionale Präpositionalphrase realisieren, erklärt sich daraus, daß diese Verben einen Ortswechsel ausdrücken, d.h. das sich bewegende Individuum befindet sich vor dem Ereignis des Ortswechsels nicht am Zielort, nach dem Ereignis aber am Zielort.

9

An die Stelle der Präpositionalphrasen können lokale Adverbien treten, worauf in diesem Rahmen nicht weiter eingegangen wird.

Fortbewegungsverben

und Ortswechsel

im

Französischen

139

Sie gehören damit in die Klasse der Zustandswechselverben (CHANGE-Verben), die in ihrer semantischen Struktur den Übergang von einem vor dem Ereignis gegebenen Zustand in den Nachzustand enthalten. Bei den zielimplizierenden Fortbewegungsverben ist der Resultatzustand der Bewegung die Lokalisation am Zielort, somit eine statische Relation. Für diese Verben ist die folgende semantische Struktur anzunehmen: CHANGE

(Loe

( x , REGION

(y)))10

Das sublexikalische Prädikat CHANGE repräsentiert den Zustandswechsel, LOC den Resultatzustand. Dazu tritt bei Verben wie accourir, affluer u.a. eine Komponente, die den Bewegungsmodus repräsentiert. Das im Verb angelegte Argument hat nach dieser Analyse, die im wesentlichen auf die von Jackendoff (1983) fur englische Verben wie to arrive vorgeschlagene zurückgeht, die thematische Rolle PLACE, nicht GOAL. Entsprechend ist die dieses Argument spezifizierende Präpositionalphrase statisch und nicht direktional. Die Kombination zielimplizierender Verben mit statischen Präpositionen ist prinzipiell nicht eingeschränkt. Restriktionen einzelner Verben gegenüber bestimmten Präpositionen können sich aber durch die Inkompatibilität einzelner Präpositionen mit den im Verb enthaltenen weitergehenden Spezifizierungen des Zielorts ergeben. Vergi.: ( 16)

*// entre devant la maison. '*Er geht vor dem/das Haus hinein.'

In metaphorischen Verwendungen, von denen viele lexikalisiert sind, können solche Beschränkungen jedoch auch aufgehoben sein: (17)

a. Le gardien plongea sur le balcon. 'Der Wächter stürzte auf den Balkon.' b. L'avion plongea sur l'objectif. 'Das Flugzeug stürzte auf das Ziel zu.'

3.2. Zielimplizierende Verben und räumliches, direktionales jusqu 'à Zwar wird das Zielargument bei den zielimplizierenden Verben generell durch eine statische lokale Präpositionalphrase realisiert. Einige dieser Verben können jedoch - wenn auch relativ selten - in Kombination mit der direktionalen Präposition jusqu 'à auftreten. (18)

10

a. La lumière de certaines étoiles η 'est pas encore arrivée jusqu 'à nous. 'Das Licht von bestimmten Sternen ist noch nicht bis zu uns gekommen.' b. Nous étions arrivés jusqu 'à la gare, lorsque la voiture nous lâcha. 'Wir waren gerade am Bahnhof angekommen, als das Auto uns absetzte.' c. II est venu jusqu 'à la porte, puis il est reparti. 'Er ist bis an die Tür gekommen und dann wieder weggegangen.'

Ähnlich im Anschluß an die Generativen Semantiker Jackendoff (1983), Levin & Rapoport (1988: 284 f.), Levin & Rappaport Hovav (1992 ,1995) u.v.a..

140

Priska-Monika

Hottenroth

d. Il est retourné jusqu 'à l'école. 'Er ist bis zur Schule zurückgegangen.' e. La foule reflua jusqu 'à la porte de secours. 'Die Menge kehrte bis zum Notausgang zurück.' f. Il accourut jusqu 'au lieu du crime. 'Er rannte bis zum Ort des Verbrechens.'

Auffallend ist nun, daß die jusqu 'à-PP in diesen Konstruktionen das im Verb angelegte Zielargument zu ersetzen scheint. Denn es müßte sonst - da es anders nicht realisiert ist mitverstanden, d.h. kontextuell rekonstruiert werden. Dies ist jedoch hier gerade nicht der Fall. Die Frage ist also: Tritt die jusqu 'à-PP hier an die Stelle des Zielarguments? Zunächst ist zu beobachten, daß eine jusqu 'à-PP nicht bei allen zielimplizierenden Verben die Wirkung hat, daß das im Verb angelegte Zielargument nicht zusätzlich kontextuell rekonstruiert wird. Bei vielen zielimplizierenden Verben bleibt - unabhängig vom Vorhandensein der jusqu 'à-PP und zusätzlich zu ihr - das im Verb angelegte Zielargument implizit erhalten, d.h. es wird mitverstanden und kontextuell rekonstruiert, so z.B. bei entrer, rentrer, s'introduire, s'immerger u.ä. Dies ist vor allem bei den Verben der Fall, die eine weitergehende Spezifizierung des Zielortes enthalten. Vergi.: (19)

a. Il est entré jusqu 'à l'antichambre. 'Er ist bis zum Vorzimmer hineingegangen.' b. Le voleur s 'est introduit jusqu 'au bureau du directeur. 'Der Dieb ist bis zum Büro des Chefs eingedrungen.' c. Il s 'est immergé jusqu 'aux épaules. 'Er ist bis an die Schultern eingetaucht.' d. Il s'enfonce jusqu 'aux genoux. 'Er sinkt bis an die Knie ein.'

Die Präpositionalphrase mit jusqu'à ersetzt hier nicht das Zielargument, sondern fuhrt zusätzlich einen Endpunkt der Bewegung ein, der entweder im Innern des (impliziten und mitverstandenen) Zielortes liegt oder aber sich auf den Körper des sich bewegenden Individuums selbst bezieht. So entspricht (20a) etwa (20b) und (21a) etwa (21b): (20)

a. Il entre jusqu 'à l'antichambre. 'Er geht bis zum Vorzimmer hinein.' b. Il entre dans la maison jusqu 'à l'antichambre. 'Er geht bis zum Vorzimmer in das Haus hinein.'

(21)

a. Le voleur s 'est introduit jusqu 'au bureau du directeur. 'Der Dieb ist bis zum Büro des Chefs eingedrungen.' b. Le voleur s 'est introduit dans le bâtiment jusqu 'au bureau du directeur. 'Der Dieb ist bis zum Büro des Chefs in das Gebäude eingedrungen.'

Die Frage ist nun, was zu diesem unterschiedlichen Verhalten der Verben jusqu 'à gegenüber fuhrt. Direktionale Präpositionen sind im Unterschied zu statischen Präpositionen wegbezogen. Die direktionale Variante der Präposition à wie auch jusqu 'à enthält einen Bezug zum Endpunkt des Weges. Jusqu 'à enthält sowohl einen dynamischen als auch einen statischen Bedeutungsanteil, was sich deutlich darin zeigt, daß an die Stelle der eingebetteten Präposition à andere statische Präpositionen treten können. Vergi.:

Fortbewegungsverben (22)

und Ortswechsel im

Französischen

141

a. Ils sont venus jusque devant la porte. 'Sie sind bis vor die Tür gekommen.' b. Ils sont retournés jusque chez lui. 'Sie sind bis zu seinem Haus/seiner Wohnung zurückgegangen.'

Von à unterscheidet sich jusqu 'à dadurch, daß die Region, die in einer mit jusqu 'à gebildeten Präpositionalphrase spezifiziert wird, nicht selbst Zielort ist, sondern die Grenze bereitstellt, an der die Bewegung endet. Anders als bei à scheint bei jusqu 'à der begrenzte Weg selbst in den Fokus zu rücken, und nicht der Zielort als Wegende. Da jusqu 'à den Weg selbst fokussiert und ihn begrenzt, verlangt es von einem Verb, mit dem es kombiniert wird, daß es eine noch nicht abgeschlossene Bewegung als Komponente enthält, wie dies in kontinuierlichen Fortbewegungsverben gegeben ist. Verben wie arriver, venir, accourir, affluer etc., bei denen die jusqu 'à-PP das im Verb angelegte Zielargument verdrängt, und Verben wie entrer, s'introduire, s'enfouir u.ä., bei denen dies nicht der Fall ist, unterscheiden sich nicht nur dadurch, daß die Verben der letzten Gruppe eine weitergehende Spezifizierung des Zielortes enthalten, sondern diese Eigenschaft geht mit unterschiedlichen Aktionsarteigenschaften, d.h. einer unterschiedlichen zeitlichen Ereignisstruktur einher, und dies ist hier entscheidend. Zwar sind beide Verbgruppen telisch, aber Verben wie venir, accourir, affluer, refluer u.ä. sind in Vendlers Terminologie accomplishments, arriver kann zumindest eine accomplishment-Interpretation haben, während entrer, s'immerger, s'introduire u.ä. nach Vendlers Klassifikation achievements sind. D.h. letztere sind bloße Übergänge von einem Vorzustand in einen Resultatzustand, d.h. von einer Lokalisation in eine andere, während erstere sich aus einem Prozeß, der dem Zustandswechsel vorausgeht und auf ihn hinausläuft, und dem Resultatzusstand zusammensetzen. Nicht nur Übergang und Nachzustand, sondern auch der vorhergehende Prozeß ist als Teilereignis, d.h. als sublexikalisches Prädikat in der Verbsemantik gegeben. In Anlehnung an Pustejovsky (1995: 73ff.) ist die folgende Ereignisstruktur für einfache Übergänge (nach Vendler achievements) anzunehmen: CHANGE STATE 1

STATE 2

Accomplishments haben dagegen die folgende Struktur: CHANGE PROCESS

STATE

Nicht zufällig gehören zu den accomplishment-Verben vor allem auch die Verben, die nicht nur die Bewegungskomponente selbst, sondern auch eine Spezifizierung des Bewegungsmodus enthalten. Die Bedingung dafür, daß eine jusqu a-PP das Zielargument verdrängen kann, ist offensichtlich, daß das Verb eine accomplishment-Struktur aufweist, weil jusqu'à die Prozeßkomponente erfordert.

142

Priska-Monika

Hottenroth

Dies genügt jedoch für seine Anbindung nicht, da bei accomplishment-Verben ein inhärent begrenzter Prozeß vorliegt. Der im Verb enthaltene, also inhärente Endpunkt der Bewegung, d.h. die Komponente "Resultatzustand der Bewegung" wäre inkompatibel mit der fokussierten Wegkomponente in jusqu 'à. Dowty (1979: 60f.) u.a. haben aufgezeigt, daß viele accomplis hment-V e rben in bestimmten Kontexten auch eine atelische (Um)Interpretation als activities, d.h. als nicht inhärent begrenzte Prozesse annehmen können. Sie drücken dann Aktivitäten oder Prozesse aus, die zwar auf das Erreichen des Ziels ausgerichtet sind, aber das Erreichen des Ziels nicht implizieren. Er zeigt dies an Beispielen wie build a house, write a letter, draw a circle u.ä. Das Ziel wird in der acímYy-Interpretation zum intendierten Ziel, und der (noch nicht abgeschlossene) Prozeß rückt in den Vordergrund. Solche Interpretationen liegen z.B. vor in Kombinationen mit dem Progressiv wie z.B. in (23) sowie auch in (24): (23)

a. Frank is building a house. b. John is drawing a circle. (24) John has been writing a book for two years.

Auch die Kombination mit jusqu'à scheint bei accomplishment-Verben eine solche Uminterpretation zu bewirken. Durch sie wird das im Verb angelegte Zielargument unterdrückt; es bleibt allein die Komponente, die die kontinuierliche Fortbewegung (in Richtung auf ein intendiertes Ziel) repräsentiert. Das erklärt, warum sowohl die Realisierung als auch die kontextuelle Rekonstruktion des im Verb angelegten Zielarguments blockiert ist. Jusqu 'à fuhrt somit relativ unabhängig von diesem ursprünglich im Verb angelegten und nun blockierten Zielargument eine Begrenzung des Bewegungsprozesses ein. Die größere Komplexität der Interpretation durch die durch jusqu 'à induzierte Uminterpretation erklärt zugleich das eher seltene Vorkommen von jusqu 'à mit Orts wechselverben sowie auch die schwankende Akzeptanz mancher Beispiele durch die Informanten.

3.3. Zielimplizierenden Verben und der Ausgangsort der Bewegung Viele zielimplizierende Verben können sich auch mit einem Argument verbinden, das sich auf den Ausgangsort der Bewegung bezieht, wobei im allgemeinen das im Verb angelegte Zielargument nicht realisiert wird, sondern implizit bleibt und aus dem Kontext rekonstruiert wird. Vergi.: (25)

a. Il est arrivé de Paris à cinq heures. 'Er ist um fünf Uhr aus Paris angekommen.' b. Ce train vient de Bordeaux. 'Dieser Zug kommt aus Bordeaux.' c. Paul est rentré de l'université. 'Paul ist von der Uni zurückgekommen.' d. Il revient du bureau. 'Er kommt aus dem Büro zurück.'

Da das Argument, das sich auf den Ausgangsort bezieht, bei diesen zielimplizierenden Verben durch eine Präpositionalphrase realisiert werden kann, ist anzunehmen, daß es ebenfalls in ihrer semantischen Struktur verankert ist, wenn auch auf andere Weise als der

Fortbewegungsverben

und Ortswechsel im Französischen

143

Zielort. Denn wenn das Ausgangsort-Argument fehlt, wird es - anders als das ZielArgument - in diesem Fall nicht mitverstanden und nicht kontextuell rekonstruiert. Aber zielimplizierende Verben enthalten einen Ortswechsel in die Zielregion. Dies impliziert einen Ausgangsort, der als Komplementregion zum Zielort in der Bedeutungsstruktur enthalten ist (dazu siehe Maienbom 1990: 89ff.). Realisiert wird dieses Argument allerdings bei den intransitiven Bewegungsverben nicht durch eine statische, sondern durch eine direktionale PP. Das bedeutet, daß dieses Argument nicht als Vorzustand, nicht als statische Lokalisation vor dem Ereignis des Ortswechsels, sondern von vornherein als wegbezogenes Argument in der semantischen Struktur angelegt ist. Jedoch ist die alleinige Realisierung des Ausgangsort-Arguments nicht bei allen zielimplizierenden Verben möglich. Es scheinen bei diesen Restriktionen ähnliche ereignisstrukturelle Eigenschaften eine Rolle zu spielen wie bei der Kombinatorik der zielimplizierenden Verben mit jusqu'à. Eine Präpositionalphrase, die den Ausgangsort spezifiziert, kann zu Verben wie arriver, venir, accourir, affluer etc., d.h. zu den accomplis hment-Verben treten; dagegen eher nicht zu achievement-Verben, die einen bloßen Übergang in eine Zielregion ausdrücken wie entrer, s'introduire, s'enfoncer, s'enfiler, s'immerger, embarquer, atterrir etc. Das heißt, daß auch fur die Realisierung des Ausgangsort-Arguments bei zielimplizierenden Verben das Vorhandensein der Bewegungskomponente, d.h. eines prozeßhaften Teilereignisses, das dem Ortswechsel vorausgeht und in ihm kulminiert, Bedingung zu sein scheint.

4. Die semantische Struktur der Verben, die das Verlassen des Ausgangsortes implizieren (SOURCE-Verben)

Zu den Verben, die das Verlassen des Ausgangsortes implizieren, gehören im Französischen z.B. partir, s'en aller, s'envoler, s'enfuir, s'échapper, s'évader, s'absenter, s'élancer (in der Bedeutung: 'losstürzen'), se retirer sowie die eher umgangssprachlichen bis familiären Verben filer, se sauver, s'ôter, déguerpir, décamper, se barrer, se défder, détaler u.ä., ebenso émigrer, déserter, déloger, démenager, débarquer, décoller, émerger, surgir, déboucher,11 jaillir, gicler, s'écouler u.v.a.

Auch in dieser Verbgruppe finden sich viele Verben, die nicht nur den Ortswechsel in ihrer semantischen Struktur enthalten, sondern auch eine Spezifizierung des Bewegungsmodus. Dies gilt wiederum besonders für die Präfixverben. Das Argument fur den Ausgangsort ist in der Bedeutungsstruktur dieser Verben ebenso angelegt wie der Zielort bei den zielimplizierenden Verben; denn, wenn es nicht durch eine Präpositionalphrase realisiert wird, wird es dennoch mitverstanden und aus dem Kontext rekonstruiert. Vergi.:

11

Siehe o. Fn. 7.

144 (26)

Priska-Monika

Hottenroth

a. Il est parti à cinq heures. 'Er ist um f u n f U h r aufgebrochen.' b. Jean vient de sortir. 'Jean ist gerade hinausgegangen.' c. Pierre s'en va sans rien dire. 'Pierre geht weg, ohne etwas zu sagen.' d. Il a filé à l'anglaise. 'Er hat sich verdünnisiert.' e. Décampe et tout de suite. 'Scher dich weg.'

4.1. Die Realisierung des Ausgangsortes Das im Verb angelegte Argument für den Ausgangsort kann - anders als das Zielargument - nur durch eine direktionale, mit der Präposition de gebildete Präpositionalphrase realisiert werden. (27)

a. Jean sortit de la maison. 'Jean ging aus dem Haus.' b. Il fila de sa cachette. 'Er rannte aus seinem Versteck.' c. Il s 'est barré de l université. 'Er ist von der Uni abgehauen.' d. Il débarqua de I 'avion. 'Er stieg aus dem Flugzeug aus.' e. L'oiseau s 'est envolé de sa cage. 'Der Vogel ist aus seinem Käfig geflogen.' f. Le voleur est échappé de la prison. 'Der Dieb ist aus dem Gefängnis entkommen.' g. Il s'extrait avec peine de la voiture. 'Er windet sich mühsam aus dem Auto heraus.' h. Le chat sortit de sous le lit/de derrière les arbustes etc. 'Die Katze kam unter dem Bett/hinter dem Gesträuch hervor.' i. J'ai déménagé de chez mes parents. 'Ich bin bei meinen Eltern ausgezogen.' (wörtl.: 'von (bei) meinen Eltern')

Anders als das Ziel, das als Resultatzustand der Bewegung durch eine statisch lokalisierende Präpositionalphrase realisiert wird, wird der Ausgangsort nicht als statische Lokalisation im Sinne eines Vorzustands, sondern bei intransitiven Bewegungsverben direktional, d.h. wegbezogen realisiert.12 Es ist nicht ganz befriedigend, wenn dies in der semantischen Strukturformel durch die Negation der Ausgangsortkomponente repräsentiert ist, wie es häufig vorgeschlagen wird: C H A N G E (ILOC (X,

12

REGION(y)))

Einige kausative Bewegungsverben verhalten sich in diesem Punkt anders, so z.B. prendre qc. dans un armoire 'etwas aus einem Schrank nehmen', boire dans un verre 'aus einem Glas trinken' u.ä.

Fortbewegungsverben

und Ortswechsel im Französischen

145

Genaugenommen repräsentiert hier die negierte Komponente die Zielregion, d.h. die Komplementregion des Ausgangsortes, was der Tatsache widerspricht, daß in diesen Verben der Ausgangsort und eben nicht der Zielort fokussiert ist. CHANGE repräsentiert hier wiederum den Zustandswechsel. Die Verben enthalten keinen Bezug zu einer dem Ortswechsel vorausgehenden und in ihm kulminierenden Bewegung, d.h. sie haben ach levem en /-Struktur, nicht accomplishment- Struktur. Sie drücken direkte Übergänge aus; ihnen fehlt die Prozeßkomponente.

4.2. Ziel bei ausgangsortimplizierenden Verben Wie die zielimplizierenden Verben auch das Argument, das den Ausgangsort der Bewegung angibt, binden können, können umgekehrt viele der Verben, die das Verlassen des Ausgangsortes implizieren, auch nur ein Zielargument bei sich haben; siehe z.B.: (28)

a. Il sortit dans le jardin. 'Er ging in den Garten hinaus.' b. Il s 'est sauvé dans la cave. 'Er hat sich in den Keller gerettet.' c. Il se retira chez soi. 'Er zog sich in sein Zimmer/seine Wohnung zurück.' d. Ils ont débarqué sur le quai. 'Sie sind am Kai ausgestiegen.' (wörtl.: 'auf den Kai') e. Ils se sont enfuis dans la forêt. 'Sie sind in den Wald geflüchtet.' f. Il s'évada dans les champs. 'Er ist in die Felder geflohen.' 13 g· Ils s'en allèrent dans le désert. 'Sie gingen (fort) in die Wüste.'

Das Ziel scheint also auch bei diesen Verben als unspezifizierte Komplementregion zum Ausgangsort in der semantischen Struktur enthalten zu sein. Da es Resultatzustand der Bewegung ist, wird es durch eine statisch lokalisierende Präpositionalphrase realisiert. Dies ist jedoch nicht bei allen Verben möglich, die den Ausgangsort enthalten, z.B. eher nicht bei partir, filer, décamper, déguerpir, n u r e i n g e s c h r ä n k t bei émigrer (z.B. in émigrer à l'étranger, émigrer en Amérique) nicht bei décoller, se détacher, s'écarter, se dégager, s'ôter u.ä.

Die Bedingungen sind nicht ganz klar. Anscheinend steht bei einigen Verben das bloße Verlassen des Ausgangsortes im Vordergrund, d.h. dies allein ist fokussiert; die Komplementregion ist dagegen so weitgehend defokussiert, daß sie nicht durch eine Präpositionalphrase realisiert bzw. spezifiziert werden kann.

13

Eine Präpositionalphrase mit pour, wie sie vor allem bei partir erscheinen kann, bezieht sich nicht auf das Ziel, das durch die Bewegung erreicht wird, sondern nur auf ein intendiertes Ziel. Vergi.: Il part pour Paris/pour la campagne 'Er zieht nach Paris/aufs Land', etc. Da sie nicht das Erreichen des Ziels beinhalten, werden diese Konstruktionen hier nicht mitbehandelt.

146

Priska-Monika

Hottenroth

5. Ortswechsel mit kontinuierlichen Fortbewegungsverben

5.1. Die semantische Struktur der kontinuierlichen Fortbewegungsverben und ihre kombinatorischen Eigenschaften - Unterschiede zu den Ortswechselverben Die wichtigsten kombinatorischen Eigenschaften der Ortswechselverben, d.h. sowohl der zielimplizierenden Verben als auch der Verben, die das Verlassen des Ausgangsortes beinhalten, sind somit die folgenden: - Die zielortspezifizierenden Argumente sind, wenn sie realisiert werden, im allgemeinen statisch lokale Präpositionalphrasen. - Nicht nur die den Ausgangsort implizierenden Verben, sondern auch viele der zielimplizierenden Verben können sich mit einer Präpositionalphrase verbinden, die den Ausgangsort realisiert bzw. spezifiziert, ohne daß zugleich das Zielargument realisiert werden muß. Mit diesen kombinatorischen Eigenschaften stehen die Verben, die das Erreichen des Zielortes bzw. das Verlassen des Ausgangsortes implizieren, im Gegensatz zu der zweiten großen Klasse von Fortbewegungsverben, den kontinuierlichen Fortbewegungsverben. Zu diesen Verben gehören z.B. im Französischen: courir, marcher, cheminer, se promener, flâner, se balader, déambuler, trotter, se trainer, trottiner, ramper, sautiller, clopiner, claudiquer, boiter, errer, rôder, circuler, vagabonder, voyager, chevaucher, galoper, pédaler, patiner, patauger, skier, naviguer, voler, nager, flotter, rouler u.v.a.

Zu dieser Klasse sind auch die reinen Richtungsverben wie avancer, proceder, progresser, s'approcher, se rapprocher u.ä. zu zählen. 14 Diese Verben enthalten in ihrer Bedeutungsstruktur nur die Komponente der kontinuierlichen Ortsveränderung, d.h. MOVE (χ), das den Prozeß der Fortbewegung selbst repräsentiert, sowie zusätzlich im allgemeinen eine Spezifizierung des Bewegungsmodus, des Mediums, Instruments der Fortbewegung u.ä. Sie enthalten jedoch im Unterschied zu den Ortswechselverben weder einen Bezug zum Zielort noch zum Ausgangsort und damit keine inhärente zeitliche und räumliche Begrenzung des Fortbwegungsprozesses; sie sind atelisch und gehören nach Vendlers Klassifikation zu den activities. Da sie keinen Ortswechsel beinhalten und somit ein Zielort nicht als Resultatzustand in der Verbbedeutung verankert ist, können sie als Zielangaben nicht statisch lokalisierende Präpositionalphrasen bei sich haben. Wenn statisch lokalisierende PP auftreten, sind sie nicht Argument, sondern Adjunkt und lokalisieren den gesamten Vorgang, d.h. sie geben den Ort an, an dem die Bewegung stattfindet; siehe z.B.:

14

Zur Ambiguität der letzteren und zu weiteren Richtungsverben wie monter, descendre ebenfalls ambig sind, siehe u. Punkt 6.1.

u.ä., die

Fortbewegungsverben

(29)

und Ortswechsel

im

Französischen

147

a. Les enfants courent dans le jardin. 'Die Kinder laufen im Garten umher.' b. Les particules de graisse flottaient sur le bouillon. 'Die Fettaugen schwammen auf der Bouillon.' c. J'aime flâner sur les grands boulevards. 'Ich schlendere gerne über die großen Boulevards.' (wörtl.: 'auf den Boulevards') d. On circule mal dans cette ville. 'In dieser Stadt läßt sich schwer fahren.'

Von dieser Regel weicht allerdings eine kleine Gruppe von Verben ab, auf die wir in Punkt 6.2 zurückkommen. Kontinuierliche Fortbewegungsverben können sich auch nicht mit einer Präpositionalphrase, die den Ausgangsort realisiert, verbinden, ohne daß das Ziel ebenfalls realisiert wird. Auch in diesem Punkt verhalten sie sich anders als die Ortswechselverben: (30)

a. *Jean marche de l'université. 'Jean läuft von der Universität weg/aus der Universität.' b. *Il a couru de la gare. 'Er ist vom Bahnhof weggerannt.' c. *Le merle a volé de l'arbre. 'Die Amsel ist aus dem Baum geflogen.' d. *Une vieille dame trottait de sa maison. 'Eine alte Frau trippelte aus ihrem Haus.'

Die Realisierung nur des Ausgangsortes bei einem Bewegungsverb scheint also davon abhängig zu sein, daß es entweder im Verb angelegt ist oder implizit als Komplementregion des im Verb angelegten Zielortarguments zur Verfugung steht, wie es bei arriver, venir u.a. der Fall ist.

5.2. Ortswechselprädikate aus kontinuierlichen Fortbewegungsverben und direktionalen Präpositionalphrasen - das eigentlich germanische Lexikalisierungsmuster Auch im Französischen können Ortswechselprädikate aus einem kontinuierlichen Fortbewegungsverb und einer direktionalen Präpositionalphrase gebildet werden. Hierfür stehen die direktionalen Präpositionen à, jusqu 'à und das Präpositionenpaar de - à zur Verfügung. (31)

SL. Il a couru à la poste. 'Er ist zur Post gerannt.' b. Il s'est traîné à la porte. 'Er hat sich zur Tür geschleppt.' c. Il a couru jusqu 'à l'église. 'Er ist bis zur Kirche gerannt.' d. Il a marché jusqu 'au sommet. 'Er ist bis zum Gipfel gelaufen.' e. Jean aime flâner jusqu 'aux quais de la Seine. 'Jean schlendert gerne bis an die Quais der Seine.'

148

Priska-Monika

Hottenroth

f. II avance jusqu 'à la porte. 'Er tritt bis an die Tür vor.' g. // clopina jusqu 'à la sortie. 'Er hinkte bis zum Ausgang.' h. Ils ont marché de Paris à Marseille. 'Sie sind von Paris nach Marseille gelaufen.' i. Il flâna du centre-ville au centre commercial. 'Er bummelte von der Innenstadt zum Einkaufszentrum.' j. Il se traîna du lit à la fenêtre. 'Er schleppte sich vom Bett bis zum Fenster.' k. Il clopina de la salle de bain à la salle à manger.15 'Er hinkte vom Badezimmer ins Esszimmer.'

Durch die Angabe der Wegbegrenzungen, besonders aber des Ziels wird der Bewegungsprozeß begrenzt, und das Prädikat wird zum Ortswechselprädikat. Die so zustandegekommenen Ortswechselprädikate haben accomplishment-StiuktuT. Für sie ist daher ebenfalls die Struktur anzunehmen, die Ortswechselverben schon im Verb selbst enthalten: CHANGE

PROCESS* (=MOVE ( x ) )

STATE ( = LOC ( x , REGION ( y ) ) )

Im Unterschied zu arriver, aber ähnlich wie bei accourir, affluer, venir u.ä. ist hier die Prozeßkomponente das fokussierte Teilereignis (siehe dazu Pustojevsky 1991: 72ff.). Das bedeutet, daß die MOVE-Komponente des Verbs hier unter das CHANGE-Prädikat subordiniert wird. Levin & Rapoport (1988) sowie auch Pustejovsky (1995) sprechen in diesem Zusammenhang von "lexikalischer Subordination" und nehmen eine Veränderung der Semantik des kontinuierlichen Bewegungsverbs an. Diese Kombinationsmöglichkeiten scheinen allerdings im Französischen stark restringiert zu sein. Nur relativ wenige Verben lassen sie zu. Am stärksten sind bei den kontinuierlichen Fortbewegungsverben die Beschränkungen gegenüber einer à-PP als Zielangabe. Zwar erlauben z.B. courir und auch se traîner eine solche Kombination, aber bei den meisten der anderen kontinuierlichen Fortbwegungsverben ist sie nicht möglich; so z.B. eher nicht bei marcher, cheminer, flâner, se promener, se balader, déambuler, trotter, ramper, clopiner, claudiquer, boiter, errer, circuler, voyager, nager, voler, naviguer, galoper, pedaler, skier u.v.a. Tritt zu diesen Verben eine à-PP, kann diese nicht Zielangabe sein, sondern nur Angabe des Ortes, an dem die Bewegung stattfindet. Die Kombination mit einer Ziel-PP, die mit direktionalem à gebildet ist, stellt somit bei kontinuierlichen Fortbewegungsverben eher eine Ausnahme dar. Präpositionalphrasen mit jusqu'à und die Kombination von Ausgangsort- und Zielangabe durch de-à-PP sind dagegen bei einer Reihe von Fortbewegungsverben möglich; allerdings gibt es auch hier Restriktionen, die uns nicht ganz klar geworden sind. Jusqu 'à kann z.B. bei courir, marcher, flotter, flâner, clopiner, avancer u.ä. erscheinen (siehe die obigen Beispiele: (31c-g)), dagegen nicht bei se promener, voyager, circuler, errer u.ä. Daraus ist abzuleiten, daß eine Bedingung für die Kombination mit jusqu 'à ist, 15

S'approcher

und se rapprocher sind auf die Präposition de festgelegt.

Fortbewegungsverben

und Ortswechsel

im

Französischen

149

daß das Verb eine kontinuierliche, lineare, d.h. gerichtete Fortbewegung impliziert. Das würde erklären, daß jusqu'à auch bei Bewegungsverben wie danser, dansoter, gigoter, sautiller u.ä. ausgeschlossen ist. Es erklärt jedoch nicht, warum die Kombination mit boiter ausgeschlossen ist, während sie mit claudiquer und clopiner akzeptiert wird; ebenso nicht, warum jusqu 'à bei se promener abgelehnt wird, dagegen aber bei flâner durchaus auftritt. Die Kombination der kontinuierlichen Fortbewegungsverben mit dem de - à-Prâpositionalphrasenpaar, das Ausgangsort und Ziel angibt, scheint ähnlichen Beschränkungen zu unterliegen. Nicht alle kontinuierlichen Fortbewegungsverben lassen sie zu. Die Bedingung scheint auch hier zu sein, daß eine lineare, kontinuierliche Fortbewegung vorliegt und damit zumindest indirekt das Wegkonzept mit seinen Teilabschnitten zur Verfügung steht.

6. Scheinbare Abweichungen von den Kombinationsregeln für kontinuierliche Fortbewegungsverben

6.1. Mehrdeutige Verben Es gibt nun eine Reihe von Verben, die intuitiv kontinuierliche Fortbewegungsverben zu sein scheinen, die sich aber in ihrer Kombinatorik mit statischen oder direktionalen Präpositionen gerade nicht verhalten wie die kontinuierlichen Fortbewegungsverben, sondern wie die Ortswechselverben. Dazu gehören z.B. die Richtungsverben monter, descendre, grimper u.ä. sowie auch tomber, dégringoler, dégouliner, débouler, dévaler u.ä. Zielangaben können bei diesen Verben durch statische Präpositionalphrasen realisiert werden: (32)

a. Nous sommes montés sur la colline. 'Wir sind den Hügel hinaufgestiegen.' b. Le matelot monta sur le pont en tâtonnant. 'Der Matrose tastete sich auf die Brücke hinauf, "(wörtl.: 'stieg tastend ... hinauf.') c. Les étudiants sont descendus dans les rues. 'Die Studenten sind auf die Strassen heruntergegangen.' d. Paul a grimpé sur un arbre/sur un chameau/dans un taxi etc. 'Paul ist auf einen Baum/auf ein Kamel/in ein Taxi geklettert.'

Sie können auch nur die Präpositionalphrase, die sich auf den Ausgangsort bezieht, bei sich haben: (33)

a. Je suis monté de la cave. 'Ich bin aus dem Keller heraufgekommen.' b. Il a déboulé du premier étage. 'Er ist vom ersten Stock heruntergekollert.' c. Le couvreur est tombé du toit. 'Der Dachdecker ist vom Dach gefallen.' d. La crème dégouline du gâteau. 'Die Sahne trieft von der Torte.'

Priska-Monika

150

Hottenroth

Tatsächlich sind diese Verben ambig, was sich selbst dann zeigt, wenn sie ohne Ziel- oder Ausgangsort-PP erscheinen: (34)

a. Je monte. 'Ich steige hinauf.' b. Je descends. 'Ich steige hinab.'

Die Verben können sich in einer Lesart nur auf die Bewegungsrichtung "aufwärts" bzw. "abwärts" beziehen. In dieser Interpretation beziehen sie sich nur auf die kontinuierliche Fortbewegung, d.h. das Verb ist atelisch. In einer zweiten Interpretation können sie die Bewegung "nach oben" (bzw. "nach unten") ausdrücken, d.h. die Bewegung zu einer oben oder unten liegenden Zielregion. Hier ist die Lokalisation "oben" bzw. "unten" als Resultat der Bewegung impliziert und wird kontextuell rekonstruiert und spezifiziert. Das Verb drückt einen Ortswechsel aus, gehört in dieser Interpretation zu den ziel-implizierenden Verben und ist telisch. Entsprechend kann das im Verb angelegte Zielortargument durch eine statisch lokalisierende Präpositionalphrase realisiert und spezifiziert werden. Wenn monter, descendre u.ä. daher mit einer statischen PP als Zielangabe auftreten, liegt die zielimplizierende Variante vor, und die Präpositionalphrase spezifiziert nur das im Verb enthaltene Zielargument. Wenn diese Verben dagegen mit jusqu'à auftreten, liegt die Interpretation als kontinuierliches Fortbewegungsverb vor, denn jusqu'à verlangt, wie oben ausgeführt, einen nicht begrenzten Prozeß, der erst durch die jusqu 'à-PP selbst begrenzt wird. Vergi.: (35)

a. Il est monté jusqu'au

gîte.

'Er ist bis zum Unterschlupf hinaufgestiegen.' b. Il est descendu jusqu 'au village. 'Er ist bis zum Dorf hinabgestiegen.'

In ähnlicher Weise wie die vertikalen Richtungsverben scheinen auch die horizontalen Richtungsverben wie s'éloigner, s'écarter, se déplacer, fuir u.ä. ambig zu sein. Sie können entweder das Verlassen des Ausgangsortes implizieren, also einen Ortswechsel, oder auch nur eine kontinuierliche Bewegung im Sinne einer Bewegung "immer weiter weg von." Die erste Variante liegt vor, wenn sie sich mit einer de-PP verbinden: (36)

a. Ne vous déplacez pas de l'endroit où vous êtes. 'Entfernen Sie sich nicht von der Stelle, an der Sie sich befinden.' b. Il voulait fuir de chez ses parents. 'Er wollte aus der Wohnung seiner Eltern fliehen.'

Das Verb ist aber ambig, wenn es ohne PP erscheint, die den Ausgangsort spezifiziert: (37)

a. Il ne se déplace qu'en taxi. 'Er reist nur mit dem Taxi.' oder: 'Er fährt nur mit dem Taxi weg.' b. Comme on le bousculait, il se déplaça. 'Weil man ihn verdrängte, hat er sich entfernt.' oder: '...ist er immer weiter gerückt.') c. Il s'écarte de quelques pas. 'Er geht einige Schritte zur Seite/weg.' oder: 'Er geht einige Schritte weiter.'

Fortbewegungsverben

und Ortswechsel im

151

Französischen

6.2. Weitere Abweichungen von den kombinatorischen Regeln für kontinuierliche Fortbewegungsverben Anders als die obigen Verben scheinen die folgenden jedoch nicht mehrdeutig zu sein, sondern nur die Interpretation als Verben der kontinuierlichen Fortbewegung zuzulassen, d.h. sie enthalten nur die MOVE-Komponente sowie eine Spezifizierung des Bewegungsmodus. Dennoch ist ihr kombinatorisches Verhalten nicht das der kontinuierlichen Fortbewegungsverben, sondern entspricht dem der Ortswechselverben. Dies gilt z.B. fur se glisser, se faufiler, foncer u.ä. Vergi.: (38)

a. L'enfant n'eut que le temps de se glisser sous le canapé. 'Das Kind hatte gerade noch Zeit, unter das Sofa zu kriechen.' b. Léon était entré au ministère furtivement et s'était glissé dans son bureau. 'Léon war heimlich ins Ministerium hineingegangen und hatte sich in sein Büro geschlichen.' c. Il s 'est glissé du fauteuil. 'Er ist aus dem Sessel geglitten.' d. Le taureau a foncé sur lui. 'Der Stier stürzte sich auf ihn.' e. Le voleur s 'est faufilé par la fenêtre dans l'appartement. 'Der Dieb hat sich durch das Fenster in das Appartement eingeschlichen.'

Ebenso verhalten sich sauter, bondir, se jeter, se lancer, se précipiter

u.a.:

(39) a. Il lui fallait sauter dans un fossé. 'Er musste in einen Graben hineinspringen.' b. Il sauta sur son vélo. 'Er sprang auf sein Fahrrad.' c. Pierre se jette sur l'adversaire. 'Pierre wirft sich auf den Gegner.' d. L'animal bondit sur la proie. 'Das Tier stürzte sich auf die Beute.' e. Il s 'est lancé du balcon. 'Er warf sich vom Balkon.' f. Il sauta d'un arbre. 'Er sprang von einem Baum.'

sowie auch glisser, couler 'fließen' u.ä. (40)

a. En tombant, il a glissé dans un trou. 'Er fiel hin und glitt in eine Grube.' b. L'eau coulait d'une source. 'Das Wasser floss aus einer Quelle.' c. La crème lui coule des doigts. 'Die Sahne tropft ihm von den Fingern.'

Es ist uns vorläufig nicht klar, was in der semantischen Struktur dieser Verben das von den übrigen kontinuierlichen Fortbewegungsverben abweichende kombinatorische Verhalten bedingt. Es ist anzunehmen, daß sie in irgendeiner Weise einen Bezug zu einem Wegkonzept, zum Ausgangs- oder Zielort der Bewegung enthalten, das die entsprechenden

152

Priska-Monika

Hottenroth

Argumente verankern kann. Denkbar ist bei den meisten m.E. am ehesten ein impliziter Bezug zum Ausgangsort, da es sich entweder um Bewegungen mit relativ heftigem, intentionalem Start handelt, wie bei sauter, se jeter, bondir, se lancer, foncer 'losstürzen', oder um nicht intentionale Bewegungen, bei denen auch der Anfang der Bewegung fokussiert zu sein scheint, wie bei glisser 'ab-, wegrutschen' und auch couler. Se faufiler und se glisser dagegen scheinen nach Auskunft muttersprachlicher Informanten eher ein 'hindurch', also den Durchgangsort zu enthalten. In zweisprachigen Lexika werden allerdings auch zielimplizierende Übersetzungen für diese Verben angegeben, z.B. fur se faufiler 'hineinschleichen', 'eindringen' fur se glisser 'sich einschleichen', was darauf hinweisen könnte, daß sie den Zielort als Argument in ihrer semantischen Struktur verankern. Abschließend soll noch eine scheinbare Ausnahme angeführt werden: (41 ) a. Napoléon a marché sur Moscou. 'Napoleon marschierte auf Moskau los.' b. Brusquement l'homme qui marchait sur lui, s'immobilise. 'Der Mann, der auf ihn zulief, hielt plötzlich an.'

