Poetic Compounds: The Principles of Poetic Language in Modern English Moetry 9783111352626, 9783484301795

Table of contents :
Preface
0 Introduction
0.1. Goals
0.2. An Outline of the Study
0.3. Data
1 Poetic Language
1.1. The Concept of Poetic Language in Linguistics and Literary Studies
1.2. Poetic Deviation
1.3. The Relationship of Poetic Language to Standard Language
2 Is Poetic Language Functionally Motivated?
3 Compounding in English
3.1. The Place of Word-Formation in the Grammar of English
3.2. Possible Compounds
3.3. The Grammar of Compounding
3.4. The Types of Compounds in English
4 Principles of Poetic Language
4.1. The Generation of Poetic Compounds
4.2. Metaphor, Metonymy, Synecdoche, Ellipsis
4.3. Poetic Compression
4.4. Foregrounding
4.5. Optional Transitivity
4.6. Ambiguity
4.7. Vagueness
4.8. Rhyme
4.9. Iconicity
5 Concluding Remarks
Appendix: Poetic Compounds in Their Context
Bibliography

Citation preview

Linguistische Arbeiten

179

Herausgegeben von Hans Altmann, Herbert E. Brekle, Hans Jürgen Heringer, Christian Rohrer, Heinz Vater und Otmar Werner

Jean Boase-Beier

Poetic Compounds The Principles of Poetic Language in Modern English Poetry

Max Niemeyer Verlag Tübingen 1987

CIP-Kurztitelaufnahme der Deutschen Bibliothek Boase-Beier, Jean : Poetic compounds : the principles of poet, language in modern English poetry / Jean Boase-Beier. - Tübingen : Niemeyer, 1987. (Linguistische Arbeiten ; 179) Einheitssacht.: Poetische Komposita (engl.) NE:GT ISBN 3-484-30179-1 ISSN 0344-6727 © Max Niemeyer Verlag Tübingen 1987 Alle Rechte vorbehalten. Ohne Genehmigung des Verlages ist es nicht gestattet, dieses Buch oder Teile daraus photomechanisch zu vervielfältigen. Printed in Germany. Druck: Weihert-Druck GmbH, Darmstadt.

CONTENTS

Preface 0

Introduction 0.1. Goals 0 . 2 . An O u t l i n e of the Study 0 . 3 . Data

1

vii

1 3 4

Poetic Language 1.1.

The Concept of Poetic Language in L i n g u i s t i c s and Literary Studies 1 . 2 . Poetic D e v i a t i o n 1.3. The R e l a t i o n s h i p of Poetic Language to Standard Language

2

Is Poetic Language Functionally Motivated?

3

Compounding in English 3.1. The Place of Word-Formation in the Grammar of English 3 . 2 . Possible Compounds 3.3. The Grammar of Compounding 3 . 3 . 1 . Introductory Remarks 3 . 3 . 2 . Percolation 3 . 3 . 3 . The R e a l i s a t i o n of Argument-Structures 3 . 3 . 4 . Stereotypes 3 . 3 . 5 . Co-ordination 3 . 3 . 6 . Location 3 . 3 . 7 . Some Semantic Restrictions 3 . 3 . 8 . The Context-Dependence of Compounds 3 . 3 . 9 . Pragmatic Principles 3 . 4 . The Types of Compounds in English 3 . 4 . 1 . NN Compounds 3 . 4 . 2 . VN Compounds 3 . 4 . 3 . AN Compounds 3 . 4 . 4 . PN Compounds 3 . 4 . 5 . NV Compounds 3 . 4 . 6 . VV Compounds

7 12 22 31

37 39 40 40 43 45 48 50 54 5S 58 63 66 66 £> 8 71 72 74 75

3.4.7. 3.4.8. 3.4.9. 3.4.10. 3.4.11. 3.4.12. 3.4.13. 3.4.14. 3.4.15. 3.4.16. 4

76 77 78 81 81 82 82 83 83 83

Principles of Poetic Language 4.1. 4.2. 4.3. 4.4. 4.5. 4.6. 4.7. 4.8. 4.9. 4.10. 4.11. 4.12.

5

AV Compounds PV Compounds NA Compounds VA Compounds AA Compounds PA Compounds NP Compounds VP Compounds AP Compounds PP Compounds

The Generation of P o e t i c Compounds Metaphor, M e t o n y m y , Synecdoche, E l l i p s i s Poeti.c Compression Foregrounding Optional T r a n s i t i v i t y Ambiguity Vagueness Rhyme Iconicity Poetic Context Contextual Cohesion Some Consequences of the Theory

Concluding Reearka

85 86 114 121 126 128 132 135 138 144 147 150 155

Appendix: Poetic Compounds in Their Context

157

Bibliography

183

PREFACE

This study is the E n g l i s h version of my German dissertation Poetische Jfomposita: Eine Untersuchung zu den Prinzipien der poetischen Sprache am BeispieJ der modernen engJischen Dichtung, (1986. Nürnberg: Regensburger M i c r o f i c h e M a t e r i a l i e n 015). Apart f r o m one or two passages it is a f a i r l y close t r a n s l a t i o n of the German v e r s i o n . It would have been an impossible task to record, at the appropriate p o i n t s in the t e x t , everyone who has i n f l u e n c e d my t h i n k i n g ; I thus w i s h to thank the f o l l o w i n g f r i e n d s and colleagues h e r e , as w e l l as all others who have in any way played a part in the w r i t i n g of t h i s study, w h e t h e r by o f f e r i n g a d v i c e and c r i t i c i s m , t a k i n g part in discussions, or g i v i n g i n f o r m a t i o n and assistance of various k i n d s : B r i g i t t e Asbach-Schnitker, E i t e l Fischer, Alan Forman, Tony Kemp, Craig Mabrey, Stephan N i e d e r w i e s e r , Peter Staudacher and e s p e c i a l l y Dieter Beier and J i n d f i c h Toman. In p a r t i c u l a r 1 should l i k e to thank P r o f . Karl H e i n z G o l l e r for his many u s e f u l h i n t s and suggestions and P r o f . Herbert B r e k l e for his constant support and guidance d u r i n g the w r i t i n g of t h i s study. It goes w i t h o u t saying that no-one except m y s e l f bears any r e s p o n s i b i l i t y for its d e f i c i e n c i e s . I should also l i k e to· thank . K e v i n Crossley-Holland permission t o p r i n t "First W a k i n g : Lumb B a n k " .

vi

for

INTRODUCTION

0.1.

Goals

The a t t e m p t to g i v e a coherent account of l i t e r a r y s t y l e , or of a p a r t i c u l a r area of l i t e r a r y s t y l e , consists in large part in an attempt to e x p l a i n our i n t u i t i o n s about the n a t u r e of p o e t r y , e s p e c i a l l y about the language of p o e t r y . It is an u n d e r t a k i n g w h i c h h a s a t t r a c t e d n o t o n l y researchers i n t h e f i e l d o f l i t e r a r y s t u d i e s but l i n g u i s t s as r e n o w n e d as Roman J a k o b s o n , M a n f r e d B i e r w i s c h , P a u l K i p a r s k y , Samuel J a y Keyser a n d M o r r i s H a l l e , t o m e n t i o n but a f e w . In recent t i m e s there has been an abundance of s t r u c t u r a l i s t and s t r u c t u r a l i s t - i n f l u e n c e d accounts of p o e t i c s t y l e as w e l l as a g r o w i n g n u m b e r of studies carried out w i t h i n the framework of generative grammar. Varied as are the t h e o r e t i c a l o r i e n t a t i o n s and the a i m s of these researchers, t h e y h a v e all been m o t i v a t e d by a sense of the i n s u f f i c i e n c y of p u r e l y d e s c r i p t i v e t a x o n o m i c a l s t u d i e s of the language of p o e t r y , by a sense of the need for an a r t i c u l a t e d theory w h i c h w o u l d account s c i e n t i f i c a l l y for the p h e n o m e n o n of the language of l i t e r a t u r e . One of t h e i r basic assumptions has been that there are c e r t a i n phenomena intuitively felt to be " p o e t i c " and that these phenomena cannot be s a t i s f a c t o r i l y accounted for w i t h o u t recourse to the grammar of standard l a n g u a g e , but t h a t , on the o t h e r h a n d , t h e grammar a l o n e w i l l n o t s u f f i c e t o e x p l a i n t h e m . T h e theory presented here is also based upon t h i s a s s u m p t i o n . It is an attempt to provide an account of one p a r t i c u l a r area of language in one p a r t i c u l a r area of l i t e r a t u r e , n a m e l y of c o m p o u n d i n g in contemporary E n g l i s h p o e t r y , in a s y s t e m a t i c way on the basis of a t h e o r e t i c a l approach to w o r d - f o r m a t i o n and g r a m m a r in the generative framework. W o r d - s t u d i e s , both of i n d i v i d u a l authors and w o r k s , and of p o e t i c s t y l e in general have been c o n s i d e r e d by many modern s c h o l a r s of p o e t i c language to f o r m an essential part of l i t e r a r y c r i t i c i s m ; t h e studies b y B a r f i e l d ( 1 9 2 8 ) , Groom (1937, 1 9 S 5 ) , Y u l e ( 1 9 4 4 ) , M i l e s ( 1 9 4 6 , 1 9 6 0 ) , V o i t l ( 1 9 6 9 ) a r e cases in p o i n t . The i m p o r t a n c e of e x a m i n i n g the words of p o e t i c a l works lies p a r t l y in the f a c t that the i n n o v a t o r y character of words is, at least intuitively speaking, immediately obvious. Thus a

1

Most of the modern t h e o r e t i c a l approaches are represented in a n t h o l o g i e s such as F o w l e r ( 1 9 6 6 ) , Chatman & L e v i n ( 1 9 6 7 ) , Freeman ( 1 9 7 0 , 1 9 8 1 a ) , Chatman ( 1 9 7 1 ) a n d C h i n g e t a l . ( 1 9 8 0 ) .

researcher concerned w i t h poetic language as c h a r a c t e r i z e d by d e v i a t i o n f r o m t h e norm w i l l f i n d ample a n d unambiguous example m a t e r i a l in the new words a poet uses. The present study is, h o w e v e r , not concerned w i t h the use of that type of word u s u a l l y marked "poetic" in the d i c t i o n a r y . D e v i a t i o n in this respect is m e r e l y a question of l e x i c a l choice w i t h its attendant considerations of c o n n o t a t i o n , and so on. W o r d - f o r m a t i o n , however, is an area of the language characterised in part by s i m i l a r phenomena to those f o u n d in the s y n t a x . It thus provides a r i c h area of study for poetic d e v i a t i o n as t h i s can exist at several l e v e l s , because the products of wordformation are the output of principles interacting at several levels. Our study has been l i m i t e d to the area of compounding because, in a discussion of a c c e p t a b i l i t y and u n a c c e p t a b i l i t y , of d e v i a t i o n and n o r m , it makes sense to l i m i t the area of study to one w h i c h can be p r e c i s e l y d e f i n e d and accounted for in g r a m m a t i c a l terms. In t h i s way it is possible to d e f i n e in e x a c t l y w h i c h way a d e v i a n t f o r m transgresses the r u l e s of standard l a n g u a g e , or w o u l d do if it occurred there, and to provide an e q u a l l y p r i n c i p l e d e x p l a n a t i o n to account for this transgression. Modern poetry has been chosen as the area of study because one c l e a r l y has the strongest intuitions about a c c e p t a b i l i t y in contemporary language. As a l i n g u i s t i c s t y l i s t i c e x a m i n a t i o n of c o m p o u n d i n g in contemporary E n g l i s h p o e t r y , t h i s study w i l l try to say why certain compound types, which the native speaker w o u l d i n t u i t i v e l y regard as u n a c c e p t a b l e , nevertheless occur f r e q u e n t l y in poems. Our f i r s t a i m , t h e n , is to show that poetic d e v i a t i o n in compounds does not occur h a p h a z a r d l y , but is governed by a set of p r i n c i p l e s . Our second aim is to show that all of the p r i n c i p l e s we need to describe poetic compounds are a p p l i c a b l e to poetic structures in g e n e r a l . There is much e v i d e n c e for t h i s , as we hope to show. We are assuming that the language of any segment of l i t e r a t u r e w i l l have to be accounted for by r u l e s of v a r i o u s types. There w i l l be p r i n c i p l e s s p e c i f i c to a c e r t a i n l a n g u a g e , age, poetic style or even a s p e c i f i c poet. But these w i l l interact w i t h more general poetic p r i n c i p l e s . It is the latter type of p r i n c i p l e s , or r a t h e r , p r i n c i p l e s w h i c h seem l i k e l y to have such status, w h i c h w i l l be our m a i n concern here. Our t h i r d a i m , t h e n , is an e x t e n s i o n of our f i r s t and second; n a m e l y , to say something of a general nature about the r e l a t i o n of poetic language to standard language. Thus the area of compounding in contemporary E n g l i s h poetry is to be seen as an i l l u s t r a t i o n of a general theory of poetic language w h i c h is assumed to have wider i m p l i c a t i o n s , not only for the w h o l e range of l i n g u i s t i c phenomena in contemporary E n g l i s h poetry but in f a c t , in certain aspects at least, for the w h o l e range of l i n g u i s t i c phenomena in

For a discussion of t h i s question in connection eighteenth century poetic d i c t i o n , see G o l l e r ( 1 9 6 4 ) .

with

poetic language in general. It remains to say a word about the data used as examples f o r the theory of poetic language put forward in this study. Poets have been selected on a theory-determined basis. That is, we have chosen to examine those poets whose works show a strong tendency to use i n n o v a t i o n in general and in p a r t i c u l a r in the area of w o r d - f o r m a t i o n . There are many modern poets whose works do not m a n i f e s t t h i s tendency very strongly ( t h o u g h rather less, perhaps, than those whose works d o ) . B u t , if the poetic p r i n c i p l e s we describe here are general in nature, we should expect to f i n d that they are e f f e c t i v e in other areas of the language in the work of those poets who do not use deviant wordf o r m a t i o n s , as they are, of course, in the work of those who do.

0 . 2 . An Outline of the Study In Chapter 1 we deal w i t h the r e l a t i o n of poetic language to standard language, whereby we attempt to say what poetic language is and why it is j u s t i f i a b l e to regard it as d i f f e r e n t f r o m standard language. In connection w i t h this q u e s t i o n we b r i e f l y discuss (in 1.1) the r o l e of poetic language studies in literary c r i t i c i s m and in l i n g u i s t i c s . In 1 . 2 , the question of poetic style as d e v i a t i o n is discussed, and we d e f i n e what we mean by the term " d e v i a t i o n " . In 1.3, we attempt to show how the s p e a k e r ' s i n t u i t i o n about the nature of poetic language can best be represented in r e l a t i o n to a model of the grammar. P o e t i c language is seen as the result of an interaction of certain poetic p r i n c i p l e s w i t h the grammar of standard language. In Chapter 2 we address the question of teleological e x p l a n a t i o n s of p o e t i c language; it is argued that the poetic p r i n c i p l e s whose i n t e r a c t i o n w i t h the grammar accounts for poetic i n n o v a t i o n are not f u n c t i o n a l l y m o t i v a t e d . Chapter 3 is a f a i r l y brief presentation of the model of wordf o r m a t i o n upon w h i c h the present study is based. It shares a number of f e a t u r e s w i t h contemporary studies on word-syntax such as Lieber (1980, 1 9 8 3 ) , S e l k i r k ( 1 9 8 2 ) , Toman ( 1 9 8 3 ) , Boase-Beier and Toman (in p r i n t ) . The position of w o r d - f o r m a t i o n w i t h i n the grammar ( 3 . 1 ) and the concept of the possible compound ( 3 . 2 ) are discussed in a d d i t i o n to various principles of the grammar ( 3 . 3 ) . In 3 . 4 , the sixteen compound-types arising f r o m the c o m b i n a t i o n of the f o u r m a j o r categories ( Ν , ν , Α , Ρ ) are examined in t u r n . By d e t e r m i n i n g how certain types of compounds are formed and why But note that in i n d i v i d u a l cases there may be d i f f e r e n c e s f r o m language to language in c o m p o u n d - f o r m s generated by the grammar. Thus, for e x a m p l e , a f o r m only possible in poetry in E n g l i s h m i g h t be a standard language f o r m in some other language.

other types do not o c c u r , it is hoped to p r o v i d e a basis for the discussion of p o e t i c compounds in Chapter 4. A thorough reading of Chapter 3 is thus a prerequisite to a f u l l understanding of Chapter 4. Chapter 4 c o n t a i n s the central thesis of the study. V a r i o u s types of compound are presented w h i c h , though u n a c c e p t a b l e , in standard l a n g u a g e , appear to be p r o d u c t i v e in the poems under discussion. On the basis of the model put f o r w a r d in Chapter 3, the exact n a t u r e of t h e i r d e v i a t i o n is d e t e r m i n e d . Those p r i n c i p l e s w h i c h seem to be at work both here and in other areas of poetic language are discussed in turn and it is shown that their interaction with the grammar of standard language accounts for the p o e t i c f o r m s in q u e s t i o n . The exact n a t u r e of t h i s i n t e r a c t i o n is e x a m i n e d and it is s h o w n , f u r t h e r m o r e , that the p r i n c i p l e s w h i c h are responsible for poetic compounds are general in n a t u r e , i . e . , they a r e n o t s p e c i f i c t o c o m p o u n d - f o r m a t i o n n o r , e v e n , to w o r d - f o r m a t i o n . Chapter 5 c o n t a i n s some r e m a r k s s u m m a r i s i n g the c o n c l u s i o n s of the study and m e n t i o n i n g several p o i n t s w h i c h could f o r m the s u b j e c t of f u r t h e r research.

0.3.

Data

As i l l u s t r a t i o n for t h i s study of poetic compounds the works of several modern E n g l i s h poets have been used. The selection of poets and works is t h e o r y - o r i e n t a t e d ; poets were chosen whose work shows a strong tendency to i n n o v a t i o n , e s p e c i a l l y in the area of w o r d - f o r m a t i o n . A selection of the compounds discussed, in the c o n t e x t s in w h i c h they occur in the poems, is appended to t h i s w o r k . An e f f o r t has been made to i n c l u d e p a r t i c u l a r l y those cases in w h i c h the context of a compound is of special r e l e v a n c e to its i n t e r p r e t a t i o n , and to g i v e the m i n i m u m context r e q u i r e d for their u n d e r s t a n d i n g . The A p p e n d i x thus f o r m s an integral part of the study. It should be stressed, h o w e v e r , that the A p p e n d i x contains o n l y m i n i m u m contexts; a f u l l u n d e r s t a n d i n g of a poetic compound c l e a r l y a l w a y s r e q u i r e s , among other t h i n g s , a reading of the w h o l e poem in w h i c h it occurs. All e x a m p l e s of compounds w h i c h appear in context in the A p p e n d i x are marked w i t h " + " when they appear as numbered e x a m p l e s in the t e x t .

4

We have also i n c l u d e d one or two poets w h o , t h o u g h o r i g i n a l l y f r o m another c o u n t r y , l i v e or have l i v e d in E n g l a n d and are in general considered to be " E n g l i s h poets".

E x a m p l e s are g i v e n in u n i n f l e c t e d f o r m in the t e x t ; the l e x i c a l categories of compound-elements are given in square brackets a f t e r each c o m p o u n d . For all examples taken f r o m the poems under c o n s i d e r a t i o n , the source is also given a f t e r the compound, in round brackets. Thus examples for w h i c h no source is given are i d e n t i f i a b l e as being of our own construction. The d e t a i l s of the source of examples serve two purposes: besides e n a b l i n g the e x a m p l e to be f o u n d in the o r i g i n a l work they serve to i d e n t i f y the e x a m p l e s in the A p p e n d i x . For t h i s reason, the texts in the A p p e n d i x are arranged a c c o r d i n g to their sources, numerically and alphabetically. The f o l l o w i n g list gives the abbrevations used throughout the work to i n d i c a t e sources. F u l l d e t a i l s of the poetic works are to be f o u n d in the B i b l i o g r a p h y . A AB ΑΤΗ BP

Peter Redgrove Peter Redgrove K e i t h Sagar (ed. ) Edward L u c i e - S m i t h (ed. ) BS D a v i d W e v i l l CL David Holbrook CW Peter Redgrove DH K e v i n CrossleyHolland DV A l a n Ross FM Anne Stevenson Fo Robert Conquest Fu John H o l l o w a y FW Seamus Heaney GH Carol Rumens HL Douglas Dunn HR Ted Hughes I D a v i d Holbrook I F David W e v i l l JA Charles Causley KP Douglas Dunn L Ted Hughes LH Norman N i c h o l s o n M John H o l l o w a y MB Ted Hughes MH G e o f f r e y H i l l MT Ted Hughes NM Terence T i l l e r NP A . A l v a r e z ( e d . ) NP8 John F u l l e r ( e d . ) PW A l a s t a i r R e i d R Ted Hughes RE Ted Hughes RG K e v i n CrossleyHolland S Peter Dale SG Peter Scupham SM Terence T i l l e r

1977 1981 1983 1970

From every C h i n k of the Ark The Apple-Broadcast The A c h i e v e m e n t of Ted Hughes B r i t i s h Poetry since 194S

1964 1978 1961 1976

Birth of a Shark Chance of a L i f e t i m e The Nature of Cold Weather The Dream-House

1980 1985 1979 1960 1979 1982 1972 1957 1960 1966 1961 1981 1960 1972 1956 1978 1971 1979 1968 1966 1982 1959 1983 1979b 1972

Death V a l l e y The F i c t i o n Makers Forays The Fugue F i e l d Work Scenes f r o m the Gingerbread House The Happier L i f e The Hawk in the R a i n Imaginings A Christ of the Ice-Floes Johnny A l l e l u i a St.Kilda's Parliament Lupercal A Local H a b i t a t i o n The M i n u t e Moon-Bells M e r c i a n Hymns Moortown Notes for a Myth The New Poetry New Poetry 8 Passwords River R e m a i n s of Elmet The R a i n - G i v e r

1968 1972 1979

Storms The Snowing Globe That Singing Mesh

SP SS StW SW TO

Peter Scupham Ted Hughes Norman Nicholson Carol Rumens Kevin CrossleyHolland TSP Charles Tomlinson UM Carol Rumens UW Charles Causley W Ted Hughes WM Peter Redgrove WNP Peter Redgrove WP Peter Redgrove WQ Peter Scupham WW John Holloway

1980 1975 1981 1983 1983

Summer Palaces Season Songs Sea to the West Star Whisper T i m e ' s Oriel

1978 1981 1968 1967 1963 1979 1968 1983 1965

Selected Poems 1951-1974 Unplayed Music Underneath the Water Wodwo At the White Monument The Weddings at Nether Powers Work in Progress Winter Quarters Wood and W i n d f a l l

1.1.