Hier liegt eine Bedeutungsvariante von sur vor, die nicht statisch lokalisierend ist, sondern in etwa der Bedeutung von vers entspricht. Die swr-PP bezieht sich hier nicht auf einen Zielort, der erreicht wird, sondern gibt die Richtung an. Das Prädikat ist kein Ortswechselprädikat, sondern drückt eine kontinuierliche Bewegung aus. Eine ähnliche Interpretation kann auch foncer in Kombination mit sur haben: (42) Le taureau s 'est mis à foncer sur nous. 'Der Stier stürzte sich plötzlich auf uns.'

7. Fazit

Die Untersuchung hat gezeigt, daß bei Verben, die das Erreichen des Ziels bzw. das Verlassen des Ausgangsortes implizieren, im allgemeinen das Zielargument als statisch lokalisierende Präpositionalphrase oder aber das Argument für den Ausgangsort als direktionale Präpositionalphrase realisiert werden kann und daß sich dieses kombinatorische Verhalten aus der semantischen Struktur der Ortswechselverben erklärt. Die Realisierung dieser Argumente ist möglich, weil sie beide in der semantischen Struktur angelegt sind, entweder unmittelbar als Ziel oder Ausgangsort - oder als Komplementregion zu der jeweils enthaltenen Region. Kontinuierliche Fortbewegungsverben lassen weder eine statisch lokale Präpositionalphrase als Zielargument, noch eine direktionale PP als Argument für den Ausgangsort (allein) zu. Ziel und Ausgangsort sind nicht in ihrer semantischen Struktur angelegt. Die wenigen Verben, die zunächst wie kontinuierliche Fortbewegungsverben erscheinen und sich aber anders, d.h. wie Ortswechselverben verhalten, scheinen - zumindest in einer Variante - eine entsprechende Struktur zu besitzen und wären somit nur scheinbare Ausnahmen.

Fortbewegungsverben

und Ortswechsel

im

153

Französischen

Bis auf wenige Einzelfälle lassen kontinuierliche Fortbewegungsverben auch eine direktionale Präpositionalphrase mit à als Zielangabe nicht zu. Allerdings kann bei einer relativ kleinen Gruppe eine räumliche jusqu 'ά-ΡΡ oder die Kombination aus einer de- und einer àPräpositionalphrase (für Ausgangsort und Ziel) erscheinen. Beide fuhren den ganzen Weg samt seiner Begrenzung ein, setzen jedoch voraus, daß das Verb eine nicht begrenzte, aber kontinuierliche, lineare Bewegung enthält. Mit anderen Worten, die ausgedrückte Bewegung muß in der Realität zu den Bewegungsarten gehören, die typischerweise zielgerichtet sind, d.h. von A nach Β führen. Durch diese Verben wird damit indirekt ein Wegkonzept mit seinen Teilabschnitten - Ausgangsort und Ziel - für die Anbindung der jusqu 'à-Prâpositionalphrase bzw. der de - à Kombination von Präposi-tionalphrasen zur Verfügung stellt. Es wurde weiter gezeigt, daß räumlich direktionales jusqu'à bei bestimmten Ortswechselverben erscheinen und das Zielargument ersetzen kann. Das setzt voraus, daß diese Verben accomplishment-Struktur, nicht achievement-Struktur haben und somit eine durch jusqu'à induzierte Uminterpretation zu activities, d.h. zu nicht begrenzten Prozessen erlauben.

Literatur

Bergh, Leonard (1948): Moyens d'exprimer

en français

l'idée de direction.

- Genève.

Doli, Ruth (1967): Der deutsche Verbzusatz als Richtungsträger und seine Wiedergabe im Französischen und Italienischen. - Disseration. Universität Tübingen. Dowty, David R. (1979): Word Meaning and Montague Grammar. - Dordrecht: Reidel. Green, G. (1973): A Syntactic Syncretism in English and French. - In: B. Kachru, H. Kahane & R. Kahane (eds.): Issues in Linguistics, 2 7 5 - 2 7 8 . Urbana, IL: University of Illinois Press. Hajdù, Judit I. (1969): Der Richtungsausdruck in der französischen und spanischen wartssprache. - Disseration. Universität Zürich. Jackendoff, Ray (1983): Semantics and Cognition. - Cambridge, MA: The MIT Press. Kaufmann, Ingrid (1995a): Konzeptuelle Grundlagen semantischer Dekompositionsstrukturen: Kombinatorik lokaler Verben und prädikativer Komplemente. - Tübingen: Niemeyer.

Gegen-

Die

- (1995b): Positionsverben und Richtung. - In: Kognitionswissenschaften 4, 154-165. Levin, Beth & Malka Rappaport Hovav (1992): The Lexical Semantics of Verbs of Motion: The Perspective from Unaccusativity. - In: I. M. Roca (ed.): Thematic Structure: Its Role in Grammar, 2 4 7 - 2 6 9 . Berlin: Foris Publications. -

(1995): Lexical Semantics and Syntactic Structure. - In: S. Lappin, S (ed..): The Handbook of Contemporary Semantic Theory, 4 8 7 - 5 0 7 . Oxford: Blackwell. Levin, Beth & Tova R. Rapoport (1988): Lexical Subordination. - In: Proceedings of the TwentyFourth Annual Meeting of the Chicago Linguistic Society 24, 2 7 5 - 2 8 9 . Maienborn, Claudia (1990): Position und Bewegung: Zur Semantik lokaler Verben. - Stuttgart: IBM Deutschland GmbH (IWBS Report 138). -

(1994): Kompakte Strukturen: Direktionale PPn und nicht-lokale Verben. - In: S. Felix, Chr. Habel & G. Rickheit (Hgg.): Kognitive Linguistik: Repräsentation und Prozesse, 229-249. Opladen: Westdeutscher Verlag.

-

(1996): Situation und Lokation: Tübingen: Stauffenburg Verlag.

Zur Bedeutung

lokaler

Adjunkte

von

Verbalprojektionen.

154

Priska-Monika

Hottenroíh

Pustejovsky, James (1991): The Syntax of Event Structure. - In: Cognition 4 1 , 4 7 - 8 1 . - (1995): The Generative Lexicon. - Cambridge, MA: The MIT Press. Rapp, Irene (1997): Fakultativität von Verbargumenten als Reflex der semantischen Struktur. - In: Linguistische Berichte 172, 490-529. Sablayrolles, Pierre (1993): A Two-Level Semantics for French Expressions of Motion. Kaiserslautern/Saarbrücken: Deutsches. Forschungszentrum fur Künstliche Intelligenz GmbH (Research Report RR-93-12). Sablayrolles, Pierre & Nicholas Asher (1995): A Typology and Discourse Semantics for Motion Verbs and Spatial PPs in French. - In: Journal of Semantics 12, 163-209. Schlyter, Suzanne (1981): 'De-à/von-zu' avec les verbes de mouvement cursifs et transformatifs. In: Chr. Schwarze, Chr. (ed.): Analyse des prépositions, 171- 89 . Tübingen: Niemeyer. Slobin, Dan A. (1997): In a Manner of Going: Influences of Lexicalization Patterns on Descriptions for Motion Events. Paper presented at the second Meeting of the Association for Linguistic Typology (ALT II), University of Oregon, Eugene, OR, Sept 11-14, 1997. Stiebels, Barbara (1991): Präpositioneninkorporierung und das Problem der Partikelverben im Deutschen. - Magisterarbeit. Universität Düsseldorf. Stöcklin, Jörg (1974): A, en, dans, sur: Essai sur la valeur intrinsèque et fonctionnelle des prépositions locales à, en, dans, et sur dans le français contemporain. - Dissertation. Universität Basle. Talmy, Leonard (1975): Semantics and Syntax of Motion. - In: J.P. Kimball (ed.): Syntax and Semantics. Vol. 4, 181-238. New York, NY: Academic Press. - (1985): Lexicalization Patterns: Semantic Structure in Lexical Forms. - In: T. Shopen (ed.): Language Typology and Syntactic Description. Vol 3: Grammatical Categories and the Lexicon, 57-149. New York, NY: Plenum Press. Vendler, Zeno (1967): Verbs and Times. - In: Z. Vendler (ed.): Linguistics in Philosophy, 97-121. Ithaca, NY: Cornell University Press. Welti, Beat Kaspar (1974): Der Bewegungs- und Richtungsausdruck in der italienischen und französischen Gegenwartssprache. - Disertation. Universität Zürich. Wunderlich, Dieter (1991): How Do Prepositional Phrases Fit into Syntax and Compositional Semantics? - In: Linguistics 1991, 29-4 [314], 591-623. Wunderlich, Dieter & Ingrid Kaufmann (1990): Lokale Verben und Präpositionen: Semantische und konzeptuelle Aspekte - In: S. Felix, S. Kanngießer & G. Rickheit (eds.): Sprache und Wissen: Studien zur Kognitive Linguistik, 223-252. Opladen: Westdeutscher Verlag. Wunderlich, Dieter & Michael Herweg (1991): Lokale und Direktionale. - In: A. von Stechow & D. Wunderlich (eds.): Semantik - Semantics: Ein internationales Handbuch der zeitgenössischen Forschung -An International Handbook of Contemporary Research, 758-785. Berlin: Mouton de Gruyter.

Part II: Cognitive-Semantic Perspectives on Prepositions

Birgitta Meex Die Wegpräposition über

1. E i n f ü h r u n g

1.1. Zielsetzung In diesem Aufsatz schlage ich eine kognitiv-semantische Analyse der Wegpräposition über vor. Ich konzentriere mich dabei auf die Struktur ihres räumlichen Verwendungsbereichs der Wegangabe; ihre nicht-wegbezogenen Verwendungen (Sie trug einen Mantel über dem Kleid; Er hängt das Bild über den Schrank) sowie ihre Verwendungen in temporalen und abstrakten Domänen 1 (Über Nacht war Regen eingetreten) bleiben in dieser Analyse außer Betracht. Ihre adverbialen (Über 100 Beamte kamen) und prädikativen (Es ist mir über) Verwendungen sowie ihre Verwendungen als Präfix (Er überfliegt die Alpen) und Partikel (Anderen geht das Herz über) werden ebenfalls nicht berücksichtigt. Ziel der Analyse ist es, einen möglichst detaillierten Überblick über die vielfaltigen wegbezogenen Verwendungen des räumlichen über zu geben, wobei ihre Bedeutungen voneinander abgegrenzt und ihre Motivation durch die Spezifizierung der sie verbindenden semantisch-konzeptuellen Relationen aufgezeigt werden sollen. In diesem Zusammenhang werden auch die verschiedenen dem wegbezogenen Gebrauch von über zugrundeliegenden Konzepte der Horizontalität und Vertikalität, des Weges, der Bewegung, der Erstreckung und der Subjektivierung erläutert. Die Relevanz der Dimensionalität soll ebenfalls überprüft werden. Schließlich wird auch der Frage nachgegangen, ob für über eine prototypische Wegbedeutung angegeben werden kann. Die Analyse erfolgt auf der Grundlage einer umfangreichen Sammlung von Belegen aus der deutschsprachigen Presse und Literatur.

1.2. Der kognitiv-semantische Ansatz Die semantisch-konzeptuelle Struktur räumlicher Präpositionen läßt sich adäquat im Rahmen der kognitiven Semantik beschreiben. Das Gebiet der Präpositionen hat schon immer eine zentrale Rolle in der kognitiven Linguistik gespielt. Als eine der ersten kognitiven Linguisten haben sich Dirven und Radden systematisch mit der Semantik von Präpositionen auseinandergesetzt (vgl. dazu ihre Beiträge in Radden & Dirven 1981). Seit den 1

Eine Untersuchung der temporalen und abstrakten Verwendungen wird in Meex (2001) vorgenommen. Dort wird gezeigt, wie die verschiedenen Lesarten des präpositionalen über ein [Continuum konstituieren, das von der Markierung der räumlichen über die temporalen bis hin zu den abstrakten Relationen fuhrt. Dabei bilden die hier zu beschreibenden spezifischen Merkmale des räumlichen über die Grundlage für die Strukturierung nicht-räumlicher Bereiche (vgl. hierzu auch Radden 1981).

158

Birgitta

Meex

wegweisenden Untersuchungen von Brugman (1983) und Lindner (1983) ist die Semantik vor allem englischer Präpositionen häufig beschrieben worden: Hawkins (1985), Herskovits (1986), Schulze (1990), Cuyckens (1991), Vandeloise (1991), Boers (1996), Lindstromberg (1997) und Claes (1998). Speziell mit der Präposition over und ihren niederländischen und deutschen Pendants haben sich folgende Arbeiten befaßt: Brugman (1983), Lakoff (1987), Vandeloise (1990), Geeraerts (1992), Deane (1993), Dewell (1994), Taylor (1995), Bellavia (1996) und Kreitzer (1997). In der vorliegenden Studie übernehmen wir den theoretischen sowie begrifflichen Apparat des kognitiv-linguistischen Ansatzes, dessen Grundbegriffe hier kurz erläutert werden. In der kognitiven Grammatik bezeichnen Präpositionen ein relationales Profil (vgl. Langacker 1987: relational profile). Dies bedeutet, daß Präpositionen in ihrer räumlichen Verwendung eine räumliche Relation zwischen zwei Entitäten profilieren. Nicht nur die Relation selbst, sondern auch die unmittelbar an ihr beteiligten Entitäten Trajektor (TR) und Landmarke (LM) werden dabei profiliert. Die Unterscheidung zwischen TR und LM entspricht dabei derjenigen von Figur und Grund. Der TR ist die Figur im relationalen Profil, die sich deutlich vom Grund, dem neutralen Rest seiner Umgebung, abhebt. Nicht nur dynamische, sondern auch statische Relationen sind durch TR und LM bzw. Figur und Grund gekennzeichnet. Dabei ist die LM das Referenzobjekt, relativ zu dem der TR lokalisiert wird. Als kognitive Domäne, hinsichtlich welcher die relationale Prädikation charakterisiert wird, fungiert der zwei- oder dreidimensional konzeptualisierte Raum. Im Folgenden soll gezeigt werden, wie die semantisch-konzeptuelle Struktur von wegbezogenem über im Rahmen des hier skizzierten Ansatzes analysiert werden kann. Der Beitrag gliedert sich folgendermaßen: Im zweiten Abschnitt wird kurz Bellavias (1996) Analyse von über vorgestellt. Der dritte Abschnitt bietet einen Überblick über die verschiedenen wegbezogenen Verwendungsweisen von über. Im vierten Abschnitt wird die Kategorie der Wegpräpositionen erörtert. Im fünften und sechsten Abschnitt werden die verschiedenen von über bezeichneten Wegrelationen in Hinblick auf den aspektuellen Charakter des Bewegungsvorgangs betrachtet. Im siebten Abschnitt werden schließlich die Subjektivierungen bei der Konzeptualisierung von Wegangaben diskutiert.

2. Die Analyse von Bellavia (1996)

Bellavia (1996) nimmt für über fünf Grundbedeutungen an, die den folgenden TR-LMKonfigurationen entsprechen: 1. Der TR befindet sich in einer gegenüber der LM vertikal höheren Position, wobei TR und LM sich nicht berühren. Dies gilt sowohl für statische Situationen (ohne Bewegung) wie in (la) als auch für dynamische Situationen (mit Bewegung wie in (lb)): (1 )

a. Die Lampe hängt über dem Tisch, b. Das Flugzeug fliegt über der Stadt.

2. Der TR legt eine Strecke entlang der horizontalen Dimension zurück, wobei er die von der LM bezeichnete Fläche überquert. Dabei kann ein Raum zwischen TR und LM

Die Wegpräposition über

159

liegen (2a), oder TR und LM können sich berühren (2b). Bellavia betont, daß der TR im letzteren Fall notwendigerweise die Grenzen der LM überquert wie in Beispiel (2b): (2)

a. Das Flugzeug fliegt über die Stadt, b. Er geht über die Straße.

3. Der TR legt eine Strecke entlang der vertikalen Dimension zurück, wobei er das von der LM bezeichnete Hindernis überwindet. Wiederum kann ein Raum zwischen TR und LM liegen (3a) oder TR und LM können sich berühren (3b): (3)

a. Er springt über ein Hindernis, b. Er klettert über die Mauer.

4. Der TR befindet sich infolge einer subjektiven Bewegung jenseits der LM: (4)

Sie wohnen über der Straße.

5. TR und LM berühren sich, und der TR bedeckt völlig die von der LM bezeichnete Fläche. Dabei kann zwischen statischen (5a) und dynamischen (5b) Situationen unterschieden werden. (5)

a. Das Tuch ist über der Nähmaschine. b. Die Frau breitet die Decke über den Tisch.

Nach Bellavia (1996), die sich in diesem Punkt der over-Analyse von Dewell (1994) anschließt, können all diese Konfigurationen einem einzigen übergeordneten Schema, dem bogenförmigen Schema, zugeordnet und darauf bezogen erklärt werden, indem jede Konfiguration jeweils einen anderen Punkt des Bogens fokussiert:

Im Folgenden möchte ich eine alternative Analyse von über vorschlagen, die sich auf seine wegbezogenen Verwendungen konzentriert. Dabei verstehen wir die Bedeutungen 2, 3 und 4 in der Analyse von Bellavia (1996) als Wegangaben. Die Wegangabe bildet einen wesentlichen Bereich der Bedeutung von über, der von Bellavia (1996) nicht bzw. Unvollständig und uneinheitlich behandelt worden ist.

3. Überblick über die Verwendungen von über als Wegpräposition

Welche verschiedenartigen Wegrelationen mit über beschrieben werden können, sei an der folgenden repräsentativen Beispielreihe in (6) illustriert:

Birgitta

160

(6)

Meex

Er hüpft über gefällte Stämme zu seinen Männern an der Motorsäge. (Bienkopp, 17) 2 Ihr PKW muss über den grossen Teich? (AMS 1/94, 66) Altöl, von einem Frachter über Bord gepumpt. (BILD 10.6.95, 2) Sie gleitet in einem Kahn über den See. (Bienkopp, 286) Vor drei Tagen lief der Äthiopier Haile Gebreselaissie einen Fabel-Weltrekord über 10000 Meter. (BILD 10.6.95, 8) f. Zur Zeit fliegt die Forschungssonde Ulysses hoch über den Sonnen-Nordpol, (bdw 8/95, 57) a. b. c. d. e.

g. Die Straßenbahnlinie 6 soll laut Zieschank auch künftig über die Leubener Straße fahren. (SZ 16.3.95,30) h. Ich fliege diesmal über Miami und Merida, Yucatan, w o man fast täglich Anschluß nach Caracas hat, und unterbreche in Merida. (Homo faber, 206) i. [Er] zieht sich das Unterhemd über den Kopf, er scheint gut ernährt. (Achim, 277) j. Jedes Kind mit Videospielgerät kann heute zu Hause den Golfkrieg nachspielen oder Dinosaurier über den Bildschirm hin- und herschieben [...]. (SPI 33/94, 138) k. Vom Atlantik zum Pazifik erstreckt sich dieses faszinierende Land über 6500 Kilometer und sechs Zeitzonen. (NKrZ 23.3.96, G27) I. Er wohnt über den Bergen.

Aus diesen Beispielen ist ersichtlich, wie unterschiedlich die durch über hervorgerufenen Wegvorstellungen sein können. Wie aus den Sätzen (6a-c) deutlich wird, kommt über in Konstellationen vor, in denen der TR ein Hindernis überquert oder eine Begrenzung überschreitet. Über ist jedoch auch möglich, wenn die Bewegung innerhalb eines bestimmen Bereichs erfolgt (6d-f) oder einen Bereich passiert (6g, h). In all diesen Fällen ist es unerheblich, ob TR und LM sich berühren oder nicht. Ebensowenig scheint die Bedeutung von über an eine horizontale Bewegungsrichtung gebunden zu sein. Wie aus (6i) bzw. aus (6j) hervorgeht, kann die Bewegung auch vertikal oder multidirektional verlaufen. Ferner kann man sich mit über nicht nur auf dynamische, sondern auch auf statische Konstellationen (6k, 1) beziehen, was in bestimmten Fällen (61) mit einem Kasuswechsel (Akkusativ-Dativ) einhergeht.3 Schließlich wird an den Beispielen in (6i, j, 1) deutlich, daß sich TR und LM nicht unbedingt vertikal zueinander zu befinden brauchen. Kurzum: Mit ein und derselben Wegpräposition kann man eine Fülle von Relationen ausdrücken, die nicht einfach als Weg bezeichnet werden können.

4. Wegrelationen

4.1. Zum Konzept des Weges Bevor wir uns den wegbezogenen Verwendungen von über zuwenden, gehen wir kurz auf das Wegkonzept ein. Zu den deutschen Wegpräpositionen gehören neben dem topologisch 2

3

Nach jedem Beispielsatz wird, insofern er belegt ist, die betreffende Belegstelle angegeben. Die Abkürzungen werden weiter unten in der Quellenangabe erläutert. Über ist also nicht auf den Akkusativ beschränkt. Insofern hebt es sich von den rein akkusativischen Wegpräpositionen durch und um ab.

Die Wegpräposition

161

über

gebrauchten über, das vom dimensional gebrauchten über deutlich abzuheben ist, durch, um, entlang und längs. Die Wegpräpositionen oder extensionalen Präpositionen im Sinne von Leys (1995) zeichnen sich dadurch aus, daß sie nicht bloß einen Ort, sondern auch ein Kontinuum von Orten bzw. Zuständen, d.h. einen Weg charakterisieren, dessen Erstreckung summativ konzeptualisiert wird, ja sogar erst auf eine kumulative Weise zustande kommt (vgl. Leys 1989, 1995). Dies bedeutet, daß die verschiedenen Positionen, die der sich bewegende nulldimensional konzeptualisierte TR nach und nach relativ zur LM einnimmt, summiert werden, so daß allmählich ein eindimensionaler Weg entsteht. Die WegVorstellung kommt also erst durch die Kumulierung der verschiedenen vom TR durchlaufenen Orte, die zusammen den Weg ausmachen, zustande:

Weg des TRs Abb. 2: summativ entstehender

Weg

Wege besitzen insofern eine Richtung, als sie kontinuierlich von einem Ort zu einem anderen verlaufen (vgl. Saile 1984). Daß die älteste Gruppe der Wegpräpositionen (durch, über, um) sämtlich den Akkusativ regiert, 4 ist in dem ihrer Bedeutung zugrundeliegenden Konzept der summativen Erstreckung begründet. 5 Wie Claes (1998) überzeugend argumentiert, unterscheiden sich die Wegpräpositionen von den sogenannten "Wechselpräpositionen" (an, auf, hinter, in, neben, nicht-wegbezogenes über, unter, vor, zwischen), die sowohl Akkusativ- als auch Dativrektion aufweisen, darin, daß der Akkusativ nicht eine zielgerichtete Bewegung auf die LM hin, sondern eine sich erstreckende Bewegung parallel zur LM signalisiert. Da Wegpräpositionen kein Ziel bezeichnen, können sie beliebig aneinandergereiht vorkommen (vgl. Saile 1984; Claes 1998): (7)

Mountainbiker wuchten ihre Geräte über Wanderwege und durch Bachläufe. (SPI 33/94, 52)

4

Entlang und längs sind jüngeren Datums und befinden sich auf dem Wege der Grammatikalisierung. Dies erklärt ihre Kasus- und/oder Stellungsalternation. Entlang kommt sowohl als Präposition wie auch als Postposition vor und verbindet sich als Präposition mit dem Genitiv und dem Dativ. Als Postposition verhält es sich wie die älteren Wegpräpositionen und tritt mit dem Akkusativ auf. Das nur noch selten gebrauchte längs weist einen im Vergleich zu entlang höheren Grammatikalisierungsgrad auf. Es kommt nur in Prästellung vor und regierte ursprünglich den Dativ, jetzt überwiegend den Genitiv (vgl. Di Meola 1999, dieser Band).

5

Nach Leys (1989, 1995) wird die Rektion räumlicher Präpositionen von ihrem summativen bzw. nicht-summativen Aspekt bestimmt. Dabei definiert Leys ein entstehendes Verhältnis als ein inhärent summatives, ein bestehendes Verhältnis hingegen als ein inhärent nicht-summatives. Im Gegensatz zu einem bestehenden Verhältnis liegt ein entstehendes Verhältnis an keinem der einzelnen, vom TR durchlaufenen Punkte als abgeschlossen vor. Die Vorstellung einer Erstreckung, die es dem TR von Wegpräpositionen ermöglicht, überhaupt in den Suchbereich der LM einzudringen, kommt somit erst ab einem gewissen Punkt in der Endphase des Weges durch die mentale Summierung der verschiedenen Punkte zustande. Daher wird das Verhältnis akkusativisch markiert.

162

Birgitta

Meex

Aus (7) wird deutlich, daß die Wanderwege nicht das Ziel der Bewegung darstellen; sie bezeichnen stattdessen die ausgedehnte Strecke selbst, auf der die Bewegung stattfindet. Den Wegpräpositionen über, durch und um ist also das Konzept der Ausdehnung gemeinsam; nur in der Art bzw. nach der Form der Erstreckung unterscheiden sie sich. Im Folgenden wird näher auf die von über geleisteten Wegspezifizierungen eingegangen.

4.2. Die Charakteristika des über-Weges Von den deutschen Wegpräpositionen drückt über die am wenigsten spezifische Pfadangabe aus (vgl. Marcq 1972, 1988). Im Vergleich zu durch (das auch adverbial wie in unter ... durch oder zwischen ... durch auftreten kann), um und entlang!längs, welche die Art der Fortbewegung in Bezug auf die LM markieren, bezeichnet über eine allgemeine, unmarkierte Wegbeschreibung. Marcq (1972) führt dazu folgendes Beispiel auf: (8)

a. b. c. d.

Wir Wir Wir Wir

fahren fahren fahren fahren

über Köln. durch Köln. um Köln (herum). unter Köln durch.

Nach Marcq (1972), dessen Ansicht wir (wenigstens hinsichtlich der Fälle in (8)) teilen, enthalten die Sätze (8b-d) drei verschiedene Realisationen der in Bezug auf die Passagerelation unmarkierten über-Phrase. Grob gesagt kennzeichnet durch die Bewegung in Bezug auf den inneren Bereich der LM, um die kreisförmige Bewegung in Bezug auf die Peripherie der LM und unter ... durch die Bewegung im negativ-vertikalen Bereich der LM. 6 Über ist dann insofern unpräzise, als man sich damit auf jede der drei spezifischeren Raumkonstellationen beziehen kann. Diese Darstellung von Marcq (1972) muß jedoch in zweifacher Hinsicht nuanciert werden. Erstens darf die bezüglich der Sätze unter (8) angestellte Beobachtung nicht so interpretiert werden, daß das von über ausgedrückte Konzept etwa vage und undifferenziert wäre. Zweitens gibt es zahlreiche Fälle, in denen über eine genauso spezifische und markierte Relation wie die anderen Wegpräpositionen beschreibt: (9)

a. b. c. d. e.

Der Der Der Der Der

Dieb Dieb Dieb Dieb Dieb

kriecht kriecht kriecht kriecht kriecht

über die Hecke. durch die Hecke. um die Hecke. unter der Hecke durch. die Hecke entlang.

Nur liegt im Falle von über nicht immer semantische Eindeutigkeit vor. So kann über in Satz (9a) ohne weiteren Kontext als eine Relation des Überquerens (TR überquert LM) oder aber als eine Relation des Verfolgens (TR verfolgt LM) interpretiert werden (vgl. auch Geeraerts 1992; Claes 1998). In der ersten referentiellen Situation bewegt sich der Dieb über die als eine Begrenzung konzeptualisierte Hecke hinweg, in der zweiten verfolgt er die längliche Dimension der Hecke, ohne die Hecke dabei zu verlassen. 6

Ähnlich markiert sind auch die in (8) unberücksichtigten Präpositionen entlang und längs, welche die Bewegung in Bezug auf die Längsseite der LM kennzeichnen.

Die Wegpräposition

über

163

Wie sich die von über bezeichneten Wege bestimmen und von den anderen Wegangaben unterscheiden lassen, soll im Folgenden gezeigt werden. Je nach dem aspektuellen Charakter der im Satz beschriebenen Situation kann man zwei Pfadarten unterscheiden, die sich nicht ohne grundsätzliche Änderung der Bedeutung durch andere Wegbeschreibungen ersetzen lassen. Wird der vom Verb bezeichnete Bewegungsvorgang imperfektiv, d.h. als unvollendet und zeitlich unbegrenzt konzeptualisiert, ist der summativ entstehende überWeg ein bedeckender Weg, ein Weg also, der die LM von über summativ bedeckt. Eine perfektive Konzeptualisierung der summativ entstehenden Bewegung ruft jedoch andere Wegarten hervor, die sich infolge ihrer Zielgerichtetheit dem AUSGANGSPUNKT-PFAD-ZIEL Schema (vgl. engl. SOURCE-PATH-GOAL) zuordnen lassen. Auf diese beiden Konzeptualisierungsmöglichkeiten wird unten im vierten bzw. im fünften Abschnitt näher eingegangen.

5. Imperfektive Bewegung: Summativ bedeckende W e g e

Als Ausgangspunkt für unsere Analyse von über als Wegpräposition betrachten wir die Beispiele unter (10), die den Gebrauch von über als bedeckenden Weg illustrieren. Auf dem Weg findet jeweils eine imperfektive, unveränderliche Bewegung statt: (10)

a. Der dicke Semo stampft über den Acker. (Bienkopp, 132) b. [...] Dann springt er wie eine lebendige Fackel lichterloh brennend aus dem Auto, wirft sich gegen die Mauer und wälzt sich über den Boden, um die Flammen zu ersticken. (AMS 1/94, 146) c. Er [...] führt den Reisenden auf exakt beschriebenen Routen über die beiden Inseln, (kosmos 10/95, 67) d. Und es kam wohl ein Wind auf, denn über die Wasserfläche fielen die Böen wie Schuppen [...]. (Parfum, 84) e. Über den Hein-Köllisch-Platz wühlte sich früher der Verkehr, jetzt atmet er eine fast gutbürgerliche Ruhe. (BRI 15/92, 190) f. Windschnittige Karossen schnurren lautlos über die blitzsauberen, palmengesäumten Boulevards. (AMS 1/94, 126) g. Der Hengst trägt den Namen des Windes, der über das australische Hochland weht. (SZ 16.3.95,21)

Aus den Beispielen unter (10) geht hervor, daß zwischen TR und LM eine topologische Relation der Kontiguität besteht. Der TR ist typischerweise ein der Gravitationskraft unterliegendes Lebewesen oder Ding, das sich relativ zu einem geographischen Objekt (der LM) fortbewegt. Geographische Objekte werden hier im Sinne von Herskovits (1986: 61) als komplexe dreidimensionale Objekte verstanden, die ein Stück Land oder eine Wasserfläche enthalten, das bzw. die mehr oder weniger begrenzt ist: Acker, Boden, Insel, Platz, Boulevard, Hochland oder Wasserfläche. Je nach der Konzeptualisierungsart ist die LM ein Teil der Erdoberfläche oder enthält einen Teil der Erdoberfläche. Die Erdoberfläche wird auf jeden Fall als eine sich horizontal erstreckende Fläche konzeptualisiert, auf der sich der TR bewegt. Überhaupt kann sie infolge der Gesetze der Schwerkraft als Default-Grund

164

Birgitta Meex

vieler statischer und dynamischer Verhältnisse angesehen werden. Wie oben angeführt, ist der von über spezifizierte Pfadkontakt mit (einem Teil) der Erdoberfläche inhärent summativ: Jeder einzelne Ort, den der TR relativ zur LM durchläuft, wird kumuliert, was einen permanenten bzw. mehrmaligen Kontakt zwischen TR und LM voraussetzt (vgl. Bouillon 1978). Allmählich bedeckt der so summativ entstandene Weg seine geographische Kontaktfläche konzeptuell und fällt schließlich mit ihr zusammen. Ob sich der Weg linear von der einen Seite der LM zur anderen erstreckt (siehe Abbildung 3a) oder mäandrisch verläuft und mehr oder weniger deckungsgleich mit der LM ist (siehe Abbildung 3b), ist vom jeweiligen Kontext abhängig. Wenn die LM selbst die Gestalt eines Weges hat (lOf), definiert sie automatisch den vom TR zu verfolgenden Weg (vgl. Geeraerts 1992). Bei flächenhaften LM-Bereichen ist dies nicht der Fall. Festzuhalten ist ferner, daß die summative Bewegung in den Beispielen unter (10) als eine imperfektive zu verstehen ist, also in ihrem zeitlich unbegrenzten und unvollendeten Verlauf fokussiert wird. Demzufolge werden Ausgangspunkt und Endpunkt der Bewegung nicht spezifiziert. Es werden auch keine Grenzen überquert, und die gebundene Fläche der LM wird nicht verlassen.

Abb. 3: über; summativ bedeckende

Wege

Eine lineare oder flächenhafte Erstreckung kann auch wie in (11) quantifiziert werden: (11)

a. Denn ohne Qualitätseinbußen läßt sich dieser Rohstoff nicht über große Entfernungen transportieren, (hb 25.8.95, K7) b. Vor drei Tagen lief der Äthiopier Haile Gebreselaissie einen Fabel-Weltrekord über 10000 Meter. (BILD 10.6.95, 8) c. Über 10.000 Meter zwangen Schmerzen sie zur Aufgabe. (SüdZ 2.8.95, 36)

Mit über kann man sich also nicht nur auf die Strecke selbst, sondern auch auf ihre Länge beziehen. Zwischen dem Konzept der Länge und dem der Strecke besteht eine metonymische Beziehung. Distanzangaben werden immer unter Bezugnahme auf zwei Punkte Ausgangspunkt und Endpunkt der Strecke - vorgenommen (vgl. Becker 1994). Mit der LM (große Entfernungen, 10.000 Meter) wird dann nicht das geographische Objekt selbst, sondern metonymisch dessen Länge angegeben. Dieser Gebrauch von über findet sich nicht selten in der Sportberichterstattung (IIb, c). Über ist sogar dann möglich, wenn jeglicher explizite Bezug auf ein Verb der Bewegung fehlt, wie in (1 lc). Aus dem Kontext geht hier natürlich eindeutig hervor, daß der TR {sie) in Bewegung ist. Auch wenn die Schwerkraft überwunden wird (12a) oder ihre Gesetze nicht länger gelten (12b), fungiert die Oberfläche der Erde oder eines anderen Körpers immer noch als der Grund, relativ zu dem die KÒer-Strecke durchlaufen wird:

Die Wegpräposition

(12)

über

165

a. Die v o m Zirpen der Grillen durchwobene Blumenwiese, über die Falter gaukeln [...]. (kosmos 1 0 / 9 5 , 4 4 ) b. Zur Zeit fliegt die Forschungssonde Ulysses hoch über den Sonnen-Nordpol, (bdw 8/95, 57)

Da TR und LM sich nicht berühren, liegt nicht länger eine Relation der Kontiguität vor. Das präpositionale Verhältnis an sich ändert sich jedoch nicht: Wie in den Beispielen unter (10) ist die LM die Grundfläche, die mental von der summativen Bewegung abgedeckt wird. Neben geographischen Objekten können auch andere ein- oder zweidimensional konzeptualisierte Objekte, die ihrerseits (direkt oder indirekt) vom Grund getragen werden, eine Basis für Wegkonstellationen bilden und daher als LM von über vorkommen: (13)

a. Über die Oberbaumbrücke fahrt derweil schon die U-Bahn-Linie 1. (BZ 2/2/96, 4) b. Mit energischen, männlich wirkenden Bewegungen fetzt sie zum schrägen Punkrocksound ihrer Band quer über die Bühne. ( B R A V O 17/95, 63) c. Eine pfenniggroße Spinne, die über den Teppichboden lief, hat das Leben einer Hausfrau zerstört. (BILD 10.06.95, 10)

Wenn die LM kein geographisches Objekt ist, brauchen sich TR und LM nicht vertikal zueinander zu befinden; entscheidend ist lediglich, daß die beiden Entitäten in Kontiguität zueinander stehen und daß die bedeckende Pfadbewegung an der äußeren Fläche der LM stattfindet: 7 (14)

a. Sie fuhr neugierig mit dem Zeigefinger über die Narbe an der Stirn eines Mannes, der sie auf dem Arm hielt. (Bienkopp, 230) b. Als Kind hatte sie von ihrem Vater einen Schlag mit dem Feuerhaken über die Stirn bekommen [...]. (Parfum, 25) c. Tränen der Erleichterung laufen Hannah über das Gesicht. (FOC 11/95, 173) d. Jedes Kind mit Videospielgerät kann heute zu Hause den Golfkrieg nachspielen oder Dinosaurier über den Bildschirm hin- und herschieben [...]. (SPI 33/94, 138)

Im Gegensatz zu den bisher besprochenen Beispielen sind die LMn von über in den Beispielen unter (14) nicht horizontal wie die Erdoberfläche, sondern vertikal oder schräg orientiert. Aus diesem Grund werden sie von Deane (1993) treffend als vertikale Flächen bezeichnet. Ebensowenig wie die Orientierung der LM spielt auch die Bewegungsrichtung selbst eine Rolle. So sind in den Sätzen sowohl die laterale Richtung (14a, b) als auch die vertikale Richtung (14c) und unbestimmte Richtung (14d) vertreten. Ausschlaggebend für den Gebrauch von über ist nur, daß der Weg relativ zur LM lokalisiert ist, und umgekehrt auch, daß die LM von der Bewegung des TRs betroffen ist, was vor allem an den Sätzen (14b) und (14c) deutlich wird. Im Gegensatz zu den Beispielen in (10)—(13) ist die LM hier

7

Anders verhält es sich, wenn zwischen TR und LM ein Abstand liegt. Man vergleiche unter diesem Gesichtspunkt die oben unter (12) aufgeführten Beispiele. In Fällen wie diesen kommt nur ein horizontal orientiertes (in der Regel geographisches) Objekt als LM in Frage, weil sich die Konstellation nur unter Bezugnahme auf die Gesetze der Schwerkraft und deren Überwindung erklären läßt. So kann man zwar sagen: Die Fliege krabbelt an der Wand entlang, aber nicht *Die Fliege fliegt über die Wand.

166

Birgitta Mee:κ

nur eine Kontaktfläche, nicht jedoch auch ein Stützpunkt für die sich erstreckende Bewegung.

6. Perfektive Bewegung: Zielorientierte Wege im AUSGANGSPUNKT-PFADENDPUNKT Schema

Im vorigen Abschnitt haben wir Beispiele erörtert, in denen der von über bezeichnete Weg imperfektiven, unveränderlichen Charakter hat. Wie oben bereits ausgeführt wurde, bedeutet dies, daß die Pfadbewegung durativ, d.h. zeitlich nicht strukturiert konzeptualisiert wird und folglich als nicht abgeschlossen gilt.8 Dieselbe sich erstreckende Bewegung kann jedoch auch veränderlich, d.h. als auf einen Endpunkt zielend, konzipiert werden (vgl. Saile 1984; Taylor 1993). Vgl. hierzu (15a) gegenüber (15b): (15)

a. Ein weißes Schiff gleitet über die silbrige Fläche, die Schweizer Fahne klirrt im Wind. (SPI 33/94, 158) b. Mit dem Tretboot über den Atlantik zu fahren, um auf die Nöte eines Indianerstammes aufmerksam zu machen - das muß einem erst mal einfallen. (BRI 15/92, 119)

Während die summative Bewegung in (15a) in ihrem kontinuierlichen Verlauf relativ zur Bezugsfläche gesehen wird, ohne daß (auch nicht etwa implizit) auf eine Ausgangs- und/ oder Zielangabe Bezug genommen wird, wird sie in (15b) als Wegschema, d.h. als Teil des einen Weg umfassenden AUSGANGSPUNKT-PFAD-ENDPUNKT Schemas profiliert. Somit ändert sich auch der aspektuelle Charakter der Situation. Die Bewegung in (15b) wird nicht länger wie in (15a) als imperfektiv, sondern als perfektiv verstanden. Im Fall von über impliziert die perfektive Lesart, daß ein außerhalb der LM liegendes Ziel erreicht wird9 oder daß wenigstens die Grenzen der LM überquert werden. In beiden Fällen wird der Raumbereich der LM verlassen, nicht jedoch der Suchbereich von über, der sich jenseits der LM unendlich weit erstrecken kann. Wenn der TR einmal in den Suchbereich von über eingetreten ist, liegt die Relation zwischen TR und LM endgültig vor. Der TR bleibt also auch dann in dem mit über assoziierten Suchbereich, wenn er sich zu einem jenseits der LM liegenden Ziel hinbewegt, das explizit genannt werden kann oder nicht. Dies erklärt übrigens auch, warum über mit dem Akkusativ verbunden wird (vgl. Draye 1996). Auf beide Möglichkeiten (Route/Passage und Überquerung) wird in 6.1 bzw. 6.2 ausführlicher eingegangen.