The Concept of Poetic Language in Linguistics and Literary Studies

In the opinion of many scholars concerned w i t h the theory of l i t e r a t u r e , studies of l i t e r a r y language in Europe represent a f a i r l y recent development in literary studies, a development w h i c h began a f t e r the First World W a r . C e r t a i n l y there were, as such scholars have almost i n v a r i a b l y pointed out, various e a r l i e r references to the language of l i t e r a t u r e , but it was only w i t h the advent of two important new c r i t i c a l methods that l i t e r a r y language gained the importance it has today for the study of l i t e r a t u r e . 2 ^ The f i r s t of the two trends in question is the method of close reading of the t e x t , the study of "the words on the page" as practised by c r i t i c s such as I . A . R i c h a r d s , F . R . L e a v i s and W . E m p s o n in the 1920s and 1930s, methods in part s i m i l a r to those used a l i t t l e later by the American proponents of New C r i t i c i s m . The second is the Continental stylistics of scholars l i k e Leo S p i t z e r , who emphasised the importance of the language of literature f r o m a p h i l o l o g i c a l point of v i e w . This development in the importance ascribed to the language of l i t e r a t u r e p a r a l l e l e d to a certain extent the development in l i n g u i s t i c s towards s t r u c t u r a l i s m and away f r o m a h i s t o r i c a l

1 2 3

4

See, for e x a m p l e . Page ( 1 9 8 4 ) . In 1957 the c r i t i c R . A . Sayce remarked that "the l i n g u i s t i c study of works of l i t e r a t u r e has become the central procedure of contemporary c r i t i c i s m " (Sayce (1957: 1 1 9 ) ) . Both English and A m e r i c a n critics of the period between about 1920 and 1960 who were convinced of the importance of the language of l i t e r a t u r e have been c a l l e d " N e w C r i t i c s " , though t h e i r approaches to poetry and c r i t i c i s m have in fact o f t e n varied quite s u b s t a n t i a l l y . For discussion of the methods of New C r i t i c i s m see, f o r e x a m p l e , Krieger ( 1 9 5 6 ) , Foster ( 1 9 6 2 ) , Lee ( 1 9 6 6 ) . For examples of the work of the B r i t i s h c r i t i c s m e n t i o n e d here see Richards ( 1 9 2 4 [ 1 9 6 7 ] ) , Empson (1930 [1965]) and Leavis ( 1 9 6 8 ) . For a discussion of the essential d i f f e r e n c e s in the C o n t i n e n t a l and A n g l o - A m e r i c a n methods see Lodge ( 1 9 8 4 : 525 6 ) . For an e x a m p l e of S p i t z e r ' s work see Spitzer ( 1 9 4 8 [1962]).

approach. The l i t e r a r y studies of R i c h a r d s , Leavis and others were in a s i m i l a r a n t i - h i s t o r i c a l t r e n d . W h i l e these latter c r i t i c s were not l i n g u i s t s , they had a deep understanding of language and an acute awareness of its importance for the study of l i t e r a t u r e . And the structural l i n g u i s t i c s of Ferdinand de Saussure (1857-1916) and his f o l l o w e r s in France had an enormous i n f l u e n c e , w h i c h of course is s t i l l in force t o d a y , upon both l i n g u i s t i c s and l i t e r a r y studies. The linguists of the Prague C i r c l e , founded in 1926 and strongly i n f l u e n c e d by Russian Formalism ( e s p e c i a l l y through Roman J a k o b s o n ) , followed a structuralist approach in both areas. In E n g l a n d , the i n f l u e n c e of s t r u c t u r a l i s m upon l i t e r a r y theory was on the w h o l e not f e l t u n t i l a f t e r the Second W o r l d W a r , where it was exerted by the work of Roman Jakobson, by t r a n s l a t i o n s f r o m the works of other Prague L i n g u i s t s - e s p e c i a l l y noteworthy is G a r v i n ' s c o l l e c t i o n of 1964 - , by translations of the work of the Russian Formalists (such as those in Lemon & Reis ( 1 9 6 5 ) ) , and also by the French scholar Roland Barthes, whose main works on l i t e r a t u r e began to be published in the 1950s and 1960s. Important post-structuralist critics such as Jacques Derrida in France, J . H i l l i s M i l l e r in America and D . L o d g e and J . C u l l e r in E n g l a n d , have continued to apply the methods of s t r u c t u r a l i s m ( o r , in a broader sense, of s e m i o l o g y , the theory of signs as proposed by Saussure and others) to the study of l i t e r a t u r e in general and of l i t e r a r y language in p a r t i c u l a r . But at about the same t i m e as s t r u c t u r a l i s t models began to gain importance for the study of l i t e r a t u r e , the new development in generative l i n g u i s t i c theory w h i c h began w i t h Noam C h o m s k y ' s Syntactic Structures in 1957 had a f u r t h e r and indeed a very strong i n f l u e n c e upon the study of literary language. The theory of generative grammar permitted a development of literary language studies or " l i n g u i s t i c s t y l i s t i c s " , as this area of study is o f t e n c a l l e d , in the d i r e c t i o n of models of l i t e r a r y competence. E x p l a n a t i o n s could be o f f e r e d for i n t u i t i o n s about l i t e r a r y language, and the question of what is possible in l i t e r a r y language as opposed to what is there in a p a r t i c u l a r

For a discussion of the p a r a l l e l development of l i n g u i s t i c s and l i t e r a r y studies see Watson ( 1 9 6 9 : 141-155). For e x a m p l e s of the work of s t r u c t u r a l i s t and structuralisti n f l u e n c e d scholars r e f e r r e d to in t h i s paragraph see Saussure ( 1 9 1 6 ) , Jakobson ( 1 9 6 0 , 1 9 7 1 ) , Barthes ( 1 9 6 4 ) , D e r r i d a ( 1 9 6 7 ) , M i l l e r ( 1 9 7 0 ) , C u l l e r ( 1 9 7 5 ) , Lodge ( 1 9 8 4 ) . S e e also Eagleton (1983). See, f o r e x a m p l e . Freeman ( 1 9 7 0 : 4 ) . See L e v i n ( 1 9 6 4 ) , B i e r w i s c h ( 1 9 6 5 ) , Abraham & B r a u n m ü l l e r ( 1 9 7 3 ) . N o t e , h o w e v e r , that the use of the term " l i t e r a r y competence" varies. For B i e r w i s c h , if we understand him c o r r e c t l y , it r e f e r s to the possession of the grammar of standard language plus a sort of " r e c o g n i t i o n grammar" for poetic structures. For L e v i n , on the other h a n d , it is a direct correlate of l i n g u i s t i c competence. See Coppay ( 1 9 7 7 ) for a c r i t i c i s m of L e v i n ' s v i e w . We do not regard as v a l i d

passage by a p a r t i c u l a r author took on central i m p o r t a n c e , though this latter type of u n d e r t a k i n g has not lost any of its value for the c r i t i c and for the s t y l i s t i c i a n concerned w i t h poetic performance. An important group of theoretical issues has arisen w i t h the increase in studies of l i t e r a r y language, and p a r t i c u l a r l y in connection w i t h those w h i c h make use of the theories of modern generative grammar. These can be expressed as f o l l o w s : What is the importance for literature of its language?, what is the importance of l i n g u i s t i c s for the study of literature and e s p e c i a l l y for the study of l i t e r a r y language?, what is the d i f f e r e n c e between the c r i t i c ' s task and the l i n g u i s t ' s task in studying literary language? The answers to these questions, w h i c h are obviously closely related to one another, vary enormously. At one extreme are statements such as that made by H . W h i t e h a l l , that "no c r i t i c i s m can go beyond its l i n g u i s t i c s " ( W h i t e h a l l (1951: 7 1 3 ) ) . At the other is the position of certain literary scholars who regard l i n g u i s t i c s as u n i m p o r t a n t or e v e n , i t seems, h a r m f u l for the study of literature. This view seems to have arisen among l i t e r a r y c r i t i c s f r o m a f e a r of h a v i n g t h e i r area of study encroached upon by "newcomers", i . e . l i n g u i s t s , and a rather understandable impatience w i t h the arrogance of the linguist who is, a f t e r a l l , " s i m p l y under-educated in the reading of poetry" ( V e n d l e r , 1966: 4 6 0 ) . The position occupied by most sylisticians lies somewhere between these extremes, i.e. they believe that l i n g u i s t i c s has something to o f f e r the study of l i t e r a t u r e indeed, it may be an essential part of the study of literature, 1 0 but it cannot say all that there is to be said about l i t e r a t u r e , nor even, as pure l i n g u i s t i c s , about literary language. The p o s i t i o n we shall be t a k i n g here has much in common w i t h this view though it deviates f r o m it in that we shall m a i n t a i n that l i n g u i s t i c s is not an essential part of the study of l i t e r a t u r e , but o n l y an essential part of the study of stylistics; literary studies may or may not include stylistics - it depends upon its aims. That is, we take the v i e w that the c r i t i c and the l i n g u i s t i c s t y l i s t i c i a n are concerned w i t h d i f f e r e n t areas of study. This is a suggestion w h i c h r e c a l l s , perhaps, a statement made by I . A . Richards to the e f f e c t that a f u l l critical statement about a literary work involves a part w h i c h is critical in the strict sense, i.e. i n c l u d i n g a value j u d g e m e n t , and a t e c h n i c a l part, d e a l i n g w i t h "the ways and means by w h i c h 11 experiences arise". H o w e v e r , our v i e w goes beyond this: the

Coppay's argument that if poetic competence were p a r a l l e l to l i n g u i s t i c competence, then everyone who could read a poem w o u l d be a poet. In f a c t , everyone who has poetic competence is a poet. The reader, or c r i t i c , may construct a model of poetic competence w h i c h enables him to read a poem. There is no c o n f l i c t here, and thus no reason to r e j e c t L e v i n ' s v i e w . 9 See also Freeman (1970: 3). 10 Compare F r e e m a n ' s assertion that "a good c r i t i c is perforce a good l i n g u i s t " (Freeman (1970: 3 ) ) . 11 I . A . R i c h a r d s ( 1 9 2 4 [1967: 1 5 ] ) .

t e c h n i c a l part of a c r i t i c ' s statement is not the same as the statement a l i n g u i s t i c s t y l i s t i c i a n w o u l d m a k e . The area of study u s u a l l y called s t y l i s t i c s or l i n g u i s t i c s t y l i s t i c s is a c l e a r l y d e f i n e d domain w h i c h , f a r f r o m s i m p l y representing common ground of l i n g u i s t i c s and literary studies, represents an area w h i c h is not the m a i n concern of e i t h e r . In this we f o l l o w W i l l i a m Hendricks ( 1 9 7 4 ) , w h o c a l l s this f i e l d " s t y l o l i n g u i s t i c s " ( P - 9 1 to e m p h a z i a e its nature as a "hyphen d i s c i p l i n e " ( p . 8 ) . S t y l o l i n g u i s t i c s , then, or stylistics, as we shall p r e f e r to c a l l it, is not s i m p l y .an overlap of l i n g u i s t i c s and l i t e r a r y studies, nor is it an a p p l i c a t i o n of l i n g u i s t i c s to l i t e r a t u r e ; it is a study of the area of their interrelationships. ^ It is, as a pure d i s c i p l i n e , not under any o b l i g a t i o n to say anything u s e f u l for l i n g u i s t i c s or l i t e r a r y c r i t i c i s m , though, of course, it w i l l a u t o m a t i c a l l y do so. The linguist need not be concerned w i t h the language of literature a n d , if he is, he is not and should not be concerned with all its aspects. The literary critic must of course be concerned w i t h the language of l i t e r a t u r e , but not necessarily in a l i n g u i s t i c sense - he does not have to be a linguist if he is concerned w i t h the language in its purely l i t e r a r y aspect. In other words, "poetic language" should not be i d e n t i f i e d w i t h poetry; a complete study of poetry can and must go beyond a study of its language. It is the s t y l i s t i c i a n , then, who w i l l be concerned w i t h l i t e r a r y language as an area for the interaction of literary studies and l i n g u i s t i c s . This is an area of study w h i c h , l i k e any other d i s c i p l i n e , w i l l have its e m p i r i c a l studies and its theoretical considerations. The former w i l l largely consist in studies of the language of i n d i v i d u a l authors, or of "poetic language" per se in a descriptive sense. The latter, by v i r t u e of d e a l i n g w i t h theories, w i l l tend not to be restricted to a s p e c i f i c author, but to make general statements. "The study of l i t e r a t u r e , as opposed to the perusal and discussion of i n d i v i d u a l works, would become an attempt to understand the conventions w h i c h make literature possible" ( C u l l e r 1975: v i i i ) - t h i s , a statement of the expectations one might j u s t i f i a b l y have of s t r u c t u r a l i s t p o e t i c a l t h e o r y , is also our expectation of the type of theory of poetic language put forward here. The present w o r k , t h e n , is to be seen as a contribution to the theory of l i t e r a r y language though the theory is, we hope, s u f f i c i e n t l y supported by e m p i r i c a l e v i d e n c e . It should by now have become clear that we are by no means presuming here to take on the c r i t i c ' s task and o f f e r f u l l i n t e r p r e t a t i o n s of poems or

12 Compare also the comment by O . T h o m a s that "the study of style is [...] a hyphenated d i s c i p l i n e , [...] stylistics is dependent upon l i n g u i s t i c theory" (Thomas (1976: 2 0 3 ) ) . 13 The term "stylistics" is f r e q u e n t l y used to r e f e r to the study of all types of style, as opposed to only l i t e r a r y style. We are using it here in the l a t t e r , more l i m i t e d sense. 14 "A poem is an object f a s h i o n e d out of the language, r h y t h m s , b e l i e f s , and obsessions of a poet and a society" ( O c t a v i o Paz (1974: v ) ) . 10

even of aspects of poems but t h a t we are concerned w i t h the nature of p o e t i c language and its r e l a t i o n to the grammar of standard l a n g u a g e .

1 . 2 . Poetic Deviation H a v i n g , it is to be hoped, established f a i r l y c l e a r l y what we are t a k i n g to be the r e l a t i o n s of both l i n g u i s t i c s and l i t e r a r y studies to the question of l i t e r a r y l a n g u a g e , we now turn to the question as to w h a t , in f a c t , l i t e r a r y language is. The answer is by no means s e l f - e v i d e n t or, at least, is not a l w a y s considered to be so. Though most studies of the language poets and novelists use have worked on the assumption - t a c i t l y made or e x p l i c i t l y stated - that there is such a phenomenon as the language of literature, t h i s is not a l w a y s the case. The v i e w s of researchers vary f r o m a categorical d e n i a l of any essential d i f f e r e n c e between standard and literary language to a view of literary language as an e n t i r e l y separate language. 1 6 Both these views have long traditions in the study of l i t e r a t u r e . To take just two w e l l known examples from earlier pronouncements on English l i t e r a t u r e , one m i g h t quote Wordsworth, who said that the language of poetry is "a selection of the language r e a l l y spoken by m e n " , and, for the opposing v i e w . G r a y , who said "the language of the age is never the language of poetry; except among the French". O b v i o u s l y these views r e f l e c t in part the b e l i e f s , the taste and the l i t e r a r y theory of t h e i r respective ages. The f o r m e r , in so far as it represented avoidance of intentional poetic devices might be c a l l e d the t y p i c a l v i e w of the romantic poets, w h i c h was in part a reaction to what were considered the excesses represented by the poetry of their predecessors, (of whom Gray is an e x a m p l e ) . The origins of these opposing views are o f t e n considered to lie in an opposition between the neo-Platonic and the A r i s t o t e l i a n t r a d i t i o n s . The romantic poets are seen as f o l l o w e r s of the neo-Platonic t r a d i t i o n because they held the v i e w - to s i m p l i f y matters greatly - that ordinary and poetic 15 See, for e x a m p l e , M c L a i n (1976: 2 4 4 ) . 16 See, f o r e x a m p l e , Thorne ( 1 9 6 5 ) . 17 See W o r d s w o r t h ' s " P r e f a c e " to the Lyrical Ballads ( ( 1 8 0 5 ) ; here Wordsworth 6. Coleridge (1976: 2 9 ) ) ; and G r a y ' s letter to West of 8 t h A p r i l 1742 (here Gray (1971: 1 9 2 ) ; the latter is quoted in Bateson (1973: 5 4 ) ) . See also G b l l e r ( 1 9 6 4 : 2 5 ) . But note that W o r d s w o r t h ' s v i e w was perhaps not as clear cut as later critics - including Coleridge, in the Biographie Literaria (1817; here Coleridge ( 1 9 5 4 : v o l . I D ) - have thought it to be. Wordsworth in f a c t says that poetic language should be "a selection of the language of men" (1976: 29; emphasis added) and observes that a properly selected subject w i l l lead to language "necessarily [ . . . ] d i g n i f i e d and variegated, and a l i v e w i t h metaphors and f i g u r e s " (1976: 3 0 ) . This fact is also noted by W i m s a t t & Brooks (1957: 3 4 7 ) .

11

language cannot be separated. According to...the t r a d i t i o n , they can and must be d i s t i n g u i s h e d .

Aristotelian

How are we to view these c o n f l i c t i n g opinions? Are we to suppose that any pronouncement on literary language can only be relevant to the age in w h i c h it is uttered? P a r t l y , yes. But it also seems possible to make statements about literary language which are independent of a particular age and c u l t u r a l area. P r i m a r i l y , this is possible because the statements of theorists and even (or e s p e c i a l l y ) of poets themselves about the language of poetry are o f t e n seen, upon examination of the work in question, not to be borne out by e m p i r i c a l evidence. As an i l l u s t r a t i o n of this type of discrepancy, compare C o l e r i d g e ' s arguments against Wordsworth's assertion that he used the language of "men in low and rustic l i f e " (Coleridge ( 1 9 5 4 : II, 4 1 ) ) . Coleridge maintains that Wordsworth could not avoid using poetic language, and adds: I r e f l e c t w i t h d e l i g h t , how l i t t l e a mere theory, though of his own w o r k m a n s h i p , i n t e r f e r e s w i t h the processes of genuine imagination in a man of true poetic genius. (Coleridge (1954: I I , 4 1 ) ) In fact this observation does not only apply to Wordsworth; it is also true of later poets who have made s i m i l a r claims. In the Introduction to his Neu Lines II, w h i c h appeared in 1963 (a sequel to the 1956 Neu Lines, which was f e l t to have established the existence of a group of poets known as "The M o v e m e n t " ) , Robert Conquest says that: Though it would be f a l s e to state that no genuine poetic e f f e c t can ever be achieved by the disruption of grammar, sense, and so on, yet this is nowadays, as always, so rare that I have thought it scarcely worth achieving a forced c a t h o l i c i t y by representing it here. (Conquest ( 1 9 6 3 : x x v i i i ) ) Conquest several times in his Introduction invokes Wordsworth, and proclaims that the poets in his c o l l e c t i o n also use "the language of men" ( p . x x v i i i ) . However, a study of the poems in Neu Lines II reveals that many of the most deceptively s i m p l e poems use rhymed l i n e s , and that there is a wealth of metaphor of various types: (1-1)

slashed clouds leaJc gold ( p . 41) iced uith a vanilla of dead unite stone ( p . 95)

18 For a u s e f u l discussion of these traditions see Hawkes ( 1 9 7 2 ) 12

a daisy g J earns as coldly as a star ( p . 134) of a l l i t e r a t i o n : (1-2)

protracted paradox of printing hands ( p . 89) a slow and stopping curve southwards ( p . 19) a seep silent all summer ( p . 55)

and of "poetic" uses of words: (1-3)

the fluting owl glides velvet ( p . 4 2 ) under the night-green boughs ( p . I l l ) glossy-with-graining pulpit ( p . 9 2 )

These f a c t s suggest that a d e s c r i p t i o n of the language of poets as "the language of m e n " , if by t h i s is meant the standai cl language , is h i g h l y inaccurate. As a f u r t h e r instance of the p o e t ' s misconceptions about h i s own w o r k , consider A . R o b b e - G r i l l e t ' s statement that "a metaphor is n e a r l y a l w a y s useless, a d d i n g n o t h i n g to the d e s c r i p t i o n " (1965: 368) and compare it w i t h the f o l l o w i n g statement by a critic: [ . . . ] however much Robbe-Gri1 let may protest against the use of h u m a n i s t i c a l l y oriented symbols and metaphors, he cannot h i m s e l f a v o i d using them. (Hagopian (1968: 4 9 ) ) These e x a m p l e s i l l u s t r a t e not o n l y the u n r e l i a b i l i t y of poets' judgements about the p o e t i c , but also the f a c t that poems in f a c t cannot be w r i t t e n w i t h o u t using poetic language. The most reasonable v i e w of the c o n f l i c t discussed above seems thus to be that the r o m a n t i c v i e w must at least, to have any v a l i d i t y , be expressed in a less extreme f o r m . It is then possible to r e c o n c i l e it w i t h the opposing v i e w , by observing that l i t e r a r y language, in any age, is the l a n g u a g e of men in the sense that it uses the grammar and l e x i s of standard language, but that it is d i f f e r e n t f r o m standard language in that it also uses devices not used in the l a t t e r . Just what these devices are and how they come i n t o being w i l l be made precise in the f o l l o w i n g pages. For the moment it is s u f f i c i e n t to say that we regard as j u s t i f i e d the assumption that there is such a t h i n g as " l i t e r a r y language" w h i c h is composed of a c o m b i n a t i o n of standard language and what

13

19

we shall in the f o l l o w i n g call "poetic language". We shall v i e w both poetic language and standard language as systems o£ p r i n c i p l e s . By "poetic language" we mean the special language commonly used in poems, novels and p l a y s , but not that c o m m o n l y used in ordinary conversation, advertisements, j o u r n a l i s m or s c i e n t i f i c works. A l l these latter d o m a i n s , w h i c h , J i k e poetic language, can be characterised by sets of p r i n c i p l e s , w i l l use poetic language at times, by consciously b o r r o w i n g . There w i l l also be cases in w h i c h poetic language a n d , say, the special language used in j o u r n a l i s m exhibit the same phenomena, because they w i l l share some of the same p r i n c i p l e s . But there w i l l be many areas in w h i c h they do not overlap. Poetic language is the result of an interaction of poetic p r i n c i p l e s w i t h standard language p r i n c i p l e s . This i n t e r a c t i o n m a y , but need not, take p l a c e . All standard language f o r m s can thus appear in poems but poetic language f o r m s o n l y appear in standard language under special circumstances. That is, there is a tendency to avoid what are regarded as "poetic" f o r m s in standard language to the extent t h a t , should they arise u n f o r e s e e n , they are u s u a l l y corrected. A s i m i l a r point to this is made by Shapiro S. Beum (1965: 93) in connection w i t h the use of a p a r t i c u l a r poetic d e v i c e in prose: [...] notice that we even take pains to avoid a l l i t e r a t i o n in prose, where it seems ? ,appropriate o n l y for a humorous or an eccentric e f f e c t . It should be noted that the d i s t i n c t i o n we have made is not so much that between prose and.-.poetry as between literary and nonl i t e r a r y types of language. 19 This is a term made famous by the Prague School l i n g u i s t s , notably MukaiOvsky; our use of it varies though somewhat f r o m his, as w i l l become clear. (See Mukafovsky (1932 [ 1 9 6 4 ] ) . ) 20 But note that d i f f e r e n t literary genres have their own p r i n c i p l e s w h i c h interact w i t h the more general poetic p r i n c i p l e s we shall be concerned w i t h h e r e . For e x a m p l e , rhyme w i l l appear frequently in l i t e r a r y prose as a l l i t e r a t i o n , but less o f t e n as f u l l rhyme and c l e a r l y cannot appear in any patterns of recurrence depending upon an arrangement in l i n e s . That many types of rhyme do not, h o w e v e r , depend upon the poetic l i n e w i l l be seen in the discussion of rhyme in Chapter 4. 21 For a discussion of the characteristics of these v a r i o u s areas of language, see, for e x a m p l e , Gu'lich & R a i b l e (1975) and Fleischer & M i c h e l ( 1 9 7 9 ) ; and f u r t h e r Sandig ( 1 9 7 1 ) for the language of j o u r n a l i s m . Spitzer ( 1 9 4 9 ) and Römer (1968) for the language of a d v e r t i s i n g and Gläser (1975) for the language of s c i e n t i f i c w o r k s . 22 Compare the statement by MukaiOvsky (1932 [ 1 9 6 4 : 19]) that "The standard language [ . . . ] avoids f o r e g r o u n d i n g [ . . . ] " ; "foregrounding" is the technique of d r a w i n g a t t e n t i o n to a l i n g u i s t i c f o r m . See the discussion in Chapter 4. 23 The d i s t i n c t i o n between prose and poetry is, h o w e v e r , one w h i c h has interested many c r i t i c s and w h i c h , however unclear 14

P o e t i c language is, t h e n , in the sense in w h i c h we shall use the t e r m , that part of the language o£ a poetic text " l e f t over when one has excluded gibberish on the one hand and ordinary discourse on the other" ( G r i n d e r & E l g i n (1973: 1 6 9 ) ) . In the f o l l o w i n g we shall in general r e f e r to poetic language as "PL" and to standard language as "SL". Having established that PL d i f f e r s f r o m SL in that it i n v o l v e s "the use of unorthodox or deviant f o r m s of language" (Leech 1966: 1 3 6 ) , let us look b r i e f l y at the f o r m s these deviations t a k e . One important source of data in support of the v i e w of PL as d e v i a n t has always been the words poets use. It has been maintained by the advocates of "negative poetic diction" (Bateson (1973: 5 2 f f . ) ) - though the context of the discussion here was rather the d i f f e r e n c e between poetry and prose - that some words should not occur in poems. J o h n s o n ' s comment on M a c b e t h , that Jcnife is s o m e t h i n g used by butchers and cooks and has no. place in poetic language, is no doubt one of the most f a m o u s . The idea that certain words are of t h e m s e l v e s poetical was an idea central to the late eighteenth century idea of poetic d i c t i o n . In modern l i t e r a r y studies, e x a m i n a t i o n s of the words an author uses have been f a i r l y common ( c o m p a r e the list, in Chapter 0 . 1 ) . The f o l l o w i n g examples show the type of words u s u a l l y considered to be i l l u s t r a t i v e of the f a c t that there is a special type of l e x i s used in poetic w o r k s : (1-4)

gloom-arrayed, daisied, broider'd, outelboued^ draun, val ley-and-God

a-dying, bleak-grovn, sterfear-surprised, dapple-davn-

Other examples given in support of the v i e w that poets use a d i f f e r e n t language have been taken f r o m the syntax of p o e t r y . Chomsky in 1963 used examples f r o m e.e. c u m m i n g s to show syntactic constructions not permissible in standard E n g l i s h such

24 25 26 27

in n a t u r e , has proved v a l u a b l e in many areas of study. As an e x a m p l e , see L e v i n (1971b) on the d i f f e r e n c e s in d e l e t i o n in prose and poetry. Compare the s i m i l a r statement by B a n f i e l d (1973a: v). The Rambler, N o . 1 6 8 , 26th October 1751; quoted in Bateson (1973: 5 3 ) . For a d e t a i l e d discussion of eighteenth century poetic d i c t i o n , see Goller ( 1 9 6 4 ) . E x a m p l e s are f r o m , r e s p e c t i v e l y : James Thomson, "The C i t y of D r e a d f u l N i g h t " (1899: 3 1 ) ; Thomas H a r d y , "The M i l k m a i d " (1902; here from Hardy (1977: 198)); William Watson, " W o r d s w o r t h ' s Grave" (1887; here f r o m Rhys (1950: 3 6 2 ) ) ; John Keats, "Hyperion" ( 1 8 2 0 ; here f r o m Keats ( 1 9 7 4 : 2 1 3 ) ) ; A l f r e d T e n n y s o n , "The S l e e p i n g B e a u t y " ( 1 8 9 4 : 105); D y l a n Thomas, "Today, t h i s insect" ( 1 9 3 6 ; here f r o m D . Thomas (1971: 1 2 4 ) ) ; W i l l i a m Shakespeare, Hamlet I , i i ( 1 6 0 3 ; here f r o m Shakespeare (1963a: 4 7 ) ) ; Gerard M a n l e y H o p k i n s , "The W o o d l a r k " (1918; here f r o m H o p k i n s ( 1 9 5 3 : 7 1 ) ) ; Salman R u s h d i e , Midnight's Children ( 1 9 8 2 : 1 2 ) . 15

as Me up at does. For f u r t h e r consider the f o l l o w i n g lines:

examples

(1-5)

[ . . . ] love cannot make smart Agfaίη this year his heart Who no heart hath.

(1-6)

Of Nineveh and distant Οphi r ue find written elsewhere

of

poetic syntax,

Poetic syntax has been examined in studies by Hammond ( 1 9 6 1 ) , Davie ( 1 9 6 6 ) , B a n f i e l d (1973a, 1973b, 1 9 8 2 ) , Freeman (1975, 1976, 1981a), Epstein ( 1975) , Austin ( 1 9 8 4 ) and many others. l Other elements of p o e t i c language w h i c h suggest that it is a language d i f f e r e n t f r o m the standard include instances of assonance and a l l i t e r a t i o n , such as the f o l l o w i n g examples illustrate: (1-7)

/ saw the large luminous living eyes 32

(1-8)

Silent fizzer, sting white in the sunlight ( T O 4 2 )

(1-9)

hopeless fear on flying

feet

of r h y m e , such as: (1-10)

From unrememJbered ages we Gentle guides and guardians he.. Of heaven-oppressed mortality

(1-11)

Children picJf painted shells to hear Long tides come hushing on the ear (SG 4 9 )

and of metaphor, as illustrated by the f o l l o w i n g examples: 28 See Fairley (1975: 2 1 6 ) . 29 Thomas Hardy, "In Tenebris" (1902; here f r o m Hardy (1977: 116)). 30 George Szirtes, "A Pheasant" (1983: 2 9 ) . 31 See also the bibliography by O z z e l l o ( 1 9 7 2 ) . 32 Thomas Hardy, " A f t e r the V i s i t " (1914; here f r o m Hardy (1977: 82)). 33 W i l l i a m G o l d i n g , Lord of the Flies (1958: 2 2 0 ) . 34 P . B . Shelley, "Prometheus Unbound" (1820; here f r o m Shelley (1975: 1 1 3 ) ) . 16

(1-12)

Cruelty has α human heart

(1-13)

[ . . . ] Bound in leather The book crauls among us

(1-14)

35

with a loud voice (AB 68)

[ . . . ] gradually her fears, her yearning anxieties for others, melted auay HJce ice-crystals in a uarm ocean.

The phenomena of a l l i t e r a t i o n , assonance and rhyme have been discussed in most handbooks on poetry and poetic t h e o r y , such as Shapiro & Beum ( 1 9 6 5 ) or Leech ( 1 9 6 9 ) . Of all aspects of p o e t i c l a n g u a g e , metaphor has probably received the most a t t e n t i o n ; it has been studied in works by Aristotle ( 1 9 7 0 ) , Coleridge ( 1 9 5 4 ) , Beardsley ( 1 9 5 8 ) , Brooke-Rose ( 1 9 5 8 ) , Black ( 1 9 6 2 ) , L i e b ( 1 9 6 4 ) , W h e e l w r i g h t ( 1 9 6 8 ) , O . T h o m a s ( 1 9 6 9 ) , Ingendahl ( 1 9 7 1 ) , Hawkes ( 1 9 7 2 ) , R i c o e u r (1975J L e v i n (1977) a n d K e l l e r - B a u e r ( 1 9 8 4 ) , t o name just a very f e w . In the present study we s h a l l be concerned w i t h all these aspects of the special nature of PL to a greater or a lesser e x t e n t ; our area of study is c o m p o u n d - f o r m a t i o n , and one of our c o n t e n t i o n s w i l l be that the p r i n c i p l e s here at work - whether g r a m m a t i c a l or p o e t i c - are also at work in other areas. ( T h i s is made e x p l i c i t in Chapters 3 and 4). If we e x a m i n e the ad hoc compounds f o u n d in p o e t i c w o r k s , we s h a l l f i n d many w h i c h are formed r e g u l a r l y . These belong to that part of PL shared w i t h SL, and could as w e l l have occurred in SL. These are compounds such as the f o l l o w i n g : (1-15)

charcoal-burning [ N A ] (StW 4 9 ) , uind-huffed [NA] (+MT 165). tall-heeled [ [ A N ] e d ] (CW 31) , hird-music [ N N ] (SP 24)38

H o w e v e r , we s h a l l also f i n d SL, cf. the f o l l o w i n g : (1-16)

compounds w h i c h w o u l d not

years-heired [ N A ] m e a n i n g "made the "made the h e i r every year"

occur

in

heir of years"

or

3 5 W i l l i a m B l a k e , " A D i v i n e I m a g e " ( 1 7 9 4 ; here f r o m B l a k e (1958: 59)). 36 George E l i o t , Adam Bede (1859; here f r o m E l i o t (1961: 1 5 7 ) ) . 37 See also the b i b l i o g r a p h y by S h i b l e s ( 1 9 7 1 ) , and c o l l e c t i o n s b y Sacks ( 1 9 7 9 ) , Ortony ( 1 9 7 9 ) , H a v e r k a m p ( 1 9 8 3 ) . 38 Here and throughout compounds are g i v e n in t h e i r u n i n f l e c t e d forms. 39 Thomas H a r d y , " H e r e d i t y " (1917; here f r o m Hardy (1977: 1 6 3 ) ) . 17

(1-17)

thought-executing [NA] with the meanings "doing e x e c u t i o n as q u i c k as thought" and "executing the 40 thought of s o m e o n e "

(1-18)

fruit-offer [ N V ] ( + N M 30) m e a n i n g like one would o f f e r f r u i t "

(1-19)

hover-search [ V V ] ( + SS 27) m e a n i n g "to hover and search"

(1-20)

uater-gifted in water"

(1-21)

horse-uorship [ N V ] ( + W N P 57) m e a n i n g "to w o r s h i p horses"

(1-22)

flouer-lovely-false [ [ N A ] A ] ( + N M 21) meaning as a f l o w e r is l o v e l y "

(1-23)

grin-Jong [ N A ] ( + R 1 0 4 ) m e a n i n g "as long as a g r i n "

"to

o f f e r something

[ N A ] (+L 46) meaning "gifted

for

swimming

"as

false

I n t u i t i v e l y , these compounds are not acceptable in standard language. It w i l l be our task to account for t h i s i n t u i t i o n more precisely, and, further, to explain, in view of this i n t u i t i o n , what it is that a l l o w s the poet to use such " i m p o s s i b l e " compounds, yet s t i l l to be understood. Among scholars who have assumed, as we are a s u m i n g , that there is a d i s t i n c t language of p o e t r y , there have been various v i e w s as to its nature and its r e l a t i o n to standard language. Some of these v i e w s are i m p l i c i t in the type of studies w h i c h have been w r i t t e n : f o r d i f f e r e n t a u t h o r s , t h e greatest d i f f e r e n c e between PL and SL lies in the f o r m e r ' s use of s y n t a x , in its use of p a r t i c u l a r l y " p o e t i c " words (or a v o i d a n c e of "unpoetic" w o r d s ) , or in its use of such l i t e r a r y d e v i c e s as symbol and m e t a p h o r . Common to all such v i e w s , whatever t h e i r e m p h a s i s , is the idea that PL d e v i a t e s f r o m SL in some w a y ; t h i s is an idea to be .found also in classical r h e t o r i c . Any v i e w w h i c h is not based on t h i s assumption d e n i e s , in e f f e c t , as we have seen, the e x i s t e n c e of poetic language as a separate e n t i t y . Among the " d e v i a t i o n - v i e w s " of p o e t i c l a n g u a g e , h o w e v e r , there are two general t e n d e n c i e s to be d i s t i n g u i s h e d . R e p r e s e n t a t i v e of the f i r s t is the f o l l o w i n g statement: [...] the distinction [i.e. that between ordinary a n d l i t e r a r y l a n g u a g e ] is not in k i n d , but in use of the 40 W i l l i a m Shakespeare, King Lear I I I , i i ( 1 6 0 8 ; here f r o m Shakespeare ( 1 9 6 3 b : 1 0 6 ) ) . These meanings are g i v e n by the editor of the l a t t e r . 41 C o m p a r e , for e x a m p l e , the o b s e r v a t i o n made by P u t t e n h a m in his r h e t o r i c of 1589, that " F i g u r a t i v e speech is a n o v e l t i e of e v i d e n t l y (and yet not a b s u r d l y ) estranged f r o m the o r d i n a r i e h a b i t e and manner of our d a y l y t a l k e and w r i t i n g " (1589 [1936: 1 5 9 ] ) ; cited b y V i c k e r s ( 1 9 7 0 : 8 5 ) . 18

l i n g u i s t i c process. The l i t e r a r y artist has at his disposal no p r i n c i p l e of language not accessible to the ordinary speaker; but the a r t i s t ' s s e l e c t i o n and manipulation of these principles are effectively extraordinary. ( C h i n g e t a l . (1980: 3 7 ) ) T h i s , in e f f e c t , is the v i e w that a poet uses the rules of SL, but uses them d i f f e r e n t l y , or "better". It is a v i e w expressed by many l i n g u i s t s ; consider C o s e r i u ' s emphasis on poetic language as "the o n l y f u l l and complete use of language" (1971: 1 8 ) , and s i m i l a r statements by l i n g u i s t s such as Dressier ( 1 9 8 1 ) . A d i f f e r e n t v i e w w h i c h at f i r s t sight seems to stand in o p p o s i t i o n to the f o r m e r is that put f o r w a r d by Russian F o r m a l i s t s , such as S h k l o v s k y (1917 [ 1 9 6 5 ] ) , by Prague School l i n g u i s t s such as MukarOvsky ( 1 9 3 2 [ 1 9 6 4 ] ) and Havranek ( 1 9 3 2 [ 1 9 6 4 ] ) , and also by later c r i t i c s and l i n g u i s t s such as Leech ( 1 9 6 6 ) , B a n f i e l d ( 1 9 7 3 a , 1973b, 1 9 8 2 ) , n a m e l y that t h e d e v i a t i o n f r o m the norm of SL w h i c h is f o u n d in poetry is not so much s i m p l y a case of d i f f e r e n c e in p r o d u c t i v i t y in PL as that i n n o v a t i o n s are f o u n d in PL w h i c h could not be f o u n d in SL, or at least, not in the language in q u e s t i o n . The f i r s t v i e w tends towards the o p i n i o n that "each component (of the grammar) has its own ' n a t u r a l ' u n i v e r s a l t e n d e n c i e s " , w h i c h the poet may not f e e l bound t o observe (Dressier 1981: 4 2 4 ) ) . T h e latter v i e w implicitly or e x p l i c i t l y v i e w s PL as h a v i n g some a d d i t i o n a l p r i n c i p l e s , w h i c h a l l o w f o r c o n s t r u c t i o n s n o t otherwise p o s s i b l e , i . e . not a c c e p t a b l e in SL, or m a i n t a i n s t h a t , in PL, rules of SL a r e "broken" ( s e e , f o r e x a m p l e . L e v i n ( 1 9 6 5 ) ) . T h e v i e w o f p o e t i c d e v i a t i o n we wish to put f o r w a r d in the present study belongs w i t h the v i e w s of the latter g r o u p . H o w e v e r , it is not q u i t e clear just in how far the two v i e w s r e a l l y d i f f e r . If a poet selects f r o m and m a n i p u l a t e s the n o r m a l l i n g u i s t i c means, he must have access to p r i n c i p l e s w h i c h a l l o w him to do so. If the nonpoet does not use these p r i n c i p l e s (- w h e t h e r or not they are "accessible to the o r d i n a r y speaker", whatever t h i s means, is not r e a l l y the p o i n t at issue here -), then they must be characterisable as something s p e c i f i c to PL. A s s u m i n g , h o w e v e r , that the f o r m e r v i e w m i g h t be seen as a less extreme f o r m of the v i e w that PL does not d i f f e r s i g n i f i c a n t l y f r o m SL and that it is also based on strong i n t u i t i o n s about the nature of PL, it seems u n l i k e l y that i t i s e n t i r e l y w i t h o u t f o u n d a t i o n . I n f a c t , w e s h a l l show that both v i e w s have a basis in the f a c t s of PL. The v i e w that a poet does not have any means at his disposal w h i c h are not a v a i l a b l e to the speaker of SL seems to be based on two i n t u i t i o n s about PL, n a m e l y that "not just a n y t h i n g goes", and that many d i f f e r e n c e s between PL and SL lie in increased p r o d u c t i v i t y in PL of m a r g i n a l SL f o r m s . The second v i e w sees a greater d i f f e r e n c e between PL and SL: there are some structures 42 T h i s is, in f a c t , i m p l i e d by proponents of t h i s v i e w : see, for e x a m p l e , N o w o t t n y ( 1 9 6 5 : 1 - 2 5 ) , Ching e t a l . ( 1 9 8 0 : 5 ) . 43 Chomsky ( 1 9 6 4 : 3 8 6 ) has also noted the f a c t that not all d e v i a t i o n is a c c e p t a b l e as p o e t i c . 19

w h i c h are possible in PL but not possible in SL. The examples given above show that t h i s is in f a c t the case. The correct v i e w of the r e l a t i o n of PL to SL seems in f a c t to be one w h i c h resolves both of the apparent c o n f l i c t s we have discussed, i . e . both the "same versus d i f f e r e n t " c o n f l i c t and the " m a n i p u l a t i o n versus innovation" c o n f l i c t . This is the v i e w that PL and SL both share the same g r a m m a t i c a l p r i n c i p l e s ( i . e . , in t h i s they are the same) but that PL has extra poetic p r i n c i p l e s ( i . e . , in t h i s they d i f f e r ) . F u r t h e r , these p r i n c i p l e s both a f f e c t t h e p r o d u c t i v i t y of certain e x i s t i n g f o r m s - in t h i s they are " m a n i p u l a t i v e " - and also a l l o w the p r o d u c t i v i t y of new f o r m s , and are thus innovative. In a d d i t i o n to t h i s , we hope to show that many of the poetic p r i n c i p l e s have counterparts in SL ( t h o u g h probably not in the grammar of S L ) , whereby the poetic principles tend to be less general in f o r m , but also much less constrained as to the type of structures they may i n v o l v e than t h e i r SL counterparts. This f u r t h e r supports the i n t u i t i o n that PL is, w i t h respect to SL, to put it s u c c i n c t l y "both the same and d i f f e r e n t " . F u r t h e r m o r e , it w i l l be observed that PL princinLes are not simply arbitrary in nature, nor u n l i m i t e d in number. Because our concern is w i t h poetic language as d e v i a t i o n f r o m the norm of the standard,, it is i m p o r t a n t that it should be clear what is a deviant f o r m . In order to establish t h i s > we need to know what f o r m s are generated by the grammar of standard l a n g u a g e . It does not make sense, h o w e v e r , to speak of d e v i a t i o n s in PL as though they were d e v i a t i o n s f r o m c o r r e s p o n d i n g f o r m s , as some researchers have attempted to do (see, for e x a m p l e . Lord ( 1 9 7 5 ) ) . Because the grammar consists of d i f f e r e n t i n t e r a c t i n g p r i n c i p l e s , d e v i a n t f o r m s can be d e v i a n t in a number of d i f f e r e n t ways. Any discussion of what the "usual" n o n - d e v i a n t f o r m w o u l d have been o n l y makes Ja^nse if some sort of statistical p r o b a b i l i t y is assumed. We are not concerned w i t h such " p r o b a b i l i t i e s " here, but o n l y w i t h w h a t , as output of the grammar of SL, is p o s s i b l e . Thus d e v i a t i o n cannot be seen in terms of a one-to-one correspondence of d e v i a n t f o r m to nond e v i a n t f o r m . D e v i a t i o n is to be f o u n d , we s h a l l argue, in the i n t e r a c t i o n of a p o e t i c p r i n c i p l e w i t h the grammar of SL. This is an i n t e r a c t i o n w h i c h o n l y takes p l a c e in PL because p o e t i c p r i n c i p l e s are by d e f i n i t i o n o n l y e f f e c t i v e here. Thus we shall d e f i n e deviation as f o l l o w s : 44 It is important to realise that the d i f f e r e n c e in p r o d u c t i v i t y cannot in any case appear a r b i t r a r i l y , but is governed by a d i f f e r e n c e in c o n s t r a i n t ; t h i s in turn must a l w a y s a r i s e o w i n g to new p r i n c i p l e s or a new i n t e r a c t i o n of p r i n c i p l e s . 45 Compare G r i n d e r & E l g i n (1973: 1 6 9 - 1 8 8 ) . 46 That is, we are concerned w i t h , to use L e v i n ' s t e r m , "external d e v i a t i o n " , as opposed to " i n t e r n a l d e v i a t i o n " , w h i c h is d e v i a t i o n f r o m the norm of an i n d i v i d u a l poem ( L e v i n ( 1 9 6 5 : 2 2 6 ) ) . S e e also L e v i n ( 1 9 6 3 , 1 9 7 1 a ) . 47 S e e B l o c h (1953: 4 2 ) , quoted i n L e v i n ( 1 9 6 3 : 2 7 9 ) , w h i c h latter essay should be consulted for a d i s c u s s i o n of and rejection of this view. See also Banfield (1973a: " I n t r o d u c t i o n " ) for a discussion of the same p o i n t . 2Ü

(1-24)

Deviation A l i n g u i s t i c f o r m is deviant by the grammar of SL.

if

it cannot be

generated

A deviant f o r m , then, is not generated by the grammar of SL a l o n e , but by this grammar in i n t e r a c t i o n w i t h poetic p r i n c i p l e s . This view has important consequences for a c r i t i c a l study of poetic language, i . e . , for the type of study which should i d e a l l y start where a stylistic study such as this leaves o f f : the s t y l i s t i c e f f e c t s of deviation are not to be sought in the contrast between the deviant form and its supposed non-deviant correlate, as this latter cannot be d e t e r m i n e d , but in the w o r k i n g of a s p e c i f i c p r i n c i p l e or set of p r i n c i p l e s , whose i n t e r a c t i o n results in the p a r t i c u l a r deviant f o r m . Obviously we are not to be taken as c l a i m i n g that the whole nature of literary language can be e x p l a i n e d in terms of d e v i a t i o n . In f a c t , a corollary of the view of PL we are proposing is that PL and literary language are two d i f f e r e n t things. Literary language contains both PL and SL, and both these aspects w i l l play art important role in any comprehensive study of literary language. But because our concern is to show the relationship of PL to SL - how PL d i f f e r s from SL, and what the two have in common - rather than to examine literary language or literature in a more general sense, the question of d e v i a t i o n w i l l be our central concern. T h i s view of PL as inherently d i f f e r e n t f r o m SL carries w i t h it the assumption that PL cannot be d e f i n e d by area of use, i.e. simply as that language used in poetry rather than in normal conversation. O b v i o u s l y , not all poetry makes use of PL on our d e f i n i t i o n and no poetry makes use of it all the t i m e . This accounts in part for the idea of some n\pdern poets, that they are trying to use a non-poetic language. T h e o r e t i c a l l y it is not impossible to w r i t e a poem in standard language, t h o u g h , as we 48 Rene W e l l e k has said "often the most commonplace, the most n o r m a l , l i n g u i s t i c elements are the constituents of literary structure" ( 1 9 6 0 : 4 1 8 ) . But it is o f t e n the case that elements w h i c h are not deviant themselves are part of a deviant pattern (such as rhyme or p a r a l l e l i s m ) . In fact it is reasonable to assume that all s i g n i f i c a n t aspects of PL are deviant in some way - either per se, or because they are p o e t i c a l l y ordered, especially p r o d u c t i v e , or stand in a s p e c i f i c r e l a t i o n s h i p to some deviant e l e m e n t . Compare also in t h i s connection J . M u k a r O v s k y ' s observation that a poetic work is a u n i f i e d w h o l e of "all [ . . . ] components, foregrounded or n o t , as w e l l as t h e i r interrelationships" (1932 [1964: 2 0 ] ) . 49 And compare W o r d s w o r t h ' s statement (see above) that he was w r i t i n g poetry w h i c h used "the language r e a l l y spoken by m e n " ; he could not have said t h i s had he f e l t that poetic language 21

have noted above, some poets who profess to do t h i s have not managed to leave all poetic elements out of t h e i r poems. Furthermore, it has, w i t h some j u s t i f i c a t i o n , been f e l t by critics and f e l l o w poets, that this type of p o e t r y , i . e . that represented by Conquest's a n t h o l o g i e s , is n o t , on the w h o l e , good poetry. A t y p i c a l o p i n i o n is that of the poet Charles T o m l i n s o n w h o , in his r e v i e w of Neu Lines I, f i n d s in the c o l l e c t i o n a "stale f e e l i n g of ordinariness", a " f a i l u r e to see things anew" ( T o m l i n s o n (1957: 2 1 5 ) ) , even a "lack of genuineness" ( p . 2 1 3 ) . "We are asked", says T o m l i n s o n , "to a d m i r e the new poetry for its negative virtues - simply, that is, for an absence of ' r h e t o r i c ' " ( p . 2 1 6 ) . And he concludes that this type of poetry is characterised, above a l l , by " m i d d l i n g n e s s [and] lack of real poetic talent" ( p . 2 1 6 ) . b u These observations are important for three reasons. F i r s t l y , they substantiate the postulation of a separate PL; if PL did not e x i s t , one could not try to avoid using it. Secondly, they substantiate the v i e w that PL is not d e f i n e d by its area of use. T h i r d l y , they show that PL, though by d e f i n i t i o n it must not necessarily be present in a poem, might nevertheless be considered essential to the success of a poem.

1.3.

The Relationship of Poetic Language to Standard Language

During our discussion in the previous section, there emerged three m a i n characteristics of poetic language, w h i c h might be summarised as f o l l o w s : (i)

poetic language deviates f r o m standard language in some way;

(ii)

this d e v i a t i o n is both a question of p r o d u c t i v i t y forms w h i c h are unproductive in SL are productive in PL - and of a c c e p t a b i l i t y - f o r m s unacceptable in SL are acceptable in PL;

(iii)

not just anything goes, i . e . the way in w h i c h PL deviates can be e x p l i c a t e d in a p r i n c i p l e d w a y .

These observations led us to regard poetic language as consisting of standard language p l u s some a d d i t i o n a l poetic p r i n c i p l e s (about w h i c h we have as yet said very l i t t l e ) . It is to be assumed that these two aspects of PL - the grammar of SL and the special poetic p r i n c i p l e s - w i l l interact w i t h one another in various ways. If we can characterise those p r i n c i p l e s of PL w h i c h are important for our area of s t u d y , namely c o m p o u n d i n g , and can 50 See a l s o , for s i m i l a r v i e w s , Holbrook (1977: Chapter 6) and the more s y m p a t h e t i c comments of Ted Hughes (Faas (1980: 201)).

22

o f f e r an account of t h e i r i n t e r a c t i o n w i t h the SL grammar of compounding, we shall have gone a long way towards accounting for the r e l a t i o n s h i p of PL to SL at least in the area of compounding, and, it is hoped, in general. In t h i s section we wish to turn our attention to a c c o u n t i n g f o r m a l l y for the i n t e r a c t i o n of poetic p r i n c i p l e s ( w h i c h we shall discuss in detail in Chapter 4) w i t h the grammar of compounding in E n g l i s h ( w h i c h we shall discuss in detail in Chapter 3). But f i r s t , it is perhaps appropriate to ask how we know that there are such things as general poetic p r i n c i p l e s w h i c h interact w i t h the grammar to produce a general poetic language. How do we know that there are not seperate rules for each w o r k , for each w r i t e r , for each movement, for each genre, or for each era? It may seem i n t u i t i v e l y obvious that this can not be the case for each work or a u t h o r , at least, but any c r i t i c or l i n g u i s t who has suggested w r i t i n g a grammar for each work or author has i m p l i e d , in e f f e c t , that each has his o w n , for c l e a r l y if each had the same grammar it would make no sense to w r i t e separate ones. The examples we have given in 1.1 show that the d e v i a t i o n s discussed can be f o u n d in a wide variety of poetic works. This suggests that there is a literary language c o m m o n , at least in some essential aspects, to poetry and prose, to d i f f e r e n t eras and d i f f e r e n t authors. This is not, though, to ignore the f a c t that this language may change so that, for e x a m p l e , what was normal for M i l t o n might be odd for Ted Hughes. This could have two reasons. The grammar of the language might change - it m i g h t acquire some rules and lose others. And the literary principles might change. However, just as it is n o r m a l l y assumed that the grammar of English has not changed e s s e n t i a l l y since the b e g i n n i n g of Modern E n g l i s h (around 1 5 0 0 ) , it also seems reasonable to assume that the l i t e r a r y p r i n c i p l e s have not changed essentially - though there "ϋΑ-> undoubtedly be inessential changes - at least w i t h i n this time." In actual f a c t , it seems probable that many p r i n c i p l e s of literary language, l i k e many grammatical p r i n c i p l e s , w i l l in the area of their a p p l i c a t i o n far exceed o n l y the period of Modern English or indeed the English language as a w h o l e . As for the variation among authors or works - one author uses more metaphor or more i n n o v a t i o n in words than another - we shall assume that this is a question of p e r f o r m a n c e . An author can choose how he i m p l e m e n t s the p r i n c i p l e s of PL. This does not a f f e c t PL i t s e l f , just as it does not a f f e c t the grammar of E n g l i s h if I choose to use a c t i v e rather than passive constructions. What we are concerned w i t h here are questions of s t y l i s t i c competence seen as 51 For a suggestion that a separate grammar be w r i t t e n for each poetic w o r k , see Thorne ( 1 9 6 5 ) . 52 Compare the statement by Paul K i p a r s k y (1973: 231): "Styles and conventions [in poetry] have s h i f t e d , but no t r u l y new forms have emerged". He goes on to suggest that t h i s f a c t is not attributable to a p a r t i c u l a r l i t e r a r y t r a d i t i o n but to "the i n t r i n s i c nature of verbal art". 53 See B a n f i e l d (1973a: xv) and ( 1 9 8 2 : 6) for a s i m i l a r point of view.

23

the i n t e r n a l i s i n g of p r i n c i p l e s of poetic language, and the nature of poetic language i t s e l f , e s p e c i a l l y in its r e l a t i o n to SL. Questions of p e r f o r m a n c e , of the choice an author makes out of means a v a i l a b l e to h i m , w i l l be of only minor concern to us here. The f a i l u r e to see this important d i f f e r e n c e between s t y l i s t i c competence and s t y l i s t i c performance has resulted in a number of unnecessary discussions as to whether the object of stylistics should be i n d i v i d u a l works or a general p o e t i c language. Typical of the view w h i c h prompts such discussion is a statement by W e t h e r i l l (1974: 139) that "stylistic description is less inadequate when it is s p e c i f i c and not related to overf o r m a l i s e d categories". W h i l e it is c l e a r l y true that descriptive studies of performance, choice and s t y l i s t i c e f f e c t are not only "less inadequate" when they are based on i n d i v i d u a l cases, but are in any other sense impossible, it is just as obviously the case that studies of s t y l i s t i c competence m u s t , by d e f i n i t i o n , r e f e r to generalised categories. Having established, t h e n , that literary language consists of standard language and poetic language principles in i n t e r a c t i o n , there remains the question of how best poetic language, and poetic language competence, are to be accounted f o r . We have already seen that a separate grammar for each author or work does not capture what is common to literary works. Furthermore, it must be objected to on the grounds that (as is also pointed out by B a n f i e l d (1973a: x i i f f . ) ) it does not a l l o w us to say a n y t h i n g d i r e c t l y about the r e l a t i o n s h i p of PL to SL, s u r e l y a central question in any study of PL. To do this we should need a separate mechanism to e x p l a i n the r e l a t i o n s h i p between the grammar of SL and that of the relevant PL. This solution would be ad hoc, uneconomical and would fail to account for important generalisations such as s i m i l a r i t i e s between PL and SL, the type of interaction between PL and SL rules, and so on. And it would f a i l to account for the central f a c t that there is no l i n g u i s t i c form ip,PL w h i c h can be explained without recourse to the grammar of S L . b 6

54 We avoid the term "grammar of poetic language" as many of the s p e c i f i c poetic p r i n c i p l e s of PL lie outside what is u s u a l l y regarded as the grammar in the strict sense. That w h i c h is r e a l l y the grammar of PL w i l l in most aspects be identical to the grammar of SL. 55 It should by now have become clear that we are regarding a model of poetic language as the model of an ( i d e a l ) a u t h o r ' s poetic competence; we thus reject the v i e w sometimes expressed (eg. in Fowler ( 1 9 6 9 ) ) that the interpretation of poetic language phenomena belongs to the aspect of p e r f o r m a n c e . See also Abraham & B r a u n m ü l l e r (1973: 1 0 5 ) , who speak of "stylistic competence" w h i c h they regard as "a d i v i s i o n of the more global concept of c o m m u n i c a t i v e competence", a v i e w we do not share. On the question of poetic competence, see also Bierwisch (1965) and Kiparsky (1973: 2 3 3 ) . 56 See also B a n f i e l d (1982: 1 9 ) .