8 9

Insofern ist das Ausmaß einer imperfektiven Bewegung unbestimmt. Auch nach Saile (1984) und Taylor (1993) impliziert das Erreichen eines Ziels einen perfektiven Prozeß.

167

Die Wegpräposition über

6.1. Route/Passage In den Beispielen unter (16) kennzeichnet über den von einem Ausgangspunkt auf ein Ziel hin verlaufenden Wegabschnitt, den wir fortan als Route bezeichnen werden und der sich graphisch wie in Abbildung 4 darstellen läßt: (16)

a. Samoiski zufolge flöß der Ölteppich über die Nebenflüsse Usa und Kolwa in die Pechora. (WELT 1.11.94, 7) b. Ginge die Fahrt über die Warschauer Brücke, so kostete die gesamte Strecke etwa 30 Millionen Mark. (BZ 2.2.96, 19) c. Über die Landstraße zur Siedlung kamen Tiefflieger am heilichten Tage und beschossen die Straßenbahn und die Radfahrer und die Wagen mit dem roten Kreuz. (Achim, 102) d. Über die Wolkengebirge schwebt eine Schar Zugvögel aus südlichen Gastländern ins Nistland zurück. (Bienkopp, 187)

Ausgangspunkt Q.Weg.des TRs

LM Abb. 4: über: summative-perfektive

Wegangabe - Route

Die vom TR verfolgte Route hat folgende Merkmale: -

-

-

Sie profiliert den intermediären Wegabschnitt 10 zwischen dem Ausgangspunkt und dem Endpunkt oder Ziel, der sich wie bei der imperfektiven Wegangabe summativ erstreckt. Im Gegensatz zur imperfektiven Wegangabe ist ihr Verlauf durch das ihr übergeordnete Wegschema gebunden: Die Bewegung ist von einem nicht notwendig genannten Ausgangspunkt und Endpunkt begrenzt und gilt als vollendet, sobald der TR den Endpunkt erreicht hat. Die Charakterisierung eines Weges als Route kann also nur unter Bezugnahme auf seinen expliziten oder impliziten Ausgangspunkt und Endpunkt erfolgen. Demzufolge wird der TR dem in der LM genannten Ort nur vorübergehend zugeordnet, was dem Satz perfektiven Charakter verleiht. Da die LM typischerweise ein geographisches Objekt mit einer wegähnlichen Gestalt ist (Nebenfluß, Brücke, Straße), liegt es auf der Hand, daß die LM vom TR als Route benutzt wird. Es ist wiederum irrelevant, ob TR und LM in einer Kontaktbeziehung zueinander stehen ( 16a,b) oder nicht ( 16c,d).

Von dem Verwendungstyp der Route ist derjenige der Passage zu unterscheiden. Beide Verwendungstypen unterscheiden sich jedoch lediglich in der Konzeptualisierung ihrer LM, die einmal eindimensional als Weg, ein andermal punktuell als Ort konzeptualisiert wird. Die folgenden Beispiele illustrieren Fälle, in denen die LM die punktuell konzeptualisierte Entität darstellt, die der TR passiert, wenn er sich von einem Ort (dem Ausgangspunkt, z.B. Frankfurt) zu einem anderen Ort (dem Ziel, z.B. Bosnien) bewegt:

10

Den Terminus verdanken wir Becker (1994).

168 (17)

Birgitta

Meex

a. Menschentrauben stehen um die Busse am Stuttgarter Omnibusbahnhof, deren Weg von Frankfurt über Stuttgart in den Krieg nach Bosnien fuhrt [...]. (RM 10.3.95, 27) b. Ich fliege diesmal über Miami und Merida, Yucatan, w o man fast täglich Anschluß nach Caracas hat, und unterbreche in Merida. (Homo faber, 206)

Hier deckt sich der von über bezeichnete Weg nicht mehr völlig mit der LM, sondern fällt nur teilweise mit ihr zusammen. In Fällen wie diesen fungiert über als unmarkierte, eine allgemeine Passagerelation als solche kennzeichende Wegangabe, die in Bezug auf die Form des Weges unterspezifiziert ist (vgl. Marcq 1972, 1988):

Ausgangspunkt °WegdesTRs

Endpunkt ·

Abb. 5: über: summative-perfektive

Wegangabe

-

Passage

6.2. Überquerung Eine zweite Möglichkeit der perfektiven Wegangabe ist die der Überquerung. In den unter dem Begriff Überquerung subsumierten Verwendungen von über ist die Bewegung als inhärent veränderlich zu verstehen, und zwar nicht, weil der TR an einem bestimmten Ziel ankommt und der Weg also abgeschlossen ist, sondern weil der TR auf seinem Weg, der von einem in der Regel unspezifizierten Ausgangspunkt zu einem in der Regel unspezifizierten Endpunkt verläuft, mit einem Hindernis oder einer Begrenzung konfrontiert wird, das bzw. die passiert werden soll. Durch das Konzept des Hindernisses ist die Überquerungslesart von über nicht nur räumlich, sondern auch funktional motiviert, wodurch sie sich von der reinen Wegangabe abhebt. Relevant ist nur, daß das Hindernis passiert wird; die Art und Weise, wie dies geschieht (z.B. springend, kletternd, fahrend, fliegend) ist unerheblich:

TR—

LM

Abb. 6: über: summative-perfektive Wegangabe Überquerung eines Hindernisses

-

Wie oben gezeigt wurde, ist die Erdoberfläche der Default-Grund, auf dem wir Menschen uns infolge der Gravitationskraft fortbewegen. Auf dieser Erdoberfläche findet man nicht selten Hindernisse vor, die es zu überwinden gilt,11 wenn man seinen Weg fortsetzen möchte. Natürliche Hindernisse sind typischerweise in der Erdkruste verankert, egal ob sie vertikal (z.B. Berge, Felswände, Hänge, Schluchten) oder horizontal (Wassermassen oder

11

Daneben gibt es noch andere Ausweichmöglichkeiten. Hindemisse können nicht nur sondern auch durchquert und umgangen oder beseitigt werden.

überquert,

Die Wegpräposition über

169

Wasserläufe mit einer horizontalen Oberfläche wie z.B. Seen, Ozeane, Buchten, Tümpel, Flüsse, Bäche, daneben auch Wege und Straßen) orientiert sind: (18) a. Über die verschneiten Berge wollte er wandern, ohne Rast, hundert Meilen weit in die Auvergne, und dort in seine alte Höhle kriechen und sich zu Tode schlafen. (Parfum, 244) b. Die Flaschen kommen übrigens im Container (zusammen mit Wein) über den Atlantik, (e&t 5/92, 110) c. Zehn Minuten dauert der schwerbewachte Trip über die Bucht von Havanna nach Casablanca am anderen Ufer. (SPI 33/94, 115) d. Wenn ich abends zu Hause war, hat er mich oft zum Bierholen in die Dorfwirtschaft über die Straße geschickt, dann konnte er ohne Lebensgefahr das Radio auf die verbotene Welle schalten. (Achim, 103) Neben geographischen Objekten im Sinne von Herskovits (1986) können auch andere befestigte oder unbefestigte Objekte dem TR die Passage erschweren. Ganz analog den natürlichen Hindernissen sind diese Objekte entweder entlang der vertikalen (19a-d) oder horizontalen (19e) Dimension orientiert: (19) a. Er hüpft über gefällte Stämme zu seinen Männern an der Motorsäge. (Bienkopp, 17) b. Er erkannte den Rest des Treppenhauses, stieg über die Trümmer langsam dorthin: Er war zu Hause. (FAZ 2.6.92, 34) c. Bei einer Preisverleihung fuhr er Prinz Charles über die Füße. (FOC 11/95, 186) d. Grenouille durchquerte den Garten und stieg über die Mauer. (Parfum, 311) e. Achim war über die Grenze gegangen und hatte Geld verbracht gegen das Gesetz. (Achim, 234) Kreitzer (1997) nimmt für das englische over an, daß ein Objekt kein Hindernis im strikt visuell-geometrischen Sinne des Wortes darstellen muß, um als LM von over in Betracht zu kommen. Es genügt, wenn die LM unter dem Aspekt der Kraftdynamik {force dynamics, vgl. Talmy 1988) schematisiert wird und einen Widerstand gegen die Pfadbewegung leistet, der nur durch das Ausüben von einer stärkeren Kraft überwunden werden kann. So stellen auch Grenzen oft Hindernisse dar, da sie nicht ohne weiteres zu überqueren sind. Gleiches gilt für über (19e), das sich jedoch in einer wichtigen Hinsicht von seinem englischen Pendant unterscheidet: Das LM-Objekt kann, muß aber nicht als Hindernis konzipiert werden. Es ist durchaus möglich, daß die LM einfach ein geographisches oder nicht-geograpisches Objekt bezeichnet, das nicht als Hindernis aufgefaßt werden kann. Relevant fur den Gebrauch von über ist nur, daß sich der TR von einem Ende der LM zum anderen bewegt und dabei die (Begrenzung der) LM überquert. Vgl. folgende Auswahl: (20) a. Sie sahen zusammengehöriger aus je weiter sie sich entfernten quer über das offene Feld aus verwittertem Asphalt. (Achim, 20) b. Und lief weg über den leergebombten Platz, die Schultasche schlug ihm gegen die Beine das Herz in die Kehle. (Achim, 139) c. Das Sonntag nacht über Hamburg hinweggezogene Sturmtief "Rotraud" hat viele Bäume entlaubt, deren Blätter zahlreiche Siele verstopften. (WELT 1.11.94, Hl) d. Wenig war geholfen mit dem Strich, den Karsch über das ungünstige Wort zog [...]. (Achim, 185)

170

Birgitta

Meex

e. [...] vierjährig lag er über Fensterbrett und Sims hinausgebeugt und starrte in die Straßenschlucht hinunter. (Achim, 94) f. Meine DC-4: nach Mexico-City, sie flog gerade über uns hinweg, dann Kurve aufs offene Meer hinaus, w o sie im heißen Himmel sich sozusagen auflöste wie in einer blauen Säure - (Homo faber, 42)

Wie die Beispiele unter (18) bis (20) zeigen, wird die perfektive Lesart des Überquerens in vielen Fällen durch den innersprachlichen Kontext hervorgerufen oder doch wenigstens verstärkt. Vergleichen wir dazu folgendes Minimalpaar: (21)

a. Ein Sturmtief zieht über Hamburg. b. Ein Sturmtief zieht über Hamburg hinweg.

Während (21a) ohne weiteren Kontext eine imperfektive Interpretation nahelegt, ist (21b) aufgrund des deiktischen Richtungsadverbs hinweg nur perfektiv interpretierbar. Über ... (hin)weg (20c, f) impliziert ebenso wie über ... hinaus (20e), daß der TR die LM hinter sich läßt und seinen Weg jenseits der LM fortsetzt. Auch mit einem Verb wie kommen (18b), sich entfernen (20a) oder schicken (18d) kann eine perfektive Wirkung erzielt werden, da solche Verben vorzugsweise eine perfektive Wegvorstellung hervorrufen.

7. Subjektivierung

7.1. Subjektive Bewegung In seinen bisher besprochenen dynamischen Verwendungen bezeichnet wegbezogenes über räumliche Konfigurationen, in denen der TR eine in der Zeit kontinuierliche Reihe von Positionen relativ zu einer LM einnimmt. In ihrer Summierung bilden diese Positionen eine Wegstrecke. Die so profilierte Wegrelation wird von dem außerhalb des Skopus der Prädikation liegenden Sprecher objektiv konzeptualisiert (vgl. Langacker 1991). Nun kann eine summativ entstehende Wegstrecke nicht nur objektiv über die sukzessive Bewegung des TRs konzeptualisiert werden, sondern auch subjektiv, d.h. durch rein mentales Scanning vom Konzeptualisierer vollzogen werden. Langacker beschreibt diesen Prozeß als subjective motion. Hinsichtlich der Bedeutung von über ließe sich dies auch als subjektivierende Bedeutungsextension beschreiben. 12 Betrachten wir etwa folgende stative Sätze mit über, die jeweils subjektive Extensionen einer der drei objektiv konstruierten Wegangaben ('bedeckender Weg,' 'Route/Passage,' 'Überquerung') illustrieren: - Bedeckender Weg: (22)

12

Dieser Student trug sogar am Sonntag ein Pflaster quer über seine Wange und schien Blut und Wunden nicht zu furchten. (Bienkopp, 55)

Die Extension ist nicht auf über beschränkt, sondern stellt ein produktives Extensionsschema nicht nur deutscher, sondern auch u.a. englischer und niederländischer Wegpräpositionen dar (vgl. Langacker 1991, Geeraerts 1992).

Die Wegpräposition

-

171

über

Route/Passage: (23)

Die über

Trnovo laufende Straße hat entscheidende Bedeutung

für den

serbischen

Nachschub. (WELT 1/11/94, 3)

-

Überquerung: (24)

[ . . . ] bunt leuchteten neben dem jungen Laub die roten Schrifttücher an den Wänden und über die Fahrbahn gespannt. (Achim, 107)

In den Beispielen unter (22) bis (24) beschreibt wegbezogenes über eine einzige, statische Konfiguration mit einem unbeweglichen, relativ langgestreckten TR, der die einen Weg ausmachenden punktuellen Orte nicht wie ein sich bewegender TR sukzessiv, sondern gleichzeitig einnimmt. Insofern besteht eine konzeptuelle Ähnlichkeit zwischen einem nulldimensional sich bewegenden TR und einem eindimensional sich erstreckenden unbeweglichen TR (vgl. Lakoff 1987). Durch den Gebrauch von über in Verbindung mit einem Verb der Bewegung oder Erstreckung 13 wird die statische Längserstreckung des TRs im Sinne einer dynamischen Bewegung entlang dessen Erstreckung konzeptualisiert. Die Subjektivierung der Bewegung erfolgt dadurch, daß der Sprecher die Erstreckung subjektiv verfolgt, so als ob er die Strecke selbst visuell und motorisch zurücklegen würde (vgl. Kreitzer 1997). Die Relation ist nach wie vor summativ; die LM wird somit akkusativisch markiert. Die Bedeutungsextension durch Subjektivierung betrifft sowohl perfektiv als auch imperfektiv interpretierte Sätze. Im Gegensatz zu den sich objektiv erstreckenden Wegen, auf denen, wie oben gezeigt wurde, entweder eine perfektive oder imperfektive Bewegung stattfinden kann, ist die Bewegung auf subjektiv konstruierten Wegen nur imperfektiv, da sie ein stabiles und zeitlich kontinuierliches Verhältnis profiliert (vgl. Langacker 1991).

7.2. Subjektiver Referenzpunkt Eine andere Art der Subjektivierung im Sinne von Langacker (1991: 326-330) liegt den Sätzen in (25) zugrunde. (25)

a. Die Zwerge in Schneewittchen wohnen über den Bergen. 1 4 b. "Früher", erinnert er sich, da habe einer schnell wieder gute Arbeit gleich über der Straße im anderen Werk gefunden [...]. (SüdZ, 31.7.97, 40) c. Beliefert wird die Fahrrad- und Maschinenbauindustrie auch über den Grenzen der BRD mit Lacksystemen für jeden Anwendungszweck. (SPI 33/94, 150)

Hier ist nicht wie in (22)-(24) von einer subjektiven, d.h. rein mental vollzogenen Bewegung, sondern von einem subjektiv konstruierten Referenzpunkt auszugehen. Nach

13

Auch wenn ein Verb der Bewegung oder Erstreckung nicht explizit im Satz erscheint, läßt es sich implizit ergänzen: z.B. Das Pflaster läuft quer über seine Wange.

14

Für diesen Gebrauch von über habe ich in meinem Korpus kaum Belege finden können. Dies verwundert nicht, wenn man bedenkt, daß wir es hier mit einer sehr markierten Konstruktion zu tun haben, sowohl im Hinblick auf die Akzeptabilität als auch die Frequenz. Standardsprachlich heißt es anstelle von über meistens jenseits.

172

Birgitta

Meex

Langacker (1991: 329) ist die Subjektivierung dann darin zu sehen, daß der Referenzpunkt mit dem ground, insbesondere mit der Position des Sprechers gleich-gesetzt wird. Der implizite Referenzpunkt in (22)-(24), von dem aus die Szene betrachtet wird, entspricht in (25) der gedachten oder aktuellen Sprecherposition und wird somit subjektiv gestaltet. Zur Lokalisierung des TRs ist in (25) neben dem Bezug auf die profilierten relationalen Entitäten (TR und primäre LM) der Bezug auf eine unprofilierte dritte Stelle (den subjektiven Referenzpunkt) notwendig. So ist auch der Unterschied zwischen (22)-(24) und (25) zu verstehen: In (22)-(24) werden die Konstellationen jeweils von einem externen, d.h. außerhalb der Szene liegenden Referenzpunkt aus betrachtet, in (25) hingegen von einem nicht explizierten subjektiven, d.h. unmittelbar an der Szene beteiligten Referenzpunkt aus. So lassen sich die Zwerge in (25a) nur relativ zur deiktischen Sprecherperspektive jenseits der Berge lokalisieren. Dies bedeutet, daß der Sprecher die Szene nicht lediglich offstage betrachtet, sondern auch onstage an ihr beteiligt ist. Die Beispiele unter (25) unterscheiden sich auch noch in anderer Hinsicht von den subjektiv konstruierten Wegangaben in (22)-(24): - Der TR ist punktuell (d.h. dimensionsneutral), nicht langgestreckt. - Die LM wird immer als ein Hindernis konzeptualisiert, das nicht einfach zu überwinden ist. - Der TR wird relativ zur Sprecherposition auf der anderen Seite der LM (Berge, Straße, Grenzen) lokalisiert (siehe Abbildung 7). - Die Verbindung mit einem perfektiven Verb der Bewegung oder Erstreckung ist ausgeschlossen. - Über regiert den Dativ und nicht den Akkusativ. Daraus können wir schließen, daß über nicht länger eine summative Wegstrecke, sondern ein nicht-summatives Lageverhältnis jenseits der LM bezeichnet. Somit ist über hier eher als Ortsangabe denn als Wegangabe zu betrachten.15 Dies erklärt, warum die LM von über dativisch markiert wird. Nach diesen Überlegungen scheint es uns auch plausibel, diesen Gebrauch von über als das dativische Gegenstück zur summativ-überquerenden Lesart in (18) aufzufassen. Während über in (18) den Fokus auf den summativen Weg richtet, auf dem sich der TR allmählich in den Suchbereich von über jenseits der LM begibt, signalisiert über in (25), daß sich der TR unverändert im Suchbereich, d.h. jenseits der als 15

Eine alternative Sichtweise bestünde darin, die Subjektivierung in (25) nicht nur auf die deiktische Sprecherperspektive, sondern zusätzlich auch auf mentales Scanning zurückzufuhren. Diese Ansicht vertritt etwa Claes (1998), nach der Fälle wie (25) die subjektive Bewegung entlang einer Wegstrecke illustrieren. Wie u.a. Lakoff (1987), Langacker (1991) und Taylor (1993, 1995) gezeigt haben, liegen beide oben angesprochenen Arten der Subjektivierung (d.h. subjektiver Referenzpunkt sowie mentales Scanning) der entsprechenden Verwendungsweise der englischen Präposition over in Ausdrücken wie Sam lives over the hill (Lakoff 1987: 440) zugrunde. Für das Deutsche scheint uns eine solche Sehweise nur gerechtfertigt, wenn der Akkusativ gebraucht wird und die Relation somit summativ kodiert wird: Die Diebe waren längst über alle Berge. Nach Leys (1995: 47) ist der Akkusativ nach sein nur möglich, wenn die üóer-Strecke objektiv, d.h. wirklich zurückgelegt wurde. Tatsächlich kann zu dem Hilfsverb des statischen Verbs sein ein dynamisch-perfektives Bewegungsverb als Partizip Perfekt (wie z.B. gegangen, gefahren, geflohen o.ä.) ergänzt werden. Leys (ebenda) begründet den Akkusativ in diesem Satz dann damit, daß die Strecke mental nachvollzogen wird.

Die Wegpräposition

173

über

(objektiver) Referenzpunkt fungierenden LM befindet, und lediglich der Blick des Betrachters auf ihn gelenkt wird.

TR

LM Abb. 7: über: Ortsangabe

: Subjektiver

Referenzpunkt

8. Schlußbemerkungen

In diesem Beitrag haben wir versucht, die verschiedenen von über bezeichneten Wegrelationen16 zu identifizieren und als kognitiv motiviert zu erklären. Dabei hat sich der Bewegungsaspekt als besonders relevant erwiesen. Die Detailuntersuchungen zur Semantik von wegbezogenem über haben in diesem Zusammenhang gezeigt, daß Wege, welche eine imperfektive, summativ-bedeckende Bewegung in Bezug auf die äußere Fläche eines Gegenstands (prototypisch die Erdoberfläche) kennzeichnen, die allgemeine, unmarkierte Wegrelation darstellen. Dies läßt sich zweifach begründen. Erstens können Wege, auf denen eine perfektive Bewegung stattfindet, oft nur über den Kontext eindeutig festgelegt werden, was dafür spricht, die Pfadbewegung primär unter ihrem durativen, unveränderlichen Aspekt zu konzipieren. Zweitens scheint es uns, aus der Überlegung heraus, daß sich eine spezifische Bedeutung in vielen Fällen einfacher aus einer allgemeineren herleiten läßt als umgekehrt, gerechtfertigt, die spezifischeren Wegrelationen der Route/Passage und der Überquerung metonymisch aus der imperfektiven Weglesart abzuleiten. Auch die Bedeckungslesart, die hier nicht diskutiert wurde, läßt sich über das Konzept der bedeckenden Erstreckung primär auf die imperfektive Lesart beziehen.

Literatur

1. Korpusmaterial Achim: AMS: bdw: Bienkopp:

16

U w e Johnson, Das dritte Buch über Achim , Frankfurt a.M., Suhrkamp Verlag (1961). auto motor und sport 1/94 bild der Wissenschaft 8/95 Erwin Strittmatter, Ole Bienkopp , Westberlin, Aufbau Verlag (1963).

Sowie ihre ortsbezogene, dativische Entsprechung.

174 BILD: BRAVO: BRI: BZ: Clown: e&t : FAZ: FOC: hb: Homo faber: kosmos : Parfum: RM: SPI: stern: Südz.: SZ: WELT:

Birgitta Meex Bild 10.6.95 bravo 17/95 Brigitte 15/92 Berliner Zeitung 2.2.96 Heinrich Boll, Ansichten eines Clowns, Köln - Berlin, Verlag Kiepenheuer & Witsch (1963). essen & trinken 5/92 Frankfurter Allgemeine 2.6.92 Focus 11/95 Handelsblatt 25/8/95 Max Frisch, Homo faber, Frankfurt a.M., Suhrkamp Verlag (1957). ¡cosmos 10/95 Patrick Süskind, Das Parfum. Die Geschichte eines Mörders , Zürich, Diogenes Verlag (1994). Rheinischer Merkur 10.3.95 Der Spiegel 33/94 stern 33/95 Süddeutsche Zeitung 2.8.95 Sächsische Zeitung 16.3.95 Die Welt 1.1.94

2. Sonstige Literatur Becker, Angelika (1994): Lokalisierungsausdrücke im Sprachvergleich: Eine lexikalisch-semantische Analyse von Lokalisierungsausdrücken im Deutschen, Englischen, Französischen und Türkischen. - Tübingen: Niemeyer (= Linguistische Arbeiten 316). Bellavia, Elena (1996): The German über. - In: M. Pütz & R. Dirven (eds.): The Construal of Space in Language and Thought, 73-107. Berlin: Mouton de Gruyter. Boers, Frank (1996): Spatial Prepositions and Metaphor: A Cognitive Semantic Journey along the Up-Down and the Front-Back Dimensions. - Tübingen: Narr. Bouillon, Henri (1978): Kontrastive Syntagmatik der deutschen Präpositionen auf und über. - In: J. Lerot & R. Kern (Hgg.): Mélanges de linguistique et de littérature offerts au Professeur Henri Draye à l'occasion de son éméritat, 3-13. Louvain: Bibliothèque de l'Université, Bureau de Recueil. Brugman, Claudia (1983): Story of Over. - M.A. thesis. University of California at Berkeley. Reproduced by the Indiana University Linguistics Club, Bloomington, IN. Claes, Karoline (1998): Das räumliche um: Semantische und syntaktische Aspekte einer Wegpräposition. - Dissertation. Katholieke Universiteit Leuven. Cuyckens, Hubert (1991): The Semantics of Spatial Prepositions in Dutch. A Cognitive-Linguistic Exercise. - Unpublished Ph.D. dissertation. University of Antwerp (UIA). Deane, Paul D. (1993): Multimodal Spatial Representation: On the Semantic Unity of Over and Other Polysemous Prepositions. - Duisburg: L.A.U.D. (Series A, Paper No. 332). Dewell, Robert B. (1994): Over Again: Image-Schema Transformations in Semantic Analysis. - In: Cognitive Linguistics 5, 351-380. Di Meóla, Claudio (1999): Adpositions in Present-day German: Prototypes, Grammaticalization, and Synchronic Variation. - Unveröff. Manuskript. Paper presented at the sixth International Cognitive Linguistics Conference, Stockholm.

Die Wegpräposition

über

175

Dirven, René (1981): Spatial Relations in English. - In: G. Radden & R. Dirven (Hgg.): Kasusgrammatik und Fremdsprachendidaktik, 103-132. Trier: Wissenschaftlicher Verlag (= Anglistik und Englischunterricht 14). Draye, Luk (1996): The German Dative. - In: W. Van Belle & W. Van Langendonck (eds.): The Dative. Vol. 1: Descriptive Studies, 155-215. Amsterdam: John Benjamins. Geeraerts, Dirk (1992): The Semantic Structure of Dutch Over. - In: Leuvense Bijdragen 81, 2 0 5 230. Hawkins, Bruce (1985): The Semantics of English Spatial Prepositions. - Ph.D. Dissertation. University of California at San Diego. Reproduced by L.A.U.T., Trier (Series A, Paper No. 142). Herskovits, Annette (1986): Language and Spatial Cognition: An Interdisciplinary Study of the Prepositions in English. - Cambridge: Cambridge University Press. Kreitzer, Anatol (1997): Multiple Levels of Schematization: A Study in the Conceptualization of Space. - In: Cognitive Linguistics 8, 291-325. Lakoff, George (1987): Women, Fire, and Dangerous Things. What Categories Reveal about the Mind. - Chicago, IL: The University of Chicago Press. Langacker, Ronald W. (1987): Foundations of Cognitive Grammar. Vol 1: Theoretical Prerequisites. - Stanford, CA: Stanford University Press. - (1991): Concept, Image, and Symbol: The Cognitive Basis of Grammar. - Berlin: Mouton de Gruyter. Leys, Odo (1989): Aspekt und Rektion räumlicher Präpositionen. - In: Deutsche Sprache 17, 97-113. - (1995): Dativ und Akkusativ in der deutschen Sprache der Gegenwart. - In: Leuvense Bijdragen 84, 39-62. Lindner, Susan J. (1983): A Lexico-Semantic Analysis of English Verb Particle Constructions with OUT and UP. - Ph.D. dissertation. University of California at San Diego. Reproduced by L.A.U.T., Trier (Series A, Paper No. 101). Lindstromberg, Seth (1997): English Prepositions Explained. - Amsterdam: John Benjamins. Marcq, Philippe (1972): Prépositions spatiales et particules "mixtes" en allemand. - Paris: Librairie Vuibert. - (1988): Spatiale und temporale Präpositionen im heutigen Deutsch und Französisch. - Stuttgart: Hans-Dieter Heinz Akademischer Verlag (= Stuttgarter Arbeiten zur Germanistik 195). Meex, Birgitta (2001): The Spatial and Non-Spatial Senses of the German Preposition über. - In: H. Cuyckens & Β. Zawada (eds.): Polysemy in Cognitive Linguistics: Selected papers from Fifth International Cognitive Linguistics Conference, Amsterdam 1997, 1-35. Amsterdam: John Benjamins. Radden, Giinter (1981): Die übertragenen Bedeutungen der englischen Raumpräpositionen. - In: G. Radden & R. Dirven (Hgg.): Kasusgrammatik und Fremdsprachendidaktik, 133-179. Trier: Wissenschaftlicher Verlag (= Anglistik und Englischunterricht 14). Saile, Günter (1984,): Sprache und Handlung: Eine sprachwissenschaftliche Untersuchung von Handhabe-Verben, Orts- und Richtungsadverbialen am Beispiel von Gebrauchsanweisungen. Braunschweig/Wiesbaden: Friedr. Vieweg & Sohn. Schulze, Rainer (1990): Sprachliche Raumwahrnehmung: Untersuchungen zu ausgewählten Präpositionen des Englischen im Rahmen der kognitiven Semantik. - Habilitationsschrift. Universität Essen, Deutschland. Talmy, Leonard (1988): Force Dynamics in Language and Cognition. - In: Cognitive Science 12, 4 9 100.

176

Birgitta Meex

Taylor, John R. (1993): Prepositions: Patterns of Polysemization and Strategies of Disambiguation. In: C. Zelinski-Wibbelt (ed.): The Semantics of Prepositions: From Mental Processing to Natural Language Processing, 151-175. Berlin: Mouton de Gruyter. - (1995 2 ): Linguistic Categorization: Prototypes in Linguistic Theory. - Oxford: Clarendon Press. Vandeloise (1991): Spatial Prepositions: A Case Study from French. - Chicago, IL: The University of Chicago Press.

Walter De Mulder & Anne

Vanderheyden

The Evolution of French Sur. Toward a Diachronic Cognitive Semantics

1. A short history of sur

According to most historical surveys, the Modern French preposition sur results from a fusion of Old French sor and sus. Old French sor (along with its variants sovre, sore, sor, sour, seur, sur, and some other less frequent forms) stems from Latin superb while sus derives from Latin susum (sursum in Classical Latin). 2 Omitting a lot of detail - cf. the large number of variants cited above - three steps can be distinguished in the in the morphological evolution of sur: 1. Until the fourteenth and fifteenth centuries, the most frequent preposition is sor (and its variants). 2. From the thirteenth century on, sus tends to be used more frequently than sor and becomes the most frequent form in the fourteenth and fifteenth centuries; Bekkers (1931: 218), for instance, signals that Froissart (a French medieval writer) uses sus ten times more than sur, 3. In the sixteenth century, sus starts to be used less and less frequently compared to sur; in the seventeenth century, it is mostly used as an adverb and only occasionally as a preposition. This paper does not deal with formal changes; rather, its aim is to investigate the semantic evolution of the forms sor, sus, and sur and to examine the relevance of diachronic prototype semantics - as presented in Geeraerts (1997) - to the study of the evolution of spatial prepositions. It has indeed been frequently stated that the polysemy of current linguistic expressions reflects their diachronic semantic evolution (Geeraerts 1997: 6, 24).3 Consequently, we want to look into the evolution of the preposition sur in order to find the cognitive links between its various meanings (Traugott 1985: 156). Our starting point will be the now widely accepted idea that the basic meaning of prepositions such as sur is spatial,4 and

1

Tobler & Lommatzsch (1971, s.v. sor) also mentions supra as the origin o f sur. According to Robert Historique (1992, 5. v. sur), sour and sor stem from super, whereas sobre and later on sore stem from supra and give rise to the uses of sur to mark domination, to express verticality without contact, and to introduce the name of a part of the body one is supported by. Etienne (1980: 167) mentions supra as well as super as the origin of sovre.

2

Waldmann (1906: 3), however, holds that in Old French sur was already used alongside sor in Norman and Anglo-Norman texts as well as in the île de France region.

3

Geeraerts (1997: 28, note 1) also refers to Brown & Witkowski (1983: 83), Sweetser (1990: 3), and Lüdtke (1985: 357).

4

Cf., e.g., Groussier(1997).

Walter De Mulder & Anne

178

Vanderheyden

that other meanings are derived from this spatial meaning b y metaphorical and m é t o n y m i e transfers and other mechanisms o f semantic change. 5

1. A s p a t i a l p r o t o t y p i c a l m e a n i n g ?

Since Old French sor is derived from Latin super, it should c o m e as n o surprise that sor has preserved the t w o spatial meanings o f super, which m a y denote a vertical relation between two objects (i) that are in contact or (ii) that are not in contact. In sor's first meaning, the trajector, the object to be localized, and the landmark, the object used to localize the trajector, are in contact. The landmark then mostly serves to support the trajector (Godefroy 1892: 518; Moignet 1976: 324): (1)

a. Aucassins fu armés sor son ceval 'Aucassin was armed on his horse' 6 b. Sur un perrun de marbre bloi se culcet 'He lies down on a block of grey marble' c. Sur piez se drecet, mais il ad grand dolur 'He got up on his feet again, but had great pain'

Aucassin et Nicolette (10, 1) Chanson de Roland ( 12) Chanson de Roland (3324) (examples cited by Moignet 1976: 324)

In its second meaning, as pointed out b y Godefroy (1892: 518), sor signals that the trajector is higher than the landmark, but not supported b y it: 7 (2)

a. Gabriels lievet sa main, sur lui fait son signacle Chanson de Roland 'Gabriel lifts up his hand, makes on him the sign of the cross' (2847) b. Li rois a fait sor aus tendre le pavillon Roum. D'Alix (f° 31 d) 'The king has made tents put up over them' (examples cited by Godefroy 1892: 518, i. v. sour)

Middle and Modern French also allow sus and sur to have both meanings, i.e. expressing a vertical relation between trajector and landmark with (3, 5) and without contact (4, 6): (3)

5 6 7

8

Li chastiaus de Haindebourch siet sus une haute roce, par quoi on voit tout le pays d'environ The castle of Haindebourch is on a high rock, from where one sees all the land around' (Froissait, Chron. II, 50, 20; cited by Bekkers 1931: 186) 8

For an overview from the standpoint of prototype semantics, see Geeraerts (1997: 84-122). The English glosses should not be read as literary translations. Sor "marque qu'un objet est au-dessus d'un autre, sans pour cela être soutenu par lui [signals that an object is higher than another one, without being supported by it]" (Godefroy 1892: 518). Bekkers (1931) analyses the use of prepositions in the first five parts of the Chroniques by Froissait, as published for the Société de l'histoire de France by Simeon Luce. Some examples are taken from the three subsequent parts of the first book.

The Evolution of French Sur

179

(4)

Et si trouvèrent plus de trois cens chaudières [...] pendues sus le feu [...] 'And there were more than three hundred similar kettles [...] hanging above the fire [...]' (Froissait, Chron. I, 71, 11 ; cited by Bekkers 1931: 187)

(5)

La tasse est sur la table. 'The cup is on the table.'

(6)

L'oiseauplane sur la vallée. 'The bird is floating above the valley.'

(Lexis 1979, s.v. sur)

These two basic spatial meanings have not been preserved in an identical fashion. In Modern French, for instance, uses of sur involving no contact between trajector and landmark are far less frequent than uses implying contact; in other words, sur normally expresses a vertical relation with contact, as is confirmed by the analyses of sur in Vandeloise (1986, 1991, 2000), where contact is almost presented as a necessary condition for the use of sur. This predominance of the uses with contact over those without is confirmed for Middle French by Bekkers (1931), whose frequency lists for sus and sur in Middle French (or rather in the Chroniques of Froissart) show that uses implying contact (467) are more frequent than those without (7). This leads him to conclude that sur is not used to express a relation on the vertical axis without contact (Bekkers 1931: 218). 9 As for Old French, a look at the uses of sor in the Romans of Chrétien de Troyes reveals that Old French also prefers sor to express contact. Thus, the use of sur to express relation with contact seems to have been more frequent from Old French on. Its use to express a relation on the vertical axis without contact has not disappeared, but is rather restricted. 10 Vandeloise (1992: 2 2 23) mentions at least two conditions, stressing at the same time the need to search for further explanations: (i) the sentence contains an expression (mostly a verb) of movement (cf. 7); (ii) the landmark is preferably inhabited, which is why (7a) seems more acceptable to Vandeloise than (8a) or (8b): (7)

a. Les oiseaux volent sur la ville. 'The birds fly over the city.' b. ILes oiseaux sont sur la ville. 'The birds are over the city.'

(8)

a. Les nuages sont sur la forêt 'The clouds are over the forest.' b. Les nuages sont sur la plaine 'The clouds are over the plain. '

The question is, of course, whether the use of sur to express contact on the vertical axis is also the prototypical one. Dubois (1983) points out that there is no necessary relation between sheer frequency and prototypicality: the prototype is the exemplar of the category that presents the most typical characteristics, those with the highest cue-validity. But since 9

10

"à la différence de sus, sur ne s'emploie pas pour exprimer la position supérieure sans contact [unlike sus, sur is not used to express higher position without contact]" (Bekkers (1931: 218); Shears (1922: 40), however, holds that the use of sur to express vertical relations without contact is more frequent in Middle French. One is left to wonder, of course, why this less frequent use still exists, especially since French audessus de seems well-suited to express a relation on the vertical axis without contact.

180

Walter De Mulder & Anne Vanderheyden

the cue-validity of a characteristic is higher if it is present in a lot of category members, 11 the very fact that sur more frequently denotes contact than no contact is evidence in favor of a semantic description of sur that accords a central status to the notion 'contact.' 12 Furthermore, at least two other criteria identifying prototypical meaning 13 suggest that sur's meaning implying contact is the more central one: prototypical meanings are the first to be learned by children, and they are the first meanings speakers use when asked to explain the meaning of a word. With regard to sur, Vandeloise (1991: 54-61) suggests, starting from a discussion of Clark (1974), that children learn the meaning of sur on the basis of a prelinguistic concept of bearer/burden; 14 in addition, evidence for on, sur's cognate in English, suggests that the meaning involving contact seems to be the first to come to mind when speakers are asked to use sur in a sentence (cf. Lindstromberg 1997: 19). However, most of the tests traditionally used to check whether a meaning has prototype status can, for obvious reasons, not be applied in diachronic semantics: there are no native speakers of Old and Middle French, nor are there children learning it. Therefore, in diachronic semantics, frequency is the determining factor of prototypicality, and is used as such by Geeraerts (1997: 35, 39). We thus conclude that the meaning of sur indicating contact on the vertical axis is the prototypical one and that the same holds for the corresponding preposition in Old and Middle French. The fact that this meaning does not undergo any changes in the course of sur' s evolution can be seen as a confirmation of this idea (Geeraerts 1997: 24, 39).

2. Monosemie or polysémie?

Vandeloise (1986: 195, 1991: 194, 2000) employs the notion 'bearer/burden' to define the basic meaning of sur in Modern French. In defining sur, this notion shows the following three characteristics: 15 (i) the bearer is generally lower than the burden; (ii) the burden is generally in contact with the bearer; (iii) the force of the bearer works against the force of gravity on the burden. These characteristics, however, should not make up a set of necessary and sufficient conditions for the use of sur. As opposed to (i), sur can express a horizontal relation as well as an 'inverse' vertical relation (where the bearer is higher than the burden): 11 12

13 14 15

For a discussion and further references, see Kleiber (1990: 75). Kleiber (1990: 108), Vandeloise (2000), and Taylor (1995: 117-118) also hold that frequency is an important characteristic of prototypical instances of a category. For a more complete list, see, e.g., Geeraerts (1986: 192-193). Mandler (1992) uses the notion 'support.' In fact, Vandeloise (1991: 194) mentions five characteristics, but his characteristic "a part of the burden is generally hidden by the burden" seems relevant for sous {under) rather than for sur, furthermore, "the bearer is generally larger than the burden" follows from the more general principle that the landmark should be more salient than the trajector.