24

Another possible way of capturing the relation of PL to SL would be to assume that the grammar of SL must account not only for standard language, but for poetic language, too. This suggestion is i m p l i c i t in discussions such as that in M c L a i n ( 1 9 7 6 ) . But it runs counter to our i n t u i t i o n s about what is the proper subject of a grammar. A n d , f u r t h e r , it f a i l s to explain our i n t u i t i o n that phenomena of PL can be separated, in a p r i n c i p l e d w a y , f r o m those of SL. S i m p l y accounting for them all in the same way w o u l d f a i l to show t h i s d i f f e r e n c e . The proposal we shall make concerning the r e l a t i o n s h i p of PL to SL is one along the lines of proposals made by Levin (1984, 1971a, 1971b) and B a n f i e l d (1973a, 1973b, 1 9 8 2 ) , namely that PL consists of an extension of the grammar of SL in certain aspects. Our reasons for suggesting this approach are t w o f o l d . F i r s t l y , as we have just shown, neither of the other two possible approaches seems s a t i s f a c t o r y . Secondly, regarding PL as an extension of SL a l l o w s us to account for just these t h i n g s we want to account f o r : namely the special nature of PL and its relationship to SL. Having decided that PL is an extension of SL, then we can proceed no f u r t h e r w i t h o u t k n o w i n g what the grammar of SL looks l i k e . We are of the opinion that, of models of grammar w h i c h have been suggested to date, a m o d u l a r l y constructed generative grammar such as that proposed, for example, in recent works by Chomsky ( e s p e c i a l l y Chomsky ( 1 9 8 1 ) ) provides the best framework for an account of the nature of PL and its r e l a t i o n to SL ( a n d , i n c i d e n t a l l y , of the nature of SL i t s e l f ) . A modular grammar of t h i s type seems at present to be best f i t t e d to serve as a basis for an account of poetic language competence and of poetic features as the result of an interaction of poetic principles w i t h the grammar of SL at c e r t a i n points. We shall therefore v i s u a l i s e the grammar on these l i n e s . If we make the further assumption that w o r d - f o r m a t i o n rules are f o r m a l l y s i m i l a r to the base rules of the syntax (see discussion in Chapter 3), then it is c l e a r l y the case that there are at least two possible views about the organisation of those parts of the grammar responsible for compounding, both of w h i c h are consonant w i t h the overall conception of a modular grammar. We could assume that the base r u l e for compounding is restricted to certain categories - this is the v i e w of compounding proposed by, for example, Selkirk ( 1 9 8 2 ) . Compounds of a categorial type not existing in normal E n g l i s h , such as VA compounds, would be thus ruled out because there is no rule of the f o r m A -» V A. Compounds of this type, h o w e v e r , are f o u n d f r e q u e n t l y in the poetry under consideration; compare the f o l l o w i n g examples: (1-25)

sag-heavy [ V A ] (MT 5 5 ) , hover-still W\} ( + R 42)

In the v i e w under discussion such examples must be taken as evidence for the assumption that PL has a d d i t i o n a l base rules of the f o r m A ·» V A. Thus one aspect of the r e l a t i o n s h i p of PL to SL would be captured by postulating a d d i t i o n a l base rules. But 25

this suggestion has certain i m m e d i a t e l y obvious disadvantages. F i r s t l y , f r o m the point of v i e w of the grammar of SL, p o s t u l a t i n g restricted base rules is not a very satisfactory suggestion, as it cannot be accounted for in any p r i n c i p l e d w a y . From the point of view of the relationship of PL to SL, this would mean that the grammar of PL is d i f f e r e n t f r o m that of SL at a very "deep" l e v e l , which does not c o n f o r m to our basic i n t u i t i o n that PL rules w i l l tend not to a f f e c t basic, i.e. " l o w e r - l e v e l " rules of the grammar. Furthermore, it is hard to see how poetic rules could work to increase the number of base rules. Whereas an interaction can be visualised as working such that constraints on overgenerated forms are l i f t e d and that thus forms generated in one part of the grammar, but u s u a l l y constrained elsewhere, are a l l o w e d , it is d i f f i c u l t to visualise an interaction w h i c h results in forms which are never generated at all somehow becoming possible. F i n a l l y , t h i s suggestion lacks a general a p p l i c a t i o n . If there were to exist a r u l e as s p e c i f i c as "in certain types of poetry, there is an a d d i t i o n a l base r u l e in the grammar of the f o r m A -» V A " , w h i c h alone on account of its s p e c i f i c i t y does not sound l i k e l y (- one w o u l d probably not wish to suggest that such is the nature of poetic principles -), then, though we could in descriptive terms account for the compounds in (1-25) above, we should lose an important generalisation about other types of constructions. In order to see this, consider examples such as the f o l l o w i n g : (1-26)

wolf-darJc

[ N A ] (GH 1 7 ) , afternoon-suJbtle

[ N A ] (TO 16)

or the f o l l o w i n g : (1-27)

the sun is a diamond

We should l i k e to m a i n t a i n that something involved in the interpretation of (1-27) is also involved in the interpretation of ( 1 - 2 6 ) . We cannot account for this f a c t with the solution proposed above. But if we can posit poetic p r i n c i p l e s of a more general nature, w h i c h work at a higher l e v e l , then these can be shown to apply ( o p t i o n a l l y ) to many areas of PL and to have, in interaction w i t h other p r i n c i p l e s , several possible results, of w h i c h the presence of a certain type of VA compound is only one possible result. However, such an approach is only f e a s i b l e if we assume that the grammar of SL does not have such arbitrary restrictions as a closed set of base rules in which only certain categories are involved. In view of these considerations, we shall assume that the base rules of the grammar, w h i c h are responsible for the generation of 57 We are here using the designation "modular grammar" to mean two things: f i r s t l y , that the grammar represents one module of a modular language system and, secondly, that the grammar itself is a modular system. 26

CO

syntactic structures and w o r d - f o r m a t i o n s , overgenerate. That is, the rule-scheme X ·* Υ Χ, which generates compounds, applies irrespective of the categories of X and Y . Unwanted f o r m s are ruled out at higher levels of the grammar by constraints. More w i l l be said about the exact nature of these constraints, which take the f o r m of both syntactic and semantic well-formedness conditions, as w e l l as pragmatic constraints, in Chapter 3, in w h i c h the grammar of compounding w i l l be discussed more f u l l y . This model of grammar and this view of the interaction of PL w i t h the grammar avoids the problems mentioned above and in a d d i t i o n has several other advantages. For one t h i n g , it explains why only certain types of d e v i a t i o n are possible in PL. If we were to accept such ad hoc measures as the extension of base rules in PL, then it w o u l d seem that there is no principled way to constrain the type of additions to the grammar of SL which PL could contain. Any extension of the grammar formulated to account for PL would thus have only descriptive adequacy; it would not e x p l a i n how certain changes in the grammar of SL are possible and others not. And it certainly seems that there are some constraints on the f o r m of compounds w h i c h apply both to SL and to PL. The f o l l o w i n g compounds, w h i c h are ungrammatical in SL, would probably be impossible in PL, too: (l-28)a.

blue-sJcy. an adjective with the meaning "as blue as the sky";

b.

pearl-yellow, a substantive w h i c h is y e l l o w " ;

w i t h the meaning "a pearl

c.

letter-under, a preposition w i t h the meaning "under, as a letter is under";

d.

JbooJi-tatile, a substantive w i t h the meaning "there is a book on the table".

In a system l i k e the one presented here, the non-existence of these (and other) compound-types is not merely coincidental, but results f r o m particular characteristics of the poetic principles. If we assume that poetic p r i n c i p l e s do not a f f e c t the headp r i n c i p l e ( i . e . , that the head or second member of a compound and the compound itself are of the same category), then this explains why compounds l i k e those in ( l - 2 8 a ) and ( l - 2 8 b ) are not acceptable. Examples ( l - 2 8 c ) and ( l - 2 8 d ) are ruled out because there is no poetic p r i n c i p l e w h i c h renders non-applicable the 58 See also Lieber (1980, 1 9 8 3 ) , Toman ( 1 9 8 3 ) , Fanselow ( 1 9 8 5 ) , Boase-Beier & Toman (in p r i n t ) . 59 It thus conforms to the requirements of X-syntax ( f o r further discussion see Chapter 3). 60 For example constraints on the relation between X and Υ which would ensure that YX is an interpretable compound (see Chapter 3 for discussion). 27

r e s t r i c t i o n s on the meanings a p a r t i c u l a r category may h a v e . * In PL, just as in SL, it is not possible f o r a preposition to be m o d i f i e d nor for a substantive to express a p r o p o s i t i o n . In a purely descriptive system, it would be d i f f i c u l t to account for the non-existence of these compound-types in a p r i n c i p l e d way and it w o u l d be impossible to view ( l - 2 8 a ) and ( l - 2 8 b ) on the one hand, and ( l - 2 8 c ) and ( l - 2 8 d ) on the other, as the result of a restriction which is in each case a p p l i c a b l e to both. A second advantage of a system along the lines proposed is that we feel that the idea of PL as having b a s i c a l l y the same grammar as SL but a grammar w h i c h is less constrained, whereby the absence of constraint at certain points is a p r i n c i p l e d matter governed by s t r i c t l y f o r m u l a b l e p r i n c i p l e s , corresponds exactly to the c r i t i c ' s , the l i n g u i s t ' s , the p o e t ' s and the reader of poetry's i n t u i t i o n about what it is that makes poetic language d i f f e r e n t f r o m standard language - a p r i n c i p l e d absence of particular constraints ( c f . L e v i n (1971a: 2 0 ) ) . We have said above that the apparent c o n f l i c t in the two views of the r e l a t i o n s h i p of PL to SL, views w h i c h both have a long and respectable t r a d i t i o n , namely the v i e w that the poet has no resources not a v a i l a b l e to the ordinary speaker, and the view that he has, can be at least partly resolved if we accept a view of PL in w h i c h there are a d d i t i o n a l resources to be used in poetic language, but in w h i c h these can be related to the grammar in a principled way and are not s i m p l y arbitrary. The foregoing discussion has now enabled us to express these two aspects of the relationship of PL to SL more p r e c i s e l y . Poetic language does not have at its disposal a d d i t i o n a l grammatical means, nor does it have less than the standard language. In speaking of PL as an extension to the grammar of SL we do not mean an extension in the f o r m of a list of additional or of n o n - a p p l i c a b l e p r i n c i p l e s . What we mean is a set of poetic p r i n c i p l e s , which render inapplicable certain restrictions in the grammar. Thus both schools of thought have, up to a p o i n t , been r i g h t . They have merely emphasised d i f f e r e n t aspects of the r e l a t i o n s h i p of PL to SL. And the v i e w , expressed by Ching et a l . ( 1 9 8 0 ) , among others, that there is no actual d i f f e r e n c e in the grammar rules f o r PL, but only in the way the rules are m a n i p u l a t e d , can also be e x p l a i n e d in the light of this approach. It rests, one could suggest, on their i n t u i t i v e recognition of the fact that grammatical principles are not simply dropped or added to in PL, but their output is d i f f e r e n t and less constrained because of the interaction of poetic p r i n c i p l e s at various levels. The exact nature of the p r i n c i p l e s of PL w h i c h interact w i t h the grammar of SL w i l l be explained in Chapter 4; s u f f i c e it to r e c a l l here that these principles w i l l be taken to be general in nature, that is, not s p e c i f i c to compounding, and not s p e c i f i c to one type of poetry, one age or even, u s u a l l y , one p a r t i c u l a r language. T h i s 61 See Fanselow ( 1 9 8 5 ) mention of general 62 But remember that, principles may w e l l grammar) in SL.

and Boase-Beier & Toman constraints on the meaning as observed above, these have counterparts ( w i t h i n

28

(in p r i n t ) for of categories. non-grammatical or outside the

is not to say that some p o e t i c p r i n c i p l e s m i g h t not tend to be implemented more by writers o£ one type of poetry than by w r i t e r s of another, and it is not to r u l e out the f a c t that there may exist some p e r i p h e r a l p r i n c i p l e s w h i c h are o n l y used in a s p e c i f i c l a n g u a g e , in a s p e c i f i c era, in a s p e c i f i c type of poetry or by a s p e c i f i c poet. But these are of less interest to us here. The p r i n c i p l e s we s h a l l be considering here w i l l be those w h i c h a p p l y to contemporary E n g l i s h poetry but we shall assume that they h a v e , on the w h o l e , universal s i g n i f i c a n c e in the language of poetry. In p r o v i d i n g evidence of the general nature of these p o e t i c p r i n c i p l e s we shall f r e q u e n t l y have recourse to p r i n c i p l e s of classical r h e t o r i c a l s t y l i s t i c s . H o w e v e r , in m e n t i o n i n g r h e t o r i c a l terms corresponding to or connected w i t h some of the principles of poetic language we propose, we are not to be taken as suggesting e i t h e r that that part of r h e t o r i c a l theory ( e s p e c i a l l y understood in its classical sense) concerned w i t h style is to be equated w i t h the study of style as s t y l i s t i c s or as part of l i t e r a r y s t u d i e s , nor that r h e t o r i c has o n l y r e l e v a n c e for poetic as opposed to standard language, nor that the central concern of rhetoric has ever n e c e s s a r i l y been w i t h s t y l e in a l i n g u i s t i c sense. There w i l l o b v i o u s l y be many areas in w h i c h a r h e t o r i c a l and a p o e t i c theory of s t y l e w i l l part company and t h e r e w i l l be areas of p o e t i c style w h i c h have l i t t l e t o d o w i t h r h e t o r i c . B u t there w i l l a l s o be areas in w h i c h the two o v e r l a p , as the use of r h e t o r i c a l terms in l i t e r a r y c r i t i c i s m , ancient and m o d e r n , i n d i c a t e s . Our concern here is to show the u n i v e r s a l i t y of the p r i n c i p l e s we propose for poetic language by s h o w i n g that they are p r i n c i p l e s wJ\i.ch have t r a d i t i o n a l l y p l a y e d a role in the discussion of s t y l e . It is not the p o e t i c p r i n c i p l e s , t h e n , w h i c h are new in t h i s study (- i n d e e d , we hope to show that t h i s is just what t h e y are not -) but the attempt to account for all p o e t i c d e v i a t i o n in a p a r t i c u l a r d o m a i n of the grammar on the basis of the i n t e r a c t i o n of a s m a l l number of such p r i n c i p l e s . A word is in order here about the status of these p o e t i c p r i n c i p l e s . W e a r e assuming that t h e i r exact f o r m ( w h a t e x a c t l y constitutes i c o n i c i t y , what is the f o r m of m e t a p h o r ) can be constructed. But t h i s is not to be t a k e n as m e a n i n g a poet goes through e x a c t l y these processes in order to w r i t e i c o n i c a l l y or to construct a m e t a p h o r . The p o e t i c p r i n c i p l e s d e f i n e a model of poetic competence, w h i c h is used by the s t y l i s t i c i e n in order to 63 For a very d e t a i l e d d i s c u s s i o n of the d i f f e r e n t aspects of r h e t o r i c and of its r e l a t i o n s h i p to logic and poetics, see Howell (1975). 64 In f a c t , one can assume that it is p a r t l y c o n s i d e r a t i o n s of t h i s n a t u r e w h i c h have prompted a number of modern studies based on classical r h e t o r i c a l s t y l i s t i c s , such as Leech ( 1 9 6 6 ) , Groupe μ ( 1 9 7 0 ) , S c h o f e r 8. R i c e ( 1 9 7 7 ) , R i c e ε. S c h o f e r ( 1 9 8 1 ) . C l e a r l y these c r i t i c s i m p l i c i t l y r e j e c t t h e r o m a n t i c h o s t i l i t y to the idea of the i m p o r t a n c e of r h e t o r i c for p o e t r y ; we endorse t h i s r e j e c t i o n . 29

r e f l e c t upon the nature of that competence, by the poet, perhaps, in order to correct what seems not to work very w e l l in his poems, in textbooks, in order to teach "creative w r i t i n g " and the analysis of poetry, and by the reader, sometimes, in order to understand a poem. In this sense we view a model of poetic competence in a way s i m i l a r to that in w h i c h models of grammatical competence are to be v i e w e d .

30

IS POETIC LANGUAGE FUNCTIONALLY MOTIVATED?

Having established that there exists such an entity as " p o e t i c l a n g u a g e " , one of the central questions in its study m i g h t appear to be the question "why is p o e t i c language d i f f e r e n t from standard language"? There are in p r i n c i p l e two possible responses to t h i s q u e s t i o n . The f i r s t lies in a d e n i a l of the v a l i d i t y of the question i t s e l f : the nature of PL is not f u n c t i o n a l l y d e t e r m i n e d , so there is no sense in asking why it is as it is. It is d i f f e r e n t "because" it is by nature d i f f e r e n t . The second response is based on the assumption that p o e t i c language is f u n c t i o n a l l y m o t i v a t e d , and takes the f o r m of the r e p l y that i n PL there are c e r t a i n ends to be a c h i e v e d find the poet c o n s i d e r c w h i c h means he can e m p l o y to a c h i e v e t h e m . In theories of standard language, and e s p e c i a l l y in the t h e o r y of g e n e r a t i v e g r a m m a r , w i t h w h i c h we are concerned in t h i s s t u d y , there of course arises e x a c t l y the same q u e s t i o n . A v i e w of standard language in w h i c h the presence and nature of g r a m m a t i c a l structures are e x p l a i n e d on t e l e o l o g i c a l grounds is u s u a l l y rejected by the generative g r a m m a r i a n , who f a v o u r s a n o n - f u n c t i o n a l i s t theory based on the innateness of l i n g u i s t i c u n i v e r s a l e . These two views are not r e c o n c i l a b l e w i t h one a n o t h e r i n s o f a r as it is not possible to e x p l a i n the same p h e n o m e n o n using both approaches. However, it is possible - and t h i s has o f t e n been attempted - to m a i n t a i n that the object of study does not represent a homogeneous w h o l e , i.e., that linguistic competence is not one p h e n o m e n o n , but several. Thus, c e r t a i n of its aspects may be g i v e n f u n c t i o n a l e x p l a n a t i o n s and others n o n f u n c t i o n a l explanations. This w o u l d in theory be possible f o r the area of PL. One m i g h t m a i n t a i n that its g r a m m a r , b e i n g i d e n t i c a l to that of SL, is i n n a t e and n o n - f u n c t i o n a l , whereas t h e special p o e t i c p r i n c i p l e s a r e f u n c t i o n a l l y m o t i v a t e d .

The v i e w that poetic language is f u n c t i o n a l l y d e t e r m i n e d has had many adherents, n o t a b l y the Prague C i r c l e l i n g u i s t s and also modern l i n g u i s t s such as H a l l i d a y . See, for e x a m p l e , the v i e w s expressed i n H a l l i d a y ( 1 9 7 1 ) . Such v i e w s have been suggested for certain areas of g e n e r a t i v e grammar. See, f o r e x a m p l e , Kuno ( 1 9 7 6 ) . T h i s was e s s e n t i a l l y the v i e w put f o r w a r d in Boase-Beier ( 1 9 8 4 ) , w i t h w h i c h I shall here take issue.

31

On the f a c e of it, it seems as though a f u n c t i o n a l e x p l a n a t i o n of these poetic language p r i n c i p l e s might be the right one, even i f ( o r especially i f ) o n e r e j e c t s f u n c t i o n a l e x p l a n a t i o n s f o r standard language. There are several reasons for t h i s . F i r s t l y , if PL is d i f f e r e n t in nature f r o m SL in any case, why should not i t , and e s p e c i a l l y that part of it which constitutes the d i f f e r e n c e , be susceptible to a d i f f e r e n t e x p l a n a t i o n as regards the f u n c t i o n a l i t y question? S e c o n d l y , poetic language appears in some sense to be "constructed", i . e . a r t i f i c i a l rather than natural, it seems that the use of poetic language can be learned up to a p o i n t , that one can perhaps learn to w r i t e p o e t i c a l l y , as one can learn to analyse poetry. T h i r d l y , it seems that poetry can be in part analysed using the categories of r h e t o r i c , and t r a d i t i o n a l l y rhetoric is regarded as the art of persuasive discourse and was in e a r l i e r t i m e s learned as a f u n c t i o n a l l y based d i s c i p l i n e . But perhaps the most a p p a r e n t l y c o n v i n c i n g reason for b e l i e v i n g that poetic language structures are f u n c t i o n a l l y motivated is that it seems r e l a t i v e l y easy to group poetic f e a t u r e s of language a c c o r d i n g to certain f u n c t i o n s w h i c h they appear to f u l f i l . Thus, for e x a m p l e , one m i g h t argue that poetic language, as d i s t i n c t f r o m standard l a n g u a g e , is m o t i v a t e d by three basic f u n c t i o n s , namely to convey as much as possible as clearly as possible and in as constrained a f o r m as possible. To the f i r s t " f u n c t i o n " , conveying as much i n f o r m a t i o n as p o s s i b l e , one might assign literary techniques such as a m b i g u i t y and m u l t i p l i c i t y of m e a n i n g , metonymy and synecdoche (on the basis that if a word is used to stand for what it is associated w i t h then there w i l l be present the meanings and connotations both of the word stated and the word it is meant to stand f o r ) , and poetic i n n o v a t i o n (on the basis that a new word or phrase conveys more and is open to more associations than a l e x i c a l i s e d one, w h i c h has a more or less f i x e d f r a m e w o r k of a s s o c i a t i o n s ) . To the second " f u n c t i o n " , that of c l a r i t y , one m i g h t assign metaphor (as it can render the u n f a m i l i a r f a m i l i a r ) , and such techniques as onomatopoeia (- murmuring of innumerable bees does not just describe the sounds the bees m a k e , but re-enacts them - ) , p a r a l l e l i s m and r e p e t i t i o n ( d r u m m i n g the point home, so to speak) and iconic syntax' (speaking of a passive person in terms of passive verbs makes the c h a r a c t e r i s t i c of passivity more o b v i o u s , for e x a m p l e ) . To the third f u n c t i o n , that of f o r m a l constraint, one might assign such poetic f e a t u r e s as rhyme and metre. This is of course only intended to be a rough o u t l i n e of what such a " f u n c t i o n a l scheme" could look l i k e . One could of course argue against this p a r t i c u l a r scheme by s a y i n g , for e x a m p l e , that a l l i t e r a t i o n can be a purely f o r m a l c o n s t r a i n t , that metaphor hinders, rather than promotes, c l a r i t y , and that metre adds to c l a r i t y if it mirrors speech or m o v e m e n t , and so on. But these would not be arguments against f u n c t i o n a l e x p l a n a t i o n s as such. It is clear that one of the main reasons why such e x p l a n a t i o n s

4 5

Tennyson, "The Princess" ( 1 8 9 4 : 2 1 3 ) . See Freeman ( 1 9 7 6 , 1981b) for demonstrations of iconic syntax. 32

the

use of

seem so attractive in the study of PL is that f o r m u l a t e d in a c o n v i n c i n g and e n l i g h t e n i n g m a n n e r .

they

can be

H o w e v e r , be that as it m a y , we shall here plead for a nonf u n c t i o n a l v i e w of poetic language, at least in the strict t e l e o l o g i c a l sense of the term " f u n c t i o n a l " . There are several important reasons for this v i e w w h i c h suggest that the arguments just given in f a v o u r of a f u n c t i o n a l approach are not sound. One reason, w h i c h constitutes a r e f u t a t i o n of the f o u r t h argument, is that it is by no means clear just what " f u n c t i o n s " poetry is supposed to f u l f i l . It is not clear that it should convey as much i n f o r m a t i o n as possible; one m i g h t argue it should convey as l i t t l e as possible in order to make the reader " w o r k " . It is not essential that poetry be c l e a r ; many w r i t e r s and c r i t i c s have favoured the v i e w that it should be obscure. And so on. A f u r t h e r reason is t h a t , even if we could pin down the f u n c t i o n s of poetry, in order to e x a m i n e the l o g i c a l consequence of the f u n c t i o n a l argument we should then have to ask what f u n c t i o n s these f u n c t i o n s serve. Why do texts in PL have to provide more (or less) i n f o r m a t i o n than texts in SL? Why does poetry have to be e s p e c i a l l y c l e a r , or especially obscure, to s t r i k e the reader w i t h the f o r c e of immediate experience, or to present a f o r m a l abstraction of such experience? Above a l l , why should it have to be constrained, packed into lines w i t h a d e f i n i t e f o r m and rhythm (and even modern, " i n f o r m a l " poetry has an outward f o r m very d i f f e r e n t f r o m that of prose)? The obvious answer to these questions is: "Because that is what poetry is l i k e " . But then the f u n c t i o n a l e x p l a n a t i o n has turned out to be a rather h a l f - h e a r t e d one, w h i c h stops short of being r e a l l y i n t e r e s t i n g . Perhaps an even stronger argument that could be brought against f u n c t i o n a l explanations is that, even if the "aims" of poetry are c l e a r , it is by no means clear just why certain poetic devices should be used to achieve certain a i m s . To take up the example g i v e n above, there is no guarantee, at least if we accept f u n c t i o n a l e x p l a n a t i o n s , that the best way of m a k i n g a person seem passive by nature is to characterise him w i t h a large number of passive verbs. Perhaps it w o u l d be even better to s i m p l y state c l e a r l y "X was a very passive person". This would not be p o e t i c , one m i g h t say. But if not, then surely we have o f f e r e d support for the v i e w that "poeticalness" is an inherent f e a t u r e of a certain type of language and not a result of its a b i l i t y to f u l f i l certain f u n c t i o n s . There must be a source other than a purely functional one for the use of passive verbs to characterise a passive person. There must be a poetical p r i n c i p l e w h i c h says, in e f f e c t : "This is the way to do it". Yet another argument against a f u n c t i o n a l view is t h a t , if the nature of PL is d e t e r m i n e d by the f u n c t i o n s it f u l f i l s , then we

6

For a discussion of obscurity in l i t e r a t u r e , and d i f f e r i n g views of critics about obscurity, see Peyre (1967: Chap. V). 33

have no p r i n c i p l e d way of e x p l a i n i n g why d i f f e r e n t means are used in SL and PL to meet the same f u n c t i o n . If a speaker of SL wanted to express a certain person's p a s s i v i t y , he would say "X is a passive person", but the poet would (or c o u l d ) , for example, use a number of passive verbs to characterise h i m . Why does SL employ one means and PL another, in order to a c h i e v e the same thing? Of course, one could answer this question by saying that SL is not f u n c t i o n a l l y determined whereas PL is and that PL uses the best ( i . e . most f u n c t i o n a l ) means. But, as we have seen, we have no way of k n o w i n g that this is in f a c t the case. The answer must be, then - if we wish o n l y to g i v e e x p l a n a t i o n s w h i c h can be adequately supported - that poetic language as such is not f u n c t i o n a l l y m o t i v a t e d . The poetic features we have mentioned above might well typically result in certain e f f e c t s - though t h i s w i l l o f t e n be a c o n v e n t i o n a l , rather than a n a t u r a l , r e l a t i o n s h i p - and w i l l o f t e n be used c o n s c i o u s l y by a poet (or even sometimes by a speaker of SL) to create e x a c t l y these poetic e f f e c t s . But t h i s belongs to the aspect of p e r f o r m a n c e and has n o t h i n g whatever to do w i t h the question of l i t e r a r y competence and the inherent nature of l i t e r a r y language. It is not the same as saying that the type of poetic features w h i c h exist in PL exist because they result in certain e f f e c t s . Poetic f e a t u r e s are s i m p l y inherent characteristics of a c e r t a i n type of l a n g u a g e , poetic language. We are thus r e j e c t i n g all statements of the type sometimes encountered, that the poet chooses to use, for e x a m p l e , i n n o v a t i o n , in order to f r e e h i m s e l f f r o m the bonds of language and to shape it to his own ends. Instead we shall say that a poet who w r i t e s poetry is " f r e e d f r o m the bonds of language" because he w r i t e s in poetic l a n g u a g e , and poetic language is by nature i n n o v a t o r y . There is no t e l e o l o g i c a l e x p l a n a t i o n for a l l i t e r a t i o n , i n n o v a t i o n , or w h a t e v e r , though a study of t h e i r e f f e c t s (and an i n d i v i d u a l p o e t ' s use of them for these e f f e c t s ) w i l l p l a y a very important part in a c r i t i c a l a n a l y s i s of a work of l i t e r a t u r e . T h i s essential d i s t i n c t i o n between the e f f e c t s of poetic features and e f f e c t s seen as f u n c t i o n s in a strict sense is important in r e j e c t i n g another argument - the second one above - in f a v o u r of a f u n c t i o n a l v i e w of PL. As we stated there, PL seems i n t u i t i v e l y to be constructed as opposed to " n a t u r a l l y occurring" standard language. But in view of the d i s t i n c t i o n between competence and p e r f o r m a n c e it can be seen that t h i s results f r o m the f a c t t h a t , in order to make a work of art using the poetic language in his competence, a poet has to work w i t h it; t h i s is the aspect of p e r f o r m a n c e . The p e r f o r m a n c e aspect w i l l probably be of greater importance for the study of PL than for the study of SL. But the " a r t i f i c i a l " nature of PL does not constitute grounds f o r considering i t f u n c t i o n a l l y m o t i v a t e d . There are many examples of poetry w h i c h support the view that PL cannot be f u n c t i o n a l l y m o t i v a t e d . Consider the phenomenon of "spontaneous" oral poetry, w h i c h is composed d u r i n g p e r f o r m a n c e It is important to realise that the term " f u n c t i o n " has o f t e n been used to mean " e f f e c t " , whereby the term is not intended to be invested w i t h t e l e o l o g i c a l s i g n i f i c a n c e . W i t h arguments based on this use of the term " f u n c t i o n a l " we have no q u a r r e l .

ο

yet conforms to strict f o r m a l criteria. Consider, too, the statements of many poets who have said they w r i t e in a "spontaneous" w a y . The f o l l o w i n g is a statement by Ted Hughes about the poems in Crow, Most of them appeared as I wrote them. They were u s u a l l y something of a shock to w r i t e . M o s t l y they wrote themselves q u i t e r a p i d l y [ . . . ] (Faas (1980: 2 0 7 ) ) If poetic language is f u n c t i o n a l l y d e t e r m i n e d , then we can e x p l a i n the f a c t that all poetry tends to use the same devices o n l y on the grounds that a poet consciously tries to use certain devices rather than others w h i c h would be t h e o r e t i c a l l y possible. But if poets w r i t e " u n c o n s c i o u s l y " , then the f a c t that t h e y , too, use just the same devices cannot be e x p l a i n e d in this w a y . Consider, too, the examples g i v e n in Chapter 1 of r o m a n t i c and modern poets who have said they wrote in everyday language, yet have not done so. They, too, have used the same devices as all other poets. And if we e x a m i n e c h i l d r e n ' s poetry, we f i n d that even poems by c h i l d r e n as young as n i n e or ten years old contain f i g u r e s such as metaphor and a l l i t e r a t i o n , w h i c h they cannot have learned consciously. F i n a l l y , consider the statement by K i p a r s k y , also quoted in Chapter 1, and representative of the i n t u i t i o n of many c r i t i c s and scholars of literary l a n g u a g e , that "no t r u l y new f o r m s have emerged" and that this state of a f f a i r s is a t t r i b u t a b l e to "the i n t r i n s i c nature of verbal art" (1973: 2 3 1 ) . All these observations suggest that PL is by nature characterisable by certain p r i n c i p l e s f r o m w h i c h , to be sure, the i n d i v i d u a l poet may deviate, but w h i c h he may neither ignore nor supplement. PL is, we shall assume, not learned: some people are poets, some are not. Learning to read poetry enables the reader to f o r m a r a t i o n a l r e c o n s t r u c t i o n of the poet's act but it does not enable him to w r i t e poetry. Poetic competence is thus p a r a l l e l to l i n g u i s t i c competence and there are no more grounds for r e l a t i n g the f o r m e r to f u n c t i o n a l e x p l a n a t i o n s than for the latter. As for the observation that r h e t o r i c , at least in the sense in w h i c h it is the art of using tropes and f i g u r e s e f f e c t i v e l y , is f u n c t i o n a l l y m o t i v a t e d , we m a i n t a i n that this does not constitute grounds for assuming poetic language to be functionally m o t i v a t e d . R h e t o r i c , and especially the f u n c t i o n a l aspect of r h e t o r i c , is an abstraction f r o m certain characteristics of p a r t i c u l a r types of language and the e f f e c t s commonly associated w i t h them (such as the language of poetry and the language of o r a t o r y ) , and as such it can be learned. But t h i s does not mean that the phenomena of w h i c h it is an abstraction are themselves functionally motivated. 8 9

See the entry for "Oral Poetry" in Preminger et a l . (1965: 591ff. ) See, for e x a m p l e , the c h i l d r e n ' s poems in Searle ( 1 9 7 2 ) .

35

In Chapter 1 we established that we did not wish to d e f i n e PL by place of occurrence. The discussion in this chapter should have made it clear that we do not wish to d e f i n e it by purpose. We shall assume that the o n l y way to d e f i n e PL is by its character of c o n t r o l l e d d e v i a t i o n .

36

COMPOUNDING IN ENGLISH

3.1.

The Place of Hord-For«ation in the Grauar of English

In the present work we are concerned with a s p e c i f i c area of word formation in poetic language, namely the area of compounding. Though there are poetic innovations in words formed by d e r i v a t i o n , by far the most productive p o e t i c w o r d - f o r m a t i o n process is that of compounding. Because, as we have established in Chapter 1, PL is best characterised as the output of an i n t e r a c t i o n of poetic p r i n c i p l e s w i t h the p r i n c i p l e s of the grammar, it is essential as a starting point in a discussion of poetic structures to give an o u t l i n e of the grammar for the domain under d i s c u s s i o n , grammar of compounding generates as output well-formed non-poetical compound words. Thus an e x a m i n a t i o n of poetic compounds w h i c h cannot be generated by this grammar, together w i t h an e x a m i n a t i o n of the grammar i t s e l f , w i l l show what a d d i t i o n a l poetic means have been employed and w i l l also i n d i c a t e the nature of the i n t e r a c t i o n between these means and the grammar i t s e l f . Before we consider the exact nature of those parts of the grammar generating compounds, we shall make some general comments about the place of w o r d - f o r m a t i o n in the grammar. One central question in recent w o r d - f o r m a t i o n studies is the question as to the nature of w o r d - f o r m a t i o n rules and their s i m i l a r i t y to or d i f f e r e n c e f r o m other rule-types in the grammar. Closely connected w i t h this is the question as to "where wordf o r m a t i o n takes p l a c e " . There are b a s i c a l l y two suggestions w h i c h can be put forward in t h i s c o n n e c t i o n . One is that wordf o r m a t i o n is not an autonomous component of the grammar and that p r i n c i p l e s at work in the syntax of the grammar are also at work in the area of w o r d - f o r m a t i o n . This was an important aspect of those early studies of w o r d - f o r m a t i o n w i t h i n the f r a m e w o r k of generative grammar w h i c h worked on the assumption that new words were generated by means of t r a n s f o r m a t i o n s (see, for e x a m p l e . Lees ( I 9 6 0 ) , Fräser ( 1 9 7 0 ) ) . I n contrast, i n those w o r d - f o r m a t i o n studies w h i c h were based on the views expressed in Chomsky ( 1 9 7 0 ) it was assumed that w o r d - f o r m a t i o n rules and syntactic r u l e s were 1

But as S e l k i r k ( 1 9 8 2 ) has pointed out, the s i m i l a r i t y between two types of r u l e does not necessarily i m p l y that they have the same l o c a t i o n . Nor does an assumption that they are d i f f e r e n t e n t a i l the assumption that they are located in d i f f e r e n t components. 37

quite d i f f e r e n t in nature and that w o r d - f o r m a t i o n was to be seen as an autonomous component of the grammar w h i c h was situated w i t h i n the l e x i c o n . These studies - among the best known are Siegel ( 1 9 7 4 ) , J a c k e n d o f f (1975, 1977), A r o n o f f ( 1 9 7 6 ) , Roeper & Siegel (1978) - are g e n e r a l l y r e f e r r e d to as " l e x i c a l i s t " studies of w o r d - f o r m a t i o n . Many of the most recent studies of w o r d - f o r m a t i o n , however, have not viewed the phenomenon of w o r d - f o r m a t i o n as the output of an autonomous l e x i c a l component of the grammar, but instead, w h i l e not assuming that words are formed by t r a n s f o r m a t i o n s , have stressed the s i m i l a r i t i e s between p r i n c i p l e s at work in wordf o r m a t i o n and p r i n c i p l e s of the syntax, semantics or pragmatics of sentences. They have shown that phenomena generally considered to f a l l under the heading of w o r d - f o r m a t i o n can be seen as the result of an interaction of such general p r i n c i p l e s . Among the more important of these studies are W i l l i a m s (1981a, 1981b), Selkirk ( 1 9 8 2 ) , Lieber ( 1 9 8 3 ) , Toman ( 1 9 8 3 ) , Fanselow (1985). The present study shares the same theoretical basis as those in the latter group. We shall assume that there is no autonomous " w o r d - f o r m a t i o n component" but that new words are formed in the lexicon - at least in part - according to p r i n c i p l e s w h i c h do not d i f f e r in nature f r o m those of sentence syntax, or from the semantics and pragmatics of sentences. E x i s t i n g words, whether s i m p l e or compound, are, we assume, listed in an inventory (the non-generative part of the l e x i c o n ) , w h i c h also contains the bound morphemes of the language. This represents an area w h i c h w i l l not be considered in detail here. As these words already exist in the l e x i c o n , it is not always clear how much they t e l l us about the processes of w o r d - f o r m a t i o n . It is w i t h new compounds w h i c h are formed as the output of p r i n c i p l e s a v a i l a b l e to w o r d - f o r m a t i o n that we shall be concerned here, and especially w i t h the d i f f e r e n c e between this output when u n a f f e c t e d by poetic p r i n c i p l e s ( i . e . the ad hoc compounds f o r m e d in SL) and when an i n t e r a c t i o n w i t h poetic rules has taken place ( i . e . ad hoc compounds formed in P L ) .

This has also been ( i m p l i c i t l y or e x p l i c i t l y ) an important aspect of t r a d i t i o n a l descriptive accounts of w o r d - f o r m a t i o n such as Bühler ( 1 9 3 4 ) or Marchand ( 1 9 6 9 ) . Whether one regards the rules for the f o r m a t i o n of compounds as part of the system of phrase-structure rules or as l e x i c a l rules w h i c h are f o r m a l l y s i m i l a r to these is not, we shall assume, of any central importance in a discussion of the nature of c o m p o u n d i n g . In this assumption we f o l l o w S e l k i r k (1982: 10). For a discussion of the notions " l e x i c a l i s e d " and "ad hoc", see Boase-Beier et a l . (1985: 6 f ) .

38

3.2. Possible Compounds Any explanatory study o£ the w o r d - f o r m a t i o n of a language is concerned w i t h the notion "possible w o r d " . That is, it attempts to show w h i c h products of w o r d - f o r m a t i o n are generated by the grammar and w h i c h are n o t , or, to put it d i f f e r e n t l y , w h i c h words are acceptable and w h i c h are not. In connection w i t h the topic of study here, there w i l l be a f u r t h e r concern, n a m e l y : what is the d i f f e r e n c e between possible complex words of SL and possible complex words of PL? The b r i e f account of the grammar of compounding in English in this Chapter (see 3 . 3 ) attempts to answer the question as to what is a possible compound word of SL. In Chapter 4 we shall look at compounds w h i c h , according to the grammar as sketched here, are not possible, but w h i c h occur in the poems under discussion. We shall argue t h a t , if the grammar does not generate t h e m , they must come f r o m somewhere else. And we shall try to establish where exactly they do come f r o m , that is, w h i c h a d d i t i o n a l poetic p r i n c i p l e s are required and in what way these interact w i t h the grammar to produce the f o r m s in question. R e c a l l that in Chapter 1 we described poetic language, in G r i n d e r & E l g i n ' s words, as the language of a poetic text l e f t over when we have excluded "gibberish on the one hand and ordinary discourse on the other" (1973: 1 6 9 ) . It is thus clear that we are concerned w i t h a t h r e e - f o l d d i v i s i o n of language, namely into: (i)

compounds generated by the grammar and hence possible both in SL and PL, i . e . compounds of "ordinary discourse" such as:

( 3 - 1 ) carpet-corner, milk-drinker, food-sales, (ii)

dance-hall',

compounds r e s u l t i n g f r o m an i n t e r a c t i o n of the grammar of SL w i t h poetic p r i n c i p l e s ; these compounds are impossible (or at least of m a r g i n a l status) in SL but possible ( a n d " n o r m a l " ) in PL:

( 3 - 2 ) grin-long [ N A ] ( + R 1 0 4 ) , flouer-lovely-false [[NA3A] ( ++ N M 2 1 ) , hover-search [ V V ] ( + SS 2 7 ) , horse-worship [ N V ] ( WNP 57); ( i i i ) compounds not generated by an interaction between the grammar of SL and poetic p r i n c i p l e s w h i c h w i l l t h e r e f o r e be impossible or of m a r g i n a l status in PL (and of course impossible in S L ) ; compare the examples in ( 1 - 2 8 ) , repeated here: (3-3) blue-sky ( a s A ) , pearl-yellow ( a s N ) , letterunder, book-table ( m e a n i n g "there is a book on the t a b l e " ) . It is important to r e a l i s e , in connection w i t h group ( i i i ) , that we do not wish to m a i n t a i n that such compounds are in f a c t a b s o l u t e l y impossible in PL. What we do m a i n t a i n is that their 39

f o r m a t i o n requires a d d i t i o n a l means w h i c h do not belong to the inventory of "normal" poetic p r i n c i p l e s and t h e i r interpretation thus requires knowledge of these extra means. These compounds hence remain outside the domain of poetic language in the sense in w h i c h we shall understand it, just as the compounds of group ( i i ) remain outside the d o m a i n of SL, although they m i g h t , under exceptional circumstances, such as conscious use of PL p r i n c i p l e s , be f o u n d there. One basic requirement of a generative grammar is that it generate all possible structures of a language and o n l y these. A central issue in connection w i t h the concept "possible compound" is thus the question as to how the grammar can generate enough types of words without a u t o m a t i c a l l y generating too m a n y . As we have seen in ( i i ) and ( i i i ) above, there are compounds w h i c h the grammar must not generate. There are p r i n c i p a l l y two ways of meeting this r e q u i r e m e n t , as we have m e n t i o n e d in Chapter 1.2. We decided there that the only p r i n c i p a l l y acceptable view is that there is no restriction on the categories combined by the base r u l e of the grammar, but that restrictions, w h i c h do not have only s p e c i f i c a p p l i c a t i o n to the domain of compounding, apply at other levels.

3.3.

The Grammar of Compounding

3 . 3 . 1 . Introductory Remarks As we have stated above, there seems good reason to v i e w the area of compounding not as an autonomous component of the grammar but as the result of an interaction among certain general p r i n c i p l e s . To account for compounding e x p l i c i t l y we shall have recourse m a i n l y to principles of the f o l l o w i n g types: phrasestructure p r i n c i p l e s ( b a s e - r u l e s ) , syntactic p r i n c i p l e s , semantic principles. We have stated, also, that we shall assume compounding takes place in a generative part of the l e x i c o n . The elements of the non-generative part of the l e x i c o n (words and bound morphemes) are s p e c i f i e d as to the grammatical category to w h i c h they belong and as to w h i c h other categories they subcategorise f o r . They also carry i n f o r m a t i o n as to the category and semantic role of the arguments they take (these two aspects of the arguments an element takes w i l l be r e f e r r e d to as the argument-structure of the e l e m e n t ) . F o l l o w i n g a number of recent studies on the structure of words, we shall assume that the rule-schema necessary for the generation of complex words is c a t e g o r i c a l l y unrestricted and S

See, for e x a m p l e . Toman ( 1 9 8 3 ) , Fanselow ( 1 9 8 5 ) , Boase-Beier & Toman ( i n p r i n t ) . 40

that it has the f o r m :

(3-4)

X

-»

Υ Ζ

This rule-schema c o n f o r m s to a restriction w h i c h ensures that the category (and the other f e a t u r e s ) of one of its elements, namely the head, are i d e n t i c a l w i t h the category and features of the whole word. This r e s t r i c t i o n on the form of the rule generating complex words is to be seen as an a p p l i c a t i o n of the X-Theory of syntax, in w h i c h the phrase-structure rules have in general the form χ -> . . . x n . . . , to the domain of w o r d - f o r m a t i o n . It is a r e s t r i c t i o n w h i c h ensures that, for e x a m p l e , a compound such as blackbird is a n o u n , that seller is also of the category Ν just as is its d e r i v a t i o n a l a f f i x , that, for example, in German, Wohnung ( f l a t ) and Arztvohnung ( d o c t o r ' s f l a t ) are both f e m i n i n e (and the latter is not m a s c u l i n e l i k e Arzt) and so on. It should be noted that in word-structure the head of a construction can not be one level lower (X n ) than the construction itself ( X n ) , as both are at the w o r d - l e v e l . Thus the head in word-syntax cannot be d e f i n e d as that element which is of the same category as the construction itself but is one level l o w e r . We shall assume that W i l l i a m s ' f o r m u l a t i o n of the Righthand Head R u l e ( R H R ) is the correct one for English: (3-5)

In morphology we d e f i n e the head of a m o r p h o l o g i c a l l y complex word to be the right-hand member of that word. ( W i l l i a m s (1981b: 2 4 8 ) )

In this sense the rule-schema ( 3 - 4 ) is always to be interpreted as X ·» Υ X. As regards structures w h i c h might appear to be l e f t - h e a d e d compounds, such as verb-particle-structures of the type eat out e t c . , it is in f a c t not clear that these actually are compounds at a l l . Although S e l k i r k (1982: 2 7 f . ) gives several grounds for regarding them as compounds, they seem to be u n p r o d u c t i v e as a type and in many cases to e x h i b i t rather i d i o s y n c r a t i c readings, w h i c h do not c o n f o r m to the usual type of interpretations compounds have ( c f . Lieber (1983: 2 5 5 ) ) . We thus tend towards the v i e w that structures of this type are not in f a c t compounds. See J a c k e n d o f f ( 1 9 7 7 ) for X-theory. For an e x p l i c i t account of the a p p l i c a t i o n of this theory in the domain of w o r d - f o r m a t i o n in G e r m a n , see Toman ( 1 9 8 3 ) . This rule is l a n g u a g e - s p e c i f i c . In other languages, e.g. French, there are left-headed compounds. A f u r t h e r problem is the status of i n f l e c t i o n a l a f f i x e s . There is no u n i f o r m i t y of o p i n i o n in the literature as to whether they are heads or not. W i l l i a m s (1981b) m a i n t a i n s they are heads, S e l k i r k ( 1 9 8 2 ) , that they are not. Because the results of the present study are compatible w i t h either v i e w , we shall not f u r t h e r consider t h i s problem here. 41

The syntactic concept of "head" has a semantic correlate: the class of objects denoted by a structure is a subclass of the class of objects denoted by its head. That is, garden-flower is a s p e c i f i c type of f l o w e r , not a type of garden. As the rule-schema generating compounds is categorially u n r e s t r i c t e d , the insertion of any l e x i c a l category (i.e. Ν , ν , Α , Ρ ) for X and Υ is possible. This results in 16 possible combinations, -namely: (3-6)

(i)

NN

VN

AN PN

(ii)

NV

VV

AV

PV

(iii)

NA

VA

AA

PA

NP

VP

AP

PP

(iv)

Because the rule-schema conforms to the Righthand Head R u l e , the r e s u l t i n g compounds of group (i) w i l l be nouns, of group ( i i ) w i l l be verbs, of group ( i i i ) adjectives and group ( i v ) prepositions. However, an e x a m i n a t i o n of i n t u i t i v e l y possible compounds in English, enables us to observe that certain of these types do not e x i s t . Just to give a few e x a m p l e s , it seems f a i r l y clear that compounds of the type A P , such as: (3-7)

It

£

green-out,

oJd-in

do not exist, just as do not compounds of the type V A , such as: (3-8)

%

try-nice,

&

consider-old.

But judgements among researchers into compounding vary as to w h i c h types of compounds are possible. For e x a m p l e , w h i l e S e l k i r k ( 1 9 8 2 ) , Lieber ( 1 9 8 3 ) and Boase-Beier & Toman (in p r i n t ) all r u l e out compounds of the type N P , VP and AP, and accept the compound types NN, V N , A N , PN, PV, NA and A A , there is no consensus of o p i n i o n about the r e m a i n i n g types, w h i c h are deemed acceptable by some of the researchers mentioned though i m p o s s i b l e by others. This could in part be because some f a c t s about compounds are l a n g u a g e - s p e c i f i c (Boase-Beier & Toman study German compounds, Lieber and S e l k i r k E n g l i s h compounds) but it also results p a r t l y f r o m d i f f e r e n t interpretations of the data i n v o l v e d . Disregarding these d i f f e r e n c e s , h o w e v e r , one point upon w h i c h all researchers 9

See also Marchand ( 1 9 6 9 : 4 0 ) . 42

are in a g r e e m e n t , is that there are certain types of compounds w h i c h do not o c c u r . As we have said above, if we assume the compounding rule-schema is c a t e g o r i a l l y unrestricted, then we must f i n d what sort of p r i n c i p l e s interact to restrict the output of the grammar at a d i f f e r e n t l e v e l . The most important areas in w h i c h the g r a m m a r , can have auch a c o n s t r a i n i n g e f f e c t on the output of compounding in the base are the p e r c o l a t i o n of f e a t u r e s (also called " i n h e r i t a n c e " ) and the assignment of t h e m a t i c r o l e s .

3 . 3 . 2 . Percolation The f e a t u r e p e r c o l a t i o n conventions are conventions w h i c h ensure that f e a t u r e s of the elements of a c o m p l e x structure (either a syntactic structure or a complex word). are " t r a n s f e r r e d " , so to speak, to the complex word i t s e l f . This can be illustrated as f o l l o w s for the German compound fiasserhlume:

N[+f]

(3-9)

(Here and h e n c e f o r t h morphological features, structures. )

^^ represents the p e r c o l a t i o n of ^ the percolation of argument-

It should be noted that we regard the process of percolation as per se f r e e f r o m c o n s t r a i n t . The f a c t that ffasserJblume is f e m i n i n e and n o t , l i k e Wasser, neuter, is not a r e f l e c t i o n of the restricted nature of percolation but of the r e s t r i c t i v e nature of the head-rule in c o n j u n c t i o n w i t h p e r c o l a t i o n . In a d d i t i o n to the percolation of such features as category name, n u m b e r , gender and case, percolation conventions also have the e f f e c t that argument-structures are i n h e r i t e d by a complex word f r o m its e l e m e n t s . An example to i l l u s t r a t e t h i s is: (3-10)

driver of the car

Here we must assume that the argument-structure of the verb drive is i n h e r i t e d by the substantive driver, f o r m e d f r o m drive + er, 10 S e l k i r k ( 1 9 8 2 ) does not make this assumption. 11 See, for e x a m p l e , S e l k i r k ( 1 9 8 2 ) , Lieber ( 1 9 8 3 ) , Toman (1983)

43

enabling the same r e l a t i o n to obtain between driver and car as in, f o r e x a m p l e , the phrase: (3-11)

he drives the car

In the case of driver, percolation has taken place not f r o m the head - er of the n o m i n a l isation driver but f r o m the non-head dri ve, As mentioned above, we are here assuming that the correct v i e w of percolation is that it is in p r i n c i p l e both f r e e and unrestricted in nature. Because the head-rule must a p p l y , however, percolation f r o m the right element of a complex word is necessary, for all those features that this element possesses. But other features then can (but need not) percolate f r o m the left-hand element. This means t h a t , in example ( 3 - 1 0 ) , the category features of -er w i l l percolate to driver but no argument-structure w i l l as - er does not possess any. Argumentstructure may percolate f r o m drive to a l l o w the driver of the car or it may n o t , to a l l o w the driver. P e r c o l a t i o n is not forced to take place because there is no r u l e corresponding to the Righthand Head. Rule for the l e f t branch. In this v i e w of percolation we f o l l o w to some extent both S e l k i r k (1982) and Toman ( 1 9 8 3 ) , i . e . , in our assumption that percolation f i r s t takes place on the right branch but any r e m a i n i n g features can percolate from the l e f t branch (assuming no c o n f l i c t arises w i t h features which have already percolated from the r i g h t ) . Features passed up from the non-head w i l l t y p i c a l l y be features of argument-structure. This is an aspect of percolation e n t i r e l y neglected by Lieber, w h o , in her account of compounding, a l l o w s only for the percolation of f e a t u r e s (whether they be category features or argument-structure features) from the r i g h t - h a n d element to .the whole compound. T h i s , however, gives wrong predictions for such structures as: (3-12)

the combiningr-process of semantic markers

because the argument-structure of combining cannot be passed on to the compound. L i e b e r ' s view is in fact p r o b l e m a t i c in a number of respects. F i r s t l y , she makes no r e f e r e n c e to the headp r i n c i p l e , and thus has to stipulate that feature percolation in compounds can only proceed f r o m the r i g h t - h a n d element. Secondly, the restriction is too strong, as example ( 3 - 1 2 ) shows. T h i r d l y , it f a i l s to a l l o w for a u n i f i e d treatment of compounding and derivation. We shall t h e r e f o r e assume that the correct f o r m u l a t i o n of percolation is the one given above. A f u r t h e r problem in L i e b e r ' s system is the restriction percolation of features to a node w i t h the same category as node from which they percolate. In other words, n o m i n a l i s a t i o n driver does not have the argument-structure of 44

of the the the

verb drive, i . e . percolation cannot proceed f r o m the V node to the N node d o m i n a t i n g it: ^ (3-13)

drive

er

This means that it is seen as a coincidence that the substantive driver has the same argument-structure as the verb drive. We shall assume that there is no such restriction upon percolation. O b v i o u s l y there are other d i f f e r e n c e s in the way in w h i c h p e r c o l a t i o n is dealt w i t h in these systems, but we have here mentioned only those w h i c h seem of greatest r e l e v a n c e . From t h i s very b r i e f discussion of the question of p e r c o l a t i o n , it is hoped that s u f f i c i e n t has been said to g i v e the reader a f a i r idea of the general issues i n v o l v e d . I n d i v i d u a l questions of p e r c o l a t i o n in compounds w i l l be dealt w i t h in section 3 . 4 , in w h i c h we s h a l l discuss the various types of compounds generated by the base.