The Evolution

(9)

of French Sur

181

la mouche sur le plafond/I 'affiche sur le murila clef sur la porte 'the fly on the ceiling/the poster on the wall/the key on the door'

Characteristic (iii) is not necessarily present either: (10)

le point sur la ligne 'the point on the line'

Old French sor as well as Middle French sus sometimes also designate a vertical relation with contact, but without support: (11)

( 12)

Soz le degret ou il gist sur sa nate 'Under the stairs where he rests on his mat' (Alexis, str. 50 a , Xies., Stengel; cited by Godefroy 1892: 518, s.v.

sour)

Quant il fu venus sus le place où il desconfi estoient, et où ses frères gisoit, si fu tous esmerveilliés 'When he arrived at the place where the defeated were, and where his brother was lying, he was really astonished' (Froiss., Chron., IV 119, 14; cited by Bekkers 1931: 188)

Vandeloise (2000: 759-760), however, holds that in comparable cases such as (13), the ground still works against the force of gravity on the stone. In his view, this is what tigers and elephants learn as they fall into a trap after having walked on the ground. (13)

la pierre est sur le sol 'the stone is on the ground'

Nevertheless, Cuyckens (1991: 199-201, 1995: 186) points out that the support relation and the idea of opposition to the force of gravity are not prominent in these cases: the ground is not first and foremost conceptualized as opposing itself to the force of gravity on the stone and Cuyckens accordingly proposes to speak of an 'underlie' relation between trajector and landmark instead of 'support.' Thus, it seems that only characteristic (ii), viz. the burden is generally in contact with the bearer, is necessarily implied by sur. This suggests that the meaning of sur is rather abstract and could be described as [contact with surface] (cf. Klein & Niise 1997: 14).16 All the other elements of meaning such as (i) and (iii), and the idea that the bearer supports the burden, would then be attributed to the context and the linguistic and extra-linguistic information associated with the trajector and the landmark that are related by sur. This kind of analysis could also account nicely for the uses of sur designating the goal or the end point of a movement.17 These can be described without having to multiply the senses of sor, sus, and sur, in that the meaning of end point or goal is mostly present when 16

Klein & Niise point out that surface has the same meaning as German Oberfläche and designates "the upper side or the largest side of an object, which is neither the back nor the underside" (Klein & Niise 1997: 14). It seems more appropriate to say that it is the active zone (Cuyckens 1991: 201) or the functionally or perceptually salient surface of the object (Zifonun et al. 1997: 2116).

17

It is a well-known property of static spatial prepositions that they can indicate a point on a trajectory - its beginning, end point, etc. - when they are accompanied by a verb of movement (Brugman 1981; Taylor 1995: 127).

182

Walter De Mulder & Anne Vanderheyden

the preposition is combined with a verb expressing movement, as in (14)—( 16); this meaning can thus be said to result from combining the abstract meaning of sur with the idea of movement expressed by the verb. ( 14) Cligès sor l'erbe qui verdoie descent a pié 'Cligès puts his foot on the green grass' (Cligès 3462-63; cited by to M.-L. Oilier 1986) (15) f...J qu'elle avoit gecté ungseau d'eau sus son amy [...] '[...] that she had thrown a bucket of water on the head of her friend [...]' (Arrêts d'amour, 112; cited by Martin/Wilmet 1980: 188) (16) Je débouche sur une place dont les efforts qu 'elle fait pour être un parc ont quelque chose d'attendrissant, d'inutile et de solennel Ί arrive at a square whose efforts (which she does) to be a park are somewhat tender, useless and solemn' (Trésor de la langue française, s.v. sur, p. 1136) Such an analysis explains, amongst other things, the difference between sus and vers in Middle French. According to Bekkers (1931: 193), sus is used in (17) because it suggests that the person or the object is close to the person who executes the action of the verb, whereas vers is used when there still is a distance to be covered to reach the place expressed by the complement of the preposition. 18 (17) [...] et passèrent derrière Raimes, et les mist en ce bois, qui regarde vers le Percelet et sus le caucie '[...] and they passed by the inferior side of Reims, and he took them into these woods, which look out onto Percelet and the road' (Froiss, Chron., II48, 7; cited by Bekkers 1931: 118, 193) The evidence thus seems suggestive of an approach to lexical semantic structure where sur has an abstract meaning which needs to be enriched by the context, i.e. a kind of monosemic two-level theory. In contrast, Cuyckens (1991, 1995) defends structuring the meanings of the Dutch preposition op - largely the Dutch equivalent of sur - as a family resemblance, even though he also defines some of op's meanings as coincidence with a surface. Cuyckens' analysis differs from the "two-level" approach in that Cuyckens also intregrates the additional notions 'support,' 'underlie,' and 'adherence' into the preposition's semantic makeup as full-blooded meanings. 19 In any case, even if one reduces the meaning proper of sur to [+ contact with a surface], as suggested by Klein & Niise (1997: 14), one still has to explain the difference between sur and other prepositions expressing 18

19

In fact, several authors hold that sur retains its static meaning when it is combined with verbs of movement and that it expresses the place where the trajector comes to rest (cf. Schifko 1977: 241, 247 and references cited there). Meyer-Lübke (1974 [1890-1906]: 486, § 433) points out that already in Latin, when the Romans had to indicate the endpoint of a movement, the state at rest following the movement was more salient than the movement. The term 'underlie' is used to characterize the relationship expressed by op in de kranten liggen op de mat/op de vloer van de woonkamer 'the newspapers are on the rug/on the living room floor' and examples such as (11 )—C12); 'adherence' characterizes the relationship in de poster zit op de muur (the poster is on the wall). There is only coincidence in an example such as ( 10).

The Evolution of French Sur

183

contact with a surface, such as à and contre. One could, of course, argue that sur expresses a different kind of contact than à and contre, but to specify the kind of contact implied, one would have to resort to the notions of verticality, support, etc. 20 Other arguments in favor of a polysémie family resemblance type of structure rather than a monosemic "abstract meaning" conception are based on Taylor (1995: 284-286): 1. We showed above that the static spatial meaning of sur designating a relation on the vertical axis with contact has a more privileged status than the one without contact. These findings seem more compatible with the prototype approach and less compatible with the abstract meaning-approach where each interpretation-in-context is derived from the same abstract meaning. 2. The specific meanings of sur are readily accessible to speakers, whereas the general, abstract meaning of sur is not. It is not at all evident, then, to hold that the general meaning would be part of the meaning of the preposition as represented in our mind, whereas the more specific meanings would not, at least if linguistic analysis is to describe the "shared stereotype" that corresponds to sur, that is, people's ideas about what people could say about it (Wierzbicka 1985: 215; Kleiber 1990: 73). 21 3. As will be shown below (Section 3), the (temporal) metaphorical extensions of the meaning of sur are derived from the concrete meanings of sur and not from its general, abstract meaning. 4. Geeraerts (1997) shows that semantic change may consist in a process whereby a more central sense gets ousted by a more peripheral one. In order to describe such a process, it is simpler to accept a model of meaning where these different meanings are already present and do not have to be derived from a more abstract meaning. 5. The meanings of prepositions, as shown by Vandeloise, are learned from specific uses, where they have a meaning that is far more precise than the general abstract meaning. Whereas it is certainly possible that speakers develop abstract meanings that group more concrete ones, it is equally probable that the development of these abstract meanings does not lead to the deletion of the earlier concrete ones. All these arguments ultimately point to a conception of the semantic structure of sur as a network with horizontal and vertical dimensions, as proposed by Langacker (1988, 1991: 266-272): "according to the model, the established senses of a word constitute the nodes of a possibly complex, extended network. The senses are linked, horizontally by relations of similarity, and vertically by the relation of a schema and its instantiations" (Taylor 1995: 286). Thus, in the following schematic representation, sense Β is a metaphoric or métonymie extension of sense A and both may be seen as instantiations of sense C, which is schematic for both A and B:

20

21

See, e.g., Vandeloise (1991: 206): "The conditions under which the preposition sur expresses contact are not arbitrary. Once again, they correspond to priorities of the bearer/burden relationship." Cf. also Vandeloise (1986: 202). At present, we prefer this formulation over the stronger claim that our analyses have to be "psychologically real"; Sandra & Rice (1995) and Rice (1996) sufficiently illustrate the problems raised by such a claim.

184

Walter De Mulder & Anne

Vanderheyden

C (schematic meaning) ν

A

instantiation

>

ν

Β

Metonymie / metaphoric relations

Applying this idea to the spatial meanings of sor, sus, and sur, we can still hold that these prepositions express in their prototypical spatial meaning a vertical relation with contact. Other spatial meanings include: - the expression of a vertical relation without contact; - the expression of additional relations such as support, underlie, and adherence (for adherence we did not find Old or Middle French examples in our corpus); - the expression of the endpoint of a movement. In addition to these meanings, Old French (18) as well as Middle French (19) sor and sus also express spatial proximity (Modern French près de 'near'): (18)

Et sour le roy Frison s 'arieste 'And he halted before king Frison' (Richars li Biaus, 2211 ; cited by Godefroy 1892: 518, s.v.

(19)

sour)

[...] li rois fu envoiiés [...] ens ou fort chastiel de Bercler, séant sus le grosse rivière de Saverne ' [ . . . ] the king was sent [ . . . ] to the castle of Bercler, situated on the great river Saverne' (Froissait, Chroniques, I 33, 13; cited by Bekkers 1931: 191)

These uses have almost completely disappeared in Modern French; they only survive in place names such as Boulogne-sur-mer ou Bar-sur-Aube (Ménard 1980: 290). Especially those proximity uses where sor or sus are followed by names or pronouns designating persons cannot be rendered by sur in Modern French. These transient changes of word meaning can easily be accounted for in a prototype model of word meaning, which views diachronic fluctuations at the boundaries of a conceptual category on a par with synchronic uncertainty regarding the delimitation of a category (cf. Geeraerts 1997: 23-24). It is not really clear, however, whether this proximity meaning should be seen as a separate meaning or whether it is merely an illustration of Herskovits' (1986: 79) quasi-rule of tolerance, which states that trajector and landmark do not have to make contact if the distance between them is not too big and if there is no other salient object that separates them. More generally, it is not always clear when one should distinguish a separate meaning (Rice 1996: 142-143). This is to be expected if this meaning, just like the additional semantic relations of support, adherence, or underlie are but specifications of the fundamental prototypical meaning with which they are associated by similarity. These considerations finally lead us to claim that sur has had the same prototypical meaning throughout its history, namely that of signaling a spatial relationship on the vertical axis implying contact, but that other meanings have extended from it, leading to the development of different more specific spatial meanings (support, adherence, underlie, relation on the vertical axis without contact) as well as to the development of a more

The Evolution

185

of French Sur

schematic meaning of contact with a surface. The goal meaning of sor, sus, and sur, then, has developed from the spatial meaning of these prepositions following a more general rule that allows static spatial prepositions to designate a place on the path of a movement when they are accompanied by a verb indicating movement (Brugman 1981; Taylor 1995: 127). In the following sections, we will take a look at the meaning extensions of sor, sus, and sor, starting from the idea that these extensions are based on the different spatial meanings distinguished above.

3. Temporal meaning

It is generally accepted that time is metaphorically conceptualized in terms of space (Lakoff & Johnson 1980: 50-54; Radden 1997).22 Consequently, spatial prepositions are frequently used to express temporal relationships, and Old French sor should be no exception. Nevertheless, it is sometimes said that Old French sor did not have a temporal meaning. Sneyders-De Vogel (1927), for instance, states that the temporal meaning only starts to spread in Middle French. This is definitely not true. Von Wartburg (1966: 431, s.v. super) mentions uses of sor meaning 'vers, environ' around 1180 and es tre sor 1 'aage (cf. ex. 21) around 1260.23 In fact, Old French sor seems to have had at least two temporal meanings: (i) "durant, environ, vers, en parlant du temps" and (ii) "un acheminement vers" (Godefroy 1892: 519): 24 (20)

a. Sor l'ajorner s 'est endormis 'At dawn he fell asleep' (file 1800; cited by Tobler/Lommatzsch 1971: 857, s.v. sor, translated by gegen) b. Sour le mangier ot moul parlé 'While eating, they talked a lot' (Sone 8253; cited by Tobler/Lommatzsch 1971: 857, s.v. sor, translated by während)

(21 )

Si revenrons au roi Phelipe qui estoit sor I 'aage de vint ans 'We returned to king Philippe who was nearly twenty' (Ménestrel

de Reims, § 22, Wailly; cited by Godefroy 1892: 519, s.v. sour)

In fact, the same meanings can be found in Froissart, for Middle French sus and sur. (22)

a. Il se partirent un jour, sus l 'eure dou 'They left one day, at diner time'

souper

(Froissart, Chroniques, I 85, 17; cited by Bekkers 1931: 197)

22

23 24

See Berthonneau (1998) (with a reply in Vandeloise 1998) and Leeman (1998) for critical remarks. Moreover, Rice (1996) concludes on the basis of experimental results that the temporal meanings of English on are just as salient as its spatial meanings. Cf. also Robert Historique (1992), s.v. sur. Thus, Old French sor already had the approximative temporal meaning that also seems characteristic of the temporal uses of Modem French sur, and that seems to oppose it to à. This difference could be a reflection of the spatial meanings of these prepositions, where sur is associated with the idea of a surface and à with the idea of a point (see also Dirven 1993).

186

Walter De Mulder & Anne

Vanderheyden

b. Li bleds sont sur le meurir 'The cornfields are on the brink of ripening' (Froissart, Chron., IV, 393, Luce; cited by Godefroid 1892: 519, s.v. sour)

What could have given rise to the idea that sor did not have a temporal meaning? A brief survey of the uses of sor in the Romans of Chrétien de Troyes shows that sor is hardly ever used temporally, which suggests that sor did have a temporal meaning in Old French, but one that was seldom used. This assumption is somewhat surprising, though, since Latin super already had a temporal meaning: according to Lewis & Short (1975), super can sometimes be translated by during or at (post-Augustal). However, there is again no mention whether super was used frequently with a temporal meaning. In any case, the fluctuations of the temporal meaning show once more that the less central meanings of the preposition do not have the stability of its central spatial meaning, thus providing further evidence for a prototypical structure of prepositional meaning (cf. Geeraerts 1997: 24). Another explanation for the idea that Old French sor had no temporal sense may be the fact that sor probably did not express immediate temporal succession, as it does in Modern French. 25 This meaning only appears in Middle French (according to Wartburg 1966, s.v. super, only from 1550 on). 26 Bekkers (1931: 199) notes that sus is frequently combined with nouns that resume a preceding situation, such as parolles, estât, etc.: (23)

Cilz consaulz et avis pleut grandement au roy de France, et dist que ensi seroit il fait. Adone, par le commandement dou ruy, sus eel arrest, se départirent li doi mareschal [...] 'This advice and this view greatly pleased the king of France, and he said that it would be done that way. Then, because of the king's command, by decree, the two Field Marshalls left [...]' (Froissart, Chroniques, V, 23, 1; cited by Bekkers 1931: 199)

Now, if the temporal meanings are not simply inherited from Latin, as seems to be the case with the meaning of immediate succession, how can their appearance be explained? Prototype theory would suggest a metaphorical extension along the lines of the conceptual metaphor TIME IS SPACE. However, Godefroy's translations of the prepositions in (20a-b) and (21) above show that the different temporal meanings do not extend from a single spatial origin: Godefroy relates sor in (20a-b) to a static spatial meaning, whereas in (21), he associates sor with a "dynamic" meaning, in which the preposition is used to designate the end point of a movement. Furthermore, according to Bekkers (1931: 197), Middle French sus is used in (22a) to specify a moment only approximately, and thus seems to be related to the meaning of spatial proximity, which sor and sus can express as shown in (18) and (19). Moignet (1976: 326) gives a comparable explanation for sor le soir 'vers le soir' and sor mer. 25

26

In Modern French, sur may also express the two meanings given above for Old French and Middle French, as shown by the examples in the Trésor de la langue française (s.v. sur)·. - temporal coincidence, with a slight sense of approximation (the meaning that roughly corresponds to (20a-b)), and - approaching a moment in time, as in aller sur ses vingt ans (cf. Old French (21 )). Sor ce is already used in Old French, but, according to Tobler & Lommatzsch, it means either "überdies" or "d'après, selon, conformément à", as in sor ce que tu diras ("je nach deiner Antwort ").

The Evolution of French Sur

187

The fact that sur' s static as well as its dynamic spatial meanings give rise to (slightly different) temporal meanings suggests that that metaphorical extensions are derived from sur's concrete meanings and not from its abstract meaning. As such, it seems preferable to incorporate both spatial meanings into sur's semantic makeup; in other words, we prefer to assign a prototypical (family-resemblance) structure to sur, rather than a single abstract meaning; cf. also Taylor's (1995: 284-286) third argument presented above. The semantic picture of temporal sur gets even more complicated when one looks more closely at sur's 'immediate succession' meaning. With regard to Middle French sus, Bekkers (1931: 199) has already pointed out that there should be a "correlation" between the two actions or situations connected by sus. Likewise, Franckel & Paillard (1997: 207) show that in Modern French, sur does not simply express temporal succession, but also introduces a subtle interplay of textual continuity and discontinuity. Briefly put, when Middle French sus and Modern French sur express succession, they often also express additional ideas. This could mean that the temporal meaning of sur and sus may have originated from other non-spatial meanings which already existed in Old French, for instance those where sor designates the "grounds" for an action and thus its causes (cf. Radden 1981: 145, 1985: 195)." Note that in this respect, Godefroy (1892: 519, s.v. sour) translates sur in (24) by "d'après, en conséquence, moyennant, en considération de": (24) Jo sui tuchez sur voz ordres. Ί am touched at your command' (Gamier, Vie de St. Thomas, Richel, 13513, Ρ 30 r°, 1783, Hippeau)

This example again shows that it is often difficult to assign a unique origin and a clear interpretation to novel uses, an observation that can again be easlily explained if one accepts a prototypical, network-like structure for these prepositions (Geeraerts 1997: 24).

4. The adversative meaning

That it is often difficult to give the precise origin of the meanings a preposition acquires can also be illustrated by the discussion of the origin of the adversative meaning of Old French sor. This meaning is exemplified in (25), where Godefroy (1892: 519) translates sur by contre, and in (26), an example from Tobler & Lommatzsch (1971: 862, i. v. sor): (25) Sur mei avez turnet fais jugement 'You have caused an unjust judgement on my behalf (Rol. 307 Müller; cited by Godefroy) (26) s'il mescente sor li [wenn er sich zu seinem Nachteil verrechnet] Ii damaces en est siens 'if he miscalculates to his own disadvantage, it is to his own detriment' (Beauman 1,10)

27

This meaning seems also in conformity with the conventional metaphor THEORIES ARE BUILDINGS (Lakoff & Johnson 1980: 46), which could then also account for this this transfer. But of course, one should first verify whether this metaphor also existed in Old French.

188

Walter De Mulder & Anne

Vanderheyden

Etienne (1980: 167) proposes deriving this new meaning from the use of sor introducing a theme ( 'au sujet de ')28 taken amiss, as in (27): (27)

La plore le filz sor le pere Plore li peres sor le fil 'There the son weeps over the father the father weeps over the son' (Cligès 2102-03; cited by Etienne 1980: 167)

Waldmann (1906: 35) and Bekkers (1931: 208), on the other hand, hold that this adversative meaning has to be derived from the dynamic spatial meaning of sor, where it designates the goal of a movement. The verb then expresses a hostile intention or gets this sense from the complement it is combined with. Thus, the adversative value would be triggered by the context, as in (28): (28)

Por che que vos avez donee Sor son defens et sor son dit [...] Sa seror a son anemi 'because you have given, despite his defence and his explicit declarations, his sister to his enemy' {Perceval, Première continuation; cited by Ménard 1980: 290).

This example shows yet again that different origins can be given for the same meaning extension, a result that is to be expected if meaning is organized in a prototypical way, since the different meanings of a word are not neatly separated from each other, but can at least partly overlap, have unclear limits, and form family resemblances (Geeraerts 1997: 11,58-60).

5. Other non-spatial meanings of sor/sur

Within the limited scope of this paper, we cannot present all the meaning extensions of the preposition sur. However, a short survey of the non-spatial meanings of sur (leaving out those already mentioned above) shows that most of sur's non-spatial meanings already existed in Old and Middle French: 1. Dominance and superiority: (29)

28

a. Oiant tuz dit apertement Ke il est reis sur tute gent 'Before all, he overtly says that he was more of a king than all others' (Evang. de Nicod., Ire vers., 49, A.T.; Godefroy 1892: 519, s.v. sour)

This meaning can also be found in Modern French, where one can discuter, méditer, rêver sur quelque chose. Here, sur introduces a cognitive domain and can be opposed to à: wheras à signals a more superficial contact, sur, in keeping with its spatial meaning, signals that its object is more profoundly affected (see also Dirven 1982: 88; Radden 1981: 144, 161; and Lindstromberg 1997: 60 on English on).

The Evolution of French Sur

189

b. Or cuidoye estre sur tous maistre 'Now I believed to be the master above all' (Pathelin, v. 1587; de même Villon, Testament, v. 1017, 1053, etc., cited by Marchello-Nizia 1992: 275) c Le pouvoir du roi de France s'exerçait sur plus de vingt millions de sujets 'The power of the king of France extended over more than twenty million subjects' (Grand Larousse de la langue française, s.v. sur)

2. Pledge, guarantee: (30) a. E li plaiez jurra sur saintz que [...] 'And the injured swore by all saints that [...]' {Lois de GuilL, 11, Chevallet.; Godefroy 1892: 519, s.v. sour) b. Et li fist jurer sus Sains que, si tost qu 'il seroit trespassés, il le feroit boulir en une caudière 'And he made him swear by the saints that, as soon as he would be dead, he would make him boil in a kettle' (Froissait, Chroniques, I 114, 13; cited by Bekkers 1931: 201) c. Jurer sur la Bible, sur la tête de sa mère 'To swear on the bible, by the head of his mother' (Trésor de la langue française, s.v. sur, IV A i e )

3. Model, conformity: (31) a. M'avoit dit et comendé Ke sor un chant ki jadis Soloit estre mult öis Chantaisse de sa bontei 'he had asked me to sing of his goodness to a tune which was often heard before' {Bern. Lhs. 309.1.; Tobler/Lommatzsch 1971: 863) b. En ce temps et en celle saison, devisa et ordonna li rois Jehans de France une belle compagnie grande et noble, sus la maniere de le Table Reonde qui fut jadis au temps de roy Artus [...] 'In those days and in that season, king Jean of France chose and prepared a beautiful, great and noble society, in the way of the Round Table which existed before in the days of King Arthur [...]' (Froissart, Chroniques, IV 126, 25, cited by Bekkers 1931: XXI) c. régler son pas sur celui de quelqu 'un, vivre sur un rythme fou 'to adjust one's step to that of someone, to live on a crazy rhythm' {Trésor de la langue française, s.v. sur, IV A2)

4. Habitual or occasional way of doing things: (32)

a. Ils oïrent autres noveles [...] qu'uns bons chevaliers lor aporte [...] qui venoit tost corani sor frain 'They heard other news, [...] that a good knight brought them [...] who arrived at a gallop holding the bridle' (G. Dole 2307; Tobler/Lommatzsch, T. 3, 2228, s.v. frein) b. Et les devoit li dus faire conduire jusques à Bourdiaus, sus son peril 'And the duke had to lead them to Bordeaux, under his command' (Froissart, Chroniques III 119, 30; cited by Bekkers 1931: 203) c. Se séparer sur un baiser 'To separate with a kiss' (Trésor de la langue française, s.v. sur, III B 1 )

190

Walter De Mulder & Anne Vanderheyden

The observation that all important meaning extensions already existed in Old French makes it impossible to use the diachronic evolution of sor in order to get an idea of the cognitive links between the different meanings of the preposition. 29 In fact, meaning changes from Old to Modern French only concern the boundaries of the categories corresponding to the different senses. Thus, the meaning of adversity (cf. 25, 28) is not explicitly mentioned in contemporary dictionaries for sur, but some uses are compatible with such an interpretation: (33) S'avancer, courir, foncer, se précipiter sur quelqu'un; lancer les chiens sur quelqu'un; s'abattre, bondir, se jeter, tomber sur une proie, sur un antagoniste '[to advance/to run to someone; to rush/to dash upon someone; to set a dog on someone; to crash down/to pounce on someone; to fall upon/to throw at a prey, an opponent' (Trésor de la langue française, 5. ν. sur, II A 2a/b) Likewise, many uses of sur that mark types of non-spatial superiority, as in (34), seem less acceptable in Modem French, but, as shown above, this does not mean that sur no longer expresses superiority; it only means that the boundaries of that meaning have slightly changed. (34) a. Oiant tuz dit apertement Ke il est reis sur tute gent 'Before all, he overtly says that he was more of a king than all others' (Evang. de Nicod., Ire vers., 49, A.T.; Godefroy 1892: 519, s.v. sour) b. Sur tuz les altres est Carles anguissus 'More than any other, Charles is terrified' (Roland·, cited by Picoche and Marchello-Nizia 1991: 279) c. Sunent eil greisle e derere e devant, Sur tuz les altres bundist li olifant 'The clarions blow behind and in front, more than any other resounds the elephant' (Roland 3118; cited by Moignet 1976: 325) Obviously, the fact that the different senses are already present in Old French does not mean that the semantic extensions cannot be explained: - Using sur to express dominance is consistent with the conventional metaphor HAVING CONTROL OR FORCE IS UP; BEING SUBJECT TO CONTROL OR FORCE IS DOWN (Lakoff & Johnson 1980: 15). - The notion of support can readily be extended to that of pledge or guarantee. On the other hand, the use of sur in expressions such as jurer sur la Bible 'to swear on the Bible' can also be grounded in the practice of putting one's hand on the Bible while making a vow and could thus be largely métonymie. - The manner and model meanings also seem to develop from the use of sur to designate the ground for an action; an example such as (32c) can also be connected to the temporal meaning of sur, but as we pointed out above, this meaning is itself probably an extension of the 'ground' meaning.

29

It might be helpful to have a look at the Latin preposition super, at the moment when it extended its uses at the expense of in (cf. Meyer-Lübke 1974 [1890-1906]).

The Evolution

of French

191

Sur

6. Conclusions

At the outset of this paper, we hoped that the study of the semantic evolution of French sur would shed some light on the cognitive links between its synchronic meanings. While this may be possible for other words and prepositions (Geeraerts 1997), it is not possible for sur, simply because most of its non-spatial meanings have already been present from Old French on. Semantic changes only seem to concern the boundaries of the different meanings; some of these meanings seem to have been more common in Old French than in Modern French (e.g. the adversative meaning), while others seem to be more widespread in Modem French (e.g. the temporal meaning). On the other hand, our analysis has corroborated some predictions and observations of Geeraerts (1997): - it is sometimes difficult to precisely locate the origin of new meanings; - the boundaries of the different meanings are in constant evolution; - at least in the case of sur, the prototypical spatial meaning does not change. Moreover, we have also shown that the study of the diachronic evolution of a lexical item can help to establish which meanings should be part of the item's network structure. As such, we argued that the dynamic spatial meaning is part and parcel of sur's semantic structure, since it allows for other extensions than the static spatial meaning (cf. our discussion of the temporal meanings of sur). While this gives us at least one partial answer to some of the questions asked by Rice (1996: 142) on the nature of prototypical meanings and networks (e.g. How many distinct elements are in the network?), a host of others are still unanswered, partly because the data are still incomplete. It would be interesting, for example, to have data on the frequency of the different meanings of lexical items, or to know which terms frequently co-occur. Thus, it seems safe to say that diachronic prototype semantics is still in its infancy.

References

Bekkers, Frans (1931): Etude sur l'emploi Amsterdam: H. J. Paris..

que Froissart

fait de la préposition.

- Ph.D. dissertation.

Berthonneau, Anne-Marie (1998): Espace et temps: Quelle place pour la métaphore? - In:

Verbum

XX, 353-382. Brown, C. H. & S. R. Witkowski (1983): Polysemy, Lexical Change, and Cultural Importance. - In: Man 18, 7 2 - 8 9 . Brugman, Claudia (1981): Story of Over. - M A thesis. Berkeley. University o f California at Berkeley. (Published as The Story of Over: Polysemy, Semantics, and the Structure of the Lexicon. N e w York, N Y : Garland. 1989.) Clark, Eve (1974): Non-Linguistic Strategies and the Acquisition o f Word Meanings. - In: 2, 1 6 1 - 1 8 2 .

Cognition

192

Walter De Mulder & Anne Vanderheyden

Cuyckens, Hubert (1991): The Semantics of Spatial Prepositions in Dutch: A Cognitive-Linguistic Exercise. - Ph.D. dissertation. University of Antwerp (UIA). - (1995): Family Resemblance in the Dutch Spatial Preposition op. - In: M. Schwarz (ed.): Kognitive Semantik/Cognitive Semantics, 179-195. Tübingen: Narr. Dirven, René (1982): Metaphors of Spatial Relations. - In: J. Hasler (ed.): Anglistendag ¡981, 6391. Frankfurt a.M.: Lang. - (1993): Dividing Up Physical and Mental Space into Conceptual Categories by means of English Spatial Prepositions. - In: C. Zelinsky-Wibbelt (ed.): The Semantics of Prepositions: From Mental Processing to Natural Language Processing, 73-97. Berlin: Mouton de Gruyter. Dubois, Danièle (1983): Analyse de 22 catégories sémantiques du français. - In: L'Année psychologique 83, 465-489. Etienne, Eugène (1980): Essai de grammaire de l'ancien français (IX-XIVe siècles). - Genève: Slatkine. Franckel, Jean-Jacques & Denis Paillard (1997): Les emplois temporels des prépositions: Le cas de sur. - In: Cahiers Chronos 3, 199-211. Geeraerts, Dirk (1986): Woordbetekenis: Een Overzicht van de Lexicale Semantiek. - Leuven: Acco. - (1997): Diachronicprototype semantics. - Oxford: Clarendon Press. Godefroy, Frédéric (1892): Dictionnaire de l'ancienne langue française. Tome septième. Nendeln/Liechtenstein: Kraus Reprint, 1969. Groussier, Marie-Line (1997): Prépositions et primarité du spatial de l'expression des relations dans l'espace à l'expression des relations non-spatiales. - In: Faits de langues 9, 221-234. Herskovits, Annette (1986): Language and Spatial Cognition: An Interdisciplinary Study of Prepositions in English. - Cambridge: Cambridge University Press. Kleiber, Georges (1990): La sémantique du prototype. - Paris: Presses Universitaires de France. Klein, Wolfgang & Niise, Ralf (1997): La complexité du simple: L'expression de la spatialité dans le langage humain. - In: M. Denis (éd.): Langage et cognition spatiale, 1-23. Paris: Masson. Lakoff, George & Johnson, Mark (1980): Metaphors We live By. - Chicago, IL: The University of Chicago Press. Langacker, Ronald (1988): A Usage-Based Model. - In: B. Rudzka-Ostyn (ed.): Topics in Cognitive Linguistics, 127-161. Amsterdam: John Benjamins. - (1991): Concept, Image, and Symbol: The Cognitive Basis of Grammar. - Berlin: Mouton de Gruyter. Leeman, Danielle (1998): La métaphore dans la description des prépositions. - In: Verbum XX, 435458. Lewis, Charlton & Charles Short (1975): A Latin Dictionary, Founded on Andrew's Edition of Freund's Latin Dictionary, Revised and Enlarged and in Great Part Rewritten. - Oxford: Clarendon Press. Lexis (1979). Dictionnaire de langue française. - Paris: Larousse. Lindstromberg, Seth (1997): English Prepositions Explained. - Amsterdam: John Benjamins. Liidtke, H. (1985): Diachronic Irreversibility in Word-formation and Semantics. - In: J. Fisiak (ed.): Historical Semantics - Historical Word-Formation, 339-354. Berlin: Mouton de Gruyter. Mandler, Jean (1992): How to Build a Baby: II. Conceptual primitives. - In: Psychological review 99, 587-604. Marchello-Nizia, Christiane (1992): Histoire de la langue française. - Paris: Dunod. Martin, Robert & Marc Wilmet (1980/· Syntaxe de l'ancien français. - Bordeaux: Sobodi (= Manuel du français du moyen-âge 2, sous la direction de Y. Lefèvre). Ménard, Philippe (1980): Manuel du français du moyen âge. I: Syntaxe de l'ancien français. - Bordeaux: Sobodi.

The Evolution of French Sur

193

Meyer-Lübke, Wilhelm (1890-1906, 1974% Grammaire des langues romanes. - Genève: Slatkine Reprints/ Marseille: Lafitte Reprints, 1974 ; translation by Auguste and Georges Doutrepont. Moignet, Gérard (1976): Grammaire de l'ancien français. - Paris: Klincksieck. Picoche, Jacqueline & Christiane Marchello-Nizia (1991): Histoire de la langue française. - Paris: Nathan. Ollier, Marie-Louise (1986): Lexique et concordance de Chrétien de Troyes d'après la copie Guiot, avec introduction, index et rimaire. - Montréal: Institut d'études médiévales/Paris: Librairie J. Vrin. Radden, Günter (1981): Die übertragenen Bedeutungen der englischen Raumpräpositionen. - In: G. Radden & R. Dirven (eds.): Kasusgrammatik und Fremdsprachendidaktik, 133-180. Trier: Wissenschaftlicher Verlag Trier (= Anglistik und Englischunterricht 14). - (1985): Spatial Metaphors Underlying Prepositions of Causality. - In: W. Paprotté & R. Dirven (eds.): The Ubiquity of Metaphor, 177-207. Amsterdam: John Benjamins. - (1997): Time is Space. - In: B. Smieja & M. Tasch (eds.): Human Contact through Language and Linguistics, 147-166. Bern: Lang. Rice, Sally (1996): Prepositional Prototypes. - In: M. Pütz & R. Dirven (eds.): The Construction of Space in Language and Thought, 135-166. Berlin: Mouton de Gruyter. Robert Historique: Dictionnaire historique de la langue française (1992): - Paris: Le Robert. Sandra, Dominiek & Sally Rice (1995): Network Analyses of Prepositional Meaning: Mirroring Whose Mind - The Linguist's or the Language User's? - In: Cognitive linguistics 6, 89-130. Schifko, Peter (1977): Aspekte einer strukturalen Lexikologie. - Bern: Francke. Shears, F. (1922): Recherches sur les prépositions dans la prose du moyen français (XlVe et Xve siècles). - Paris: Honoré Champion. Sneyders-De Vogel, Κ. (1927): Syntaxe historique du français. - Groningen: Wolters. Sweetser, Eve (1990): From Etymology to Pragmatics. - Cambridge: Cambridge University Press. Taylor, John (1995 [1989]): Linguistic Categorization: Prototypes in Linguistic Theory. - Oxford: Clarendon Press. Tobler, Adolf & Erhard Lommatzsch (1971): Altfranzösisches Wörterbuch. - Wiesbaden: Franz Steiner Verlag. Traugott, Elizabeth Closs (1985): On Regularity in Semantic Change. - In: Journal of Literary Semantics XIV, 155-173. Vandeloise, Claude (1986): L'espace en français. - Paris: Seuil. - (1991): Spatial Prepositions: A Case Study from French. - Chicago, IL: The University of Chicago Press. - (1992): Structure of Lexical Categories and Family Resemblances. - Duisburg: L.A.U.D. Paper no. 325. - (1998): Les domaines des prépositions avant/après. - In: Verbum XX, 383-394. - (2000): Plus de clarté dans la relation porteur/porté. - In: M. Coene, W. De Mulder, P. Dendale & Y. D'Hulst (eds.): Traiani Augusti vestigia pressa sequamur: Studia Linguistica in honorem Lilianae Tasmowski, 753-769. Padova: Unipress. Waldmann, A. (1906): Die begriffliche Entwicklung des lateinischen "Super" ("Supra") und "Sursum " im Französischen. - Leipzig. Wartburg, Walther von (1966): Französisches Etymologisches Wörterbuch. - Basel: Zbinden Druck und Verlag. Wierzbicka, Anna (1985): Lexicography and Conceptual Analysis. - Ann Arbor: Karoma Publishers. Zifonun, Gisela, Ludger Hoffman, Bruno Strecker, et al. (1997): Grammatik der deutschen Sprache. - Berlin: Mouton de Gruyter.

Maaike

Beliën

Force Dynamics in Static Prepositions: Dutch Aan, Op, and Tegen

1. Introduction 1

The paper discusses the semantics of the Dutch prepositions aan 'on,' op 'on,' and tegen 'against.' In Cuyckens (1991) - the most extensive account to date of these, and other, prepositions in Dutch - aan, op, and tegen are each described in terms of a (different) set of senses that are related by family resemblance. This approach to word meaning has been common practice in cognitive-linguistic analyses of prepositional meaning since Brugman's (1981) and later Lakoff s (1987) studies of English over. I will present an analysis of Dutch aan, op, and tegen that differs from Cuyckens' analysis in two respects. Rather than treat prepositional meaning in terms of a family-resemblance network of interrelated senses, I will propose a more unified account that describes a preposition's meaning in terms of a single Platonic concept, which may adapt under contextual pressure (cf. Van der Leek 1996a, 1996b, 2000). In addition, unlike Cuyckens (1991), I will account for the differences between aan, op, and tegen in terms of their forcedynamic properties. More specifically, the semantics of aan, op, and tegen will be characterized in terms of Platonic concepts that each involve contact between two entities, but that differ crucially with respect to force dynamics.

2. Cuyckens' (1991) treatment of aan, op, and tegen

2.1. Op In Cuyckens' analysis, Dutch aan, op, and tegen can all be used to express a relation of between an entity χ and an entity y. For op, Cuyckens distinguishes three kinds of COINCIDENCE relation, namely COINCIDENCE & SUPPORT, COINCIDENCE & ADHERENCE, and plain COINCIDENCE (cf. 1991: 216-219). In example (la), op expresses COINCIDENCE & SUPPORT: y, the table, has a horizontal surface that supports x, the books, from underneath. In (lb), op expresses COINCIDENCE & ADHERENCE; this sense of op involves a surface of y to which χ adheres and which can have any orientation (i.e. vertical, downward, or upward). Op in example (lc) expresses COINCIDENCE only. COINCIDENCE

1

The research presented here is part of a project (200-50.083) funded by the Netherlands Organization for Scientific Research (NWO). I would like to thank Hubert Cuyckens, Theo Janssen, and Frederike van der Leek for their comments on an earlier version of this paper.

196 (1)

Maaike

Beliën

a .De boeken liggen op de taf el. (Cuyckens' example 30b, 1991: 194) 2 the books lie on the table 'The books are on the table.' b. De hagedis zit op de muur. (30c, 194) the lizard sits on the wall 'The lizard is on the wall.' c. Mozes tikte met zijn staf op de rotó. (53d, 210) M o s e s tapped withhis staff on the rock ' M o s e s tapped on the rock with his staff.'

Cuyckens further distinguishes between uses of op as a non-path preposition and as a path preposition. In the examples in (1), op is used as a non-path preposition, i.e. χ has a "static or invariable location" with respect to y (cf. Cuyckens 1991: 116). Path prepositions express a path or a trajectory of χ with respect to y; cf. examples (2a) and (2b). (2)

a. Hij gooide 'He threw b. Hij gooide 'He threw

zijn sleutels his keys on de modder the mud on

op de tafel. (60a, 221 ) the table.' op de muur. (60b, 221 ) the wall.'

As a path preposition, Cuyckens argues, "op assigns a series of locations or a path to χ that starts at some vague point and at whose terminus χ enters into one of the following relat i o n s w i t h the s u r f a c e a s s o c i a t e d w i t h Y: ( a ) COINCIDENCE & SUPPORT . . . ( b ) COINCIDENCE & ADHERENCE . . . ( c ) COINCIDENCE" ( 1 9 9 1 : 2 2 1 ) . In ( 2 a ) , the t e r m i n a l r e l a t i o n is o f the COINCIDENCE & SUPPORT t y p e ; in ( 2 b ) , the t e r m i n a l r e l a t i o n is o f the COINCIDENCE

&

ADHERENCE t y p e .

2.2. Aan According to Cuyckens, aan can be used to express two kinds of spatial relation: CONTIGUITY a n d COINCIDENCE. In a r e l a t i o n i n v o l v i n g CONTIGUITY, Χ is p a r t o f y ' s r e g i o n o f

interaction; whether x is actually in contact with y is not considered relevant (cf. Cuyckens 1991: 231). In sentence (3), for instance, the house(s) in which the people referred to by zij 'they' live may border the avenue, but they might equally well be located at a slight distance from the avenue. (3)

Zij wonen aan de Jan van Rijswijcklaan. (75a, 231) they live on the Jan van Rijswijck-avenue 'They live on the Jan van Rijswijcklaan.'