3.3.3. The Realisation of Argument-Structures We turn now to the question of the r e a l i s a t i o n of argumentstructure in compounds. There are two important points to be made in t h i s c o n n e c t i o n . The f i r s t is t h a t , for the interpretation ot compounds such as those in ( 3 - 1 4 ) , we need to know something about the a r g u m e n t - s t r u c t u r e of the f u n c t i o n a l elements i n v o l v e d : (3-14)

fruit-sale, book-lover,

hovercraft

In other words, s o m e t h i n g i n v o l v e d in the i n t e r p r e t a t i o n of the phrase to sell fruit w i l l also be required to interpret fruitsale. The same goes for to love books and took-lover, a cratt hovers and hovercraft.

12 See Lieber (1983: 2 6 8 ) . We have used a somewhat d i f f e r e n t representation. However, as this has no consequences for the p o i n t under d i s c u s s i o n , we s h a l l not go into this question here. We shall s i m p l y assume that it makes sense to represent a f f i x e s as above. See also Toman (1983: S O f f . ) on t h i s p o i n t . 45

The second point to be made here is that the question of the realising of argument-structures can be seen, together w i t h the p r i n c i p l e of percolation o u t l i n e d above, as a constraint on the output of compounds generated by an unrestricted base-rule. L i e b e r , for e x a m p l e , argues that if we assume that argumentstructures must be r e a l i s e d , then we can account for the absence of such compounds as put-ύοχJay the same token as we can account for the absence of a phrase he put the box ( c f . Lieber (1983: 256)). Now though it c e r t a i n l y m i g h t be expected to be the case that the same e x p l a n a t i o n accounts for the absence of both structures, we shall assume here, in contrast to Lieber, that the assignment of t h e m a t i c roles ( θ - r o l e s ) , i . e . , the r e a l i s a t i o n of argumentstructures, is fijee in a p p l i c a t i o n , just as are other p r i n c i p l e s of the grammar. This a l l o w s us to f o r m a compound street-sale, for e x a m p l e , in w h i c h the object of the verb to sell is not realised and also need r^ot be r e a l i s e d c o m p o u n d - e x t e r n a l l y . That a corresponding phrase he sells in the street is not possible is, we shall assume, o w i n g to some a d d i t i o n a l p r i n c i p l e of sentence-syntax whic^ s i m p l y does not apply in words. And the f a c t that the word put-ύοχ is also i m p o s s i b l e is o w i n g to the fact that some words are so marked in the l e x i c o n that they require their argument-structures a l w a y s to be r e a l i s e d . This is the case for verbs l i k e put and a d j e c t i v e s l i k e fond. T h i s , t o o , is a p r i n c i p l e independent of θ-assignment as such. A p r i n c i p l e of r e a l i s a t i o n of a r g u m e n t - s t r u c t u r e together w i t h a p r i n c i p l e of p e r c o l a t i o n c a n n o t , t h e r e f o r e , be enough to account for the whole range of possible compounds. There w i l l still be many compounds f o r m e d w h i c h must be r u l e d out by other, independently motivated p r i n c i p l e s . It is in any case our v i e w that it is p r e f e r a b l e to assume that i n d i v i d u a l p r i n c i p l e s are themselves unrestricted in t h e i r a p p l i c a t i o n and that the pattern of possible structures w h i c h arises is e x p l a i n e d by the interaction of these p r i n c i p l e s . Having established that .θ-role a p p l i c a t i o n is f r e e in n a t u r e , there remain two f u r t h e r questions to be considered, n a m e l y what happens when θ-role assignment does not take p l a c e , and the question of the "direction" of fl-role a p p l i c a t i o n in a compound. To consider the latter question f i r s t , because it m i g h t seem i n t u i t i v e l y "odd" ( i . e . , not what one expects given the t y p i c a l r e l a t i o n between θ-role assignment and the head of a construction in sentence-syntax) that the head of a compound can be an argument of the non-head, one might propose that θ-role assignment in compounds can o n l y proceed f r o m the head to the non-head. T h i s would mean t h a t , for the largest group of 13 We are assuming that the r e a l i s a t i o n of argument structures i n v o l v e s the assignment of t h e m a t i c - r o l e s . S e l k i r k , for instance (1982: 30ff.), speaks of the r e a l i s i n g of g r a m m a t i c a l r e l a t i o n s rather than θ-roles.

compounds w h i c h appears to contradict compounds such as the f o l l o w i n g : ( 3 - 1 5 ) a . hovercraft, b. draubridge,

this

proposal,

i.e.

float-stone, sawmill, rattlesnake throu-stick, pushbutton, scatter-cushions

in w h i c h the head of the compound is the subject of the non-head ( 3 - 1 5 a ) , or the o b j e c t ( 3 - 1 5 b ) , a d d i t i o n a l explanations must be g i v e n , for example that the head in the (3-15a) compounds is not i n t e r p r e t e d as the subject of the verb, that the subject does not receive a θ-role in any case, or, for ( 3 - 1 5 b ) , that these compounds are e x c e p t i o n s , and so on. However, in Lieber ( 1 9 8 3 ) and Boase-Beier and Toman (in p r i n t ) , it is argued that θ-role assignment may proceed both f r o m the head and the non-head, and that yet more p r i n c i p l e s interact w i t h θ-role assignment to account for the absence of c e r t a i n types. T h i s is the p o s i t i o n we shall adopt h e r e . The exact consequences of t h i s p o s i t i o n , as of the assumption that θ-role assignment is f r e e , w i l l become clear in section 3 . 4 , where the various compound types are discussed. As regards the question of what happens when θ-assignment does not take p l a c e , most researchers have been in agreement that i f the non-argument-taking e l e m e n t of the compound is not an argument of the other, then it must be interpreted as an adverbial of t i m e or p l a c e , i . e . , as an argument for w h i c h a verb does not subcategorise. In the f o l l o w i n g cases, there is a contrast between examples ( 3 - 1 6 a ) , in w h i c h the compounds have a direct object in the n o n - h e a d - p o s i t i o n , and the corresponding (3l & b ) e x a m p l e s , in w h i c h they have an argument not subcategorised f o r ; for t h i s type of argument we s h a l l adopt the term "semantic a r g u m e n t " , used by Lieber (1983: 2 5 7 ) . ( 3 - 1 6 ) a . flouer-seller, lorry-driver, pain-healer b. street-seller, Saturday-driver,

faith healer

In order to contrast arguments of the type in (3-16a) w i t h semantic arguments, we shall on occasion r e f e r to the f o r m e r as "proper arguments". Proper arguments are arguments subcategorised for by a verb. It should be noticed that an account of these data along these l i n e s does not a c t u a l l y hinge in any way d i r e c t l y on the question of whether 0-role assignment is f r e e or n o t . Obviously an 14 T h i s is the e x p l a n a t i o n g i v e n compounds. 15 This is L i e b e r ' s e x p l a n a t i o n compounds (see Lieber 1 9 8 3 ) . 47

by S e l k i r k for

the

(1982)

for

these

presence

of

these

assumption that argument-structure must be realised cannot be taken to mean that the compounds in (3-16b) are i m p o s s i b l e . Their presence c o u l d , f o r e x a m p l e , be e x p l a i n e d by saying that the argument-structure of sell does not percolate to the next highest node, because p e r c o l a t i o n is only possible between nodes of the same category. Thus seller has no a r g u m e n t - s t r u c t u r e , so the compound street-seller is interpreted l i k e any other NN compound. This is the suggestion put f o r w a r d , for e x a m p l e , by L i e b e r (1983: 2 6 5 f f . ) . In such a system, the f a c t that flouer-seller means "someone who sells f l o w e r s " is accounted for by assigning it a d i f f e r e n t structure, n a m e l y : (3-17)

ilflover}^

lseJJlv

erf^

In this work we shall not take over L i e b e r ' s assumptions in this respect. We shall assume that p e r c o l a t i o n can proceed to a node of a d i f f e r e n t category. Thus seller, though it is a n o u n , has the same argument-structure as the verb sell. Because t h i s argument-structure can be r e a l i s e d , but does not n e c e s s a r i l y ha,ve to be, seller can occur both in compounds such as car-seller and in compounds such as street-seller. To put the case somewhat differently, while ignoring pragmatic considerations, the compound car-seller m i g h t mean either "someone who sells cars" or "someone who sells something in cars".

3.3.4. Stereotypes There is, h o w e v e r , yet a t h i r d type of i n t e r p r e t a t i o n possible for the compound car-seller. Again i g n o r i n g , for the sake of c l a r i t y , pragmatic considerations, we m i g h t (theoretically) interpret car-seller as m e a n i n g "a seller (of something) who comes to work by car". In t h i s case the correct interpretation depends upon the f a c t that there is a new r e l a t i o n i n v o l v e d . This is the r e l a t i o n "to drive" or "to come to w o r k " or w h a t e v e r , w h i c h might be called a stereotype-relation of the noun car*, in other words, it is a r e l a t i o n w h i c h is associated in the l e x i c o n w i t h the noun car. In this case the argument-structure of the verb sell plays no role at a l l , as the element car is n e i t h e r argument nor semantic argument of the verb sell. This use of what we have c a l l e d a stereotype-relation of a noun occuring in a compound plays a very large role in the interpretation of compounds, in p a r t i c u l a r of those consisting of two substantives. Because there is no r e l a t i o n a l element in compounds of t h i s t y p e , such as the compounds in (3-18) b e l o w , the r e l a t i o n e x i s t i n g between the two elements of the compound cannot be found without recourse to other than s t r i c t l y grammatical i n f o r m a t i o n , c f . : (3-18)

caJse Jcnife,

shoe cupboard, glue factory,

flower shop

We s h a l l argue here, in k e e p i n g w i t h assumptions we have made e l s e w h e r e and w i t h those of other researchers, that what is required to interpret a large m a j o r i t y of NN compounds is a stereotype-relation. The i n t e r p r e t a t i o n of a compound AB is possible o n l y if a s e m a n t i c reotesentaion of the f o r m " r e l a t i o n ( A , B ) " can be constructed. If the r e q u i r e d r e l a t i o n is not contained in the c o n t e x t of the c o m p o u n d , it w i l l in most cases be a stereotypical r e l a t i o n of one of the nouns in the compound. In other w o r d s , when we interpret cake knife as "a k n i f e w h i c h c u t s / f o r c u t t i n g cakes", we are using the r e l a t i o n " c u t " , w h i c h is associated w i t h the noun knife in the letter's lexical entry. 1 8 S i m i l a r l y shoe cupboard is i n t e r p r e t e d as "cupboard c o n t a i n i n g shoes", glue factory as " f a c t o r y w h i c h produces g l u e " , flover shop as "shop w h i c h sells f l o w e r s " by means of l e x i c a l - s e m a n t i c k n o w l e d g e about the nouns cupboard, factory and shop w h i c h we shall in f u t u r e r e f e r to as stereotypes. It w i l l be obvious that stereotypes do not by any means a l w a y s take the f o r m of r e l a t i o n s . I n d e e d , the p r o t o t y p i c case of a stereotype is a q u a l i t y , such as "red" w h i c h is the stereotype of blood, roses e t c . , " c o l d " , w h i c h is the stereotype of ice, snou, "being at an advanced stage" w h i c h is the stereotype of evening, autumn, old age, and so on. We s h a l l r e t u r n to the q u e s t i o n of one-place stereotypes later but at the moment it is w i t h t w o - p l a c e stereotypes w h i c h we are m a i n l y concerned as they provide the r e l a t i o n r e q u i r e d in the semantic representation of a n o n - r e l a t i o n a l compound. Just as for compounds w h i c h c o n t a i n a r e l a t i o n a l e l e m e n t , there w i l l be three types of i n t e r p r e t a t i o n for compounds w h i c h c o n t a i n a noun w i t h w h i c h a r e l a t i o n is stereotypically associated. These three types of i n t e r p r e t a t i o n are as f o l l o w s : ( 3 - 1 9 ) a . shoe-factory w i t h the m e a n i n g "a f a c t o r y w h i c h produces shoes b. vinter-factory w i t h the m e a n i n g "a f a c t o r y w h i c h produces s o m e t h i n g o n l y in the w i n t e r " 16 See, for example, Fanselow (1981), Boase-Beier (1984), F a n s e l o w ( 1 9 8 5 ) , Boase-Beier e t a l . ( 1 9 8 5 ) . 17 That is, there are v a r i o u s s e m a n t i c p r i n c i p l e s w h i c h make possible the c o n s t r u c t i o n of t h i s semantic r e p r e s e n t a t i o n (part o f t h e l e v e l u s u a l l y c a l l e d ' l o g i c a l f o r m ' , c f . , f o r e x a m p l e , Chomsky ( 1 9 8 0 : 1 4 3 ) . The semantic r e p r e s e n t a t i o n is then assigned an i n t e r p r e t a t i o n . 18 For a d i s c u s s i o n of the n a t u r e of stereotypes and t h e i r r e l a t i o n to the words of w h i c h they are stereotypes, see Putnam ( 1 9 7 5 ) , D a h l g r e n ( 1 9 7 8 ) , E i k m e y e r & R i e s e r ( 1 9 8 1 ) and Schwarze ( 1 9 8 2 ) . 19. It is not of course to be i m a g i n e d that the stereotype is an a c t u a l verb or a d j e c t i v e . It is s i m p l y a r e l a t i o n , whose semantic content can be c o n v e n i e n t l y represented as a p a r t i c u l a r verb or a d j e c t i v e . 49

c. forest-factory forest"

with

the

meaning

"a

factory

in

the

In e x a m p l e ( 3 - 1 9 a ) , the element of the compound w h i c h does not supply the stereotype-relation is an argument of the element w h i c h does. In ( 3 - 1 9 b ) , the element w h i c h does not supply the stereotype-relation q u a l i f i e s t e m p o r a l l y t h e r e l a t i o n "produce" provided by factory. In ( 3 - 1 9 c ) , a new r e l a t i o n "stand in, be located i n " , obtains between forest and factory. It is important to see that there is a vast d i f f e r e n c e between examples (b) and (c) in that in ( b ) , the e l e m e n t winter m o d i f i e s the stereotyper e l a t i o n "produce" whereas in (c) t h i s r e l a t i o n plays no role whatsoever. Forest-factory does not mean " f a c t o r y w h i c h only produces things in the f o r e s t " . We shall thus assume that the i n t e r p r e t a t i o n of an NN compound and also of a r e l a t i o n - c o n t a i n i n g compound in w h i c h the r e l a t i o n plays no r o l e , can be assigned by recourse to a stereotype-relation. Compounds l i k e that in ( 3 - 1 9 c ) , in w h i c h the stereotype-relation f a i l s to be realised are context-dependent. Stereotypes are of course not only of r e l e v a n c e for the i n t e r p r e t a t i o n of c o m p o u n d s . It has been shown that they play an important r o l e in l e x i c a l o r g a n i s a t i o n and in the interpretation of many types of l i n g u i s t i c u t t e r a n c e . In p a r t i c u l a r , the i n t e r p r e t a t i o n of u n d e r d e t e r m i n e d structures w i l l r e q u i r e the h e l p of stereotypes; consider the f o l l o w i n g e x a m p l e s : (3-20)

Peter's booJf Mr Smith 's car

One possible reading of both these examples is c l e a r l y the possessive o n e . But if t h i s does not a p p l y , Peter's book w i l l be interpreted as either "the book w h i c h Peter is r e a d i n g / has read" or "the book w h i c h Peter is w r i t i n g / has w r i t t e n " . "Read" and " w r i t e " are stereotypical r e l a t i o n s of book. S i m i l a r l y , Mr Smith's car w i l l be i n t e r p r e t e d as "the car Mr S m i t h d r i v e s " , as "drive" is a stereotypical r e l a t i o n of car.

3.3.5. Co-ordination There is, however, a f u r t h e r type of possible interpretation of compounds w h i c h is not o b t a i n a b l e either f r o m a r e l a t i o n a l element nor f r o m one associated w i t h a stereotypical r e l a t i o n , and w h i c h does not appear to be context-dependent. This is the i n t e r p r e t a t i o n of a compound as a co-ordinated s t r u c t u r e . It seems we have to do w i t h two d i s t i n c t types, w h i c h are 20 For a very d e t a i l e d Hatcher ( 1 9 S 1 ) .

study

50

of

co-ordinated compounds,

see

exemplified as follows: ( 3 - 2 1 ) a . Belgium-Holland,

fifty-six

b. house-boat, hoy-child In the f i r s t case, the type e x e m p l i f i e d in ( 3 - 2 1 a ) , c l e a r l y the compound cannot be interpreted in the way compounds usually are: fifty-six is not a type of six but a c o m b i n a t i o n of fifty and six. The compounds in ( 3 - 2 1 b ) , h o w e v e r , seem to be quite r e g u l a r . That is, a house-boat r e f e r s to a type of boat w h i c h is also a house, a boy-child is a c h i l d who is a b o y . Though the two types are not a l w a y s as c l e a r l y d i f f e r e n t i a t e d as in the above e x a m p l e s , it does seem o b v i o u s that there are two d i f f e r e n t types of i n t e r p r e t a t i o n i n v o l v e d . This semantic consideration leads us to the proposal that there must be two d i f f e r e n t sources for these compounds. F o l l o w i n g Toman ( 1 9 8 5 ) , we shall assume that the base r u l e of c o - o r d i n a t i o n w h i c h gives rise to structures w i t h co-ordinated constituents is also operative in compounding, and that compounds of the type in ( 3 - 2 1 a ) are a result of this r u l e , rather than of the base r u l e w h i c h r e s u l t s in such structures as r e c t i o n a l compounds. Because the f o r m e r type of compound is very r e s t r i c t e d , o c c u r i n g u s u a l l y o n l y w i t h proper names or numbers, we shall l a r g e l y be concerned w i t h the second type. This type, we shall m a i n t a i n , is the result of the assignment of a coo r d i n a t i v e i n t e r p r e t a t i o n to a compound f o r m e d by the normal compound rule-schema X ·» Υ X. We shall r e f e r to this type of compound as h a v i n g a "basic r e l a t i o n " of c o - o r d i n a t i o n . The interpretation of such compounds w i l l be expected to result f r o m a c o m b i n a t i o n of the s e m a n t i c s of co-ordination and the s p e c i f i c semantics of compounds. But f i r s t we note that there is a r e s t r i c t i o n upon c o - o r d i n a t i o n of the f o r m : (3-22)

O n l y i d e n t i c a l categories can be co-ordinated.

This is not of course a special s t i p u l a t i o n for compounds but is a r e s t r i c t i o n of a general nature, a p p l y i n g both to other coordinators and to other co-ordinated structures. The f o l l o w i n g examples serve as i l l u s t r a t i o n : (3-23)a.

£

He uas young, handsome and travelling to worA by bus.

21 See Fanselow ( 1 9 8 5 ) . 22 For a discussion of this point see, for e x a m p l e , Lang (1977: 3 9 f f . ) . This is o b v i o u s l y a necessary but not a s u f f i c i e n t c o n d i t i o n , as the two co-ordinated elements must also have the same "semantic f u n c t i o n " (see Schachter (1977: 90) quoted in A s b a c h - S c h n i t k e r (1978: 3 2 ) ; see the latter for a discussion of the l i t e r a t u r e on c o - o r d i n a t i o n ) . 51

b. c.

He valued slouly, up the stairs, and quietly.

%

He is paying attention or tired.

This r e s t r i c t i o n w i l l mean that ( 3 - 2 4 a ) below can be assigned a whereas those in ( 3 - 2 4 b ) cannot:

compounds such as those in co-ordinative i n t e r p r e t a t i o n ,

( 3 - 2 4 ) a . hoy-child, poet-painter, blue-blacJi b. hovercraft,

blackbird, underwear

In order to discover a satisfactory i n t e r p r e t a t i o n for coordinated compounds such as those in ( 3 - 2 4 a ) , it is necessary to discuss the semantics of " a n d " , the only possible c o - o r d i n a t i v e r e l a t i o n w h i c h could obtain between the e l e m e n t s of a coordinated compound. In the l i t e r a t u r e on c o - o r d i n a t i o n , t h e r e _ h a s been a f a i r amount of discussion about the m e a n i n g of "and". It has o f t e n been assumed that "and" has several i n t e r p r e t a t i o n s w h i c h can be reasonably c l e a r l y d i f f e r e n t i a t e d , for e x a m p l e : (i)

a d d i t i v e , as in he bought bread and meat

(ii)

s e q u e n t i a l , as in he ran out and fell

( i i i ) a m a l g a m a t i v e , as in it is both beast and man Now it m i g h t be m a i n t a i n e d that t h i s variety lies, in the f a c t that there are d i f f e r e n t semantic types of "and". Contrary to views of this type is that expressed by, for e x a m p l e , Dik (1968: 2 6 9 ) , that "and" does not of i t s e l f have a variety of m e a n i n g s , but that the d i f f e r e n c e s in the f i n a l i n t e r p r e t a t i o n of structures co-ordinated by "and" are e i t h e r "inherent in the contents combined" or added on the grounds of " i n t e r p r e t a t i o n a l possibility". We shall assume that this is in f a c t the correct v i e w . However, it is clear that a co-ordinated compound could nevertheless not have a v a r i e t y of readings corresponding to d i f f e r e n t i n t e r p r e t a t i o n s of "and". A compound poet-painter can i n t u i t i v e l y not mean "a poet and a painter" - i . e . two persons as in (i) above nor "a painter who p r e v i o u s l y was a poet", as in ( i i ) above. The f o r m e r reading is the correct one for coordinated compounds generated by the separate p r i n c i p l e of co-

23 See, for e x a m p l e , Dik ( 1 9 6 8 ) , Lang ( 1 9 7 7 ) , Brettschneider (1978). 24 For a discussion of t h i s v i e w and the contrasting o n e , see Dik (1968: 259-270). 25 See in this connection also B ü h l e r ( 1 9 3 4 : 4 0 9 ) , as cited in D i k (1968: 2 6 7 ) . 52

o r d i n a t i o n , such as those in (3-21a) above, n a m e l y compounds l i k e Belgium-Holland, but not f o r compounds of the type we are at present discussing. The second reading could not apply if we assume w i t h Dik that the semantics of "and" do not include sequential or other aspects but that these are added in context, because compounds do not have the a b i l i t y to express temporal aspects. The correct r e a d i n g for a compound such as poet-painter seems to be something s i m i l a r to ( i i i ) above. B u t , remembering the s p e c i f i c semantics of headed structures such as compounds, we r e c a l l that the class of e n t i t i e s denoted by the whole structure is a subclass of the class of entities denoted by the head. In other words, poet-painter must denote a type of painter. This f a c t , in c o n j u n c t i o n w i t h the type of co-ordinative reading in ( i i i ) above, w i l l mean that a compound poet-painter is assigned the meaning "painter who is a poet", boy-child w i l l mean "child who is a b o y " , and so on. This corresponds to our i n t u i t i o n about the meaning of such compounds. The assumption that t h i s is the o n l y possible type of reading w h i c h can be assigned to a compound w i t h a basic r e l a t i o n of coo r d i n a t i o n is borne out by the type of interpretations AA compounds can be g i v e n . If our assumption is correct, we should expect a compound red-orange u s u a l l y to have the reading "orange, w h i c h is red" ( i . e . "reddish-orange"). This w o u l d in f a c t seem to be the correct reading for t h i s compound. The semantics of heads would exclude the reading "red and orange" except as the interpretation of a compound of type (3-21a) above, w h i c h , it w i l l be noted, has a quite d i f f e r e n t stress pattern - namely level stress - f r o m the type of compound we are discussing, where the accent is on the f i n a l e l e m e n t . In t h i s latter reading, w h i c h is in any case the usual one as the former type is, as noted above, very r e s t r i c t e d , the compound must denote a subclass of orange. Thus we should not n o r m a l l y be able to say a red-orange table and mean, for e x a m p l e , "a table w h i c h has red legs and an orange top". Further, the semantics of heads w i l l e x c l u d e any reading for a compound red-orange such as "a colour w h i c h is both red and orange" as the compound must denote a type of orange not of colour. (Whether this d i s t i n c t i o n in meaning would a c t u a l l y represent a distinction in the entities referred to is not the point at issue h e r e ) . That t h i s is the only type of reading u s u a l l y possible for an AA compound is c o n f i r m e d by the observation that compound adjectives such as black-unite, yelloupurple are much less common than compounds l i k e blue-green, redgold, grey-black. If both types of reading for coordinative compounds were common, then there w o u l d be no reason for this d i f f e r e n c e in p r o d u c t i v i t y . But since compounds l i k e blacJc-uhite can c l e a r l y only mean "black and w h i t e " and not " b l a c k i s h w h i t e " , they must belong to a d i f f e r e n t type f r o m compounds such as bluegreen. The v a r i a t i o n in p r o d u c t i v i t y is thus not a c h a r a c t e r i s t i c of a p a r t i c u l a r type of AA compound but a r e f l e c t i o n of the varying p r o d u c t i v i t y of two d i f f e r e n t processes, w h i c h both result in co-ordinative compounds. In view of these considerations, then, we shall assume that the i n t e r p r e t a t i o n of a compound AB, whereby the u n d e r l y i n g r e l a t i o n is c o - o r d i n a t i o n .

53

O f.

is in general of the f o r m "B w h i c h is A " , and when we r e f e r to co-ordinated compounds in f u t u r e we shall in general be speaking of t h i s type. Co-ordinated compounds, perhaps because of the restrictions on t h e i r f o r m a t i o n , are not very productive in E n g l i s h , though the second type is more productive than the first.

3.3.6.

Location

A f u r t h e r possible "basic r e l a t i o n " w h i c h can obtain in the interpretation of a compound is the r e l a t i o n of location. Most compounds having such a relation are interpretable by means of stereotypes, that is, when the r e l a t i o n o b t a i n i n g in a compound AB is "A is the l o c a t i o n for B" such as in the compound village shop, then the r e l a t i o n " l o c a t i o n " is a stereotype of A. There are, however, compounds in w h i c h the local r e l a t i o n cannot be f o u n d by means of stereotypes, because n e i t h e r element of the compound is s t e r e o t y p i c a l l y a l o c a t i o n . Consider the f o l l o w i n g examples: ( 3 - 2 5 ) a . JbooJi chair b. chair book

meaning " c h a i r on w h i c h there are books" meaning "book w h i c h is on a chair"

c. sill-geranium meaning "geranium on a s i l l " N o t i c e that many compounds w h i c h m i g h t at f i r s t sight appear to be of this type are r e a l l y probably best considered as interpreted by means of a d i f f e r e n t stereotype. Thus, for e x a m p l e , the f o l l o w i n g compounds: (3-26)

sveet-box, vase cupboard,

toy shelf

are all interpreted by means of the stereotype "contains" w h i c h is associated w i t h words such as box, bay, cupboard, shelf, uardrobe etc. In the compounds in examples ( 3 - 2 5 ) , however, there are no such "container-names" or other nouns associated w i t h a p a r t i c u l a r stereotype-relation. In v i e w of these f a c t s , we shall assume that there is a r e l a t i o n of l o c a t i o n w h i c h , l i k e the r e l a t i o n of c o - o r d i n a t i o n , may obtain in a compound and is i n t e r p r e t a b l e w i t h o u t stereotypes. H o w e v e r , t h i s seems, at least in standard l a n g u a g e , to represent a m a r g i n a l compound type. A l m o s t ail compounds of t h i s type (and this c e r t a i n l y applies to ( 3 - 2 5 a ) a n d ( 3 ~ 2 5 b ) above) f a l l under t h e h e a d i n g "contextdependent" .

26 T h i s is a r e l a t i o n commonly assumed in studies on wordf o r m a t i o n to f o r m the basis tor i n t e r p r e t a t i o n of n o m i n a l compounds. See, f o r e x a m p l e , Levi (1978: 2 8 2 ) . 54

3.3.7. Some Semantic Restrictions B e f o r e we t u r n , h o w e v e r , to the question o£ context-dependency and c o n t e x t - e m b e d d i n g , it is important to see that the syntactic and semantic p r i n c i p l e s g i v e n thus f a r w i l l not serve to r u l e out all impossible compounds, because their i n t e r a c t i o n is too unrestricted. We must assume that there are a d d i t i o n a l semantic principles which will rule out any r e m a i n i n g cases of overgeneration. Thus a p r i n c i p l e concerning the semantics of categories w i l l ensure that a compound such as the f o l l o w i n g is not f o r m e d : (3-27)

behind-under m e a n i n g "under ( s o m e t h i n g ) , whereby the r e l a t i o n ' u n d e r ' is compared to the r e l a t i o n ' b e h i n d ' "

Though a compound l i k e this m i g h t be " a l l o w e d " in every other sense, i.e. be generated and given a reading in the c o n t e x t , it will be unacceptable because prepositions represent nonm o d i f i a b l e relations. Some sort of restrictions of this nature w i l l also say, for e x a m p l e , that prepositions and verbs cannot serve as semantic arguments, because there is no meaning assignable to them w h i c h allows them to serve as l o c a l , temporal or instrumental m o d i f i e r s . In f a c t , it seems f a i r l y clear that the o n l y category w h i c h can represent any of these m o d i f i c a t i o n s is N. Thus one might f o r m u l a t e a restriction of the f o r m : (3-28)

Only words of category N can be semantic arguments.

It is interesting to compare in t h i s connectio t n a r e s t r i c t i o n to the e f f e c t that [ + V ] categories cannot be arguments, g i v e n by Kayne (1982: 2 4 ) ; in Boase-Beier & Toman (in p r i n t ) , it has·been argued that this r e s t r i c t i o n must be extended to exclude the assigning of β-roles to any but n o m i n a l categories. In view of the f a c t that this r e s t r i c t i o n appears to apply to both semantic and proper arguments, it seems rather that it has to do w i t h the nature of arguments and not w i t h the question of θ - a s s i g n m e n t . A c c o r d i n g l y , we shall here assume that the r e s t r i c t i o n is to be stated as f o l l o w s : (3-29)

R e s t r i c t i o n on the Nature of Arguments O n l y words of category N can be a r g u m e n t s ( s e m a n t i c or proper).

27 See Fanselow ( 1 9 8 5 ) for other constraints f o l l o w i n g semantics of categories.

55

f r o m the

This principle can probably be motivated by semantic considerations. In connection w i t h this issue there is the question of what can serve as a semantic argument. We shall assume that semantic arguments consist of a closed class, namely { l o c a t i o n , t i m e , instrument}. This set w i l l thus not c o n t a i n , for example, "material", "basis for comparison", "things not r e l a t e d " , "cause", and so on, just to give a few examples. That is, the compounds in (3-30a) but not those in (3-30b) w i l l be possible: (3-30)a.

night-flying in the meaning " f l y i n g by night" parJc-suimming in the meaning " s w i m m i n g in the park" (to) Jcnife-cut in the meaning "to cut with a k n i f e "

b. ^^paper-sui'aiming in the meaning " s w i m m i n g made of paper" ^flower-wash in the meaning "to wash l i k e a f l o w e r " thirst-drink in the meaning "to d r i n k because of ^thirst" Jsnife-cut in the meaning "to cut w i t h o u t a k n i f e " C l e a r l y , semantic considerations r u l e out the f i r s t two types of semantic argument illustrated in ( 3 - 3 0 b ) , and the f o u r t h type might be ruled out on semantic considerations of a d i f f e r e n t t y p e , namely that that w i t h o u t w h i c h one does something is not usually considered relevant enough to be the means of m o d i f i c a t i o n and that, if t h i s sort of a reading could e x i s t , it would result in too many i n t e n s i o n a l l y d i f f e r e n t compounds denoting the same action: Jcnife-cut, scissor-cut, penJcnife-cut, razor-cut could all denote the same type of cutting. The absence of semantic arguments of the t h i r d type, namely "cause", is e m p i r i c a l l y established; such compounds do not e x i s t , but at present we are not in a position to say w h y . Another semantic p r i n c i p l e w h i c h restricts the acceptability of compounds works at the level of the construction of a semantic representation for a non-relational compound. This p r i n c i p l e was put forward in Boase7Beier & Toman (1985) as a p r i n c i p l e of context organisation.'' We here assume a s l i g h t l y d i f f e r e n t status for this p r i n c i p l e and f o r m u l a t e it as f o l l o w s : (3-31)

The N o - M i x i n g P r i n c i p l e In the semantic representation " R ( A , B ) " of a compound AB, A and B must be assigned arguments of the same relation.

The f o l l o w i n g examples illustrate this p r i n c i p l e : 28 For r e l a t i o n a l compounds, the p o s s i b i l i t y that two relations apply w i l l be ruled out by the p r i n c i p l e s of percolation and f u l f i l m e n t of argument-structures. 56

(3-32)

In the dar Je forest grou uhite roses. But the rose forest is dangerous.

(3-33)

In the dark forest Jives a Jcindly magician uho eats f/hite roses. But the rose forest is dangerous.

The second example is c l e a r l y not acceptable in the intended r e a d i n g . The two elements of the compound are arguments of two d i f f e r e n t r e l a t i o n s : rose is an argument of the structure "eat (rose, m a g i c i a n ) " and forest of the structure " l i v e ( f o r e s t , m a g i c i a n ) " . In the f i r s t example both rose and forest are arguments of the same r e l a t i o n , namely "grow i n " , and the compound is thus acceptable in this m e a n i n g . It can be shown that this p r i n c i p l e is general in nature. It a p p l i e s , for e x a m p l e , to g e n i t i v e constructions such as the night of the accident, w h i c h cannot mean "the night in w h i c h a man who had had an accident disappeared." It seems reasonable to assume, in f a c t , that this p r i n c i p l e is part of an i n t e r p r e t i v e p r i n c i p l e of more general n a t u r e . Other restrictions on the r e l a t i o n a compound (or other underdetermined structure) can have are that the r e l a t i o n may not be a n e g a t i v e one and that there may not be two r e l a t i o n s , even if they share the same arguments. For the f i r s t r e s t r i c t i o n , consider t h e f o l l o w i n g examples: (3-34)

^uater-plant meaning "a plant w h i c h does not need water" snake-man meaning "a man who has nothing to do w i t h snakes" book-table meaning "table on which there are no books"

For the second restriction, namely u n i q u e n e s s , consider the f o l l o w i n g : (3-3S)

that

of

relationary

rose-forest meaning "forest in which roses grow and in w h i c h the trees are rose-coloured" snake-car meaning "car used to transport snakes and ^ w h i c h has a snake painted on it" gold-ship meaning "ship made of gold w h i c h transports gold"

In the case of compounds such as those in ( 3 - 3 4 ) and ( 3 - 3 5 ) , it is important to r e a l i s e that the d i f f e r e n c e between meaning and r e f e r e n c e must not be o v e r l o o k e d . Of course we do not mean to say that a compound book-table, for e x a m p l e , cannot be used to r e f e r to a table on w h i c h there are no books. It can be used in this way if it r e f e r s to a table on w h i c h there are u s u a l l y books. What we w i s h to m a i n t a i n is that the compound cannot mean this. The same applies to examples of the second group, such as roseforest. C l e a r l y , t h i s compound can be used to r e f e r to a forest in w h i c h roses grow and in w h i c h the trees are rose-coloured, but it cannot mean t h i s . In a case such as this the compound s i m p l y means something l i k e "forest w h i c h has to do w i t h roses". The 57

interpretation of a compound m a y , t h e n , be v a g u e , as long as this vagueness is resolved in the context. But it may not be ambiguous. These three restrictions on interpretation: that there may not be a m i x e d argument-structure, a negative relation or more than one r e l a t i o n , m i g h t be considered s p e c i f i c subclasses of a general interpretive p r i n c i p l e such as, for e x a m p l e , that a nonexpressed r e l a t i o n must be recoverable. This would not be the case if two relations (as in m i x i n g and a m b i g u i t y ) or a negative relation were used. However, the question as to whether these are r e a l l y separate p r i n c i p l e s or all part of the same p r i n c i p l e is not of great importance here. What is important is that all these semantic restrictions ( w h i c h do not only apply to compounds) are responsible for restricting the output of the semantic component, in that any compounds not r e c e i v i n g an interpretation because they transgress these restrictions w i l l not be a l l o w e d . It is assumed that semantic restrictions of this nature are necessary. Note that in a system such as that proposed by Fanselow ( 1 9 8 5 ) , for e x a m p l e , they would not need to be stipulated because there is a general interpretive p r i n c i p l e w h i c h ensures that any compound which cannot be interpreted by one of his semantic principles (functional application, conjunction, plural q u a n t i f i c a t i o n , inducing a stereotype) is not interpretable. However, if we add to this list, as seems necessary, the p o s s i b i l i t y of inducing a context-relation, then it is necessary to restrict the possible context-relations. As we do not wish to see these restrictions as s p e c i f i c to context-relations, we have f o r m u l a t e d them as general semantic restrictions.

3.3.8. The Context-Dependence of Compounds If a compound can be interpreted on the basis of θ-assignment f r o m the relation contained in the compound or obtained via stereotypes, or by means of a basic r e l a t i o n , we shall r e f e r to it as " l o c a l l y interpretable". Compounds w h i c h are not l o c a l l y interpretable are context-dependent. That is, the relation between the two elements of the compound must then be given in its context. This p o s s i b i l i t y is given both when the compound does not contain any overtly r e l a t i o n a l element and when it does contain such an element but the r e l a t i o n contained is not that used in the construction of the semantic representation of the compound. Furthermore, we shall assume t h a t , for a compound w i t h two or more possible readings, there is o f t e n a "preferred" reading and that this w i l l be the one assigned the compound in absence of i n f o r m a t i o n to the contrary. An example of such a " p r e f e r r e d " reading is the reading whereby θ-assignment has taken place as opposed to that where it has not. Thus a compound factory sale, w h i c h could t h e o r e t i c a l l y mean either "the sale of a factory" or "the sale of something in a f a c t o r y " w i l l be 29 For the question of local i n t e r p r e t a b i l i t y and text-embedding, see also W i l d g e n ( 1 9 8 2 ) . 58

assigned the f o r m e r reading if there is no c o n t e x t - i n f o r m a t i o n w h i c h indicates that the latter is the correct reading. If the latter reading a p p l i e s , such c o n t e x t - i n f o r m a t i o n can be m i n i m a l , m e r e l y consisting of the object of sale, as in the f o l l o w i n g example: (3-36)

The f a c t o r y sale of extremely p r o f i t a b l e .

recycled

paper

was found

to

be

As recycled paper is the object of sale, then factory cannot receive t h i s r o l e , but must f u n c t i o n as a term of location. However, it is context-dependency of the first-mentioned type w h i c h w i l l be of most interest to us here, namely the case in w h i c h an NN compound occurs which contains no overtly or i m p l i c i t l y relational e l e m e n t , or in w h i c h a new relation is supplied by the context. Examples of this are: (3-37)

sJcychair meaning "a chair having the colour of the sky; a sky-blue chair"

(3-38)

car-Jbuyer meaning "a buyer (of computers) for the COMPEX, who comes to work by car".

firm

N o w , as w i l l i m m e d i a t e l y be seen, there is nothing in the compounds themselves to suggest what the correct reading is. The f i r s t compound could, among other things, mean "a special chair used by an astronomer when looking at the sky", "a chair which goes up into the sky ( i . e . perhaps a c h a i r - l i f t ) " and so on. This compound can only be interpreted if the relation - for example, that the c h a i r , at least w i t h respect to its colour, is l i k e the sky - is stated e x p l i c i t l y in the context of the compound. In the compound in ( 3 - 3 8 ) , the non-context-dependent reading w o u l d , of course, be "buyer of cars". But a new relation "come to work" is the correct r e l a t i o n , and this must be e x p l i c i t l y stated in the context. A g a i n , notice the d i f f e r e n c e between this compound and the above example factory sale (3-36) in connection w i t h the question of context-dependence. Example (3-36) is not a case of context-dependence in the strict sense. It is rather the case that the context makes clear that a marginal type of reading (in w h i c h factory is a semantic argument) is the relevant one instead of the usual reading (in which factory is a proper a r g u m e n t ) . In the example cai—buyer (3-38) however, the arguments car and buyer are arguments of a c o m p l e t e l y new r e l a t i o n . We have shown that the context-dependence of compounds involves the presence of the applicable relation in the context. This r e l a t i o n , together w i t h the constituents of the compound, f o r m s the basis for the semantic representation of the compound. 59

The way in which this r e l a t i o n is expressed in the context is subject to f u r t h e r restrictions. In order that the relation be i d e n t i f i a b l e , there must be present in the context two elements corresponding to the constituents of the compound. The required relation is then the relation between the two elements in the context. The condition that the elements corresponding to the constituents of the compound must be present in the context can be stated as f o l l o w s : (3-39)

The Principle of Contextual Counterparts If a structure is not locally interpretable, counterparts of the elements of this structure must occur in a l o c a l l y interpretable f o r m elsewhere in the context of the non-locally interpretable structure.

For a compound the e f f e c t of this principle can be represented schematically as f o l l o w s : (3-40)

R

...

A'

...

B'

CAB]

...

AB is the compound, and A ' and B' are the corresponding elements ( w h i c h must not necessarily be l e x e m a t i c a l l y identical to A and B) in the context, and R is the relation between A ' and B' which is to be induced between A and B. The relation of Α ' , Β ' to A, B must be such that it is possible to establish that the former are in fact the counterparts of the latter. We can thus assume the f o l l o w i n g context-principle: (3-41)

The P r i n c i p l e of I d e n t i f l a b i l i t y of Counterparts For a structure which cannot be l o c a l l y interpreted, the counterparts of its elements must be i d e n t i f i a b l e as such.

But these two principles of context-embedding w i l l not be s u f f i c i e n t to ensure that the counterparts A ' , B' in the context can actually be i d e n t i f i e d . We shall thus assume that the structure of the text is in this respect subject to at least two f u r t h e r principles. The f i r s t can be f o r m u l a t e d as f o l l o w s : (3-42)

The Principle of Contextual Distance Counterparts to l i n g u i s t i c elements of a n o n - l o c a l l y interpretable structure must not be at too great a distance from the elements they are counterparts of.

It is to be assumed that this p r i n c i p l e is valid for both spoken and written texts. It is thus a question of both temporal 60

distance (in spoken t e x t s ) and spatial distance (in w r i t t e n t e x t s ) . That is, to g i v e an e x a m p l e , a speaker may not use a compound only interpretable by r e f e r e n c e to something he said ten m i n u t e s ago (unless he has special reasons f o r assuming the listener has remembered i t ) . The p r i n c i p l e has i n t e n t i o n a l l y been f o r m u l a t e d vaguely because the d e f i n i t i o n of "too great a distance" w i l l vary f r o m situation t o situation. N o r m a l l y everything which goes beyond the short-term memory of a listener w i l l be "too great a distance". But if the speaker has stressed something as i m p o r t a n t , the capacity of the listener to recall it w i l l c l e a r l y be extended. S i m i l a r l y , important events may serve as contextual counterparts over longer periods of time than unimportant ones. The actual e f f e c t s of the p r i n c i p l e w i l l thus vary according to how distance is d e f i n e d . The second p r i n c i p l e we shall discuss here is the p r i n c i p l e of " s p e c i f i c - g e n e r a l order", as g i v e n in Boase-Beier & Toman ( 1 9 8 5 ) . If we look at examples of compounds whose context contains the relevant r e l a t i o n and the two corresponding elements A ' , B' we see that not all cases are e q u a l l y acceptable: (3-43)

Every evening in Atlantis the white eagle brought red roses to the priests. When the flouer eagle died, the people of Atlantis mourned for seven months.

(3--44)

?Every evening in Atlantis, the uhite eagle brought red flouers to the priests. When the rose eagle died, the people of Atlantis mourned for seven months.

(3-45)

?Every evening in Atlantis, the uhite bird brought red roses to the priests. When the flouer eagle died, the people of Atlantis mourned for seven months.

A close e x a m i n a t i o n of these examples reveals that the r e l a t i v e u n a c c e p t a b i l i t y of ( 3 - 4 4 ) and ( 3 - 4 5 ) as compared to ( 3 - 4 3 ) originates f r o m the f a c t that the compound rose eagle ( 3 - 4 4 ) and flouer eagle ( 3 - 4 5 ) one of the elements ( n a m e l y rose in ( 3 - 4 4 ) and eagle in ( 3 - 4 5 ) is more s p e c i f i c than its counterpart. A comparison of all possible combinations of the lexemes rose. eagle, flouer and bird in compound and context substantiates this: all those cases in w h i c h A ' , B ' are hyponyms of A and B or are identical w i t h them are acceptable, whereas all cases are unacceptable in which A ' , B' are hyperonyms of A, B. We can thus formulate the following principle: (3-46)

The P r i n c i p l e of S p e c i f i c - G e n e r a l Order In the case of a t e x t u a l s t r u c t u r e : . . . A'

. . . B'

. . . AB

the r e l a t i o n of the counterparts A ' , B ' to the compoundelements A, B must f o l l o w the order s p e c i f i c - g e n e r a l .

61

This means that A ' , B' must be identical to A , B or be hyponyms of them. Aa stated in Boase-Beier & Toman ( 1 9 8 5 ) , this p r i n c i p l e seems actually to be only a special instance of a much more general pragmatic p r i n c i p l e , namely that the speaker gives JLS much i n f o r m a t i o n as necessary but not more than necessary. Thus, when new i n f o r m a t i o n is given for the f i r s t t i m e it is as s p e c i f i c as necessary in order to ensure r e f e r e n c e to the intended object but when it is taken up later in a text it need only be so s p e c i f i c as to ensure i d e n t i t y of r e f e r e n c e w i t h the previous t e r m . The order g e n e r a l - s p e c i f i c is thus not possible as, for e x a m p l e , flower as a f i r s t instance does not c o n t a i n enough i n f o r m a t i o n if the object r e f e r r e d to is a rose and rose as a second instance is too s p e c i f i c to be a t a k i n g - u p of flauer. J1 A f u r t h e r p r i n c i p l e governing the interpretation of contextdependent compounds concerns the amount of i n f o r m a t i o n the context may c o n t a i n . This could be f o r m u l a t e d as the f o l l o w i n g principle: (3-47)

The P r i n c i p l e of Observance of Semantic P r i n c i p l e s Context-embedding may not v i o l a t e semantic p r i n c i p l e s .

What is meant by this is that the context may not. c o n t a i n i n f o r m a t i o n i n d i c a t i n g that the r e l a t i o n between the e l e m e n t s of the compound is "mixed" or n e g a t i v e , or ambiguous, or that a preposition means what o n l y a substantive can m e a n , and so on. This w i l l ensure that the p r i n c i p l e s of context-embedding do not interact w i t h the semantic p r i n c i p l e s in such a way as to " a l l o w " s e m a n t i c a l l y i l l - f o r m e d compounds. If t h i s were to h a p p e n , the text would perforce be i l l - f o r m e d , s i m p l y because the amount of i n f o r m a t i o n necessary to ensure interpretation of a p r i n c i p a l l y uninterpretable compound would be so great as to render the text impossibly c o m p l e x . For e x a m p l e , if the compound water-plant were r e a l l y to mean "plant w h i c h does not need w a t e r " , a normal context, in w h i c h the counterparts of water and plant and the r e l a t i o n "does not need" are contained, w o u l d not s u f f i c e to ensure interpretability, even if the counterparts are l e x e m a t i c a l l y identical to the e l e m e n t s of the compound. Compare the following example: (3-48)

?There are many plants which do not need water-plants are very easy to maintain.

water.

Such

30 In this connection see also the m a x i m s put f o r w a r d by Grice (1975: 4 5 f f . ) . 31 For a discussion of some examples w h i c h seem to contradict this p r i n c i p l e but w h i c h can be shown to be of a d i f f e r e n t nature, see Boase-Beier & Toman ( 1 9 8 5 ) . 62

This context is o b v i o u s l y not s u f f i c i e n t to enable the compound to be understood as having the required m e a n i n g because it v i o l a t e s the s e m a n t i c p r i n c i p l e w h i c h says that the u n d e r l y i n g r e l a t i o n in a compound may not be n e g a t i v e . The o n l y possible context w h i c h w o u l d a l l o w t h i s w o u l d be one in w h i c h the f a c t that t h i s reading is meant is a c t u a l l y e x p l i c i t l y stated, but such a text w o u l d presumably not be w e l l - f o r m e d . If a compound is not l o c a l l y i n t e r p r e t a b l e , it must r e c e i v e a context-dependent r e l a t i o n , w h i c h then serves as the basis for its i n t e r p r e t a t i o n , a c c o r d i n g to these p r i n c i p l e s . This s i t u a t i o n i s expressed i n t h e f o l l o w i n g p r i n c i p l e : (3-49)

The Interpretation P r i n c i p l e All l i n g u i s t i c structures must be interpreted e i t h e r by θ - r o l e - a s s i g n m e n t or by the assignment of e i t h e r a stereotype-relation, a basic r e l a t i o n or a c o n t e x t dependent r e l a t i o n .

This p r i n c i p l e has the e f f e c t that out of the s t r o n g l y over generated output of the base only those compound^ w i 1 1 a c t u a l l y occur w h i c h can be interpreted by the means g i v e n .