Beside CONTIGUITY, aan can also express COINCIDENCE relations; they are of two kinds: MINIMAL ATTACHMENT a n d TRANSIENT CONTACT. T h e e x a m p l e s in ( 4 ) i n v o l v e MINIMAL ATTACHMENT.

2

The examples in this article and their translations are from Cuyckens (1991). I have added glosses where they diverge considerably from the translations.

Force Dynamics

(4)

in Static

a. het schilderij

197

Prepositions

aan de muur (65b, 2 2 5 )

'the painting on the wall' b. de

druppel

the drop

water aan zijn vinger (67b, 2 2 6 ) water on

his

finger

'the drop of water on his finger'

In (4a), the painting is attached to the wall by means of some attachment device, e.g. a combination of a nail in the wall and a small hook in the frame of the painting. In (4b), the water stays attached to the finger because of its adhesive properties (it cannot be supported by the finger from underneath; in that case op would be used, cf. Figure 7 in Section 6 below). Cuyckens argues that the COINCIDENCE & ADHERENCE sense of op involves MAXIMAL ATTACHMENT, which means that x, or at least major parts of x, cover (part of) y. ATTACHMENT expressed by aan, on the other hand, only involves MINIMAL ATTACHMENT. Cuyckens explains the difference by means of example (5): " O p can lexicalize th[e] spatial relation between de trui and de muur when the whole, or at least very substantial parts, of the sweater coincide with the wall, for instance, when they are on display in clothing stores. Aan does not impose this requirement. It is enough that the sweater is attached to the wall by means of one or other device for aan to be used" (Cuyckens 1991: 195). (5)

De trui hangt aan/op de muur. (32, 194) 'The sweater hangs on the wall.'

The second kind of COINCIDENCE relation that can be expressed by aan is TRANSIENT which Cuyckens describes as "brief touching" (1991: 228), an example of which is given in (6).

CONTACT,

(6)

Hi] voelde

aan zijn baard. (70b, 228)

he felt on his beard 'He touched his beard.'

As a final remark about aan before we turn to tegen, Cuyckens does not distinguish nonpath and path uses of aan. He does not discuss examples in which χ describes a path or a trajectory. I take this to mean that his position is that aan can only be used as a non-path preposition.

2.3. Tegen Tegen, in Cuyckens' account, can be used to express COINCIDENCE relations of two kinds: and plain CONTACT. CASUAL CONTACT is a "COINCIDENCE relation between χ and y in which SUPPORT (between χ and y), ADHERENCE, or ATTACHMENT are not at stake" (Cuyckens 1991: 263). CASUAL CONTACT relations can be further subdivided into two types. In the first subtype, the entity χ is in contact with the vertical surface of y, while at the same time χ is also supported from underneath by an upward facing horizontal surface. In (7), for instance, the table is in contact with the wall, while supported by the floor. CASUAL CONTACT

198 (7)

Maaike

Beliën

De tafel Staat tegen de muur. (37a, 2 6 2 ) 'The table stands against the wall.'

In the second subtype of CASUAL CONTACT, χ is not supported from underneath and is in contact with a surface of y that can have any orientation, as in the examples in (8). In (8a), χ is in contact with a vertical surface, in (8b) with a horizontal surface that has a downward orientation, and in (8c) with a horizontal surface that has an upward orientation. (8)

a. Het schilderij the painting

hangt tegen

de

hangs against the

muur. (33b, 2 6 2 ) wall

'The painting is hanging on [against] the wall.' b. De

lamp

hangt tegen

de

zoldering.

the lamp hangs against the

(38a, 2 6 3 )

ceiling

'The lamp is on the ceiling.' c. Mel zijn oor tegen

de

grond

hoordehijde

with his ear against the ground heard

Indianen naderen.

he the Indians

(42c, 2 6 7 )

approach

'With his ear against the ground, he heard the Indians approach.'

Sentence (9) illustrates tegen symbolizing a COINCIDENCE relation where there is "CONTACT between an entity χ and the boundary line of a 2dim entity y" (Cuyckens 1991: 268) and where the notion CASUAL does not seem relevant. (9)

Mijn

perceel

ligt tegen

my

plot

lies against the his

het zijne. (43a, 2 6 8 )

'My plot is next to his.'

Tegen, like op, can also be used as a path preposition, in which case "x is assigned a path that starts at an unspecified point and whose terminus is the vertical surface associated with y" (Cuyckens 1991: 222). At that terminus, Χ is in a relation of CASUAL CONTACT with y. It is left unspecified whether or not χ attaches or adheres to y (Cuyckens 1991: 268). Sentences (10a) and (10b) illustrate the use of tegen as a path preposition; in (10a), the entity χ does not attach to entity y at the end of its path, while in (10b), it does. (10)

a. Hij gooide

de vaas tegen de muur. (44a, 2 6 8 )

'He threw the vase against the wall.' b. Hij gooide de modder tegen de muur. (44b, 2 6 8 ) 'He threw the mud against the wall.'

2.4. Summary I will summarize this section by enumerating the different senses that Cuyckens distinguishes for each of the three prepositions. Op can be used to express COINCIDENCE & SUPPORT, COINCIDENCE & ADHERENCE ( i . e . MAXIMAL ATTACHMENT), COINCIDENCE, o r a

path ending in any one of these three

COINCIDENCE

relations. Aan can be used to express or CONTIGUITY. Tegen can be used to (either with or without support from underneath), or a

MINIMAL ATTACHMENT, TRANSIENT CONTACT,

express CONTACT, CASUAL CONTACT path ending in CASUAL CONTACT.

Force Dynamics in Static

Prepositions

199

3. The Platonic concept approach to word meaning

My analysis of the semantics of aan, op and tegen, which I present in Section 5, differs crucially from Cuyckens' analysis in that, unlike Cuyckens, I do not describe the semantics of a preposition in terms of a family-resemblance network of senses. In a familyresemblance network, the senses need not share a set of necessary and sufficient features. Instead, they are related just like members of a family: one member will share some features with another member of the category, but does not need to share all its features, and there may be members in the category that are only indirectly related and that have no features in common. There are two serious problems associated with the family-resemblance network approach to meaning. First, it does not explain how the language user knows which sense to select from the network of senses associated with a lexical item (e.g. a preposition) when (s)he hears the item used. For instance, how can a language user know which one of the senses of op to select (COINCIDENCE, COINCIDENCE & SUPPORT, COINCIDENCE & ADHERENCE, or the path senses), when (s)he is confronted with a sentence in which op is used? Second, the family-resemblance approach cannot sufficiently explain how language users can interpret a novel sense of a word, if they have to rely on a list of related senses. In general, what the family-resemblance network of senses approach neglects is what the other words in the sentence and the wider linguistic and non-linguistic context contribute to the interpretation of a preposition, if it is at all possible to speak of the interpretation of a preposition without considering the context's contribution. In the approach to word meaning followed here, which has been developed by Van der Leek (1996a, 1996b, 2000), context plays a crucial role. Van der Leek claims that a word's meaning is constituted by a Platonic concept, a single maximally schematic concept that language users abstract from experience through their perceptual and cognitive apparatus. 3 Van der Leek points out that "'Platonic' in this context is not to be seen in Plato's original metaphysical sense, but in the sense that any realization of this (mental) Platonic concept is by definition more concrete, hence no longer Platonic" (2000: 318). Meaning construction in Van der Leek's approach is highly dynamic, creative, and compositional. When a word is used, the language user will negotiate its Platonic concept with the Platonic concepts associated with the other words in the sentence. A Platonic concept is schematic; combining it with other concepts will narrow down the possible interpretations. The interpretation that a word thus receives in a particular context can be called a "sense in context" (due to Pustejovsky 1995: 60), which "is not the sense of the [word] in its own right, but the sense that is symbolized through its association with other words (or phrases) that bring in their own meaning" (Van der Leek 2000: 318). Language, in this approach, is considered to be highly underdetermined (cf. Fauconnier 1994 [1985], 1997; Verhagen 1997): the information associated with the lexical items and the grammatical structure of a sentence uttered only constrain the kinds of interpretations that are possible.

3

The term "Platonic concept" is taken from Hofstadter and McGraw's (1995: 412) analysis of letter concepts. Van der Leek (1996a,b) uses the term "skeletal meaning" and "core sense" to refer to essentially the same notion.

200

Maaike Beliën

4. The role of force dynamics in prepositional meaning

My analysis of aan, op, and tegen differs from Cuyckens' (1991) analysis in another respect as well. Unlike Cuyckens, I will systematically take into account the forces that are at work in the spatial relations expressed by the three prepositions. The suggestion that force dynamics may play a role in the semantics of prepositions can also be found in Regier (1996) and Bowerman (1996), who both refer to the experimental study by Freyd, Pantzer & Cheng (1988), in which it is suggested that force dynamics and potential motion are part of mental representations of static spatial scenes. Freyd, Pantzer & Cheng's study does not concern language, but if mental representations of static spatial scenes indeed include forcedynamic properties, force dynamics may very well play a role in the semantics of prepositions that are used to express static spatial relations. Regier (1996) discusses the English preposition above. He shows that a characterization of above such as 'the center of mass of the trajector is located higher in the visual field than the center of mass of the landmark' cannot be correct, in light of the scenes depicted in Figure 1 (his Figure 5.1, 1996: 82). While above can be used for a description of the scene on the left, it cannot be used for a description of the scene on the right. Regier suggests that "part of what above denotes is the possibility of the trajector striking the landmark if it is allowed to fall under the influence of gravity" (1996: 83). In the scene on the left, the trajector would strike the landmark if it were allowed to fall under the influence of gravity, which makes it an example of above. In the scene on the right, the trajector does not strike the landmark, which does not make it a good example of above.

O O Figure I: The role offorce dynamic considerations in above (taken from Regier 1996: 82)

Bowerman's (1996) study is even more relevant to the present study in that it suggests that force dynamics may play a role in the semantics of the Dutch prepositions op and aan. Bowerman contrasts op and aan with English on, to illustrate how different languages categorize spatial relations in different ways. She characterizes op and aan in the following way: "a figure op a ground is perceived as stable, i.e. not in any salient way resisting an underlying force that pulls it away from the ground, while a figure aan a ground is seen as being prevented from manifesting a tendency towards separation" (1996: 153). While these characterizations seem to go a long way, they do not cover the whole spectrum of uses of op and aan. Bowerman's characterization of op does not hold in light of example (lc) above, in which there is no stable situation. Nor can her characterization of aan be correct in light of example (3) above, in which there is no tendency toward separation. Still, I fully agree with her suggestion that "[i]t is plausible that underlying 'force

Force Dynamics in Static

Prepositions

201

dynamic' considerations like these are relevant to the semantics of preposition choice even in seemingly static situations" (1996: 153-154).

5. The Platonic concepts associated with aan, op, and tegen

Before presenting my analysis of the semantics of the Dutch prepositions aan, op, and tegen in terms of force-dynamic Platonic concepts, I will first consider two terminological issues. Remember that Cuyckens (1991) characterizes these three prepositions in terms of a COINCIDENCE relation between an entity χ and an entity y. My characterization of what the prepositions share semantically differs from Cuyckens' description in two ways. First, I will use the term 'contact' instead of 'coincidence.' To my mind, the term 'coincidence' seems too broad in that it may denote spatial relations between two entities that involve more than just contact, namely substantial overlap. Secondly, I will use Langacker's (1987) notions "trajector" and "landmark" instead of Cuyckens' (entity) χ and (entity) y, since "trajector" and "landmark" bring out the asymmetry between the two participants in the relation expressed by a preposition.4 The participant that is called the trajector "has special status and is characterized as the figure in the relational profile" (Langacker 1987: 217). The landmark is the other salient entity in the relation, which provides a "point [...] of reference for locating the trajector" (1987: 217). In general, then, the semantics of the prepositions aan, op, and tegen can be characterized in terms of Platonic concepts that involve contact between a trajector and a landmark, but that differ with respect to their force-dynamic properties. The force-dynamic notion that plays a crucial role in distinguishing the semantics of the three prepositions will be labeled 'stick to,' for lack of a better term. The notion 'stick to' denotes the property of a trajector being attached or fixed to a landmark; 'stick to' includes not only adhesion, but any kind of force (adhesive, magnetic, gravitational, etc.) that causes one entity to be attached, fixed, or stuck to another entity. The prepositions op and aan differ from tegen in that the kinds of contact symbolized by op and aan are different from the kind of contact symbolized by tegen. While op and aan symbolize relations in which the trajector 'sticks to' to the landmark at the area of contact, tegen symbolizes a relation in which the trajector does not stick to the landmark at the area of contact. Op, in turn, is different from aan in that, in the case of op, the trajector sticks to the landmark at the area of contact due to a force that is directed from the trajector toward the landmark, while in the case of aan, such a force does not play a role. In (11)—(13), I present definitions of the Platonic concepts associated with aan, op and tegen. Figures 2—4 are visual representations of the Platonic concepts involved, in which the lighter grey circle represents the trajector and the darker grey circle represents the landmark. Let us first consider the Platonic concept associated with aan, as given in (11) and Figure 2.

4

In fact, Cuyckens uses the notions "trajector" and "landmark" in his later work (cf., e.g., Cuyckens 1994).

202 (11)

Maaike

Beliën

The Platonic concept associated with aan: 'a relation of contact between two entities, a trajector and a landmark, in which at the area of contact the trajector sticks to the landmark.'

Figure 2: The Platonic concept associated

with aan

Figure 2 shows that the trajector (the lighter grey circle) is in contact with and positioned to the left of the landmark (the darker grey circle). The property 'stick to' is represented by six little lines connecting trajector and landmark. Even though Figure 2 shows the trajector to the left of the landmark, the absolute position of the trajector with respect to the landmark is not specified: as long as the trajector is in contact with, and sticks to, the landmark, it can obtain any position with respect to the landmark. A definition of the Platonic concept associated with op is presented in (12); Figure 3 is its visual representation, in which the black arrow represents a force. ( 12)

The Platonic concept associated with op: 'a relation of contact between two entities, a trajector and a landmark, in which at the area of contact the trajector sticks to the landmark due to a force that is directed from the trajector toward the landmark.'

tr

lm

Figure 3: The Platonic concept associated

with op

Op differs from aan in that there is a force directed from the trajector toward the landmark which is responsible for the trajector sticking to the landmark. Just as for aan, the position of the trajector with respect to the landmark is not important, just as long as they are in

Force Dynamics in Static

203

Prepositions

contact. The direction of the force, however, is cruciai: it is directed from the trajector toward the landmark. In (13), I present the definition of the Platonic concept associated with tegen·, Figure 4 is its visual representation. (13)

The Platonic concept associated with legen: 'a relation of contact between two entities, a trajector and a landmark, in which at the area of contact the trajector does not stick to the landmark.'

tr

1m

Figure 4: The Platonic concept associated with tegen

Tegen differs from op and aatt in that trajector and landmark do not stick to one another. This is represented in Figure 4 by the short lines radiating out from the area of contact. Here again, as we have seen for aan and op, the exact position of the trajector relative to the landmark is not important, just as long as the trajector is in contact with the landmark and, at the area of contact, does not stick to it.

6. Platonic concepts in context

I will now show that an analysis in terms of these Platonic concepts produces a more unified and satisfactory account than an analysis in terms of family-resemblance networks of senses. First of all, notice that the notion of 'path' is not part of any of the Platonic concepts defined above. Yet, Cuyckens (1991) distinguishes a separate 'path' sense for such examples with op and tegen as (2) and (10). (2)

a. Hij gooide 'He threw b. Hij gooide 'He threw

zijn sleutels op de tafel. (60a, 221 ) his keys on the table.' de modder op de muur. (60b, 221 ) the mud on the wall.'

(10)

a. Hij gooide 'He threw b. Hij gooide 'He threw

de vaas tegen de muur. (44a, 268) the vase against the wall.' de modder tegen de muur. (44b, 268) the mud against the wall.'

204

Maaike

Beliën

In the Platonic concept approach, distinguishing a separate path sense for op and tegen is not necessary. The notion of a path is not part of the semantics of op or tegen, but arises in context. The Platonic concept associated with gooien can be characterized as involving a process that requires two participants, in which one participant moves his/her hand, which holds the second participant, and then releases the second participant, which then travels along a path through the air. In (2) and (10), this Platonic concept combines with a PPcomplement that involves contact between two entities. One interpretation that is allowed by this combination involves a process of throwing, in which the thrown object follows a path that ends in contact with the entity described by the NP-complement of the preposition (cf. also Van der Leek's treatment of English throw at, 1996b: 369). The notion of path should therefore not be considered to be part of the semantics of op or tegen·, it is part of the sense in context. Secondly, the Platonic concept proposed for op in (12) and Figure 3 above obviates the need to distinguish the three kinds of COINCIDENCE relation for op that Cuyckens (1991) proposes: COINCIDENCE & SUPPORT for examples such as (la), COINCIDENCE & ADHERENCE for examples such as (lb), and COINCIDENCE for examples such as (lc). (1)

a .De

boeken liggen

the books

lie

op de

tafel. {30b, 194)

on the table

'The books are on the table.' b. De

hagedis zit

the lizard

op de

muur. (30c, 194)

sits on the wall

'The lizard is on the wall.' c. Mozes

tikte

met zijn staf

op de

rots. (53d, 2 1 0 )

Moses tapped with his staff on the rock 'Moses tapped on the rock with his staff.'

What unifies these three types of contact relations is that there is a force directed from the trajector toward the landmark such that the trajector sticks to the landmark. In (la), it is the force of gravity that acts on the books, so that their weight presses down on the table. In (lb), it is the adhesive properties of the lizard's paws that cause it to stay attached to the wall. In (lc), it is the combination of gravity and the force exerted by Moses that causes the staff to come into contact with the rock. At first sight, one might wonder whether the interpretation of (lc) really involves any 'stick to' relation at all; admittedly, the staff does not stick to the rock in the sense that it stays there. This is due, however, to the activity identified by the verb tikken 'to tap,' which involves only brief contact. Notice, however, that if Moses were to stop his tapping, the staff would remain on the rock due to gravity. If tegen, which does not involve the 'stick to' relation, is used instead of op (see lc' below) a different conceptualization is evoked. Moses does not tap his staff on the upper side of the rock, which would trigger op, but against one of its vertical sides. If Moses were to stop his tapping, gravity would pull the staff to the ground and the staff would not stay in contact with the rock, i.e. would not stick to the landmark. ( l c ' ) Mozes

tikte

met zijn staf tegen

de rots.

Moses tapped with his staff against therock 'Moses tapped against the rock with his staff.'

Force Dynamics

in Static

205

Prepositions

The same difference can be observed between examples (2b) and (10b) above. In (2b), the mud sticks to the wall at the area of contact, i.e. the area where the mud hits the wall, and it remains there. This conceptualization is different from the one associated with (10b), where the mud comes into contact with the wall and starts sliding downwards; in other words, at the area of contact, i.e. the area where it hits the wall, the mud does not stick to the wall, but starts to move downwards. Thirdly, it is no longer necessary to distinguish several senses for tegen. Recall that in Cuyckens' (1991) account of tegen, a distinction is made between CONTACT relations, cf. (9), and CASUAL CONTACT relations. (9)

Mijn

perceel

ligt legen

my

plot

lies against the his

het zijne. (43 a, 268)

'My plot is next to his.'

Cuyckens distinguishes several CASUAL CONTACT relations: (i) relations in which the trajector is supported from underneath and is in contact with a vertical surface of the landmark, cf. (7); and (ii) relations in which the trajector is not supported from underneath, but in contact with a surface of a landmark that can have any orientation, cf. (8). (7)

De tafel Staat tegen de muur. (37a, 2 6 2 ) 'The table stands against the wall.'

(8)

a. Het schilderij the painting

hangt tegen

de

muur. (33b, 2 6 2 )

hangs against the wall

'The painting is hanging on [against] the wall.' b. De

lamp

hangt tegen

de

the lamp hangs against the

zoldering.

(38a, 263)

ceiling

'The lamp is on the ceiling.' c. Met zijn oor tegen

de

grond

hoorde hij de

with his ear against the ground heard

Indianen naderen.

he the Indians

(42c, 2 6 7 )

approach

'With his ear against the ground, he heard the Indians approach.'

I will argue that the examples in (7)-(9) can all be accounted for if the semantics of tegen is characterized as proposed in (13) and Figure 4 above, i.e. as symbolizing 'a relation of contact between two entities, a trajector and a landmark, in which at the area of contact the trajector does not stick to the landmark.' 5 Let us consider examples (7) and (8c) first, in which the absence of the notion 'stick to' is very clear. In (7), the table is in contact with the wall, but at the area of contact the table does not stick to the wall: the table is supported by the ground. If the ground were to move, be lowered for example, the table would move too; it would not stick to the wall. Similarly in (8c), the ear is in contact with the ground, but does not stick to the ground. At first sight, (8a) and (8b) may seem to involve the 'stick to' relation. Both (8a) and (8b) allow interpretations in which the trajector is attached to the landmark: (8a) allows interpretations in which the painting hangs on a hook in the wall, and (8b) allows interpretations in which the lamp is screwed to the ceiling. Yet, while these interpretations 5

Notice that Cuyckens' ( 1991: 2 6 3 ) definition o f CASUAL CONTACT ("COINCIDENCE relation between χ a n d y i n w h i c h SUPPORT ( b e t w e e n χ a n d y ) , ADHERENCE, o r ATTACHMENT a r e n o t at s t a k e " ) i s i n

fact very similar to m y definition o f the Platonic concept associated with tegen.

206

Maaike Beliën

involve attachment, attachment is not what is symbolized by (8a) and (8b); what these sentences do symbolize is a kind of contact that is not of a 'stick to' nature. Sentence (8b) cannot be used for a situation in which the lamp hangs from the ceiling only by means of a cord - cf. the scene depicted by Figure 5 (left); in a situation like that, aan would have to be used. Instead, the lamp has to be a ceiling lamp, which is not only in contact with the ceiling at its point(s) of attachment, but also at other places - cf. the scene depicted by Figure 5 (left); and it is at those places that the relation 'stick to' does not hold. Note that if the scene in Figure 5 (right) is interpreted as a relation of attachment, the preposition aan would be used.

Figure 5: A lamp aan (left)/tegen (right) a ceiling

Similarly, the kind of contact relation that is designated by (8a) is not one of attachment. If the painting hangs from a hook in the wall, what is symbolized is the contact between the wall and other parts of the painting (e.g. the back of the lowest part of the frame). This is illustrated by Figure 6 (left), where the arrow identifies the 'non-stick-to' relation. Yet (8a) allows more interpretations, one of which is the scene depicted in Figure 6 (right). In that situation, the painting hangs from a cord from the ceiling and touches the wall with one of its sides. In that case, too, the painting is in contact with the wall, but does not stick to it. These are good examples of what Langacker (1987) calls construah the ability of human beings to structure or portray a situation in different ways for purposes of thought or expression. The scene depicted in Figure 6 (left), for example, can be construed in several ways, even if we restrict ourselves to relations of contact between the painting as trajector and the wall as landmark. Either the relation of attachment, at the hook, is profiled, which would be conveyed by means of aan. Or, the 'non-stick-to' relation between the lower part of the frame and the wall is profiled, which would be conveyed by means of tegen.

Figure 6: A painting tegen a wall

I would like to treat examples such as (9), which in Cuyckens (1991) constitute a separate category denoting 'non-casual contact', in the same vein as examples (7) and (8). While in objective reality, the two plots of land are not separate entities, but part of a larger, con-

207

Force Dynamics in Static Prepositions

tinuing stretch of land, they may be construed as constituting two separate entities that border on one another, very much in the same way as two paving stones: in contact, but not sticking to one another. 6 Finally, also Cuyckens (1991) various senses of aan, CONTIGUITY (cf. 3), MINIMAL ATTACHMENT (cf. 4), and TRANSIENT CONTACT (cf. 6), can be subsumed under one Platonic concept. (3)

Zij wonen aan de Jan van Rijswijcklaan. (75a, 231) they live on the Jan van Rijswijck-avenue 'They live on the Jan van Rijswijcklaan.'

(4)

a. het schilderij aan de muur (65b, 225) 'the painting on the wall' b. de druppel water aan zijn vinger (67b, 226) the drop water on his finger 'the drop of water on his finger'

(6)

Hij voelde aan zijn baard. (70b, 228) he felt on his beard 'He touched his beard.'

What the uses of aan in (3), (4) and (6), share, in my view, is that, unlike examples with tegen, they all designate a 'stick to' relation between trajector and landmark; furthermore, unlike examples with op, the 'stick to' relation does not result from a dominant force directed from the trajector to the landmark. This does not mean, however, that interpretations of examples with aan never involve any forces other than the 'stick to' relation. In a context such as (4b), and the same holds for (4a), gravity will also be part of the interpretation. In the scene depicted in Figure 7 (left), the drop hangs from the finger; the adhesive properties of the drop prevent it from falling. This interpretation includes the force of gravity which operates on the drop, pulling it away from the finger. This force is not, however, part of the semantics of aan (pace Bowerman 1996: 153, cf. Section 4 above). Instead, it is part of the context that the language user thinks of when (s)he tries to come up with an interpretation that involves a drop of water sticking to a finger.

C

C Figure 7: A drop aan (left)/·op (right) a finger

Yet, if in addition to the 'stick to' relation, there is a dominant force that is directed from the trajector toward the landmark, aan cannot be used. Sentence (4b) cannot be used, for example, to describe the scene in Figure 7 (right): a drop of water that rests on a finger. In that case, the force of gravity exerts a downward force on the drop causing the trajector (the drop) to stick to the landmark (the finger). To describe this scene, op has to be used. Regier 6

Instead of using tegen in (9), aan is also possible, but again, that would evoke another construal, one which profiles the attachment between two adjacent pieces of land.

208

Maaike Beliën

(1996) makes an important point in this respect when he states that children learn the spatial concepts associated with prepositions by taking "every positive instance for one spatial concept to be an implicit negative instance for all other spatial concepts being learned" (1996: 62). The existence of op in the system of prepositions pre-empts the use of aan in the situation just described. This means that that situation will not be an option if the language user tries to find an interpretation for a drop of water aan a finger. An example such as (6) - TRANSIENT CONTACT in Cuyckens' (1991) account - is another good example of a sense in context: the notion TRANSIENT should not be ascribed to aan, but results from negotiating the Platonic concept associated with aan with the other linguistic information in the sentence. In (6), the man establishes contact between his hand and his beard, either just once or repetitively (when stroking his beard, for instance). In either case, the hand sticks to the beard for some time, which warrants the use oí aan. The fact that this contact only lasts for a limited amount of time - which may be one reason why Cuyckens calls this TRANSIENT CONTACT - is information we know about people touching their beards (and as such, it is not inherent in the semantics of aan). With respect to the CONTIGUITY cases oí aan, I think, unlike Cuyckens (1991), that the notion of contact is relevant and present in the conceptualization. Cuyckens distinguishes the use of aan in (3) from the other usages, because in (3) the trajector (the inhabitants, or the house) is not necessarily in contact with the landmark (the avenue); rather, the trajector is part of the landmark's region of interaction. While there may not be objective contact between the house and the avenue in that the house may be at a distance from the avenue, I think that the scale of the house and the avenue are such that the relation between them are naturally construed as one of contact. Consider in this respect Langacker's (1987) discussion of close to, whose instances in (14) he claims to be "essentially identical if one abstracts away from the absolute distances involved" (1987: 118: his examples 5a and 5d). (14)

a. The two galaxies are very close to one another. b. The sulphur and oxygen atoms are quite close to one another in this type of molecule.

Similarly in (3), the scale of the trajector and landmark involved, i.e. a house and an avenue, naturally evokes a certain distance from which the relation between the two is viewed, and from which it can easily be construed as a relation of contact.

7. Concluding remarks

In this paper, I have argued that an analysis of the semantics of the Dutch prepositions aan, op, and tegen in terms of force-dynamic Platonic concepts produces a more unified and satisfactory account than an analysis in terms of family-resemblance networks of senses such as Cuyckens (1991), which does not recognize the systematic importance of force dynamics. I have described the semantics of aan, op, and tegen in terms of Platonic concepts, single maximally schematic concepts that language users abstract from experience through their perceptual and cognitive apparatus and that may adapt under contextual près-

Force Dynamics in Static

Prepositions

209

sure. In the case of aan, op, and tegen, the Platonic concepts each involve a static relation of contact between a trajector and a landmark, but the concepts differ crucially with respect to force-dynamic properties.

References

Bowerman, Melissa (1996): The Origins of Children's Spatial Semantic Categories: Cognitive versus Linguistic Determinants. - In: J. Gumperz & S. C. Levinson (eds.): Rethinking Linguistic Relativity, 145-176. Cambridge: Cambridge University Press. Brugman, Claudia (1981): Story o/Over. - M.A. thesis. University of California at Berkeley. Cuyckens, Hubert (1991): The Semantics of Spatial Prepositions in Dutch. A Cognitive-Linguistic Exercise. - Unpublished Ph.D. dissertation. University of Antwerp (UIA). -

(1994): Family Resemblance in the Dutch Spatial Preposition Op. - In: M. Schwarz (ed.): Ergebnisse, Probleme, Perspektiven, 179-195. Tübingen: Narr. Freyd, Jennifer J., Teresa M. Pantzer & Jeannette L. Cheng (1988): Representing Statics as Forces in Equilibrium. - In: Journal of Experimental Psychology: General 117, 395-407. Fauconnier, Gilles (1994) [1985]: Mental Spaces: Aspects of Meaning Construction in Natural Language. - Cambridge: Cambridge University Press. - ( 1997): Mappings in Thought and Language. - Cambridge: Cambridge University Press. Hofstadter, Douglas & Gary McGraw (1995): Letter Spirit: Esthetic Perception and Creative Play in the Rich Microcosm of the Roman Alphabet. - In: D. Hofstadter (ed.): Fluid Concepts and Creative Analogies, 407-466. New York, NY: Basic Books. Lakoff, George (1987): Women, Fire, and Dangerous Things. What Categories Reveal about the Mind. - Chicago, IL: The University of Chicago Press. Langacker, Ronald W. (1987): Foundations of Cognitive Grammar. Vol 1: Theoretical Prerequisites. - Stanford, CA: Stanford University Press. Van der Leek, Frederike (1996a): Rigid Syntax and Flexible Meaning: The Case of the English Ditransitive. - In: A. Goldberg (ed.): Conceptual Structure, Discourse and Language, 321-332. Stanford, CA: CSLI Publications. - (1996b): The English Conative Construction: A Compositional Account. - In: Proceedings of the Annual Meeting of the Chicago Linguistic Society 32, 363-378. - (2000): Caused-Motion and the 'Bottom-Up' Role of Grammar. - In: A. Foolen & F. van der Leek (eds.): Constructions in Cognitive Linguistics, 301-331. Amsterdam: John Benjamins. Pustejovsky, James (1995): The Generative Lexicon. - Cambridge, MA: The MIT Press. Regier, Terry (1996): The Human Semantic Potential. - Cambridge, MA: The MIT Press. Verhagen, Arie (1997): Context, Meaning, and Interpretation, in a Practical Approach to Linguistics. - In: L. Lentz and H. Pander Maat (eds.): Discourse Analysis and Evaluation: Functional Approaches, 7-39. Amsterdam/Atlanta: Rodopi.

Ignasi Navarro i Ferrando Towards a Description of the Meaning of At

1. Introduction

This paper offers a new, cognitive-linguistic view of the meaning of the preposition at Standard accounts of the meaning of English prepositions are based on geometric, or topological, descriptions (Lindkvist 1950; Leech 1969; Bennett 1975; Quirk et al. 1985; Herskovits 1986). Accordingly, an explanation of the contrast between the prepositions at, on, and in, for instance, is looked for in the geometric configuration of the prepositions' landmarks as a point, surface or line, and area or volume, respectively. In this paper, it will be argued that the semantic description of at cannot (only) hinge on a?s landmark configuration. As Talmy (1983) suggests, there are other aspects that language takes into account in order to express spatial relationships, such as the trajector's geometry, site, path or orientation, the conceptualizer's perspective and point of view, the scope and reference frame of the scene, and, finally, force-dynamic patterns of interaction between the participants. According to Deane's (1993) multimodal conception of meaning, language users perceive and conceptualize three aspects of space. (i)

Visual space images, which represent spatial relationships in terms of separation, contiguity, angle of vision, and any aspect related to the position of entities in relation to each other; visual space images thus refer to topological relationships. (ii) Manoeuvre space images, which pertain to information based on motor control and the capacity to interact with other people and manipulate objects as well as the body; manoeuvre space images thus refer to functional relationships. (iii) Kinetic space images, which provide information necessary to calculate forcedynamic interaction in terms of paths, directions, axes, gravity, or the relative orientation of the participants (cf. Talmy 1988); kinetic space images thus refer to force-dynamic relationships. In our experience of space, these perceptual/conceptual aspects or relations tend to cooccur. As such, they are likely to co-constitute the particular conceptualizations invoked by prepositions as well (in terms of trajector-landmark relationships). It seems safe to say, then, that it is the topological, functional, and force-dynamic aspects of space which together form the essence of a preposition's (in particular, of at's) semantic makeup and thus make up its conceptual schema.

1

I am grateful for the helpful comments and suggestions provided by the two editors of this volume, Hubert Cuyckens and Günter Radden. I also gratefully acknowledge the comments of two anonymous referees.

212

Ignasi Navarro i Ferrando

It is my contention that this conceptual schema constitutes the core of a polysemy network, whereby the core comprises the sense of a linguistic unit that is most economically related to the peripheral senses. As such, it is sanctioned, at least partially, by all the senses of the linguistic unit. In this paper, at's conceptual schema will be represented in a graphic form that is reminiscent of the schema's gestalt nature. As a gestalt, a conceptual schema is decomposable, but its parts are only meaningful within the whole they are part of. The senses of at will be presented as a radial category, with the conceptual schema at the centre. As Sandra & Rice (1995) point out, a radial category not only shows the various senses of a polysemous morpheme, but also which of them is deemed to be the central one and which ones are peripheral. Most importantly, it reveals which senses are directly linked to each other, i.e. it shows the conceptual distance between different senses as well as the mapping of their extensions. Finally, I will briefly look at the prototype of at. In this respect, it will be pointed out that a conceptual (or central) schema does not necessarily coincide with the prototype of the category. Prototypical members of a category (i.e. prototypical senses) are, amongst others, those members/senses that show the highest cue validity - and that maximize differences with respect to the prototypes of other categories.2 However, rather than focus on at's prototype, I will attach more importance to topological, functional, and force-dynamic aspects and the way they determine conceptual regions. Conceptual regions are not discrete, but merge into each other. Thus, senses in one region may be close to, or far from, other senses depending on their location and their distance to other regions.

2. Methodology

2.1. Corpus analysis vs. psycholinguistic experimentation Cognitive linguists have strongly criticized the overwhelming use of introspection in linguistic methodology (Geeraerts, Grondelaers & Bakema 1994; Sandra & Rice 1995). The fact that theorists turn to introspective knowledge of the phenomena studied may lead to ad hoc conclusions. The two alternatives to introspection currently applied in cognitive linguistics are psycholinguistic experimentation and corpus analysis. Psycholinguistic experiments allow researchers to elicit language users' reactions to certain phenomena such as judgments about the prototypical or peripheral status of a particular sense of a word. Corpus analysis provides large-scale information about linguistic phenomena. For example, a highly frequent sense of a word in the corpus is indicative of a high degree of entrenchment of this particular sense in the linguistic community. Taking this into account, I shall attempt to determine prototypical uses of at on the basis of their frequency in the corpus, which in turn shows the cue validity of the most relevant perceptual/conceptual aspect of each sense.

2

Since only at, and no competing prepositional categories will be analyzed in this paper, only the first of these factors will be considered in this analysis of at

Towards a Description of the Meaning of At

213

2.2. The corpus The corpus used for the analysis of at is the Brown Corpus of American English compiled by Francis & Kucera (1961). A total of 2,238 instances of at were analysed. They correspond to all the ar-examples found in 210 different texts evenly distributed over fifteen text types. 3 Using a corpus with different text types has the advantage that the sentences studied represent a high diversity of speakers and situational and linguistic contexts. The influence of particular idiolects, topics, and social or geographical dialects can therefore be kept to a minimum.

2.3. Aim and procedure The aim of the analysis was to investigate the polysemy of at within the network model outlined above. To that purpose, the following steps were taken: -

Collection of the data: All the instances of at found in the corpus were retrieved by using the Word Cruncher software provided by the ICAME Collection CD ROM (1991). The data was obtained by a KWIC concordance of the whole sample, whereby each item selected was accompanied by a maximum of 100 characters to the left and to the right so as to provide sufficient context for the task of disambiguation. - Semantic classification: This task involved distinguishing patterns of meaning according to total or partial sanction by the conceptual schema. - Determining prototypical senses: Prototypical senses of at were established on the basis of absolute frequencies of particular senses. Higher frequency was assumed to imply a higher degree of cue validity for a particular dimension of perceptual space as well as a higher degree of entrenchment of the sense in the linguistic community. The results offered in this paper are limited to the data found in the corpus utilized. Analyses based on other corpora or other varieties of English might yield different results.

3. T h e landmark configuration of at

3.1. The landmark as a point The idea that the landmark of at is to be conceptualized as a point has become established in the semantic description of this preposition. It appears pervasively in Lindkvist's (1950) study, which presents a thorough review of contexts in which at appears. Lindkvist (1950: 3 In order to avoid skewed results, the same number of texts (i.e. fifteen) were selected from each text type, except from those types that contain less than fifteen texts in the BC, which was the case for the text types of Science Fiction (six texts) and Humor (nine texts). The sentences are referenced according to the BC coding; for instance, "A 13: 56" refers to text type, text number, and line number.

214

Ignasi Navarro i Ferrando

129) states that "[a/] can only be used to indicate a point in proximity to something, and not a line or any other thing thought of as having real extension." Herskovits (1986) describes prepositional semantics in terms of "ideal meanings," i.e. "irreducible idealfs] associated with preposition[s]," from which prepositional usage may deviate according to pragmatic tolerance or linguistic convention. In particular, ideal meanings are unitary descriptions based on geometric relations that apply to various geometric figures (points, lines, surfaces, volumes) associated with objects. The ideal meaning proposed for at is "for a point to coincide with another" (Herskovits 1986: 128). Thus, an expression with at asserts that two points, each specified by a different description, overlap in space, the first being typically mapped onto an object and the second onto a fixed earth location. In addition, Herskovits distinguishes a number of use types (i.e. sets of usages) for at, which are seen as deviations from this ideal and which relate to the ideal through various sense shifts. According to Herskovits (1986: 128), the use type "spatial entity at location" involves an object or event "on, in or very close to another object which defines a location. The two objects are conceptualized as coincident points." Possible locations include entities denoted by the words place, spot, location, etc. and geographical locations or geometrical points as in the examples The book is at the place where you left i f , Paul is at Yosemite; There are lounging chairs at the beach·, There is a Christmas party at the office (Herskovits 1986: 128). The question is why this sense is described as an object on, in, or close to another object. Is at synonymous with these prepositions? How can we conceptualize a person at Yosemite if the valley is to be seen as a point? Similarly, is the office where a Christmas party is celebrated a point that coincides with a party as another point? Similar problems arise with respect to other of Herskovits' (1986: 133ff.) use types such as "spatial entity 'at sea'" as in The Titanic will never be at sea again, "spatial entity at generic place" in the context of leisure activities as in He likes to spend his vacations at the seaside, "person at institution" as in My son is at the University, "person using artefact" as in Maggie is at her desk. With respect to this last use type, for instance, Herskovits admits that Maggie in Maggie is at her desk is understood as using, or interacting with, the desk, but she attributes this interpretation to pragmatic inference, not to the meaning of the locative. One might rightly wonder, then, why the use of by or on does not produce the same inferences in this context. It seems that Herskovits' analysis of examples is based on circularity: examples with at are explained as having the "reference object" seen as a point, but no reason is given why a point configuration should apply for valleys, the sea, universities, a medium, etc. The only justification seems to be the use of the preposition at itself. If two prepositions are possible in the same context, the preference for one over the other is attributed to the geometric configuration: at applies to a point, on to a surface of support, and in to a volume or container. But the question arises why a speaker decides in favor of one or another preposition in each case. Cienki's (1989) account of at comprises a set of "preference rules"4 that define prototypical senses. They include a coincidence sense, where the landmark is seen as a point, a functionality sense, where the landmark is seen as an object of activity, and an adlative sense. The adlative sense of at is defined as follows: "This usage does seem to reflect the point-apprehensibility of the L-r that is required for at in the Loc sense. The motion 4

For the notion of 'preference rule ' see Jackendoff ( 1983).

Towards a Description of the Meaning of At

215

involved is on a path to the L-r as a whole (as a point), and the resultant impression is of motion about to overwhelm the L-r as target" (Cienki 1989: 149).5 While Cienki's proposal distinguishes three prototypical senses, it does not account for the possible relationship between these various senses. It also adheres to the view that landmark configuration - as a point or as an object of activity - is essential for the meaning of at. However, according to our frequency analysis based on the Brown Corpus, the number of instances of at where the landmark is configurationally a point is very small. On the other hand, a word such as point as a landmark does not guarantee that the preposition at is used, since point also co-occurs with on and in.

3.2. Landmark as indeterminate The other trend that gives pre-eminence to the landmark configuration as the central feature of the semantic structure of at characterizes it as "indeterminate" (Hawkins 1984) "with insignificant dimensionality" (Leech 1969). By itself, this characterization does not sufficiently distinguish at from other prepositions (e.g. by, via, to, toward), which also feature an indeterminate landmark. It therefore needs to be supplemented with additional, relational information, e.g. "juxtaposition/contiguity" (Leech 1969) or "coincidence" (Hawkins 1984). Hawkins' indeterminate landmark may enter into a coincidence relationship (i) with an indeterminate trajector or (ii) with a path trajector. The first schema is illustrated by He sat at the roll-top desk and the solar observatory at Mt. Wilson, where a state verb or noun gives a stative character to the whole expression. The second schema is instantiated by Buck rushed at the splintering wood and They poked sticks at him·, it requires a verb of motion and a terminative path trajector. It should be pointed out, though, that this schema cannot account for all dynamic usages of at, since the spatial relation denoted by at does not always express the termination of the path. Furthermore, it seems to overlook other relevant aspects of the spatial relation denoted by dynamic at, namely the trajectorlandmark orientation and the force-dynamic axis.

4. The role of function

Miller & Johnson-Laird (1976) and Dirven (1989) emphasize the role that function plays in the meaning of at. In Miller & Johnson-Laird's account, "x at y" implies that χ (the referent/trajector) is included in the region of y (the relatum/the landmark), i.e. "where χ can interact with y socially, physically, or in whatever way *'s conventionally interact with / s . " (Miller & Johnson-Laird 1976: 388). For designations with at in relativistic space,6 judgments involving the region of an object y combine perceptual with functional informa5 6

In Cienki's text L-r stands for Localizer (= landmark) and SpE for Spatial Entity (= trajector). For the terms "absolute space" and "relativistic space", see Miller & Johnson-Laird (1976: 3 8 0 382.

216

Ignasi Navarro i Ferrando

tion about the use of that object. For designations with at in absolute space (as in at ¡3,000 feet and at 0 degrees longitude), the region of interaction is the region around the value denoted by the landmark y,7 i.e. absolute at need not be absolutely precise. Their proposal for a schema of at is the following: AT (x, y): A r e f e r e n t Χ is at a r e l a t u m y if: (i) INCL (x, REGION (y))

This schema puts particular restrictions on referent and relatum: (i) the referent must not be too large to be included in the region of the relatum; (ii) relata should be less mobile and larger or more salient than the referents they locate, but small objects can serve the function of a relatum too if they are perceptually salient. Furthermore, the inclusion relation should not be symmetrical, which is especially relevant when referent and relatum are commensurate in size and mobility. So, the schema is expanded with a second condition: (ii) n o t (INCL (y, REGION (*))

In our analysis below, we will take into account some important aspects pointed out by Miller & Johnson-Laird such as the functional nature and the asymmetry of the relationship, as well as the agentive character and orientation of the trajector.

5. The conceptual schema underlying at

The conceptual schema of a prepositional concept is formed by our bodily experience and perceptual space as well as by linguistic input and interaction. On the whole, language users do not understand the notion of "point" as a place or object whose extension and internal structure are ignored (and we should therefore ban this notion from semantic description and avoid points in graphic representations of the meaning of at). Indeed, children do not have direct experience of points as such. Furthermore, even though people may ignore certain aspects of objects or places for linguistic purposes, they do not conceptualize places or objects as points, i.e. as having no extension and no internal structure. The representation of landmarks as points and trajectories as lines is the product of métonymie thought which highlights the geometrical schematization of an event by abstracting away from other aspects which are also relevant for the meaning denoted by a preposition (cf. Talmy 1983). Speakers have to master the conceptual schema of a preposition before they can use the concept on their own. The acquisition of a conceptual schema is a process that starts with the first linguistic input, in particular when children start associating their elders' use of the preposition with dimensions of their own experience with objects and places (cf. Bowerman 1996). As a result of this type of bodily and linguistic experience, a series of 7

It is interesting to point out that Miller & Johnson-Laird correlate the character of the landmark with the distinction between designations with at in relativistic space and in absolute space: (relatively concrete) landmark objects are associated with relativistic designations with at, while landmarks that have a more abstract character (e.g. at 13,000 feet) link up with absolute designations.

Towards a Description of the Meaning of At

217

aspects from the physical domain of perception are internalized that determine the meaning of at for that domain. At the final stage in the process of its acquisition, the following spatio-physical dimensions define the conceptual schema of the preposition at: 1. Functional front of the trajector. The trajector of at is a human being or an entity that can be conceptualized as having a functional front. By analogy with this human (functional front) model, many non-human entities can be conceived of as having a front with which they interact with the landmark. For example, by analogy with the human front, a chair is conceived of as having a front with which it "interacts" with a table. Hence, the phrase the chair at the table is acceptable, but not the phrase *the ball at the table. The reason is that a chair may be placed in such a way that it appears to interact with a table, with one of its sides oriented towards the table. If the chair is placed in a different position with respect to the table, it may be said to be by, near, in front of, etc. the table, but not at it. Objects such as balls, however, lack a side that may be specifically oriented for interacting with tables in a way analogous to chairs and humans. Therefore, spatial relations between balls and tables are lexicalized by means of by rather than at. 2. Functional nature of the landmark. The landmark is an artefact that offers the trajector an accessible side for interaction. Here an artefact is understood as any object or living being conceived of in such a way that human beings - or by extension other entities conceptualized as having a functional front - can interact with them, either in following some natural or culturally acquired behavior or by adopting an attitudinal position. Thus, the man at the tree can be understood as 'the man picking fruit,' 'the man taking care of the tree,' etc., as long as the parts of the tree addressed are perceived of as being accessible for use, manipulation, or operation. Similarly, in the expression the woman at the window, the woman is conceptualized as (possibly) using the window for a certain purpose such as looking out of it. This functionality aspect associated with the landmark provides an important semantic contrast between at's conceptual schema and the conceptual schema evoked by prepositions such as by. 3. Contiguity of trajector and landmark. With respect to their topological configuration in visual space, trajector and landmark are, or tend to be, in a relation of contiguity. The perceptual contiguity expressed by at does not necessarily imply contact nor does it imply absence of contact. Note that contiguity implying contact is, for instance, expressed by on, while proximity, which implies absence of contact is, for instance, expressed by the preposition by. 4. Interaction between trajector and landmark. On the basis of their functional configuration in "manoeuvre space," trajector and landmark are expected to interact in a face-toface pattern. Thus, a chair placed with its back towards a table is not at the table; rather, it is seen as by the table, since the preposition by does not require the functional front of the trajector to point towards the landmark. The trajector may also show a certain amount of intentionality in using, manipulating, or affecting the landmark. Thus, A at Β indicates that A is using, manipulating, or affecting Β in a certain canonical way. This canonical way may be defined either by the biological and physical configuration of the participants, or by cultural usages and customs of the linguistic community. Due to the trajector's intentionality and functional orientation, the relation is asymmetrical.

218

Ignasi Navarro i Ferrando

5. Directionality. On the basis of their force-dynamic configuration in kinetic space, the positions of trajector and landmark relative to each other define a common axis, along which their relationship adopts a certain directionality. This axis is defined by the functional front of the trajector and its orientation towards the landmark. Prototypically, this is the horizontal axis as found in the canonical position of humans as standing upright. According to this axis, a chair that is lying on the floor could not possibly be construed as at a table, nor can a chair be at a table if the table is lying on the floor, since it does not offer its canonically accessible side for interaction. So, the axis of motion, or the directionality of relative positions, is defined by a line which is typically perpendicular to the axis of the trajector that derives from its upright position 6. Active zones. Trajector and landmark present recognizable "active zones" which roughly coincide with the sides they offer to each other. Langacker ( 1 9 9 1 : 1 9 0 ) defines active zones as "those portions of a trajector or landmark that participate directly in a given relation." 8 Thus, in a construed situation where a man is picking fruit at a tree, the man's active zones are his face and the palms of his hands, while the active zone of the tree are the accessible boughs where the fruit is hanging. In short, the trajector's active zone is defined by its functional front, and the landmark's active zone by any of its areas accessible to use or manipulation by the trajector. 7. Scale. Trajector and landmark are conceived of as being located on the same scale. 9 Thus, under normal circumstances a chair is not construed at a galaxy or a molecule at a table. The conceptual schema established by such spatio-physical dimensions will be referred to as the E N C O U N T E R Schema. Strictly speaking, dimensions 1 and 2 specify trajector and landmark configurations as contextual elements, whereas dimensions 3 to 7 specify aspects of the relationship expressed by at. It should be noted that there is no reference to the geometric dimensionality of the participants. The E N C O U N T E R Schema underlying the use of at defines the basic sense 1 of at and is illustrated in Figure 1. Expressions which instantiate the E N C O U N T E R Schema are: at the altar, at the wheel, at the desk/table, at anchor, at a keyboard (piano, organ), at the mirror, at the counter, at the window, at the bar, at bat, at the helm, at the well, etc., all of which express situations where people use or manipulate tools, machines, or other kinds of objects. Certain verbs like pull, tear, clutch, tug, yank, push, suck, etc. typically co-occur with at in this sense and incorporate the "encounter" semantics of at.

8

The concept of active zone is related to the concept of search domain, as used by Miller & Johnson-Laird (1976). For further illustration of this concept see Langacker's account (1991: 189—

9

The term scale is used here in the sense of that dimension of imagery used in conceptualization, as described by Langacker (1987: Section 3.3.1; 1991: 7).

201).

219

Towards a Description of the Meaning o/At

functional space

active zones

Kinetic axis

trajector Frontal orientation contiguity Figure 1:

ENCOUNTER

Schema

6. Partial sanction of the ENCOUNTER Schema

6.1. Topological configuration senses In these senses, the relation of contiguity is perceptually prevalent over functional or forcedynamic dimensions. The following subsenses are distinguished (see Figure 2, p. 227): (i) coincidence and (ii) definite contiguity. 6.1.1. Coincidence (Sense 2a) In this sense, the idea of contiguity prevails. Still, the functional and force-dynamic dimensions, which remain in the background, may provide a contrast with other lexical units in certain contexts, as can be seen in the following examples: (1) (2)

At General Power's seat in the balcony there is also a gold phone. (G03: 84) Ordinary Carey Williams, armed with a pistol, stood at the polls to insure order. (A01: 76)

Though contact is not implied, there is the possibility of interaction. Thus, in (1) the phone at the seat implies that it is ready to be used by the person in the seat. Similarly, in (2) the guard is understood as performing some function in relation to the polls, and not as merely located 'by' or 'near' that place. At in this sense often collocates with words like location, place, or point. The idiom at hand may have emerged from this sense, indicating that the trajector is construed as coincident with the location of the hand and ready to be used or manipulated, as in (3):

220 (3)

Ignasi Navarro i Ferrando He felt his empty pocket and knew that Roberts had retrieved the only weapon at hand. (L06: 14)

In keeping with the TIME IS S P A C E metaphor, periods of time landmarks can be understood as spans with which another entity, event, action, or state may coincide, and they may therefore trigger the preposition at (cf. Wierzbicka 1993). This temporal 'coincidence' sense is present in several collocations indicating periods of time such as parts of the day (e.g. at night, at dawn, at dusk, at noon, at midnight, at day time) or other periods like at present, at times, at ... time, at Christmas, etc. This sense also appears with less idiomatic expressions of time such as period, date, interval, moment, turn, etc. The relation of the trajector with the period of time is one of mere coincidence, as opposed to on, which implies duration and a progressive view of the period, or in, which implies inclusion in the period. In this sense, at is used with lexical units that mean events or periods, e.g. at birth, at death, at maturity, at infancy, etc. The following examples from the Brown Corpus illustrate this usage: (4) (5) (6) (7)

The executions took place at dawn only a few hours after Havana radio announced their conviction by a revolutionary tribunal [...] (A21: 78) Marquis Pick [...] seems to be the pick of the stable at the present time. (E09: 49) It was probably at this period that Littlepage got his first good look at the ordinary Russian soldier. (G50: 87) Many other (probably nearly all) snakes at maturity are already more than half their final length. (J 11: 1)

This sense of at also underlies idioms like at present, at the moment, or at once. The time referred to is conceived of as coincident with the present time. In the case of at once, the unit once is understood to refer to a very short period of time.

6.1.2. Definite contiguity (Sense 2b) This sense of at is illustrated by the following example: (8)

Straight vertical edges, such as those at the back seam of a sock, can be woven together invisibly. (E15: 150)

In this sense of at, the landmark expresses a part of an entity like end, beginning, top, bottom, front, back, extreme, middle, centre, peak, level, head, edge, ridge, limit, plane, extremity, rim, etc., thus functioning as if it were an active zone of that entity. The trajector is seen as contiguous to this part. Perceptual contiguity is even stronger here than in the coincidence sense. The functional and force-dynamic dimensions of perceptual space are backgrounded in the conceptualization, and at begins to undergo a process of grammaticalization, which turns the lexical unit at into a more schematic unit. Furthermore, this process causes a syntactic construction of the form at (the part of) the entity to become conventionalized; as such, it may also be used in domains other than the physical domain, where at then loses its semantic force. The word following at is then more salient semantically, as shown in the following example: (9)

It is sex that obsesses them, sex that is at the basis of their aesthetic creed. (G13: 55)

Towards

a Description

of the Meaning

of At

221

The TIME IS SPACE metaphor also utilizes this syntactic construction with some of the expressions listed above as well as others which indicate parts of periods of time like start, outset, onset, conclusion, commencement, etc. The time referred to is part of a larger period as in the following example: (10)

[...] it would produce 17 million dollars to help erase an anticipated deficit o f 63 million dollars at the end o f the current fiscal year next A u g & 31. (A02: 5)

This sense comprises collocations of at with stage, phase, etc., which designate coincidence with particular periods of a process, action, or event, as in: (11)

T w e l v e projects proposed by private groups are at the contract-negotiation stage [...] (A07: 55)

At in this sense indicates coincidence with numbers and other numerical expressions that refer to a degree on a scale of, for example, temperature, distance, height, speed, length, weight, depth, wave-length, size, volume, rpm, rate, rhythm, humidity, pressure, latitude, astronomical units, age, value, propulsion, percentages, noise, capacity, density, time, etc., where the scale itself needs not to be mentioned, as exemplified in: (12) (13)

Mrs. Albert Quell is in charge o f admittance for the dancing at 9 P.M. ( A l 8: 33) Besides the well-known hydrogen line at 21 cm wavelength, the spectra o f extraterrestrial radio sources m a y contain sharp lines [...] (HI 1:19)

From the "degree on the scale" usage of at we can infer expressions that indicate a certain quantity such as at least, at most, at large and their variant forms at the least, at the very least, etc. Other idiomatic expressions indicate a part of any type of scale like at best and at worst, which indicate locations on a scale of goodness. Coincidence with temporal measurements are expressed by idioms like at first and at last or at the very first, at long last, as well as by the standard expression at... o 'clock, as illustrated below: (14)

M o v i n g past the presidential viewing stand and Lafayette Square will be at least 4 0 marching units. (A08: 2 6 )

(15)

But they, naturally, kept his secret well, and the public at large knew only of a great excitement in musical and court circles. (K08: 99)

(16) ( 17)

At best only an approximation could be arrived at. (D14: 16) Caught at last, he w a s sentenced to prison. (F08: 3 7 )

The "degree on the scale" usage also occurs with expressions that denote a magnitude on a scale such as at a ... height/speed/pressure/ rpm, etc. The magnitude may also be left implicit. For example, at length means 'more than expected or necessary,' at a distance is understood in the sense of 'at a great distance,' and at any rate is equivalent to 'it does not matter how.' The following examples illustrate this usage: (18)

They mounted up and rode slowly behind the others at a safe distance. (N02: 146)

(19)

At any rate, it s h o w s us h o w immaterial w e are. (D13: 58)

Common uses in colloquial speech are expressions of amounts of money like at... cost, at... fee, at... price, at... tax, at... salary and other expressions that explicitly or implicitly indicate an amount of money, understood as a degree on the scale of amounts of money as in:

222 (20)

Ignasi Navarro i Ferrando

The bill is designed to provide a special schooling for more deaf students in the scholastic age at a reduced cost to the state. (A02: 18)

Similarly, the numerical scale is used to refer to places identified by means of correlative numbers, like addresses or streets, as in the following example: (21)

In addition to these activities, the NAEBM, with headquarters at 420 lexington [sic] Avenue, New York City [...] (E06: 77)

The grammaticalization process implies that speakers are no longer aware of the physical coincidence and assume that at is no more than the adequate morpheme for numbers and scalar expressions. At this stage of semantic bleaching, the functional and kinetic dimensions are already completely absent from the conceptualization.

6.2. Functional configuration senses At in these senses profiles functional space - manoeuvre space - whereby the relation of interaction (functional orientation and intentionality) is highlighted. The following more specific senses can be distinguished: (i) operation and (ii) rough coincidence/unspecific function.

6.2.1. Operation (Sense 3) In this sense, the canonical interaction is conceived of as the use or manipulation of the landmark, or an attempt to affect the landmark. Though the topological and force-dynamic dimensions remain in the background, there is still a fair degree of coincidence, and the trajector's intentionality and orientation define a potential axis of motion towards the landmark. The landmarks in this sense denote places where people perform activities or participate in events for which those places are specifically conceived; these landmarks are usually buildings or public places. The trajectors in this αί-relation are the people who use, manipulate, or affect these buildings/public places, or simply act in relation to them. Trajectors may also be concepts that denote the activities - or the events that these activities constitute - that usually take place in public places or buildings. At appears in this sense in situations where people act, according to their function, typically or canonically in relation to a particular place which is designed for that particular purpose or function. The following patterns are found: - Actors, dancers, orchestras, companies, etc., and their audiences, who perform in places designed for that purpose, e.g. theatres, auditoriums, halls, cinemas, playhouses, opera houses, etc., or parts of them, such as the stage, the screen, the arena, etc. Also included are their actions and attitudes as well as the events that these actions constitute, like plays, concerts, performances, films, representations, premieres, etc., or the objects and equipment typically used for them, as for example: (22)

Miss Sutherland first sang Lucia at Covent Garden in 1959. (C02: 49)

Towards a Description

of the Meaning of At

223

- Sportsmen, teams, players, and their audiences as well as their actions, activities, or attitudes, i.e. games, matches, races, competitions, the desire to win, etc., in relation to the place where they are carried out, like stadiums, race circuits, sports halls, courts, rings, etc., or parts of these places, such as the track, the lawn, etc., as in the following example: (23)

Hansen will engage in his first workout at Miami Stadium prior to the opening tomorrow night of a two-game weekend series... ( A l 1: 4 2 )

- Doctors, nurses, and other types of personnel, and their patients or guests as well as their equipment, actions, and activities, or the events that these constitute in relation to the places where they typically occur, namely hospitals, health centres, etc., or parts of these, like the operating theatre, as in the following example: (24)

[...] as time passes the demand for medical care at VA hospitals will grow proportionately as age fosters illness. (B15: 86)

- Professors, teachers, researchers, faculty, and other types of personnel, and students as well as their actions, activities, and the events these constitute, in relation to the places where they are expected to occur, like schools, universities, research institutes, conservatoires, foundations, etc., or parts of these like classrooms, laboratories, etc., as in example (25): (25)

I am a sophomore at Mount Pleasant High School. (B15: 42)

- Workers, managers, directors, etc., and their clients as well as their actions, activities, and the events that these constitute, in relation to the places where these typically occur, like a company, a factory, a shop, a mine, etc., or parts of these like offices, assembly lines, etc., as in: (26)

The colored people are getting employment at Kent House [...] (K04: 95)

- Priests, monks, nuns, and other members of the clergy, and their parishioners as well as their actions, activities, and the events that these constitute, like services, mass, weddings, etc., in relation to the places where these are carried out, such as church, cathedral, etc., or parts of these, like chapels, etc. as in: (27)

[...] And don't tell me you didn't at church Sunday. (P03: 63)

- Waiters, chefs, cooks, room service, maids, receptionists and other types of personnel, and their clients as well as their actions, activities or the events these constitute, like eating, having a meal, or spending a holiday, in relation to the places where these events usually occur, such as restaurants, cafés, hotels, resorts, etc., as illustrated in example (28):: (28)

The dinner will be held at the Hotel Pierre. (A 14: 53)

- Ships, sailors, fishermen, submarines, etc., as well as their actions, activities, and the events these constitute, like fishing, coastal watching, etc., in relation to the place these are to take place, typically the sea (where sea is culturally conceived of as a place for human activity), as in:

Ignasi Navarro i Ferrando

224

(29)

Additionally, on the many ships at sea and in the smaller naval stations [...] (H10: 65)

Other patterns are less frequent but they follow the same semantic criteria; for instance, military personnel at their locations, sellers of all kinds and their customers at their shops, painters and paintings at galleries, dwellers at a residence, etc.

6.2.2.

Rough coincidence/unspecific function (Sense

2-3)

Consider the following examples: (30) (31 ) (32)

[...] another Big Four summit session at Paris [...] (A34: 18) Player gradually increased his lead over Palmer in winnings and added one more tournament victory at Miami. (A38: 42) Goldberg fell into line with Mr. Hodges' appraisal, though there has been some reluctance to do so at the White House. (B05: 65)

Both topological and functional configurations apply here, so that the conceptual distance between 'coincidence' and 'operation' is reduced: there is coincidence and the trajector bears a functional relationship with the place expressed by the landmark. The physical boundaries of the landmark are blurred in such a way that the trajector may be subsumed within the space of the landmark. The scope of predication of the landmark should not be too large as compared with the scope of the trajector. This constraint explains why the landmark of at is rarely a concept that denotes large expanses, such as countries or big islands. The landmark is usually the name of a place, building, town, city, small piece of land, or small island. Accordingly, the trajector is human or some kind of social or institutional activity, event or entity. Thus, the trajector is not only physically contiguous to the landmark, but also performs some action, carries out some activity, or participates in some event or process for which the location is especially relevant. Aspects of functional space are not fully profiled in sense 2-3, because landmark places are not specifically conceived for a particular function or purpose canonically associated with the trajector. The people, events, activities, etc. that play the role of trajectors are not particularly expected to be, act or happen at those places. On the other hand, the conceptual distance to force-dynamic aspects of the ENCOUNTER Schema is remarkable, since sense 2 - 3 does not indicate any motion along the interactional axis.

6.3. Force-dynamic configuration senses Force-dynamic senses of at focus on the interaction between trajector and landmark and the trajector's directionality, with the topological relation of contiguity and functional orientation remaining in the background. Thus, at is compatible with other linguistic units that express motion along an axis or a tendency to reach a goal, even though the final contact with that goal is not necessarily achieved. The trajector's direction of motion is determined by its functional front as well as by the landmark's accessible active zone. Three more specific senses can be distinguished: (i) search for contiguity, (ii) off contiguity, and (iii) achieved contiguity.

Towards a Description of the Meaning of At

225

6.3.1. Search for contiguity (Sense 4a) This sense of at is illustrated in the following examples from the Brown Corpus: (33) (34) (35) (36)

[...] he felt a fluttering object brush his face. He snatched at it savagely. (F02: 88) Every time I closed my eyes I saw Gray Eyes rushing at me with a knife. (N04: 73) A small Indian dived at Montero, who caught him with a swift upward stroke of his rifle butt. (N04: 152) The sharp wind slapped at him and his feet felt like ice [...] (K10: 49)

In these sentences, at focuses on kinetic aspects of the relation, i.e. its direction. As a result, at in this sense is typically used with motion verbs. The trajector searches for contiguity with the landmark while following the line of the interactional axis. Accordingly, the trajector's functional front and the accessible side of the landmark are relevant for the construal of this a/-relationship. The relation is asymmetrical due to the trajector's intentionality. The context for this sense of at requires verbs which express the notion of 'directing at s o m e t h i n g ' s u c h as beat, charge, come, dab, direct, dive, fire, flail, fly, go, jump, lash, lunge, make, run, rush, shoot, skip, slap, slash, smash, spit, spring, swagger, swing, throw,

thunder, etc. At in the sense of 'search for contiguity' may also be understood as implying aggression or an aggressive attitude. This sense is, thus, useful for establishing a semantic contrast with other relational lexical units in which the element of aggressiveness is missing: throw at vs. throw to, run at vs. run to, go at vs. go to, etc. The meaning of 'aggression' cannot be attributed to the verb, since verbs such as throw, run, or go do not normally convey this aspect of meaning. With verbs of contact like bite, jab, lash, pierce, poke, slap, slash, stab, etc., at in the sense of 'search for contiguity' invites the implicature of 'attempt to' when contiguity might not be achieved as in sentence (37), which is seen in the context of non-reality: (37)

Does each tentacle of the octopus of City government lash at whatever it dislikes [...]? (B19: 66)

6.3.2. Off contiguity (Sense 4b) The following examples illustrate sense 4b: (38) The horses moved at a clump; (El 1: 64) (39) At a nod of his head they let go, [...] (F05:86)

This sense of at also profiles directionality, which, however, points away from the landmark so that the relation of contiguity tends to disappear. The force-dynamic pattern is thus reversed, but the functional orientation of trajector and landmark remains. It is by virtue of this particular orientation that the force-dynamic axis is maintained. Thus, the conceptual distance is enhanced with respect to those senses where contiguity is the most salient aspect, and reduced with respect to senses where function is the most salient aspect. The trajector's intentionality disappears, but the relevance of its functional front and orientation remains. This sense is found in expressions like at a gesture, at a remark, at an order, at a suggestion, etc., where the trajector is an entity whose reaction is conceived of as prompted

226

Ignasi Navarro i Ferrando

by the contiguity relationship with the landmark. Dirven (1993) refers to this sense as a causative usage of the preposition. The causative relation can be noticed in examples like the following: (40) Still more time was consumed while the pilot, at the radioed suggestion of Continental president Robert Six, tried to persuade the armed pair [...] (A42: 44) (41) [...] 23 operators and office personnel stand ready to move into action at a minute's notice. (B20: 87)

6.3.3. Achieved contiguity (Sense 2 - 4 ) This sense is illustrated in the following examples from the Brown Corpus: (42) [...] the band arrived at a small clearing, [...] (B25: 61) (43) Place sleeve seam at center underarm and center of sleeve cap at shoulder seam. (El 5: 147) These sentences involve both motion and achieved contiguity focusing on the end of the path. In particular, trajectors are frontally oriented so that a particular directionality and the asymmetry of the relationship is apparent. Furthermore, the trajector's terminative path coincides with the landmark. This sense of at occurs when it collocates with verbs like arrive, come, place, put, set, etc. Finally, this sense of at is characterized by the reduction of conceptual distance between between those senses for which the kinetic axis of interaction is prevalent and those senses where contiguity is prevalent.

7. A radial category o f at

The radial network shown in Figure 2 represents the conventional meaning of at in adult usage, based on the adult written variety of the Brown Corpus. Senses directly linked to the ENCOUNTER Schema (Senses 2a, 3,4a, and 4b) emerge from highlighting one of the perceptual dimensions that conflate in the central schema. Sense 2 - 4 involves a reduction of the conceptual distance between groups 2 and 4, due to the simultaneous highlighting of topological as well as kinetic dimensions; similarly, sense 2 - 3 combines topological and functional dimensions. Moreover, senses 2 - 3 and 2 - 4 do not derive directly from the central schema because they fail to incorporate the third conceptual region. Thus, in sense 2 - 4 there is no trace of canonical operation. Similarly, in sense 2 - 3 there is a complete lack of motion or directionality. On the other hand, senses 4a and 4b both derive directly from the centre; there is no link between them, since they differ only in the kind of directionality towards, or away from, the landmark, which shows that they have appeared independently. Finally, sense 2b moves towards the periphery due to semantic bleaching of both functional and kinetic dimensions. Perceptual dimensions define conceptual regions that merge into each other resulting in intermediate uses. Extensions from one sense to another are gradual, to the extent that many instances found in the corpus are difficult to classify. The network is assumed to be a graphic representation of linguistic usage with no psychological reality itself. Nevertheless, the network

Towards a Description of the Meaning of At

227

is compatible with authentic usage, since it can be used to make correct predictions of actual language use. Topological conceptual region

Force-dynamic conceptual region

encounter

2-3 \ rough \ coincidence/ unspecific J function /

I

/ /

\

contiguity

3 operation

\ \

Figure 2: A radial category of at

The network proposed meets the requirements stated by Sandra & Rice (1995) for discrimination capacity, relational structure, and conceptual distance. Sense discrimination is achieved by presenting the schema of at as a multimodal system, from which subsenses can be derived through partial sanction. Af s relational structure is shown by links between senses. Finally, conceptual distance correlates with the extent to which senses share semantic/conceptual information and is graphically shown by the location of senses in conceptual regions. The ENCOUNTER Schema does not necessarily identify the prototype of the category. While it shows the highest degree of semantic overlap in that it constitutes the central sense that can

228

Ignasi Navarro i Ferrando

most easily be linked to all the others, it is not the most frequent one. High frequency of occurrence makes a sense become more prominent, conventionalized, and entrenched in adult usage. It is assumed by the linguistic community to be the main sense of a word, and it might lead linguists to identify it with a hypothetical core sense. In the analysis of at, this is the case of senses 2a and 2b, where contiguity of trajector and landmark is profiled. Sense 2b is the most frequent one in adult written language (25.28%), since it includes the use of at with numerical expressions in science, economy, and other abstract domains. In the physical domain, other senses like 4a are also highly frequent. But if temporal uses are included in group 2, instances profiling physical and temporal contiguity amount to 56% of the total number of occurrences. Thus, if cue validity is taken as the main criterion to define the prototype, contiguity shows the highest value. Table 1 shows the absolute frequencies and the percentages of the senses analysed in the corpus. SENSE

ABSOLUTE

PERCENTAGE

FREQUENCY

2a Coincidence 2a Coincidence-time 2b Definite contiguity 2b Definite contiguity-time 2-3 Rough coincidence/unspecific function 3 Operation

277 136 260 566 94 171 180

12.38% 6.08% 11.62% 25.28% 4.20% 7.64% 8.04%

4a Search for contiguity

362

16.18%

4b Off-contiguity

85

3.80%

2-4 Achieved contiguity unclassified 10

29 78

1.30% 3.48%

2,238

100.00%

1 ENCOUNTER s c h e m a

TOTAL

Table I: Absolute frequencies

and percentages

of at-senses in the corpus

analysed

8. Conclusions

It has been shown that the semantic characterization of at cannot (only) be provided by af s landmark configuration ("point" or "indeterminate landmark"). Furthermore, at does not express a relation of general location, but a finely specified one, as described by the ENCOUNTER SCHEMA (cf. Section 5). In particular, rather than only a landmark configuration, three dimensions of perceptual space are relevant for the spatial relationship, as determined by topology, function, and force-dynamic interaction. The framework can also be adopted in the analysis of other prepositions. On the basis of these three dimensions, at expresses: 10

This group comprises 66 instances of the expression at all, plus 12 ambiguous instances.

Towards a Description

229

of the Meaning of At

(i) (ii)

a topological relation of coincidence; a functional relation of use or manipulation of the landmark by the trajector (intentionality); (iii) a force-dynamic pattern that follows a horizontal axis defined by the frontal orientation of the trajector; (iv) a relationship between two participants on the same scale whose active zones are the functional front (trajector) and a side accessible to manipulation (landmark). Identifying a prototypical sense constitutes a controversial issue. On the one hand, the ENCOUNTER Schema shows the highest degree of semantic overlap and constitutes the central sense, i.e. the one that is most easily linked to all the others. On the other hand, the highest degree of cue validity attributed by speakers to visual aspects brings the contiguity senses to a 56% of the instances found in the corpus, sense 2b being the most frequent (25.28%). Our proposal of a radial semantic structure should not be taken as a mere reordering of senses of at, since: (i)

it does more than provide a mere list of contexts and their meanings; it explains what the semantic contribution is of a? to those contexts; (ii) it explains why these contexts appear, and not others, i.e. the motivation of uses; (iii) it provides a scheme that explains the conceptual link between the different uses observed, and can motivate new ones; (iv) it explains why a particular lexical item is acceptable in certain contexts, e.g. person at place of employment, person at sea, etc., and what semantic difference there may be between it and other prepositions (on, in) in these contexts; (v) it refutes the idea that landmark configuration is the clue to the meaning of at, which has been assumed in previous semantic accounts of this preposition. Finally, no claims are made on the route of acquisition of the different senses posited. What is offered is an account of adult usage in written language. In order to test the radial category proposed here, further research is required in the direction of psycholinguistic experimentation of the kind carried out by Sandra & Rice (1995) and Rice (1996). Developmental studies in the line of Tomasello (1987) should also be carried out to compare children's speech with adult language, as well as to establish ontogenetic patterns of meaning acquisition and evolution.

References

Bennett, David C. (1975): Spatial ficational Bowerman,

and Temporal

Uses of English

Prepositions:

An Essay

in

Strati-

Semantics. - London: Longman. Melissa (1996): Learning how to Structure Space for Language: A

Crosslinguistic

Perspective. In: P. Bloom (ed.): Language and Space, 385-436. Cambridge, MA.: MIT Press. Cienki, Alan J. (1989): Spatial

Cognition

and the Semantics

of Prepositions

in English,

Polish

and

Russian. - München: Otto Sagner. Deane, Paul D. (1993): At, by, to, and Past: An Essay in Multimodal Image Theory. - In: of the Annual Meeting

of the Berkeley Linguistics

Society

19, 1 1 2 - 1 2 4 .

Proceedings

230

Ignasi Navarro i Ferrando

Dirven, René (1989) : Space Prepositions. - In : R. Dirven (ed.): A User's Grammar of English: Word, Sentence, Text, Interaction, 519-550. Frankfurt a.M.: Lang. - ( 1 9 9 3 ) : Dividing up Physical and Mental Space into Conceptual Categories by means of English Prepositions. - In: C. Zelinsky-Wibbelt (ed): The Semantics of Prepositions, 73-97. Berlin: Mouton de Gruyter. Francis, W.N. & H. Kucera (1961): Brown Corpus: A Standard Corpus of Present-Day American English. - Brown University. Geeraerts, Dirk, Stefan Grondelaers & Peter Bakema (1994): The Structure of Lexical Variation. Berlin: Mouton de Gruyter. Hawkins, Bruce W. (1984): The Semantics of English Spatial Prepositions. - Ph.D. dissertation. University of California San Diego. Herskovits, Annette (1986): Language and Spatial Cognition·. An Interdisciplinary Study of the Prepositions in English. - Cambridge: Cambridge University Press. ICAME (1991): ICAME Collection of English Language Corpora. CD ROM. - Bergen: Norwegian Computing Centre for the Humanities. Jackendoff, Ray (1983): Semantics and Cognition. - Cambridge, MA.: MIT Press. Langacker, Ronald W. (1987): Foundations of Cognitive Grammar. Vol 1: Theoretical Prerequisites. - Stanford, CA: Stanford University Press. - ( 1991 ): Concept, Image, and Symbol: The Cognitive Basis of Grammar. - Berlin: Mouton de Gruyter. Leech, Geoffrey Ν. (1969): Towards a Semantic Description of English. - London: Longman. Lindkvist, Karl-Gunnar (1950): Studies on the Local Sense of the Prepositions IN, AT, ON and TO in Modern English. - Lund: Berlingska Boktryckeriet. Miller, George & Philip Johnson-Laird (1976): Language and Perception. - Cambridge, MA: Harvard University Press. Quirk, Randolph, Sydney Greenbaum, Geoffrey Leech & Jan Svartvik (1985): A Comprehensive Grammar of the English Language. - London: Longman. Rice, Sally (1996): Prepositional Prototypes. - In: M. Pütz & R. Dirven (eds.): The Construal of Space in Language and Thought, 135-165. Berlin: Mouton de Gruyter. Sandra, Dominiek & Sally Rice (1995): Network Analysis of Prepositional Meaning: Mirroring Whose M i n d - The Linguist's or the Language User's? - In: Cognitive Linguistics 6, 89-130. Talmy, Leonard (1983): How Language Structures Space. - In: H. Pick & L. Acredolo (eds.): Spatial Orientation: Theory, Research and Application, 225-281. New York, NY: Plenum Press. - ( 1988): Force Dynamics in Language and Cognition. - In: Cognitive Science 12, 49-100. Tomasello, Michael (1987): Learning to Use Prepositions: A Case Study. - In: Journal of Child Language 14, 79-98. Wierzbicka, Anna (1993): Why Do We Say in April, on Thursday, at 10 o'clock? In Search of an Explanation." - In: Studies in Language 17,437-454.

Günter Radden & Elizabeth Matthis Why Similar To, but Different Front?

1. Introduction*

The prepositions used for describing similarity and difference have a spatial basis. The abstract notions of similarity and difference are in general understood in terms of the metaphors SIMILARITY is CLOSENESS and DIFFERENCE is DISTANCE. These complementary metaphors also apply to a number of randomly selected languages. A major challenge of this paper is to discover the cognitive motivation for these metaphors and to explain the particular use of prepositions in English. A notorious case of seemingly erratic prepositional usage is that of from, to, and than with different. These prepositional variants are often claimed to be interchangeable. However, this paper argues that these prepositions convey differences in meaning which derive from conceptual schemata typically associated with them.

2. Metaphorical understanding of similarity and d i f f e r e n c e as space

Notions of similarity and difference are often understood metaphorically in terms of space. Similarity is metaphorized as closeness, difference as distance. 1 These two complementary metaphors are illustrated in the examples under (1) and (2): (1)

SIMILARITY IS CLOSENESS a. Maroon and crimson are close in color, but they are by no means identical. b. Buying land on the m o o n borders

on the absurd.

c. B e n is nearly as tall as me. d. The cotton shirt is approximately

the same price as the rayon shirt.

Sentence (la) may be interpreted spatially or metaphorically. In the former reading, maroon and crimson are located close to each other on the color spectrum. In the latter reading, the two colors are understood as being similar to each other. Sentence ( l b ) would only be interpreted metaphorically in the sense "buying land on the moon is similar to being absurd." The adverbs nearly and approximately in sentences ( l c ) and (Id) involve spatial closeness on a scale, but can also be seen as metaphors of similarity. This can be *

W e w o u l d like to dedicate this article to Ekkehard König on the occasion o f his 60th birthday.

1

Cf. Lakoff et al. (1994), w h o describe the metaphors as SIMILARITY IS CLOSENESS and DIFFERENCE BETWEEN IS DISTANCE BETWEEN, and Grady (1997: 283), w h o lists t w o spatial metaphors for similarity, viz. SIMILARITY is PROXIMITY and SIMILARITY is ALIGNMENT, as in There parallels

between

are

stunning

these two novels·, the latter metaphors, however, will not be discussed here.

232

Günter Radden & Elizabeth

Matthis

illustrated by rewording the same statements as Ben's height is similar to mine and The price of the cotton shirt is similar to that of the rayon shirt. The S I M I L A R I T Y IS C L O S E N E S S metaphor is also found in the etymology of many words expressing similarity such as nearly, next to, approximately), approach, and affinity, and is at the base of the words same and similar ('at the same time, simultaneous' = temporal closeness). (2)

DIFFERENCE IS DISTANCE

a. b. c. d.

Red and green are far apart. The difference between red and green is vast. Your idea of friendship and my idea of friendship are worlds apart. Hard work separates the men from the boys.

Like sentence (la), sentence (2a) may be understood either spatially or metaphorically, while sentences (2b-d) are only understood metaphorically in the sense of indicating a difference. The D I F F E R E N C E IS D I S T A N C E metaphor also accounts for expressions such as far from the truth, widely different, generation gap, and poles apart. It is also found in the etymology of many words expressing difference such as different (from differre 'bear apart'), distinguish, separate, deviate, diverge, and depart from.2 All of these words derive from roots meaning 'separate.' In the metaphorical mapping, states of similarity or difference are metaphorized as situations of closeness or distance, as in His performance is nearly perfect and His view is widely different from mine. By entailment, changes to states of similarity or difference are metaphorized as motion towards closeness or distance, as in His performance is approaching perfection and His view diverges from mine. States of similarity and difference are, however, also expressed by using the motion prepositions to and from. Thus, we find the following correspondences between the spatial situations described in (3) and the metaphorical ones described in (4): (3) (4)

a. b. a. b.

Fred's Fred's Fred's Fred's

house house house house

is close to Gerald's house. is far away from Grant's house. is similar to Gerald's house. is different from Grant's house.

In these sentences, the Goal preposition to and the Source preposition from make us see the scene as motional. We trace a mental path from one entity to the other. These situations are similar to those analyzed by Langacker (1991: 157-160) as subjective motion, in which motion verbs are used to describe states as in The roof slopes steeply upward. Here the conceptualizer is thought of as mentally moving along a path. Subjective motion is different from physical motion. In physical, or objective, motion, the mover permanently changes his position leaving one position behind when reaching the next position. In subjective motion, the positions along which the conceptualizer mentally moves are added in such a way that they are still present in the conceptualizer's mind and not left behind. This dynamic view of a static situation is achieved by what Langacker calls summary scanning. 2

Cf. also Sweetser's (1987: 450) analysis of the vocabulary of mental states. The etymologies of de-cide (< Lat. de+caedo 'cut off from') and distinguish (Lat. di(s)- 'apart') suggest to her "that mental choices and mental distinctions or categorizations have been metaphorically structured as physical separation processes."

233

Why Similar To, but Different From?

Summary scanning results in a summary view of a path as a whole in "oriented space," i.e. in conceived directionality. This paper argues that these two aspects of subjective motion, summary viewing and oriented space, provide the conceptual foundation for both the spatial understanding of closeness and distance and their metaphorical extensions, similarity and difference. When we view the whole path, we automatically perceive its length or the distance between its endpoints, i.e. summary viewing provides a natural basis for spatial distance and metaphorical distance. When we scan a scene, we automatically impose directionality on the path taken by our eyes. As will be seen below, directionality is an essential aspect distinguishing closeness from distance and, metaphorically, similarity from difference.

3. Closeness and distance in oriented space

Let us first look at the spatial situations described in (3a) and (3b), which can be diagrammed as shown in Figure 1. Closeness

Distance

(a) Fred's house is close to Gerald's house, (b) Fred's house is far (away) from Grant's house. Figure 1: Closeness and distance in oriented space

Following standard practice, we will describe the moving entity as the figure (Fig) and the stationary entity as the ground (Grd). The different sizes of the houses indicate that the smaller figure is to be located with respect to the larger ground. The houses drawn in dotted lines represent the starting point of the conceptualizer's subjective motion, and the arrows indicate the direction of subjective motion. In the spatial situation (3a), Fred's house is close to Gerald's

house, the Goal prepo-

sition to leads us to mentally move along a path from the figure, Fred's house, to the ground, Gerald's house, and focus on the end point of the path. The path mentally covered is short, and the houses are thus in close proximity. The spatial arrangement in (3b), Fred's house is far (away) from Grant's house, describes the complementary situation: the Source

preposition from has us mentally move along a path from the ground, Grant's house, to the figure, Fred's house, and focus on the end point of the path. The path mentally covered is long, and the houses are thus distant from each other. The method English uses to describe spatial situations of closeness and distance is fairly widespread across languages. Tables 1 and 2 list the construals for spatial closeness and

Günter Radden & Elizabeth

234

Matthis

distance found in a random selection o f languages. 3 The focus here is on the spatial marking irrespective of whether it is coded by a preposition, a postposition, or another spatial gram accompanying an expression o f closeness or distance. The static notions of 'be at/near/by/on' and 'be with' are described as Place and Accompaniment, respectively, and the dynamic notions of 'close to/onto' and 'away from' are described as Goal and Source, respectively. 'Closeness'

Languages

Place/Accompaniment Goal

German, Danish, Dutch, Afrikaans/Korean English, Italian, Spanish, Russian, Persian, Hungarian, Japanese French, Spanish 4 , Chinese, Japanese 5 Polish, Serbo-Croatian, Persian, Finnish

Source no spatial marking

Table 1: Construals of 'spatial closeness' in different

Turkish,

languages

'Distance'

Languages

Goal Source

Kurdish, Arabic German, Dutch, Afrikaans, Danish, English, Italian, Spanish, French, Polish, Russian, Serbo-Croatian, Persian, Hungarian, Finnish, Turkish, Japanese, Chinese, Korean Persian 6

no spatial marking

Table 2: Construals of 'spatial distance ' in different

languages

Spatial closeness is coded in different ways: as a Place, as a Goal, as a Source and without any additional spatial marking apart from the proximity expression. The predominant pattern is that of marking closeness by means of a Goal marker, i.e. to see it as subjective

3

4

5

6

We should like to express our thanks to the following people, most of whom kindly responded to a query posted in LINGUIST and provided us with data on 'close to,' 'different from,' 'similar to' and 'different from' in various languages: Tanja Amini, Angeliki Athanasiadou, Claudia Borgonovo, Szilvia Csabi, Francisco Dubert, Lance Eccles, Rosa J. Garcia, Zeki Hamavand, Naoko Hayase, Esme Jansen van Vuuren, Sean Jensen, Kristine Jensen de López, Arsalan Kahnemuyipour, Young-Key Kim-Renaud, Jeong-Hwa Lee, Rick McCallister, Naida Mehmedbegovic, Victor Pekar, Asya Pereltsvaig, Marc Picard, Edward Plaisance, Karl Reinhardt, and Claudia Soria. Unfortunately, the sentences provided in the languages and comments given about them are too many to present here. In Spanish and Galician, cerca takes the Source preposition (Galician: A miña casa està/é cerca da tua), while próximo goes with the Goal preposition in Spanish {Mi casa está muy próxima a la suya) and with the Goal or Source preposition in Galician (A miña casa está próxima á/da túa). Japanese has both Goal {ni) and Source {kara) markers with proximity, the former, however, being the more unmarked variant. Persian uses different constructions depending on the elements focused upon: if the focus is on the entities related, the Goal marker ¿>e is used {khaneye man be khaneye to nazdik ast 'house my to house your close is'), if the focus is on the notion of closeness, no additional marker is used {khaneye man nazdike khaneye to ast 'house my close house your is').

235

Why Similar To, but Different From?

motion to the ground entity. Two of the languages listed, Spanish and Japanese, construe closeness either as a Goal or as a Source and therefore appear twice in Table 1. In contrast to the wide range of expressions for spatial closeness, spatial distance (cf. Table 2) is coded highly consistently across languages by means of a Source marker, i.e. it is seen as subjective motion away from the ground entity. The only languages in our corpus that do not conform to this pattern are Kurdish and Arabic, which use the same Goal marker for both closeness and distance. One of the languages, Persian, has either a Source marker or no spatial marking. From a cognitive view of language, these observations pose several questions: (i)

Why are static situations of spatial closeness or distance conceived of as subjective motion? (ii) Why is the direction of subjective motion reversed with closeness and distance? (iii) Why is the Source construal for distance more uniform across languages than the Goal construal for closeness?

3.1. Static situations as subjective motion Why are static situations of spatial closeness or distance conceived of as subjective motion? One would expect that a static spatial situation describing the closeness or distance of two objects is expressed by means of a static construction. This is in fact the way the languages listed under Place/Accompaniment in Table 1 (i.e. German, Danish, Dutch, Afrikaans, and Korean) describe situations of closeness. 'My house is close to your house,' for example, is rendered in German as in (5) and in Korean as in (6), i.e. the two houses are seen as being in a near-contact or comitative relation, respectively: (5) (6)

Mein

Haus

ist nah

an Deinem

my:NOM house

is close on

na-uy cip-un

ne-uy

Haus.

your:DAT house 7

cip-kwa

kakkap-ta

I-GEN house-Top you-GEN house-coM ('with') Close-DECL

These spatial conceptualizations seem to be motivated in a straightforward fashion by our perceptual experience: objects that are close together are in the same visual field and, hence, may be perceived simultaneously. But even within the same visual field, they may be perceived successively by letting one's eye travel from one object to the other. In this case, the static situation is subjectively construed as motional and either the Goal or Source marker is chosen in describing the situation. Situations of 'distance' are never conceived of in a static way but only dynamically in terms of subjective motion. This fact may also be explained on perceptual grounds: objects which are at a greater distance from one another are not in the same visual field and hence cannot be perceived together. In order to see both of these objects, the observer needs to scan the path from one location to the other location. Let us now consider the issue of directedness in subjective motion. 7

The following abbreviations are used in these glosses: NOM nominative; COM comitative; DAT dative; DECL declarative; GEN genitive; TOP topicalizer.

236

Günter Radden & Elizabeth

Matthis

3.2. Direction of subjective motion Why is the direction of subjective motion reversed with closeness and distance? Spatial situations which are scanned allow for two directions of scanning: from the figure to the ground or from the ground to the figure. As a rule, the direction of scanning is fixed. For example, in English, we can neither speak of *X is close from Y nor of *Y is far to 8 We should expect to find a cognitive motivation for this asymmetry. Let us first look at the preferred direction associated with closeness. In the languages listed in Table 1, closeness is predominantly expressed by means of a Goal marker as in English close to or in Russian blizkiy k 'close onto.' The use of a Source marker with closeness is much rarer, and in at least French and Spanish, the "all-purpose" preposition de has taken over so many functions that it is no longer uniquely associated with its original sense of 'source.' When we conceive of objects which are close together, we thus typically scan from the figure to the ground. This preferred viewing arrangement may also have a perceptual basis: although the two objects are in the same visual field, they are of course not equally in focus at the same time. The viewer begins his or her subjective motion from the more salient entity, the figure, and scans over to the less salient entity, the ground. Since the viewer has both objects in his or her visual field, s/he may from the very beginning focus his or her gaze on the ground as the goal, which gives rise to the situation illustrated in Figure la. The preferred direction associated with distance leads from the focused ground to the figure. In the languages listed in Table 2, the use of a Source marker is almost the only option available for coding situations of distance.9 This linguistic situation might have the following perceptual analog: Objects which are far apart form each other are typically not in the same visual field, and hence we may not be able to scan from the figure to the ground because the distant goal may not be visible as a target. A much safer strategy to use would be to scan in the direction from the ground to the salient figure as illustrated in Figure lb. Here the observer takes the viewing position of the ground. In language, this is typically achieved by choosing a deictic alignment in which the speaker appears as the ground as in (7a), which is in fact felt to be a more natural description than a description in which the speaker functions as the figure as in (7b). (7)

a. He (Fig) lives far from me (Grd). b. I (Fig) live far away from him (Grd).

The construal chosen in (7a) with the speaker as the ground is remarkable because in our normal alignment of figure and ground the speaker tends to identify with the salient figure. The cross-linguistically preferred directionality of scanning from the ground to the distant figure may thus have its motivation in a folk model of a viewpoint taken when looking at distant objects. 8

9

Great distances, however, may be described by to as in It is very far to the shops, which is understood in the sense of 'It is very far to walk to the shops.' The only languages in this corpus which are exceptional with respect to their coding of distance are Kurdish and Arabic, which use a Goal marker for both closeness and distance, viz. la- and min, respectively. The distinction between distances is only made by the adjective and neutralized with respect to the spatial marker, which seems to be semantically bleached.

Why Similar To, but Different From?

237

3.3. Distance as Source Why is the Source construal for distance, i.e. figure is far from ground, more uniform across languages than the Goal construal for closeness, i.e. figure is close to ground? The Source pattern for distance apparently seems to be better motivated than the Goal pattern for closeness. One reason may be seen in the natural direction of scanning especially for distant objects as discussed above. Another motivating factor favoring the figure is far from ground construal may be an interactional aspect. Let us illustrate this aspect by way of the following pair of sentences: (8)

a. How close are we to the airport now? b. How far (away) are we from the airport now?

In a situation involving humans and objects or other humans, the human will typically interact with the object(s) or other human(s). Thus, the close io-question in (8a) suggests that we are heading towards the airport and want to reach it soon, i.e. we want to "interact" with the object, which is within reach. An away /row-question, by contrast, has the opposite effect: (8b) suggests that we are heading away from the airport and no longer want to "interact" with the distant object. It is only a short way from spatial closeness and distance to metaphorical closeness and distance. The spatial expressions close to and far (away) from may be extended to describe metaphorical closeness, i.e. similarity as in (9a), and metaphorical distance, i.e. difference as in (9b): (9)

a. This is close to the truth, b. This is far from the truth.

Like closeness and distance in oriented space, judgments of similarity and difference are thus understood in terms of subjective motion; and as with spatial closeness and distance, the directionality of scanning tends to be fixed: the figure is similar to the ground and not *similar from the ground, and the ground is different from the figure and not *different to the figure. In the following, we will first look at the conceptual impact of judgments of similarity and then examine in what way metaphorical extensions from oriented space to 'similarity' and 'difference' are motivated.

4. Judgments of similarity and difference vs. acts of comparison

Judgments of similarity or difference are similar to acts of comparison. Langacker (1987: 101-109) describes acts of comparison by the schematic form S > Τ = V, where S refers to the standard of comparison, Τ to the target of comparison, > to the operation of scanning in a particular direction, and F to a value for the vector of scanning in some domain. Whenever we compare two entities, we thus have in mind a standard entity and a target entity, and by mentally moving from the standard entity to the target entity, we register discrepancies between the two entities in some domain, and the degree of divergence between the

238

Günter Rodden & Elizabeth

Matthis

entities compared can be expressed as the "value" of comparison. To what extent is Langacker's model of acts of comparison compatible with judgments of similarity or difference? Acts of comparison and judgments of similarity/difference are similar in that they both require the conceptualizer's scanning from one entity to another, i.e. they both involve directed motion. Acts of comparison and judgments of similarity/difference are, however, different in at least the following respects. First, acts of comparison are based on specific frames of reference 10 or dimensions which are typically expressed explicitly as in Henry is as tall as Leopold or Henry is taller than Leopold. Here, the two people are compared with respect to the dimension of height, and the "discrepancy value" between their heights may be measured by means of their different positions on the scale of height. This is not the case with judgments of similarity or difference. We typically judge two things as being similar or different on a fairly global basis. Thus, we can be unspecific about a judgment of similarity, as in (10a), but we can hardly do so with an act of comparison, as in (10b). Conversely, we may ask for the frame of reference with similarity judgments by saying In what way?, as in (10c), but we cannot do so with comparisons, as in (lOd): (10)

a. b. c. d.

Sheila looks like her sister, but I just can't say why. ?Sheila is taller than her sister, but I just can't say why. Sheila looks like her sister. - In what way? Sheila is taller than her sister. - ?In what way?

SIMILARITY COMPARISON SIMILARITY COMPARISON

Second, acts of comparison lead to the discovery of discrepancies between the two entities compared. If no discrepancies between two events are registered, an act of comparison leads, as a limiting case, to recognition. Judgments of similarity or difference, on the other hand, are based on recognizing sameness or no sameness in two events. We do not go about comparing all kinds of things in order to find two entities which are similar or different. Instead, we judge things as being similar or different on the basis of their global gestalt. Third, following Langacker, acts of comparison involve a directed path from the standard of comparison to the target. The kind of comparison Langacker seems to have in mind is a conscious act by which the conceptualizer first activates a standard model and then matches it against some target as, for example, in looking for some item in a shop. Acts of comparison, however, are also triggered by a stimulus which is compared to a standard entity. This situation typically applies to recognition by means of identification criteria. Aitchison (1994: 65) illustrates the impact of identification criteria with the following delightful example: A farmer put a ring through a cow's nose and people automatically wrongly identified it as a "bull." 11 Here, the appearance of a particular thing was compared 10

11

The notion of reference frame with judgments of similarity is used by Tversky (1977), who does not, however, distinguish between 'similarity' and 'comparison.' The bull story cited by Aitchison goes back to a newspaper report. A farmer w h o had trouble with people breaking down fences so that his cattle escaped put a ring through the nose of one of the cows. People mistook the cow for a bull and the local police called the farmer up: Police·. Farmer:

You've got a bull in the park - it's illegal. I'm sorry but we have no bull in the park.

239

Why Similar To, but Different From?

to the appearance of the prototypical "standard" member of a category, and, in this case, people's similarity judgments obviously resulted in a faulty categorization. Acts of comparison and judgments of similarity or difference are, thus, closely related but clearly different cognitive processes. We will describe the schematic meaning evoked by an act or expression of comparison as a "comparison schema." In the following section we will also specify the schematic meanings associated with the notions of 'similarity' and 'difference.' Up to this point we have looked at the cognitive process leading to judgments of similarity and difference. We will now look at the spatial metaphors used in different languages to express notions of similarity as opposed to difference and attempt to account for them in terms of a folk model of similarity and difference. As a final step, we will also try to account for the different prepositions used in English to denote the notion of difference.

5. Metaphorical construals of'similarity' and 'difference'

The complementary metaphors SIMILARITY IS CLOSENESS and DIFFERENCE IS DISTANCE are widely attested across languages. The distribution of the Goal and Source patterns is even more systematic with these metaphorical usages than with the literal spatial usages listed in Tables 1 and 2. In the random sample of languages chosen, similarity and difference are expressed in terms of space as shown in Tables 3 and 4 (other non-spatial construals are not listed). We will look at spatial construals of 'similarity' first. 'Similarity'

Languages

Place/Accompaniment Goal

Dutch 12 , Afrikaans/Chinese, Korean Danish, French, English, Italian, Spanish, Polish, Russian, Greek, Persian, Hungarian, Turkish, Hebrew, Japanese German, Danish 13 Finnish Russian 14 , Serbo-Croatian, Persian 15

Similarity Comparison no spatial marking

Table 3: Construals

12

13

14 15

of 'similarity ' in different

languages

( ...continued) Police: I've seen it myself - I saw the ring in its nose. Farmer: You'd better go and look at the other end. In Dutch 'similarity' can be expressed by the spatial prepositions aan, op, and met, depending on the adjective or verb the preposition collocates with: (i) gelijk aan, gelijkend op 'similar on', vergelijkbaar met 'similar with'; (ii) lijken op 'to be similar on.' In Danish the choice of the preposition to express 'similarity' depends on the adjective: thus, 'my house is similar to yours' translates in Danish as Mit hus er lige som ('like') dit or Mit hus er mage til ('to') dit. Russian h a s p o h o z h i y na 'similar on,', shozhiy s 'similar with,' and podobniy 'similar' + DAT. For the Persian prepositional and zero variants expressing 'similarity' the same distinction applies as for those of 'closeness.'

240

Günter Radden & Elizabeth

Matthis

As with 'closeness,' the preferred coding of 'similarity' is not as a static situation, but as a dynamic situation of subjective motion. The preferred direction is that of the figure's motion to the ground as the goal - the direction from a Source does not even occur in this corpus. Instead, forms marking the Place, Accompaniment, Goal, Similarity, Comparison, and spatially unmarked forms are used. Let us compare these data to the forms used to construe the notion of 'difference.' 'Difference'

Languages

Accompaniment Goal Source

Persian, Chinese, Japanese, Korean English, Turkish English, Danish, Dutch, Afrikaans, French, Italian, Spanish 16 , Polish, Russian, Serbo-Croatian, Greek, Hebrew, Hungarian English, German, Danish, Dutch 17 , Afrikaans, Finnish

Comparison

Table 4: Construals of 'difference ' in different

languages

As with 'distance,' the notion of 'difference' is exclusively coded in a dynamic way, and the preferred direction is away from the focused ground, the Source, to the figure. The Goal is only focused upon in Kurdish, which uses the Goal preposition for both similarity and difference, and English, which, as will be discussed below, uses different to as a variant of different from. Like some other languages, English makes also use of the comparison schema as in different than. Chinese and Korean are interesting in that they split up the conceptual space in a different way: they use the same Comitative marker for both 'similarity' and 'difference' and other markers for 'closeness' and 'distance.' The overwhelming metaphorization of 'difference' as 'source' is also supported by sign languages. In American Sign Language, the sign for 'different' is made in the following way: "Beginning with both extended index fingers crossed in front of the chest, palms facing forward, bring the hands apart from each other with a deliberate movement." 18 How can these preferred metaphorizations of 'similarity' and 'difference' be represented and explained conceptually? A situation of similarity as in example (4a), Fred's house is similar to Gerald's house, is metaphorically conceived of as a figure's motion towards a ground, i.e. its goal. We may think of this metaphorical conceptualization as follows: the conceptualizer moves an image of Fred's house to an image of Gerald's house and finds global commonalities in both houses, which allows him to subsume both of them as members of the same category of house. At the same time, the conceptualizer notices slight differences between the houses so that they are not judged to be identical. To indicate that

16

17

18

The Goal form diferente a is also used especially in conversational Mexican Spanish. It is also listed in dictionaries and frowned upon by purists. Both Danish and Dutch have, depending on the spatial adjective, a Source or a Comparative preposition: Dutch verschillend van ('from') and anders dan ('than'), Danish forskellig fra ('from') and anderledes end ('than'). Random House American Sign Language Dictionary (1994). The sign for 'similar, alike, identical,' however, does not seem to correspond to anything found in verbal language: "Move the right y hand, palm facing down, from side to side with a short repeated movement in front of the body."

241

Why Similar To, but Different From?

the houses are similar, but not identical, the speaker might have said, The houses are nearly the same, which means the conceptualizer stopped before reaching the identity stage. A situation of difference like in sentence (4b), Fred's house is different from Grant's house, is conceived of as a figure's motion away from a ground entity, i.e. its source. We may think of this metaphorical conceptualization in the following way: The images of the two houses are first compared and then judged not alike enough to be members of the same category. Then, as a result, the figure entity is metaphorically pushed away. Figure 2 is a simplified illustration of these two processes. As in Figure 1, the arrows indicate the direction of the conceptualizer's subjective motion and the houses drawn in dotted lines represent its end point and starting point. Similarity

Difference

(a) Fred's house is similar to Gerald's house,

(b) Fred's house is different from Grant's house.

Figure 2: Similarity and

difference

Not surprisingly, the conceptual schémas of similarity and difference resemble those of closeness and distance in oriented space. Their metaphorical correspondences are listed under (11). (11)

a. b. c. d. e.

Similar things correspond to close things. Different things correspond to distant things. Judgments of similarity/difference correspond to motion. Judgments of similarity correspond to motion toward a goal. Judgments of difference correspond to motion away from a source.

Similar things tend to be close to each other. Grady (1997: 129) conjectures that this mapping is motivated by our experience of finding similar objects near each other in our environment: blades of grass cluster together in a lawn, rocks cluster together on the ground, and clouds cluster together in the sky. Further motivating factors might be that "it is easier to make comparisons and perceive similarity when objects are close together" and that visual backgrounds "are likely to be more similar for objects near each other than for two which are separated by a greater distance." By analogy, we may expect different things to be distant from each other. Judgments of similarity or difference are mental processes and, like other mental processes, are metaphorized as motion: THINKING IS M O V I N G . 1 9 In judging whether two 19

Cf. Lakoff & Johnson (1999: 236-238): Examples of the metaphor THINKING IS MOVING are My mind was racing·, My mind wandered for a moment; to reach a conclusion ; to drive at a crucial point in the discussion, etc. Likewise, Jäkel (1997: 195), in his study of metaphors for mental activity, lists a metaphor THINKING IS A JOURNEY, and Sweetser (1970: 60), in her analysis of modality, analyzed reasoning processes as being metaphorically modelled as a journey through space.

242

Günter Rodden & Elizabeth

Matthis

things are similar or not, the conceptualizer does not move himself but rather, as in spatial closeness and distance, remains at his position and only performs a subjective, i.e. mental act of motion. As in the case of spatial distances, the directionality in the process of subjective motion is important. In judging two things to be similar, we mentally move to the ground thing, and in judging two things to be different, we move away from the ground thing. But, unlike things which are spatially close or distant, things which are similar or different are seen as being moved by an internal force: Similar things are attracted to each other, while different things are repelled from each other. The proverbial expression Birds of a feather flock together exemplifies the folk view that similar things are attracted to each other - the Goal preposition to in together lends further support to this view. The proverbial expression Oil and water don't mix, by contrast, demonstrates the folk view that different things separate themselves from one another. These two complementary folk models might be described as "attraction schema" and "repulsion schema," respectively. The attraction and repulsion schemata represent idealized metaphorical conceptualizations of similarity and difference. The attraction schema is not only found in expressions denoting similarity but also in expressions denoting related concepts; it is even used with notions of identity. 20 Table 5 lists groups of symmetric adjectives in English that are construed with the Goal preposition to. They all involve notions of similarity, in the case of identity "complete" similarity. The repulsion schema underlies predicates denoting 'difference' and related concepts such as 'distinction' and 'separation' or 'division' as shown in Table 6. Attraction schema

Expressions

Similarity Association 21 Relation

similar to, close to, next to, comparable to linked to, tied to, connected to related to, parallel to12, equivalent to, proportionate to, corresponding to, akin to complementary to, senior to, junior to, superior to, inferior to opposed to, contrary to equal to, identical to

Complementarity Opposition Identity

Table 5: Attraction schema with English predicates of similarity and related

Repulsion schema

Expressions

Difference Distinction Separation

different from, differ from distinguish from, distinct from, apart from separate from, diverge from, depart from

concepts

Table 6: Repulsion schema with English predicates of difference and related concepts 20

21

The problem of distinguishing between the prepositions with and to with symmetric predicates (parallel with/to, identical with/to, but only simultaneous with, analogous with) shall not be dealt with here. More specifically, this metaphor may be, following Grady (1997: 129), described as ASSOCIATION IS CONNECTION.

22

Following Grady (1997: 283), this metaphor might be described as SIMILARITY IS ALIGNMENT.

Why Similar To, but Different From?

243

English seems to reflect these two folk models perfectly in its use of to and from with predicates denoting 'similarity' and 'difference.' There are, however, some inconsistencies. First, the negated forms of similar, i.e. not similar or dissimilar, are not, or not only construed with the Source preposition from, but also, and more commonly, with the Goal preposition to as in the following examples.23 (12)

a. Japanese has a grammatical category of tense which is not too dissimilar to that of English. b. [...] the study of the properties of objects in the social sciences is quite dissimilar to equivalent studies in physics or chemistry [...]

The negation of dissimilar in sentence (12a), which results in the positive meaning 'fairly similar,' might be responsible for the use of the proposition to. This does not, however, apply to sentence (12b) and similar sentences with dissimilar, which seem to sound more natural with to than with from. Thus, it is not the semantics of dissimilar that accounts for the prepositional construal but the presence of the form similar. The same applies to the reverse situation: the negation of different, i.e. no(t) different, does not necessarily affect the use o f the preposition from as in I'm no different from any man.

A more complicated situation arises in varieties of English in which different may also be construed with to or than. The remainder of this paper will focus on explaining the distribution and motivation of these three prepositions with different.

6. Different with from, to, and than

In different varieties of English two or three of the following forms may be used: (13)

a. Your hair is different from my hair.

b. Your hair is different to my hair. c. Your hair is different than my hair.

The use of the Source marker from in (13a) reflects the repulsion schema. On the basis of our cross-linguistic data, this construal is expected to be the normal and unmarked usage. The use of the Goal marker to with the notion of 'difference' in (13b) is an exception across languages. It makes us see the difference between things in terms of the attraction schema.24 The use of the preposition than in (13c) makes us see the difference between things in terms of the comparison schema, even though the particular dimension is not stated explicitly. We may, for example, understand the difference between your hair and my hair in (13c) in the sense of 'your hair is blonder than my hair' or 'your hair is more 23

24

These and the following examples were taken from the BNC. Examples with dissimilar from in the BNC corpus include The pleasures of the romance novel are not dissimilar from those of the chocolate bar; naughty but nice. Another example of an adjective which allows alternation between from and to is immune as in Diplomats are immune from/to punishment, which we also tested on our questionnaire (see Section 6.2).

244

Günter Rodden & Elizabeth

Matthis

curly than my hair,' etc. The use of the comparison schema in conceiving of difference is, as shown in Table 4, not unique to English but is also found in other languages such as German, Dutch, Danish, and Finnish. The preference speakers of English have for one of the prepositional alternatives with different seems to be determined by at least three factors interacting with each other: (i) the prescriptive tradition, (ii) geographical distribution, and (iii) the schematic meaning associated with each preposition. 25

6.1. The prescriptive tradition The prescriptive tradition is probably still an important factor in the speaker's choice of the preposition with different. According to prescriptivists, from is the "correct" preposition to go with the original meaning of the prefix dis- 'away from.' 2 6 Arguments given in favor of different from rely on "logic": "[...] logic supports established usage: one thing differs from another, hence, different from" (Strunk & White 2000: 44). By the same token, "than is sometimes defended with the argument that other and otherwise - logically equivalent to different and differently - are idiomatically followed by than" (Barzun 1966: 167). Even if contemporary dictionaries no longer "prescribe" different from, the notion of from being the correct form still seems to be very much present in people's minds. An amusing illustration of this prescriptive attitude is found in H.F. Ellis' short story Preparing for the West (1982: 320): 'How's it go then?' he said. 'How d'you find it?' 'Find it?' I asked. 'All this,' he said. 'New York. Different to London, eh?' 'Yes, indeed. Yes. Oh yes. Different from London certainly.' I agreed, taking care not to stress the corrected preposition. [...]

People's awareness of the prescriptive rule may contribute to their preference for the preposition from in more formal contexts and in the written form. A study on prepositions used with different in New Zealand English found that from is the predominant preposition used in written discourse but not in spoken discourse, where also to and than are widely used. 27 A similar situation applies to the United States: than is much more frequently used with different in spoken English than in written English. 28

25

26

27

See Rohdenburg (this volume) for a further factor, the Complexity Principle, determining the choice of the preposition following different. The Writing Handbook by Kammer & Mulligan (1953: 160), for example, admonishes the student to "Use from, not than or to, after different" and corrects the prepositions in the sentences Oh, she is entirely different than other girls and Smollett's reply was very different to what the chemist expected into from. Interestingly, the same attitude applies to Spanish diferente a, which is widely used and at the same time frowned upon by prescriptivists. Laurie Bauer analyzed the prepositions with different in the Wellington Corpus of Written N e w Zealand English and Robert Sigley the corresponding frequencies in the Spoken Corpus:

245

Why Similar To, but Different From? 6.2. Geographical distribution

Prepositional use with different is often associated with geographical distribution, both on the national scale of British, American and Commonwealth English, and on the regional scale of dialectal differences within these larger areas. Corpus studies on the distribution of the prepositions with different show that from is the preferred preposition in present-day British and American English both in their written and spoken modes. 29 Speakers of British English use both different from and different to and, more rarely, also different than. Speakers of American English allow for two prepositional variants: different from and different than. Different to is very rare or not used at all in America. Reference grammars and dictionaries are well aware of this difference between British and American usage. For example, Murphy's English Grammar in Use observes that British English has both different from and different to as in The film was different from (or to) what I'd expected (p. 260), while American English has both different from and different than as in It was different from (or than) what I'd expected (p. 283). But neither American English nor British (... continued) Wellington Corpus of Written NZ English different from different to different than

28

47 8 2

Wellington Corpus of Spoken NZ English

(82.4%) (14.0%) (3.5%)

30 21 5

(53.6%) (37.5%) (8.9%)

Sigley pointed out by email that older speakers are more likely to prefer different to and that New Zealand English seems to be standardizing on different from. We would like to express our thanks to David Minugh, Robert Sigley, and Stefan Gries, who drew our attention to statistical data on prepositional usages with different. Milroy & Milroy (1999: 14) contrasted the usages of the prepositions with different in two American newspapers, on a radio station, and in an Australian paper. The Table below shows that from predominates in the written mode of the two American papers but is only slightly more frequent than than in the spoken mode of US Broadcast News. New York Times 96 different from different to different than

917 25 155

(84%) (2%) (14%)

Los Angeles Times 95 654 25 255

(70%) (3%) (27%)

US Broadcast News 95 1,129 83 907

(53%) (4%) (43%)

Sydney Morning Herald 95 431 216 16

(65%) (33%) (2%)

Kenndey (1998: 194f) extracted the prepositions following different in four corpora and computed the following distribution expressed as tokens and percentages:

different from different to different than

LongmanLancaster Corpus

Brown Corpus (written US)

LOB Corpus (written UK)

1,193 75 34

40 0 12

38 4 2

(91.6%) (5.8%) (2.6%)

(76.9%) (0%) (23.1%)

(86.4%) (9.1%) (4.5%)

London-Lund Corpus (spoken UK) 31 3 3

(83.8%) (8.1%) (8.1%)

246

Günter Rodden & Elizabeth

Matthis

English nor Commonwealth English are homogeneous with respect to the prepositional alternatives used with different. In the United States, for example, the use of than with different is almost twice as common on the West Coast as on the East Coast.30

6.3. Schematic meanings associated with prepositions Speakers generally assume that, within one such geographical variety, the prepositional variants with different are interchangeable. For example, both Your hair is different from mine and Your hair is different than mine are equally acceptable to speakers of American English and apparently convey the same meaning. In our view, however, each of the prepositions expresses a meaning of its own. The meanings associated with from, to, and than in conjunction with different relate to the conceptual schemata discussed above: The Source preposition from evokes the repulsion schema; in conjunction with different, from is therefore expected to express substantial differences. The Goal preposition to is found in similarity expressions and evokes the attraction schema; in conjunction with different, to is therefore assumed to apply to differences that are slight or non-existent. Than is used in comparative constructions and evokes the comparison schema; in conjunction with different, than is expected to be used to describe differences that apply to one dimension only. In his or her use of language, the speaker has to opt for one of the prepositional construals. Provided that all other variables are held constant, the speaker's choice of the preposition is in all likelihood determined by the specific meaning associated with 'different.' Our basic assumption was that the prepositions from, to, and than are not used interchangeably but rather express subtle differences in meaning. In order to test these assumptions, we carried out a small empirical investigation.

7. Empirical investigation on the use of prepositions with different

We tried to delimit the meanings of the particular prepositions following different based on judgments by native speakers. We developed a questionnaire of thirteen test sentences with different in which native speakers were to fill in the preposition they would use in each case. To distract the participants' attention from the issue of different, we included seven more sentences with other adjectives, which are not considered in this study. The questionnaire is attached as Appendix 1. The numbers of the sentences refer to those of the questionnaire. The sentences in our survey were selected to convey situations which fit each of the three conceptual schemata: situations conveying substantial differences were chosen as examples of the repulsion schema, situations conveying slight differences were chosen to fit the attraction schema, and situations conveying differences along one dimension were 30

Compare the different frequencies of from and than in the New York Times and the Los Angeles Times in the Table of footnote 28. David Minugh, who kindly provided these data, suggests "a US gradient from formal (NYT) to less so (LAT) to spoken (BN)."

247

Why Similar To, but Different From?

assumed to evoke the comparison schema. The aim was to test if native speakers associated the conceptual schemata with the respective prepositions. Our informants were twentythree British, eighteen American, and six Australian speakers of English so that we were also able to elicit geographic differences in prepositional usages. Due to the interaction of the factors - prescriptive, geographic, and semantic - which jointly play a role in the use of these prepositions, usage was expected to be highly divided. The results of the empirical investigation are presented in the following sections. Since we are only interested in the usages of the three prepositions, we have disregarded a few other responses (in, with, when, or, and it's) and computed the frequencies of the three prepositions relative to each other. We will first discuss the overall values of the prepositions with different in American, British, and Australian English and then the use of the prepositions with respect to the conceptual schemata.

7.1. Overall values of prepositions used in American, British, and Australian English The results of the responses by American, British, and Australian speakers on the choice of the prepositions following different are listed in Table 7. American English from to than sum

105 0 126

(45%) (0%)*** (55%)*** 231

Table 7: Frequencies

British English 140 106 56

(46%) (35%)*** (19%)***

Australian English 43 26 19

302 of responses to prepositions

(49%) (29%) (22%)

sum 288 132 201

88 following

different31

The distribution represented in Table 7 deviates highly significantly from the distribution that would be expected on the basis of pure chance (χ 2 =135.43; df=4; ρ < 0.001 ***).32 However, it is important to note that the overall significance results from four of the nine cells only. A configuration frequency analysis33 shows that the observed frequencies for to and than in American and British English deviate strongly from the expected values: In American English, to does not occur at all although, according to chance distribution, it should have occurred 49 times. On the other hand, than occurs 126 times although only 74 occurrences were expected. In British English, to was found 106 times, while 64 occurrences were expected, whereas than occurs much less often than was to be expected: 56 instances instead of 97 were obtained. These results confirm the general tendencies observed for American and British English: absence of to and widespread use of than in 31 32 33

Two and three asterisks represent findings significant at the 1% and 0.1% levels, respectively. We would like to express our thanks to Stefan Gries for helping us analyze the data statistically. A configuration frequency analysis is a statistical technique for analyzing multidimensional contingency tables. It is based on an overall Chi-square test and tests whether the deviation of the observed value from each cell in an n-dimensional table from an expected value is significant. The p-value on which this significance test is based is adjusted for multiple post hoc tests by the socalled Bonferoni correction.

248

Günter Rodden & Elizabeth

Matthis

American English and little use of than and widespread use of to in British English. No significant differences were found in the usages offrom in British and American English or in the distribution of the three prepositions in Australian English.

7.2. Use of prepositions with respect to the conceptual schemata In this section we will briefly discuss the test sentences used for each of the three conceptual schemata and the preferred prepositional choices of each of the three geographical groups of native speakers. The results of the correlation frequency analysis of the three variables -schemata, prepositions and geographical varieties - are summarized in Appendix 2.

7.2.1. Repulsion schema The repulsion schema is associated with substantial global differences. The four test sentences listed in Table 8 were chosen as examples of substantial differences and received the following ratings by the speakers of the three geographical varieties. American #

Test sentences

Australian

British from

to

than

from

to

than

7

18

4

1

4

2

0

from

to

than

12

0

(3)

Sounds waves are very different water waves.

(6)

Her new hairstyle is different anything I have ever seen.

11

0

7

13

8

2

5

1

1

Their lives are rather different those of other people.

11

0

7

15

7

1

2

3

1

16

0

2

19

4

0

5

2

0

23

65**

36

16

8

2

(9)

(16)

Third World countries are different one another.

50

o«**

4**1·

Table 8: Sentences illustrating the repulsion schema and their ratings

Sentence (3) conveys a contrast between two apparent members of a category. Due to their common inheritance within a taxonomy, members of a category are in general taken to be very similar rather than different. Water waves and sound waves, however, do not belong to the same category of "waves" because water waves are literally waves while sound waves are only metaphorically waves. The contrast between the literal expression and the figurative one needs to be emphasized to offset the apparent similarity suggested by the term waves. The use of the preposition from is an appropriate choice to magnify their substantial difference. Sentence (6) makes a statement about a new hairstyle that is unlike any hairstyle the speaker has seen before. It is as if the speaker is flipping through his or her mental log of

Why Similar To, but Different From?

249

hairstyles in an attempt to find one to compare with this one, but comes up with nothing the new hairstyle lacks a grounding entity for comparison. The high ratings for different from probably reflect the substantial difference the subjects saw between two things that have nothing in common to compare. Sentence (9) also expresses a global difference, but the difference can be interpreted as substantial or slight depending on the emphasis given to the intensifier rather, putting stress on rather underscores a large difference, whereas not stressing it lessens the apparent difference. This would account for the slightly lower overall percentage o f f r o m . Sentence (16) describes a reciprocal situation. In comparing any two countries, we mentally switch our vantage points and direction of viewing and see each of two entities once as a figure and once as a ground. Conceptually a reciprocal situation may thus be said to have a "doubling effect," which may account for the high ratings o f f r o m . The speakers of all three geographical varieties are unanimous in their preference of from with the repulsion schema in these sentences; the ratings of the British speakers are even significantly higher than expected (p < 0.001). Conversely, to and than are less frequently chosen than expected; in British English the value for than is even significantly lower than expected (p < 0.01). The subjects' predominant, in part significant, choice of from conforms with our expectation.

7.2.2. Attraction schema The attraction schema is associated with the notion of 'similarity' and slight differences. We hypothesized that different should therefore also be construed with to in situations of slight differences in varieties that have different to, e.g. British and Australian English, but not American English. Sentences which we expected might best match such situations are ones which negate different and thus semantically convey the notion of 'similarity.' Thus, the use of to in I am no different to you is semantically equivalent to I am similar to you and may lead subjects to opt for to. In American English, no different is expected to evoke the comparison schema rather than the repulsion schema, i.e. be used with than. The test sentences listed in Table 9 were chosen to test the impact of negation on the choice of the preposition following different. Our expectations about the choice of prepositions with the attraction schema are well confirmed in British English, where the values of to are significantly higher (p < 0.001) and those of than significantly lower than expected (p < 0.001). In British English, the attraction schema as expressed by to is thus clearly distinguished from the comparison schema as expressed by than. In American English, than occurs only slightly, but not significantly more frequently than from ·, in Australian English, from is also only slightly, but not significantly more frequent than either to or than.

250

Günter Rodden & Elizabeth

American

#

Test sentences

(4)

Your reaction was no different __ anyone else's

(7)

It was Christmas but no different __ any other day except that the shops were closed.

from

British

Matthis

Australian

to

than

from

to

than

from

to

than

8

0

10

1

15

3

4

3

0

8

0

10

10

11

2

3

1

3

(14)

His accent is no different the way he behaves.

8

0

8

13

8

2

3

2

0

(15)

I am no different you.

5

0

13

8

13

2

4

3

0

(20)

He is not at all different the other kids.

13

0

4

11

13

0

4

2

1

45

49

18

11

4

42

0***

60***

Table 9: Sentences illustrating the attraction schema and their

ratings

Sentence (20) is the only test sentence that does not conform to the overall pattern - in fact without this sentence a much clearer preference for than with the attraction schema in American English would emerge. Since American English does not have the option of to and the sentence does not exhibit the comparison schema, from is the natural choice. However, in British English the sentence does conform, although not significantly, to the attraction schema. We also tested a further sentence with no different: (19) Our job now is no different it was seven years ago. This sentence contains the element of comparison and will therefore be presented together with the sentences displaying the comparison schema.

7.2.3. Comparison schema A type of situation that we assumed exemplifies 'difference' in the sense of the comparison schema is that of a comparison of the same referent at different times. Three of the sentences in our survey describe such a situation, in which an entity's present state is compared to an earlier state; one of the sentences, sentence (11), expresses earlier and later times as part of different referents. We expected that the preposition to be used with such situations should be than. The sentences and the subjects' responses to these sentences are listed in Table 10. Sentence (1) compares the same referent in time with respect to its appearance. The difference between the earlier and later state might be explicated in terms of a comparative statement such as 'she looks better than before,' and the comparison schema and the preposition than are obviously well-suited to express this situation of difference.

251

Why Similar To, but Different From?

American # (1)

Test sentences She looks different before.

The situation of an older woman is different that of a younger woman. (13) He is fundamentally different he used to be. (19) Our job is no different it was seven years ago.

Australian

British

from

to

than

from

to

than

from

to

than

2

0

16

8

2

14

2

1

5

9

0

9

12

10

2

2

3

2

2

0

16

4

6

12

2

2

3

0

0

17

2

5

16

3

1

3

58***

26

23

44

9

7

13

(11)

3*** o***

Table 10: Sentences illustrating the comparison schema and their ratings Sentence (11) seems to present a conflicting situation to the subjects: The comparative adjectives older and younger evoke the comparison schema and than, while the different referents may suggest a greater difference and, hence, favor the use of from. The fact that, although the referents are different, they are both women and thus belong to the same category could lead to the use of to. This would explain why the ratings show that this sentence evokes the comparison schema far less readily, especially in British English. Sentences (13) and (19) also fit the comparison schema. They involve a comparison of the same referent in time, but the ground is not expressed as a noun phrase, but as a sentence. According to prescriptive school grammar, the use of than should be avoided. Many subjects felt insecure about which form to use and opted for complex expressions (from what, from how, from the way, to what, to how). These sentences obviously present a conflicting situation between the comparison schema and the prescriptive factor. One British informant even refused to fill in the gap and crossed out the sentence with an "X" as if to say that sentence could not exist. The negation of different in sentence (19) would explain the use of to by some informants (especially British, see 7.2.2). In American English, sentences (1), (13), and (19) are clear cases of the comparison schema with than occurring significantly more frequently (p < 0.001) and to and from significantly less frequently than expected (p < 0.001). Than is also predominantly, though not significantly, more frequently found in British and Australian English in these sentences. Sentence (11) is the exception in this group, with a significant preference for from in British English answers, a tie in American English, and no significant preferences in Australian English.

7.3. Discussion The results of the empirical investigation confirm our assumption that the prepositions used with different are not interchangeable but carry meanings associated with their respective conceptual schema. Several prepositional usages significantly diverge from the expected

252

Günter Radden & Elizabeth Matthis

values, others only display such tendencies. Significant results were found in British and American English, not in Australian English, which is probably due to our small number of informants. Speakers of British English use all three prepositions with different, and the overall usage of to is significantly higher than expected, while that of than is significantly lower. Moreover, British speakers associate most of the prepositions with one of the types of difference: substantial global differences significantly correlate positively with from and negatively with to, and slight differences correlate positively with to and negatively with than. However, no significant correlations were found for the comparison schema in British English. Speakers of American English only use from and than with different, and the overall usage of than is significantly higher than expected. American speakers tend to prefer from with substantial differences and than with slight differences, but not at any level of significance. They do, however, associate the comparison schema with prepositions: especially differences related to the same referent over time significantly correlate positively with than and negatively with from.

8. Conclusion

We presented evidence for a metaphorical understanding of similarity and difference in terms of closeness and distance, respectively. The spatial basis is, amongst others, reflected in the use of spatial prepositions or other spatial markers. A number of randomly selected languages were compared with respect to the types of spatial marker used for closeness and distance on the one hand and similarity and difference on the other hand. Cross-linguistically, the predominant way of expressing these static notions is dynamically in terms of directed subjective motion: both closeness and similarity tend to be expressed by means of a Goal marker as in close to and similar to, and both distance and difference are almost exclusively expressed by means of a Source marker as in far from and different from. These linguistic patterns point to a folk model in which we see close and similar things as being attracted and distant and different things as being repulsed. We described these two aspects of this folk model as the "attraction schema" and "repulsion schema," respectively. A notion closely related to similarity and difference is that of comparison. While judgments of similarity or difference involve global aspects of the things compared, acts of comparison highlight the property of comparison as in X is taller than Y, i.e. they involve one specific dimension. We referred to this situation as the "comparison schema." Typically, each of these conceptual schemata is associated with its own preposition: similarity with to, difference with from, and comparison with than. English is unique among the languages studied in that the expression different may be construed with the Source preposition from, the Goal preposition to, and the Comparison preposition than. In the past, the issue of the choice of the preposition with different has mainly been looked at from the prescriptive point of view and as a matter of geographical variation - apart from that, the prepositions following different are generally claimed to be

Why Similar To, but Different From?

253

interchangeable. While not disregarding these factors, we chose to assume that the prepositions used with different convey different meanings and that their choice is determined by the conceptual schemata associated with them: thus different from is expected to convey the idea of substantial difference, different to the idea of slight difference or even similarity, and different than difference with respect to a specific dimension. These assumptions were empirically tested by presenting a questionnaire including thirteen sentences with different to American, British, and Australian informants and asking them to fill in the preposition that came to mind. The test sentences were chosen according to the three conceptual schemata associated with the prepositions from, to, and than. The results based on 47 responses largely confirm our assumptions: different from is the favored choice for substantial global differences in all three varieties and significantly so in British English, different to is the most frequently chosen preposition with slight global differences and significantly so in British English, and different than is widely chosen in all three varieties for differences along a specific dimension and significantly so in American English.

References

Aitchison, Jean (19942): Words in the Mind: An Introduction to the Mental Lexicon. - Oxford: Blackwell. Barzun, Jacques (ed.) ( 1966): Wilson Follett: Modern American Usage. - N e w York: Warner. Ellis, H. F. (1986): Preparing for the West. - In: Humorous Stories. London: Octopus Books. Grady, Joseph (1997): Foundations of Meaning: Primary Metaphors and Primary Scenes. - Ph.D. dissertation. University of California at Berkeley. Ikegami, Yoshihiko (1987): 'Source' vs. 'Goal': A Case of Linguistic Dissymmetry. - In: R. Dirven & G. Radden (eds.): Concepts of Case, 122-146. Tübingen: Narr. Jäkel, Olaf (1997): Metaphern in abstrakten Diskurs-Domänen. - Frankfurt a.M.: Lang. Kammer, Michael P. & Charles W. Mulligan. (1953): Writing Handbook. - Chicago, IL: Loyola University Press. Kennedy. Graeme (1998): An Introduction to Corpus Linguistics. - London: Longman. Lakoff, George et al. ( 1994): Master Metaphor List. - University of California at Berkeley. Lakoff, George & Mark Johnson (1999): Philosophy in the Flesh. - N e w York: Basic Books. Langacker, Ronald W. (1987): Foundations of Cognitive Grammar. Vol. 1: Theoretical Prerequisites. - Stanford: Stanford University Press. - (1991): Concept, Image, and Symbol: The Cognitive Basis of Grammar. - Berlin: Mouton de Gruyter. Milroy, James & Lesley Milroy (1999): Authority in Language: Investigating Standard English. London: Routledge. Murphy, Raymond (1944 2 ): English Grammar in Use: A Self-Study Reference and Practice Book for Intermediate Students. - Cambridge: Cambridge University Press. Costello, Elaine (comp.) (1994): Random House American Sign Language Dictionary. N e w York: Random House. Strunk, William Jr. & E. B. White (2000 4 ): The Elements of Style. - Needham Heights, MA: Allyn and Bacon.

254

Günter Radden & Elizabeth

Matthis

Sweetser, Eve (1987): Metaphorical Models of Thought and Speech: A Comparison of Historical Directions and Metaphorical Mappings in the Two Domains. - In: Proceedings of the Annual Meeting of the Berkeley Linguistics Society 13, 446-459. Tversky, Amos (1977): Features of Similarity. - In: Psychological Review 84, 327-352.

Appendix 1 : Questionnaire

We are doing some work on prepositions and would really appreciate your help. Please fill in the gaps with the preposition that first comes to mind. We are interested in your native speaker intuition and not grammar. Thanks! 1. She looks different before. 2. Are the chairs identical the set in the living room? 3. Sound waves are very different water waves. 4. Your reaction was no different anyone else's. 5.1 often use English work. 6. Her new hairstyle is different anything I have ever seen. 7. It was Christmas but no different any other day except that the shops were closed. 8. He is not typical Indians in Britain. 9. Their lives are rather different those of our people. 10. Diplomats are immune punishment. 11. The situation of an older woman is different that of a younger woman. 12. He doesn't know his head a hole in the ground. 13. He is fundamentally different he used to be. 14. His accent is no different the way he behaves. 15.1 am no different _ _ _ you. 16. Third World countries are different one another. 17. Microsoft seems to be immune foreign competition 18. My thoughts are identical yours. 19. Our job now is no different it was seven years ago. 20. He is not at all different the other kids. Please fill in the following

to help us with our statistics.

What country are you from? (e.g. Canada, Australia, etc.) Is English your native language? Have you ever lived in another English-speaking country for more than 6 months? If yes, where?

Why Similar To, but Different From?

255

Appendix 2

Results o f a configuration frequency analysis o f conceptual schemata, prepositions and varities CONSTELLATION OF VARIABLES

OBSERVED

EXPECTED

P-VALUE

FREQUENCY

FREQUENCY 3 4

(CFA)35

ATTRACTION

TO

BRITISH

60

245

O***

COMPARISON

THAN

AMERICAN

58

232

0***

REPULSION

THAN

BRITISH

4

301

G*** 0*** 0**#

ATTRACTION

THAN

BRITISH

8

373

COMPARISON

FROM

AMERICAN

13

333

ATTRACTION

TO

AMERICAN

0

187

O***

REPULSION

TO

AMERICAN

0

151

0***

COMPARISON

TO

AMERICAN

0

153

0***

REPULSION

FROM

BRITISH

65

431

0.001**

COMPARISON

FROM

BRITISH

26

435

0.002

ATTRACTION

THAN

AMERICAN

45

285

0.002

REPULSION

FROM

AMERICAN

50

329

0.002

REPULSION

THAN

AUSTRIAN

2

88

0.007

COMPARISON

THAN

BRITISH

44

304

0.010

ATTRACTION

THAN

AUSTRALIAN

4

109

0.016

ATTRACTION

TO

AUSTRALIAN

11

71

0.107

COMPARISON

THAN

AUSTRALIAN

13

89

0.112

COMPARISON

FROM

AUSTRALIAN

9

127

0.185 0.196

REPULSION

FROM

AUSTRALIAN

16

126

REPULSION

TO

AUSTRALIAN

8

58

0.222

REPULSION

TO

BRITISH

23

197

0.257

COMPARISON

TO

BRITISH

23

200

0.273

ATTRACTION

FROM

BRITISH

49

534

0.291 0.300

ATTRACTION

FROM

AUSTRALIAN

18

156

COMPARISON

TO

AUSTRALIAN

7

58

0.364

ATTRACTION

FROM

AMERICAN

42

409

0.451

REPULSION

THAN

AMERICAN

23

230

0.529

621

621

SUMS

34 35

The expected frequencies were rounded to integers. While the traditional critical p-value for standard tests of significance is 0.05, note that, for this table, the p-value had to be adjusted for multiple post hoc tests. Since 3^=27 constellations of variables were possible (three conceptual schémas, three prepositions and three varieties), the common threshold value of 0.05 had to be divided by 27. As a result, the adjusted p-value indicating a result significant on the 5% level is 0.05:27=0.0019.

Hubert

Cuyckens

Metonymy in Prepositions

1. Introduction 1

In their research into the semantics of prepositions, cognitive linguists have become increasingly convinced that the meaning of prepositions can effectively be described in terms of a network of senses or usages which are related to each other by means of general cognitive principles such as metaphor, metonymy, generalization, specialization, and image-schema transformations. As such, prepositional research has shown that there is far less arbitrariness or idiomaticity in prepositions than was hitherto accepted. This should not be surprising, given the attention that has been paid to prepositions in cognitive semantics in the past two decades, where research into the factors motivating semantic relations between the senses/usages of a lexical item have been a central concern. What explains the attention of cognitive semantics for prepositions? In cognitive linguistics, which advocates an experientialist approach to meaning (Lakoff 1987), the semantics of linguistic expressions, and in particular of lexical items, reflects the conceptualization or mental representation of entities or states of affairs in extralinguistic reality (cf. Taylor 1993: 131). The conceptualization of the spatial domain, i.e. of the space surrounding us, of our movements in space, and of (the relation between) objects in space takes up a priviliged place in that it constitutes a basic domain in terms of which more abstract domains (e.g. the domain of emotion) can be conceptualized (Lakoff 1987; Johnson 1987; Lakoff & Johnson 1980; Lakoff & Turner 1989). It should not be surprising, then, that lexical-semantic research in cognitive linguistics has focused first and foremost on lexical items that express spatial relations and which, through semantic extension, can be used to lexicalize relations in more abstract domains. A second factor is the renewed interest from cognitive linguistics for lexical semantics, the recognition that polysemy is "the natural order of things" (Langacker 1988: 50), and concomitantly, the fact that cognitive linguistics was able to come up with an attractive representational model for polysemous lexical items (in terms of a network of interrelated, motivated semantic distinctions). In this scientific climate, it seems quite acceptable that attention would go first to those words that are polysemous par excellence, viz. prepositions. Finally, within the framework of grammaticalization2 research, various researchers have drawn attention to (the cognitive motivation of) the semantic evolution of linguistic

An earlier version of this paper was first published in Dutch as "Metonymie in Preposities"; in S. Gillis, J. Nuyts & J. Taeldeman (eds.) (2000): Met Taal om de Tuin Geleid., 6 3 - 7 3 . University of Antwerp. Permission to publish this slightly revised English version was kindly granted by the editors.

258

Hubert

Cuyckens

expressions (e.g. Sweetser 1990; Heine, Claudi & Hünnemeyer 1991a, 199b; Traugott & König 1991; Hopper & Traugott 1993; Jurafsky 1996). In light of the view that synchronic, polysemous lexical semantic structures (containing both concrete and abstract, grammaticalized usages) somehow reflect the diachronic evolution of word meaning, linguists have looked at the question to what extent a word's spatial usages underlie, from a diachronic perspective, its more abstract usages. As such, it appears that grammaticalized usages of prepositions can often be traced back to their initial spatial meanings. (Genetti 1991; Kilroe 1994; Kabata & Rice 1997; Cuyckens 1999).3 One of the most important principles underlying the relation between the various usages of a lexical item is metaphoric extension (see the very extensive literature after Lakoff & Johnson's Metaphors we Live By (1980); Gibbs & Steen 1999; Dirven 1993, 1995). Recently, however, metonymy has received increasingly more attention as a motivating principle (see Traugott & König 1991; Schwenter & Traugott 1995; Kövecses & Radden 1998; Panther & Radden 1999; Barcelona 2000). In this article, I want to show, by means of the partial semantic analysis of a number of prepositions, 4 that metonymy plays an important part as a motivating factor in semantic extensions and deserves at least as much attention as metaphor-based semantic extensions. Along with Kövecses & Radden (1998: 39), I would like to define metonymy as follows: Metonymy is a cognitive process in which one conceptual access to another conceptual entity, the target, within the definition does not mention 'contiguity', or 'proximity' métonymie relationship. This is because the entities are 'contiguously' related.

entity, the vehicle, provides mental same domain, or ICM5 [...] [T]his between the entities involved in a part of an ICM, and as such are

In what follows, I will show the importance of metonymy for the description of synchronic semantic extensions in the prepositions with and for and for the description of semantic change in the prepositions to and by.

2

3

4

5

In Kurylowicz's classical definition, grammaticalization is "the increase of the range of a morpheme advancing from a lexical to a grammatical or from a less grammatical to a more grammatical status" (1975: 52). In addition, grammaticalization research of prepositions also looks at the development of (spatial) prepositions from lexical content words such as nouns, verbs, adjectives, and adverbs (see, e.g., Kahr 1975; Heine, Claudi & Hünnemeyer 1991a, 1991b; Svorou 1994). It is not the purpose of this article to provide a full semantic description; only those semantic aspects that are significant for the concept of metonymy will be focused on. An ICM is an "Idealized Cognitive Model." Lakoff (1987: 68) describes ICMs as mental structures that organize reality. Analogous definitions can be found in Ungerer & Schmid (1996: 47), who characterize ICMs as "all the stored cognitive representations that belong to a certain field." In a similar vein, Kövecses & Radden (1998: 41) state that "we have ICMs of everything that is conceptualized, which includes the conceptualization of things and events, words and their meanings, and categories of things and events."

Metonymy

in

259

Prepositions

2. The preposition with

The preposition with can express the following relations, exemplified in (l)-(5): ACCOMPANIMENT

(1)

I will walk with m y sister.

(2)

Make sure you cut the meat with a sharp knife.

INSTRUMENT

(3)

T h e y defused the bomb with the utmost care.

MANNER

(4)

I cannot hear y o u with all the w i n d o w s open.

CIRCUMSTANCE

(5)

He trembled with fear.

CAUSE

Dirven (1993: 81; 1995: 101) holds that the abstract usages of with, as illustrated in (2)-(5), should be understood as metaphorical extensions of its spatial 'accompaniment' usage in (1). In other words, there exists a mapping relation between the spatial relation/notion 'accompaniment' and the abstract relations/notions 'instrument,' 'manner,' 'circumstance,' and 'cause': "manner or instrument are [...] seen as companions to fulfil an assignment; circumstance is a companion standing by and cause [is] a[n accompanying] feeling or emotion leading to visible effects" (Dirven 1993: 81). Each of the relations between 'accompaniment' and 'instrument/manner/circumstance/ cause' can indeed be viewed as a metaphorical mapping between two domains, i.e. a concrete, spatial and an abstract domain. Various usages of with, however, may also be linked metonymically. As such, with in (4) not only denotes the accompanying circumstance of the state of affairs ("I cannot hear anything") expressed in the sentence; at the same time, this state of affairs is also caused by the accompanying circumstance. In (6), this causal meaning is present even more clearly: (6)

With the drug lords safely behind bars, the people finally feel safe again.

Although the use of with in (4) and (6) can be understood as an instantiation of the metaphor CAUSE IS CIRCUMSTANCE, whereby with lexicalizes the mapping relation from the abstract domain 'circumstance' to the abstract domain 'cause,' it cannot be denied that both the notion 'circumstance' and 'cause' are part of one event-ICM. As part of one ICM, they are in a conceptual contiguity relation which, in turn, reflects the extralinguistic contiguity of accompanying circumstance and causal circumstance. Indeed, an accompanying factor in a particular event may easily turn into a causal factor in that event. In light of this (conceptual) contiguity, the meaning 'accompanying circumstance' of the preposition with may either blend with the notion 'causal circumstance' so that both notions are simultaneously present (as in (4)), or it may fade so that only the notion 'causal circumstance' remains (5). In sum, then, when describing the relation between 'accompanying circumstance' and 'causal circumstance' in terms of contiguity, we are describing a métonymie relation, which can be captured as CIRCUMSTANCE FOR CAUSE.

260

Hubert

Cuyckens

3. The preposition for

Consider the following sentences: (7) (8)

I bought a new suit for my brother's wedding. The dogs fought for the bone. (cf. Radden 1985: 193)

(9)

A certain amount must be deducted for depreciation. ( L D C E ) 6

Radden (1985: 193-194) describes the use offor in these sentences as instantiations of the metaphor CAUSE IS GOAL, where 'goal' should be interpreted in "in its spatial sense of destination" (1985: 194). Rather than metaphoric, the relation between the abstract domains of 'goal' and 'cause' is in the first instance métonymie. In sentences (7)-(9), the landmark of the relation lexicalized by for can be interpreted as the goal of the action expressed in the sentence and at the same time as the factor causing that action. In (7), for instance, the new suit is being bought with a view to wearing it at the wedding party. At the same time, if it had not been for the wedding party, the subject (I) would not have bought the suit; in other words, the wedding party is the causal factor in purchasing the suit. Sentences (8) and (9) can be analyzed analogously. The fact that both interpretations are present simultaneously in for reflects that he conceptual notions 'goal' and 'cause' should in the first instance be regarded as two components of one and the same event-ICM, which are in a (conceptual) contiguity relation. Given this contiguity relation, the notions 'goal' and 'cause' may not only blend (as in (7)-(9)), but the notion 'goal' may fade, giving rise to instances offor in which only the causal meaning is present: (10)

We could hardly see for the mist. ( L D C E )

While metaphor may undeniably be a motivating factor in the usages of for in (7)-(9), metonymy seems at least equally important in accounting for the semantic relation between for's 'goal' and 'cause' senses, because this relation is based on contiguity: entities functioning as the goal of an event or action can at the same time play a causal role in that event, or take on a causal role exclusively. This métonymie relation can be captured as GOAL FOR CAUSE.

So far, I have described metonymy as a motivating factor in the the synchronic usages of a lexical item. In what follows, I will illustrate that metonymy may also operate as a diachronic extension principle, i.e. as a factor of semantic change.

4. The preposition to

The English preposition to can express allative and dative relations, as in (11) and (12):

6

In this article, the following abbreviations have been used: LDCE Contemporary English; OED = Oxford English Dictionary·, c = circa.

= Longman

Dictionary

of

261

Metonymy in Prepositions

(11) (12)

I am not going to the market tomorrow. Could you please give this book to John?

ALLATIVE DATIVE

At the same time, the infinitive in English is often preceded by a particle that is formally identical to the allative-dative preposition to: (13) (14)

I went to the story to buy some cigarettes. I want you to do the cooking tonight.

Although the allative-dative preposition to and the infinitival particle to are homonyms at first sight, they are demonstrably related. How can this relation best be described and what mechanisms play a role in deriving one usage from the other? The first step in the development from allative-dative to to infinitival to is the extension from allative to purposive to in Old English. Indeed, in Old English, the preposition to also served as a preposition before the nominalized verb in the dative, i.e. the verb stem suffixed with the dative case marker, indicating "motion, direction, inclination, purpose, etc. toward the act or condition expressed by the infinitive" {OED). (15)

Ac Heáand Christ of heofonum me spreec to and sende me to bodigenne his lare. 'But the Savior Christ from heavens spoke to me and sent me to preach his teachings.'

The purpose reading of the Old English preposition to introducing an infinitive not only occurred in contexts such as (15), but also following complement-taking verbs such as forbid, bid (manipulative verbs) and desire, want, prefer (desiderative verbs) (for a complete description of this evolution, see Cuyckens &Verspoor 1998). The link between the allative and the purposive usages of to is often described in terms of the metaphors PURPOSES A R E DESTINATIONS or PURPOSES A R E PHYSICAL GOALS (cf. Lakoff & Turner 1989; Johnson 1987: 114-117). However, while the metaphorical link between purpose and spatial goal undoubtedly exists, it is in the first instance based on the (conceptual) contiguity of the notions 'spatial goal' and 'purpose'; the conceptual contiguity, in turn, reflects an extralinguistic contiguity in that in our everyday experience we usually have an intention in getting to a particular place. This can be seen in sentences (16)(19): (16) ( 17) (18) (19)

They came to our rescue. They sat down to dinner. We were out to breakfast. I am going to be married, (cf. Hopper & Traugott 1993: 82)

In (16)—(19), the allative function of ίο indicating the trajector's motion in the direction of and reaching the landmark combines with the notion of purpose, in other words, both notions cooccur. In (16), for instance, the trajector {they) traverses a path ending in the place where we need to be rescued, but at the same time, the very intention of reaching that spatial goal is rescuing us. The contiguity between 'spatial goal' and 'purpose' allows us to say that the relation between these two notions is initially métonymie. While to was initially used as an allative preposition, to quickly came to denote 'spatial goal + purpose' (cf. (15) in Old English and ( 16)—( 19) in Present-day English), and later developed an exclusively purposive function

Hubert Cuyckens

262

(cf. 13-14). This contiguity-based semantic change can be described as métonymie transfer (see also Geeraerts 1997: 70-72).7

5. The preposition by

Metonymy-based semantic change also plays an important role in the development of the preposition by from its use as a spatial preposition to its use as an agent marker in the passive. Next to its 'proximity' usage, one of by's earliest spatial usages is to express the trajector's motion along (i.e. over) a particular path (the landmark). (20) 3fic pe leede be pam weje (OED, c. 888) If I you led by this way 'If I led you by/along this way' (21) Per com a prost bi pe weie. (OED, c 1175) There came a priest by the way 'There came a priest along the way.'

From early on - the first attestations in the OED date back to 1205, the beginning of the Middle English period - by denotes a path over which motion occurs and at the same time a means of transit; in other words, the 'path' reading and the 'means of transit' reading of the landmark cooccur. While the notions 'path over which motion occurs' and 'means of transit' are present to the same degree in such early instances of this usage as (22), the 'means of transit' reading dominates in later usages such as (23): (22)

Comen ... bi sœ & bi londefeole ...leoden. (OED, c 1205) Came by sea and by land many people 'Many people came by sea and by land.'

(24)

Why not send the parcel by rail?

The semantic cooccurrence of by marking a 'path' landmark and a 'means of transit' landmark is clearly based on the cooccurrence in conceived reality of entities functioning at the same time as paths to a particular destination and as the means by which that destination can be reached. Given this contiguity, the preposition by, which initially only denotes 'a path over which motion occurs,' later blends with the notion 'means of transit' and through time comes to denote almost exclusively 'means of transit. ' As in the case of to, this type of semantic change, which arises out of contiguity (or concomitance), can be classified as metonymical transfer.

7

The fact that one sense/usage of a lexical item fades and passes into another sense/usage does not necessarily mean that the initial sense gets lost. Indeed, in addition to the more recent grammaticalized usages of to, the original allative-dative usages still exist. In contrast, the initial spatial meaning 'along' of the preposition by, from which by's use as an agent marker has developed, has not been preserved.

Metonymy in Prepositions

263

A second metonymy-based semantic change occurs in the development from by marking 'means/instrument' to by marking 'cause.' Consider the following sentences: (24)

Christe draweth soules unto hym by his bloudy sacrifice. {OED, c 1548) 'Christ drew souls onto him by his bloody sacrifice.'

(25)

Sooneafte by this synne he fylle. (OED, 1483) Soon after by this sin he fell 'He fell soon afterwards because of this sin.'

(26)

By the growth of his beard and grey hairs, I did not know him. (OED, 1667)

In (24), for instance, by's landmark denotes the means by which Christ draws souls to him, but at the same time, it constitutes the cause of, or reason for, that event. In (25) and (26), the notion 'cause' is predominant. Again, the change from 'means/instrument' to 'cause' can be described as an instance of contiguity-based metonymical transfer, in that there exists a (conceptual) contiguity relation between an entity's role in an event as a means and as a cause/reason. As a result of this contiguity, the landmark of the preposition by, which initially only denotes 'means,' then blends with the notion 'cause/reason' and through time comes to denote almost exclusively 'cause/reason.'

6. The relationship between metonymy and metaphor

As the previous analyses show, both metaphor and metonymy may motivate synchronic semantic extensions and diachronic semantic change. When trying to distinguish between metaphor- or metonymy-based semantic extensions and change, it is important to keep in mind whether the related concepts belong to two different conceptual domains (ICMs) or one. However, in itself, this is not sufficient as an operational criterion that enables us to draw a clear dividing line between metaphorical and metonymical semantic extensions or change. In the preposition for, for instance, the notions 'purpose' and 'cause' can be seen either as constituting two different domains or as being part of one event-ICM. Similarly, the semantic relations discussed about the other prepositions can be motivated metaphorically and metonymically. It seems to me, though, that when the semantic relation between two concepts can be motivated metaphorically as well as metonymically, the métonymie relation, which is based on contiguity within one domain, is more fundamental/basic than the metaphoric relation, which is based on the mapping between two conceptual domains, precisely because conceptual contiguity often reflects a physical contiguity in extralinguistic reality. Furthermore, it seems plausible that two concepts may much more readily be viewed as being related when they are components of the same domain than when they constitute (or are part of) different domains. This amounts to saying that metaphoric relations between different domains, if not derived from métonymie relations, are at least less fundamental than métonymie relations. That metonymically-based semantic relations have a more fundamental status than metaphorically-based relations is nicely illustrated in Lakoff and

264

Hubert Cuyckens

Johnson's well-known metaphor MORE IS UP (cf. Lakoff & Johnson 1980: 15f. and Lakoff 1993: 240, who do not, however, explicitly mention its métonymie basis). This metaphor has, in fact, a physical basis in that the more things are put on top of a pile, the higher it becomes, or the more fluid is poured into a glass, the higher the level of the fluid in the glass. In other words, there exists experiential contiguity between larger amount (MORE) and higher level (UP) (see also Taylor 1995: 138 and Radden 2000 for the métonymie basis of MORE IS UP, and Kövecses & Radden 1998: 61-62 for more examples of the more fundamental status of métonymie to metaphoric relations). The fact that metonymically based relations between different usages/readings/senses of a lexical item are often categorized as metaphoric has to do with the fact that over time, and in particular when a semantic relation between two concepts occur frequently, each of the two concepts may become increasingly important in its own right (or gain independence) so that each of them starts constituting a domain of its own (and is no longer seen simply as a component in one overarching domain; see also Schwenter & Traugott 1995: 264-265).

7. Conclusion

In this article, I have tried to show through the analysis of semantic relations in prepositions that metonymy is as least as important a factor in motivating semantic extensions or semantic change as metaphor. In the semantic relations discussed, the metonymically-based relations turned out to be more fundamental than the metaphor-based ones. Of course, this does not mean that metaphor-based relations should take a back seat. As Kövecses & Radden (1998: 62) show, metaphors fulfill certain cognitive functions that metonymy does not: mapping the structure of the source domain on the target domain, while maintaining the inferential patten of the source domain. Still, there seems plenty of evidence that some metaphoric relations may only arise through métonymie relations.

References

Cuyckens, Hubert (1999): Historical Evidence in Prepositional Semantics: The Case of English By. In: G. A. J. Tops, B. Devriendt & S. Geukens (eds.): Thinking English grammar: To Honour Xavier Dekeyser, Professor Emeritus, 15-32. Leuven: Peeters. Cuyckens, Hubert & Marjolijn Verspoor (1998): On the Road to To. - In: J. van der Auwera, F. Durieux & L. Lejeune (eds.): English as a Human language: To Honour Louis Goossens, 57-72. München: Lincom Europa. Dirven, René (1993): Dividing Up Physical and Mental Space into Conceptual Categories by means of English Prepositions. - In: C. Zelinsky-Wibbelt (ed.): The Semantics of Prepositions, 73-98. Berlin: Mouton de Gruyter.

Metonymy in

-

Prepositions

265

(1995): The Construal of Cause: The Case of Cause Prepositions. - In: J. R. Taylor & R. E. MacLaury (eds.): Language and the Cognitive Construal of the World, 95-118. Berlin: Mouton de Gruyter. Fries, C.C. (1952): The Structure of English: An Introduction to the Construction of English Sentences. - New York, NY: Harcourt Brace. Geeraerts, Dirk (1997): Diachronic Prototype Semantics. - Oxford: Clarendon Press. Genetti, Carol (1991): From Postposition to Subordinator in Newari. - In: E. C. Traugott & Β. Heine (eds.): Approaches to Grammaticalization, 227-256. Amsterdam: John Benjamins. Gibbs, Raymond W., Jr., & Gerard J. Steen (1999): Metaphor in Cognitive Linguistics. - Amsterdam: John Benjamins. Heine, Bernd, Ulrike Claudi & Friederike Hünnemeyer (1991a): Grammaticalization: A Conceptual Framework. - Chicago, IL: The University of Chicago Press. - (1991b): From Cognition to Grammar: Evidence from African Languages. - In: E. C. Traugott & Β. Heine (eds.): Approaches to Grammaticalization, 149-187. Amsterdam: John Benjamins. Hopper, Paul J. & Elizabeth C. Traugott (1993): Grammaticalization. - Cambridge: Cambridge University Press. Johnson, Mark ( 1987): The Body in the Mind: The Bodily Basis of Meaning. - Chicago, IL: The University of Chicago Press. Jurafsky, Dan (1996): Universal Tendencies in the Semantics of the Diminutive. - In: Language 72, 533-578. Kabata, Kaori & Sally A. Rice (1997): Japanese Ni: The Particulars of a Somewhat Contradictory Article. - In M. Verspoor, Kee Dong Lee & E. Sweetser (eds.): Lexical and Syntactical Constructions and the Construction of Meaning, 107-127. Amsterdam: John Benjamins. Kahr, Joan C. (1975): Adpositions and Locationals: Typology and Diachronic Development. - In: Working Papers on Language Universals 9, 21-54. Kilroe, Patricia (1994): The Grammaticalization of French à." - In: W. Pagliuca (ed.): Perspectives on Grammaticalization, 49-61. Amsterdam: John Benjamins. Kövecses, Zoltan & Günter Radden (1998): Metonymy: Developing a Cognitive Linguistic View. In: Cognitive Linguistics 9, 37-77. Kurylowicz, Jerzy (1975 [1965]): The Evolution of Grammatical Categories. - Reprinted in Esquisses Linguistiques II, 38-54. München: Fink. Lakoff, George (1987): Women, Fire, and Dangerous Things: What Categories Reveal about the Mind. - Chicago, IL: The University of Chicago Press. - (1993):. The Contemporary Theory of Metaphor. - In A. Ortony (ed.): Metaphor and Thought, 202-251. Cambridge: Cambridge University Press. Lakoff, George & Mark Johnson (1980): Metaphors We Live By. - Chicago, IL: The University of Chicago Press. Lakoff, George & Mark Turner (1989): More than Cool Reason: A Field Guide to Metaphor. Chicago, IL: The University of Chicago Press. Langacker, Ronald (1987): Foundations of Cognitive Grammar: Theoretical Prerequisites. - Stanford, CA: Stanford University Press. - (1988): A View of Linguistic Semantics. - In: B. Rudzka-Ostyn (ed.): Topics in Cognitive Linguistics, 49-90. Amsterdam: John Benjamins. Lindstromberg, Seth (1998): English Prepositions Explained. - Amsterdam: John Benjamins. Radden, Günter (1985): Spatial Metaphors Underlying Prepositions of Causality. - In: W. Paprotté & R. Dirven (eds.): The Ubiquity of Metaphor, 177-207. Amsterdam: John Benjamins. Radden, Günter (2000): How Metonymie are Metaphors? - In: A. Barcelona (ed.): Metaphor and Metonymy at the Crossroads: A Cognitive Perspective, 93-108. Berlin: Mouton de Gruyter.

266

Hubert

Cuyckens

Schwenter, Scott A. & Elizabeth C. Traugott (1995): The Semantic and Pragmatic Development of Substitutive Complex Prepositions in English. - In: A. H. Jucker (ed.): Historical Pragmatics: Pragmatic Developments in the History of English, 243-273. Amsterdam: John Benjamins. Svorou, Soteria (1994): The Grammar of Space. - Amsterdam: John Benjamins. Sweetser, Eve (1990): From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. - Cambridge: Cambridge University Press. Taylor, John (1993): Prepositions: Patterns of Polysemization and Strategies of Disambiguation. - In: C. Zelinsky-Wibbelt (ed.): The Semantics of Prepositions, 151-175. Berlin: Mouton de Gruyter. - (19952): Linguistic Categorization: Prototypes in Linguistic Theory. - Oxford: Clarendon Press. Traugott, Elizabeth C. & Ekkehard König (1991): The Semantics-Pragmatics of Grammaticalization Revisited. - In: E. C. Traugott & Β. Heine (eds.): Approaches to Grammaticalization, 189-218. Amsterdam: John Benjamins. Ungerer, Friedrich & Hans-Jörg Schmid (1996): - An Introduction to Cognitive Linguistics. London: Longman.

Addresses of Contributors

Maaike Beliën Vrije Universiteit Amsterdam De Boelelaan 1105 NL-1181 HV Amsterdam The Netherlands Email: [email protected]

Elizabeth Matthis Carl-Petersenstr. 40 D-20535 Hamburg Email: [email protected]

Hubert Cuyckens Department of Linguistics University of Leuven Blijde-Inkomststraat 21 B-3000 Leuven Belgium Email: [email protected]

Birgitta Meex University of Antwerp (UFSIA) Prinsstraat 13 B-2000 Antwerp Belgium Email: [email protected]

Walter de Mulder Université d'Artois UFR Lettres Modernes 9 rue du Temple BP 665 F-62030 Arras Cédex France Email : waiter. demulder@univ-artois. fr, [email protected]

Britta Mondorf Universität - GH Paderborn Fachbereich 3 Anglistik/Sprachwissenschaft Warburger Str. 100 D-33098 Paderborn Germany Email: [email protected]

Claudio Di Meóla Università degli Studi di Roma "La Sapienza" Gruppo di Lingue e Letterature Germaniche Via Carlo Fea 2 1-00161 Roma Italy EMail: claudio.dimeola@uniromal .it

Ignasi Navarro I Ferrando Departament de Filologia Anglesa i Romànica Universität Jaume I 12.071 Castellò de la Plana Spain Email: [email protected]

Dagmar Haumann Universität Erfurt Anglistische Linguistik PF 307 D-99006 Erfurt Germany Email: [email protected]

Günter Radden Universität Hamburg Institut für Anglistik und Amerikanistik Von-Melle-Park 6 D-20146 Hamburg Germany Email: [email protected]

268 Priska-Monika Hottenroth Universität Hamburg Institut für Romanistik Von-Melle-Park 6 D-20146 Hamburg Germany

Addresses of Contributors

Anne Vanderheyden University of Antwerp (UFSIA) Departement Romaanse Talen Rodestraat 14 B-2000 Antwerpen Belgium Email: [email protected]

Email: [email protected] Gisa Rauh Bergische Universität, GH Wuppertal Fachbereich 4 Gaußstrasse 20 D-42097 Wuppertal Germany Email: [email protected] Güner Rohdenburg Universität - GH Paderborn Fachbereich 3 Anglistik/Sprachwissenschaft Warburger Str. 100 D-33098 Paderborn Germany Email: [email protected]

Niina Ning Zhang Zentrum für Allgemeine Sprachwissenschaft Jägerstr. 10-11 D-10117 Berlin Germany Email: [email protected]