3.3.9. Pragmatic Principles Besides s y n t a c t i c , s e m a n t i c a n d c o n t e x t - e m b e d d i n g r u l e s f o r c o m p o u n d s , there w i l l also b e pragmatic p r i n c i p l e s w h i c h a f f e c t the type of compounds f o r m e d in standard E n g l i s h . These are by nature less strict than s y n t a c t i c or s e m a n t i c p r i n c i p l e s and can more e a s i l y be b r o k e n . H o w e v e r , it is reasonable to assume t h a t , if a pragmatic principle is not observed, then the i n t e r p r e t a b i l i t y of a compound w i l l be made more d i f f i c u l t . E i t h e r the compound w i l l be u n a c c e p t a b l e or its correct i n t e r p r e t a t i o n w i l l have to be ensured by the presence of context-information, because, as we have said a b o v e , only i n t e r p r e t a b l e structures a r e produced. P r a g m a t i c p r i n c i p l e s d e a l w i t h a s p e c i f i c aspect of the i n t e r p r e t a b i 1 i t y of l i n g u i s t i c s t r u c t u r e s , n a m e l y , w i t h t h e i r appropriateness in a g i v e n c o n t e x t (or c o t e x t ) . From a p r a g m a t i c point of v i e w , o n l y a p p r o p r i a t e 32 Note that t h i s is not the same as F a n s e l o w ' s i n t e r p r e t a t i o n p r i n c i p l e m e n t i o n e d above (see Fanselow (1985: 2 9 5 ) ) . 33 Whether there are other r e s t r i c t i o n s on the i n t e r p r e t a t i o n of compounds we are not in a p o s i t i o n to say at the m o m e n t , b u t it is not i m p o s s i b l e that there are. F a n s e l o w ( 1 9 8 5 ) discusses one or two such possible c o n s t r a i n t s . 34 Compare Levinson ( 1 9 8 3 : 2 4 ) , discussed in Stohr (1985c: b ) . What we are here r e f e r r i n g to as p r a g m a t i c p r i n c i p l e s have been studied as text-constructional and text-embedding p r i n c i p l e s b y W i l d g e n (1983a, 1 9 8 3 b ) . 63

structures w i l l be r e a d i l y interpretable and only r e a d i l y interpretable structures will be sanctioned by pragmatic p r i n c i p l e s . Pragmatic p r i n c i p l e s can roughly be d i v i d e d into three groups. Those of the f i r s t group w i l l interact w i t h the semantic component in that they w i l l render interpretable structures w h i c h w o u l d otherwise be of peripheral status. That is, pragmatic factors, by rendering such structures appropriate, w i l l ensure t h e i r i n t e r p r e t a b i l i t y . A t y p i c a l example of this type of pragmatic p r i n c i p l e is that of contrast. As we have said above, NN compounds c o n t a i n i n g a r e l a t i o n a l element in the headposition are u s u a l l y assigned a r e a d i n g in w h i c h the non-headelement is an argument of the head. The non-argument reading is possible, but must be assigned a p e r i p h e r a l p o s i t i o n . H o w e v e r , if contrast is used, these compounds immediately become more acceptable. Consider the f o l l o w i n g e x a m p l e : (3-50)

car-transportation (of animals)

The usual reading of the compound car-transportation is "transportation of cars". If the o b j e c t ot transportation is given or is to be understood as animals, then the compound must have the reading "transportation i n / b y means of cars". But the compound in t h i s m e a n i n g , though possible, is somewhat odd. However, consider the f o l l o w i n g sentence: (3-51)

As transporting the animals by aeroplane has proved to be too expensive, the Ministry has suggested cartransportation as an alternative.

In this sentence the compound "transportation of animals in acceptable.

car-transportation meaning cars" has become fully

The contrast m a y , as in this e x a m p l e , take place w i t h i n the text i t s e l f , or it may have the f o r m of contrast between an existing l e x i c a l i s e d compound or phrase and one w h i c h is introduced in the t e x t . Another pragmatic factor of this nature is that of analogy, either to an existing lexicalised compound or phrase, or to a compound or phrase formed in the foregoing text. Secondly, there are pragmatic factors w h i c h do not a c t u a l l y a f f e c t the output of the semantic component but which s i m p l y render ( s e m a n t i c a l l y acceptable) compounds more l i k e l y . An example of this type of pragmatic p r i n c i p l e is that known as the M i n i m a x P r i n c i p l e , ° w h i c h in e f f e c t states that there is a 35 See Stbhr (1985a) for the role of contrast in nominal compounding. 36 For a discussion of this p r i n c i p l e see Carroll & Tanenhaus (1975: 51).

tendency to supply as much i n f o r m a t i o n in as l i t t l e space as possible. T h i s p r i n c i p l e has o f t e n been seen as important for word-formation, e s p e c i a l l y for the question of the use of wordf o r m a t i o n processes in contrast to the f o r m a t i o n of other structures. F u r t h e r general p r a g m a t i c p r i n c i p l e s of t h i s type are the principle of r e i f i c a t i o n , the principle of f i l l i n g ( s u b j e c t i v e or objective) l e x c a l gaps, and the p r i n c i p l e of c l a r i t y of r e f e r e n c e . P r i n c i p l e s such as these w i l l render the use of compounds more l i k e l y b u t , because they are m a i n l y concerned w i t h the q u e s t i o n of p e r f o r m a n c e and do not interact d i r e c t l y w i t h the semantics of compounding, they w i l l not f o r m part of our central concern here. A p r a g m a t i c p r i n c i p l e of a t h i r d type is the p r i n c i p l e of pree m p t i o n as f o r m u l a t e d by Clark & Clark ( 1 9 7 9 : 7 9 8 ) for d e n o m i n a l verbs. The observance of t h i s p r i n c i p l e ensures i n t e r p r e t a b i l i t y ; if it is not observed, u n a c c e p t a b l e structures r e s u l t . In its most general f o r m , t h i s p r i n c i p l e can be stated as f o l l o w s : (3-52)

The P r i n c i p l e of Pre-Emption A new word cannot be f o r m e d if there exists in the l e x i c o n a phrase or another word ( s i m p l e or c o m p l e x ) w i t h e x a c t l y t h e same m e a n i n g .

For the q u e s t i o n of i n t e r p r e t a t i o n , the e x i s t e n c e of t h i s p r i n c i p l e means that a new compound w i l l in general a l w a y s be interpreted as though it means s o m e t h i n g d i f f e r e n t f r o m an a l r e a d y e x i s t i n g one. (Of course we are l e a v i n g aside the s i t u a t i o n in w h i c h a speaker s u f f e r s f r o m a permanent or temporary gap in his l e x i c o n . ) As we have seen, the f o r m a t i o n of a compound must be v i e w e d as the r e s u l t of an i n t e r a c t i o n of v a r i o u s d i f f e r e n t general p r i n c i p l e s . In the i n t e r p r e t a t i o n of a compound v a r i o u s types of k n o w l e d g e i n t e r a c t ; these are connected to v a r i o u s types of p r i n c i p l e . Some of t h i s k n o w l e d g e is at a s t r i c t l y g r a m m a t i c a l level ( k n o w l e d g e about the f o r m a t i o n of compounds, the h e a d - r u l e , - m a r k i n g of a r g u m e n t s ) , other processes i n v o l v e d are s e m a n t i c in nature and r e f e r to k n o w l e d g e stored in the l e x i c o n about the m e a n i n g s of words and categories, and s e m a n t i c k n o w l e d g e about

37 See, for e x a m p l e , B r e k l e (1978: 73) and Stöhr ( 1 9 8 5 c : 9). 38 For a discussion of the use of compounds in contrast to s y n t a c t i c phrases, s e e F l e i s c h e r ( 1 9 7 0 ) . 39 On the role in c o m p o u n d i n g of r e i f i c a t i o n see B r e k l e ( 1 9 8 0 : 7 5 f . ) a n d Stöhr ( 1 9 8 5 c : 1 3 ) , o f t h e f i l l i n g o f l e x i c a l gaps Geckeler ( 1 9 7 7 ) , Brekle (1980: 7 3 f f . ) , Stöhr ( 1 9 8 5 c : 1 3 ) , a n d of c l a r i t y of r e f e r e n c e D o w n i n g (1977: 8 3 9 ) and Stöhr ( 1 9 8 5 b ) . 40 For a summary of e x p e r i m e n t s in w h i c h the r o l e of these p r a g m a t i c f a c t o r s has been t e s t e d , see Stöhr ( 1 9 8 5 d ) . 65

the i n t e r p r e t a t i o n of argument-structures. Yet a g a i n , there are further principles connected with pragmatic knowledge or k n o w l e d g e about the structure of texts.

3 . 4 . The Types of Compounds in English As we have seen in the last s e c t i o n , it is possible to account for the phenomenon of compounding in E n g l i s h in such a way that all possible compounds, and o n l y these, are generated by the i n t e r a c t i o n of v a r i o u s components of the l a n g u a g e . In this section we s h a l l look at the compound types generated by the base rule of compounding and try to ascertain what the f i n a l output of t h e grammar w i l l b e , a f t e r i n t e r a c t i o n w i t h t h e s e m a n t i c component has taken p l a c e . In t h i s way we hope to establish w h i c h f o r m s are i m p o s s i b l e . It w i l l be r e c a l l e d t h a t , for our study of the r e l a t i o n s h i p of PL to S L , our m a i n concern w i l l be w i t h those compounds w h i c h are u n a c c e p t a b l e in SL but acceptable in PL; because they represent the p o i n t of d i v e r g e n c e of PL f r o m SL, it is e s p e c i a l l y w i t h r e f e r e n c e to them that our theory of additional principles required for PL w i l l be formulated. As we saw at the b e g i n n i n g of t h i s c h a p t e r , there are 16 possible c o m b i n a t i o n s r e s u l t i n g f r o m t h e u n r e s t r i c t e d c o m p o u n d i n g r u l e - s c h e m a X -» Υ Χ, n a m e l y : (3-53)

NN

VN

AN

PN

NV

VV

AV

PV

NA

VA

AA

PA

NP

VP

AP

PP

We shall now c o n s i d e r each of these types in t u r n .

3 . 4 . 1 . NN Compounds Compounds w i t h two n o m i n a l e l e m e n t s are p r o b a b l y the most frequently-occurring in E n g l i s h . Among NN compounds it is necessary to d i s t i n g u i s h between those c o n t a i n i n g a verbal e l e m e n t (a n o m i n a l i s a t i o n ) or an element w h i c h is otherwise r e l a t i o n a l , and those w h i c h do not c o n t a i n a r e l a t i o n a l e l e m e n t . NN compounds w h i c h do not c o n t a i n a r e l a t i o n a l e l e m e n t w i l l be i n t e r p r e t e d , it w i l l be r e c a l l e d , e i t h e r by means of a stereotype of one e l e m e n t ( e . g . shoe factory, where the stereotype is "produce") or by means of a basic r e l a t i o n ( e . g . hoy chJJd ) or by means of a r e l a t i o n expressed in the c o n t e x t ( e . g . snake man m e a n i n g "the man who caught a s n a k e " ) . C l e a r l y there can be no r e s t r i c t i o n s f r o m the p r i n c i p l e s of p e r c o l a t i o n and r e a l i s a t i o n of a r g u m e n t - s t r u c t u r e on compounds whose e l e m e n t s do not have

66

a r g u m e n t - s t r u c t u r e , so in these cases restrictions w i l l be of the nature of the semantic r e s t r i c t i o n s m e n t i o n e d above upon the r e l a t i o n between the elements. Compounds w h i c h v i o l a t e these r e s t r i c t i o n s ( e . g . snake man in the meaning "a man who has n o t h i n g to do w i t h snakes") w i l l be ruled out. Other compounds m i g h t be r u l e d out because they are not i n t e r p r e t a b l e by any of the means g i v e n , i . e . , they v i o l a t e the I n t e r p r e t a t i o n P r i n c i p l e . For e x a m p l e , the f o l l o w i n g compound in the g i v e n context w i l l be impossible ( a s s u m i n g the text given represents the complete c o n t e x t ) : (3-54)

/έ was midnight entered.

and the room was coJd.

The snake-man

F u r t h e r m o r e , an NN compound m u s t , as a n o u n , mean something a noun can mean ( i . e . net a f u n c t i o n or a p r o p o s i t i o n ) . Any NN compound w i t h o u t a r e l a t i o n a l e l e m e n t , w h i c h does not v i o l a t e any of these p r i n c i p l e s , w i l l be a c c e p t a b l e . It w i l l be r e c a l l e d that NN compounds in w h i c h the head is relational (e.g. brother, seller, seeking e t c . ) are to be interpreted such that the non-head-element is e i t h e r an argument or a s e m a n t i c argument of the r e l a t i o n a l head, or such that a new r e l a t i o n o b t a i n s . R e l a t i o n a l nouns such as brother w i l l behave e x a c t l y l i k e deverbal nouns in t h i s respect. R e l a t i o n a l and deverbal n o u n s in the n o n - h e a d - p o s i t i o n can, it w i l l be r e c a l l e d , realise t h e i r argument-structure outside the compound in some cases. We repeat e x a m p l e ( 3 - 1 2 ) here as ( 3 - 5 5 ) : (3-55)

the combining-process of semantic markers

W h i l e ( 3 - 5 5 ) m i g h t be considered a l i t t l e odd to some speakers of E n g l i s h , it is acceptable to others. Clearly, for these speakers, the argument-structure of combine has percolated to the compound-node and the o b j e c t of combine is e x t e r n a l l y r e a l i s e d . To account for the f a c t that such c o n s t r u c t i o n s are severely restricted, it is necessary to e x a m i n e the type of noun w h i c h can occur as the head of a compound in such a c o n s t r u c t i o n . The f o l l o w i n g data suggest that o n l y certain types of head are possible: ( 3 - 5 6 ) a . Spring is the healing-time of all ills. b. The waiting-period for neus of very trying for all concerned.

the

trapped

miners uas

41 Q u e s t i o n i n g of n a t i v e speakers of E n g l i s h has shown t h i s to be the case. 67

c. There were various questions about process of mercury with gold.

the

amalgamateng-

( 3 - 5 7 ) a . The dandelion is α healing-plant of many ills. £. b. There uas a special waiting room for news of the miners. £

c. There were questions aJbout mercury with gold.

the

amalgamating

dish

of

C l e a r l y the argument-structure "heal X" in ( 3 - 5 6 a ) is not prevented f r o m being realised by the head time, but it is prevented f r o m being realised by plant in ( 3 - 5 7 a ) . The same applies to period in ( 3 - 5 6 b ) and room in (3-57b) w i t h regard to the a r g u m e n t - s t r u c t u r e "wait for X", and to process in ( 3 - 5 6 c ) and dish in ( 3 - 5 7 c ) w i t h regard to the argument-structure "amalgamate X w i t h Y " . In the f i r s t case of each of these pairs, the θ-role can be assigned by the v e r b , in the second t h i s is not possible. F o l l o w i n g Toman ( 1 9 8 3 : 6 1 ) , we shall r e f e r to heads l i k e those in ( 3 - 5 6 ) as transparent. Transparent heads are l e x i c a l e l e m e n t s of a very g e n e r a l , abstract semantic nature l i k e time, period and process - w h i c h apparently do not b l o c k Θassignment and w h i c h , t y p o l o g i c a l l y s p e a k i n g , could be s u f f i x e s in other languages. The examples in ( 3 - 5 7 ) , however, represent the unmarked case: the constructions are unacceptable because the assignment of the S-role is "blocked" by the head of the compound.

3 . 4 . 2 . VN Compounds

If we assume, w i t h L i e b e r ( 1 9 8 3 : 2 6 1 ) , that the verbal root of a VN compound cannot supply any of its f e a t u r e s to the compound as a w h o l e , because p e r c o l a t i o n is o n l y d e f i n e d for the r i g h t hand branch, but that argument-structure must be f u l f i l l e d , then we can i m m e d i a t l y e x p l a i n why t h i s compound type is f a i r l y u n p r o d u c t i v e in E n g l i s h . If the verb cannot pass its argumentstructure to the compound, but argument-structure must be s a t i s f i e d , then the second e l e m e n t must be an argument of the f i r s t (as in drawbridge) , or a s e m a n t i c a r g u m e n t , if the f i r s t has no argument-structure to f u l f i l (as in pay-booth). S e l k i r k ( 1 9 8 2 : 2 5 ) , on the other h a n d , m a i n t a i n s just the opposite, namely that the head of" a compound cannot be the argument of a verb in the n o n - h e a d , and e x p l a i n s that compounds in w h i c h t h i s i n t u i t i v e l y seems to be the case are not instances of argumentstructure f u l f i l m e n t . T h u s , for e x a m p l e , search party is "a party w h i c h h a s some r e l a t i o n t o searching" ( p . 2 5 ) . ( I n L i e b e r ' s

42 It should be n o t e d , h o w e v e r , that the English data are less c o n c l u s i v e t h a n , f o r e x a m p l e , c o r r e s p o n d i n g data f r o m German. N e v e r t h e l e s s , speakers of E n g l i s h n o t i c e a clear contrast between the a c c e p t a b i l i t y of examples in ( 3 - 5 6 ) and those in (3-57). 68

system, too, compounds i n w h i c h t h e head-element i n t u i t i v e l y seems to be the s u b j e c t are also e x p l a i n e d independently of argument-linking. ) We shall m a i n t a i n that the correct e x p l a n a t i o n is d i f f e r e n t f r o m either of these. U n l i k e Lieber and S e l k i r k , we shall assume that the subject also receives a θ-role f r o m the v e r b , and, u n l i k e S e l k i r k , that the head can be an argument of the non-head, thus in cases l i k e hoverfly, the head-noun fly receives a θ-role f r o m the verb hover. H o w e v e r , the types a l l o w e d by our system w i l l be r o u g h l y the same. Either the second e l e m e n t w i l l be an argument of the verb in the non-head-position or it w i l l be a semantic argument. At t h i s point it is necessary to say something of a general nature about compounds w i t h a r e l a t i o n a l element in the non-headp o s i t i o n , in order to make clear w h i c h readings can e x i s t , both for VN and other compounds of t h i s t y p e . Because both θ-role assignment and p e r c o l a t i o n are assumed to be f r e e , and the " f o r c i n g " of percolation by the head-rule is i r r e l e v a n t in the case of r e l a t i o n a l non-heads, then there are at f i r s t sight t h e o r e t i c a l l y 4 possible states of a f f a i r s : (i)

argument-structure percolates,

head is an argument

(ii)

argument-structure percolates, head is a s e m a n t i c argument

( i i i ) argument-structure does not p e r c o l a t e , head is an argument (iv)

argument-structure does not p e r c o l a t e , semantic argument

head

is a

Let us consider each of these cases in t u r n . (i) If we look at n o m i n a l i s e d verbs, we see that the argumentstructure of the non-head percolates to the compound, w h i l e the m o r p h o l o g i c a l f e a t u r e s percolate f r o m t h e h e a d , i n keeping w i t h the h e a d - r u l e : (3-58)

(obj.)

-τ I I

V

N

drive

-Jr

V Ν

car

(Here and in the f o l l o w i n g , ---» represents θ-role a s s i g n m e n t . ) 69

The verb drive realises its object outside the compound. However, if the n o m i n a l head were not a s u f f i x , as here, but a n o u n , it could only be the case that the noun external to the compound and the noun in the head-position receive d i f f e r e n t roles, as the same role cannot be assigned twice because of the β - c r i t e r i o n (see Chomsky (1981: 3 6 ) ) . That is, one might expect a construction such as: (3-59)

* pay-master of the wages

w i t h the structure: (3-60)

^_J> N—^""

(obj.)

1 I I I I

V N

wages

Here, pay assigns a subject-role to master c o m p o u n d - i n t e r n a l l y and an object-role to wages compound e x t e r n a l l y . H o w e v e r , the structure in ( 3 - 6 0 ) is not acceptable. We assume that this is because, though percolation can take place as far as the dominating N node, the compound-head-noun blocks 9-role assignment to wages; this assumption is discussed above. In other words, case (i) is not inherently impossible, but it has no consequences as, even if the argument-structure percolates, θ-roles cannot be assigned to elements outside the compound. Because s u f f i x e s are taken to be "transparent", i . e . , not to block θ-role assignment, there is no problem w i t h structures l i k e the driver of the car, nor w i t h structures in w h i c h a "semantically empty" head does not block θ-role assignment, as discussed above. But as it does not seem possible that a head could be both " s e m a n t i c a l l y empty" and an argument of the nonhead-verb, such cases w i l l not occur. ( i i ) The same considerations would apply to this case as to case ( i ) . The head-noun blocks θ-role assignment and so a construction such as: (3-61)

pay-booth of the wages

where booth possible.

is

a

local

semantic

argument

of

pay,

are

not

Most VN compounds (and others w i t h r e l a t i o n a l non-heads) w i l l 70

thus be of types ( i i i ) and ( i v ) : (3-62)

pay-master

(type ( i i i ) )

(3-63)

pay-booth

(type ( i v ) )

A f i f t h state of a f f a i r s is also possible, it w i l l be remembered; n a m e l y , that θ-marking does not occur and both elements are arguments of another ( c o n t e x t u a l ) r e l a t i o n . However, because, as mentioned above, we assume that neither verbs nor a d j e c t i v e s nor prepositions can be arguments, this p o s s i b i l i t y is a u t o m a t i c a l l y ruled out for all except NN compounds. It might f u r t h e r be expected that VN compounds could exist in which the second element is also r e l a t i o n a l , cf . : (3-64)

#

run-vaJJfj'ng·,

4t

sJt-Jyj'ngr

In these compounds, if the argument-structure of the head is inherited by the compound, then the argument-structure of the compound must, in a syntactic structure, be f u l f i l l e d e x t e r n a l l y . This in itself presents no problems. However, the verb in the non-head-position would then have to be a semantic argument. Because verbs cannot be semantic arguments, such compounds should be i m p o s s i b l e , as, in f a c t , they are.

3 . 4 . 3 . AN Compounds There seem to be r e l a t i v e l y few restrictions upon the f o r m a t i o n of AN compounds, which are f a i r l y productive in English, c f . : (3-65)

blackbird, shortwave, hardware, hothouse ^

The question as to whether a d j e c t i v e s are or are not θ-role assigners has been treated d i f f e r e n t l y in the l i t e r a t u r e . Lieber (1983) states that a d j e c t i v e s are non-argument-taking elements. In Boase-Beier & Toman (in p r i n t ) , it is argued that there is no reason not to regard a d j e c t i v e s as on a par w i t h i n t r a n s i t i v e verbs, thus as θ-assigners. This is the v i e w we shall take here. 43 It is o f t e n d i f f i c u l t to d i s t i n g u i s h AN compounds f r o m AN phrases, partly because the stress patterns o f t e n assumed to i n d i c a t e whether the structure in question is a compound or a phrase are not always r e g u l a r . We s h a l l , h o w e v e r , not discuss t h i s question here as it is not d i r e c t l y r e l e v a n t to our topic. For a discussion, see B r e k l e ( 1 9 6 6 ) . 71

We can thus d i s t i n g u i s h two types of AN compounds a c c o r d i n g to whether 0-assignment has taken place or n o t . In the f i r s t type, the noun is an argument of the a d j e c t i v e , or at least, the a d j e c t i v e is predicated of the noun: (3-66)

JblacJfbird,

shortwave

It w i l l be r e c a l l e d that this is u n p r o b l e m a t i c because the argument-structure of an element in non-head-position need not percolate to the h i g h e r node and can thus be r e a l i s e d i n t e r n a l l y . In the second t y p e , the a d j e c t i v e is not predicated of the n o u n : ( 3 - 6 7 ) a . chronic ward, wanted list b. fast-eater,

slow-mover

In the examples in ( 3 - 6 7 a ) , the a d j e c t i v e m o d i f i e s an argument of the relation inherent in the head-noun ( e i t h e r because it is derived or is associated w i t h a s t e r e o t y p e - v e r b ) , in the examples in (3-67b) the f i r s t element is an adverb and m o d i f i e s the relation in the head. We shall assume that the latter are not compounds but structures l i k e the f o l l o w i n g : (3-68)

llfast-eat

]„ er ]„

The examples in (3-67a) are, on the other h a n d , e x a c t l y p a r a l l e l to VN examples such as pay-Aooth and dance-floor. That is, the noun is a semantic argument of the a d j e c t i v e and the compounds are interpreted as: "ward housing chronic cases", "list cont a i n i n g wanted persons".

3 . 4 . 4 . PN Compounds An e x a m i n a t i o n of PN compounds in E n g l i s h reveals that their correct interpretation poses several problems. C o n s i d e r , for e x a m p l e , the compound undercarpet. Lieber. (1983: 2 6 1 ) m a i n t a i n s that in compounds of t h i s type the p r e p o s i t i o n assigns an o b j e c t role to the n o u n . But, supposing that p r e p o s i t i o n s are a b l e to assign 0-roles - w h i c h is not e n t i r e l y clear - t h i s appears not to apply here because then the compound w o u l d mean something l i k e "under the carpet". But t h i s is, f i r s t l y , not what the compound means a n d , secondly, this interpretation stands in direct c o n f l i c t w i t h the h e a d - p r i n c i p l e , according to w h i c h the compound must denote a type of carpet. The actual i n t e r p r e t a t i o n required

44 For a discussion of the v i e w that prepositions do not assign θ - r o l e s , see Boase-Beier & Toman ( i n p r i n t ) .

72

for the compound appears rather to be "carpet which is under a carpet". H o w e v e r , given the f a c t s about percolation as we have stated them here, one would not expect PN compounds to have such m e a n i n g s . In order to have t h i s interpretation, the compound would have to have the f o l l o w i n g structure: (3-69)

N

under

(obj.)

carpet^

carpet2

The object-role of under would have to be inherited by the upper N node. But the external assignment of the object-role to carpet2 is not possible, because it is blocked^ by the head-noun carpetf, for this reason there is no phrase undercarpet of the carpet corresponding to t h i s structure. If we assume as an alternative p o s s i b i l i t y that the argument-structure of under does not percolate and that under assigns the head-noun carpetj a subjectr o l e , then the corresponding reading is "carpet w h i c h is under something". This is, however, not the correct reading, as is indicated by an e x a m i n a t i o n of the relevant data. Overblanket cannot denote a blanket w h i c h is over a c h a i r , but only a blanket w h i c h is over other blankets. Afterglou may not denote a glow f o l l o w i n g something w h i c h is not at least metaphorically a glow. That is, we cannot - unless we aim at oxymoron - say the afterglow of the darJiness. Examples from German substantiate these observations: Vorzimmer ( ' b e f o r e - r o o m ' ) is the room before another room, n o t , for e x a m p l e , before a house; Unterbett ( ' u n d e r - b e d ' ) is the bed (-cover) under another bed ( - c o v e r ) , and n o t , for e x a m p l e , under a t a b l e , and so on. It seems, t h e n , that in compounds of this type the class of possible objects of the preposition is restricted in an unexpected manner: the object must belong to the same class as the subject. The only way to account for this fact seems to be to assume that we have here to do w i t h an element w h i c h is, at least f u n c t i o n a l l y , an a d j e c t i v e rather than a preposition. Otherwise there appears to be no way in w h i c h the object of the preposition could be restricted in this w a y . It is, however, as pointed out in Boase-Beier & Toman (in p r i n t ) , a typical characteristic of adjectives that they specify particular members of a set of i n d i v i d u a l s . Thus undercarpet would have the same meaning as the under (most) carpet (of a series of carpets); this is in fact exactly what it does mean. I f , however, there are prepositions w h i c h , for s e m a n t i c reasons, cannot f u n c t i o n as a d j e c t i v e s , then we should expect them not to form compounds. A preposition of 73

t h i s type #is between; note that there is no corresponding adjective betweenmost, and it is i n t u i t i v e l y d i f f i c u l t to i m a g i n e what such an a d j e c t i v e could m e a n . In f a c t , it is clear that compounds such as betveen-room are unacceptable in English. It is our v i e w t h a t , instead of speaking of p r e p o s i t i o n s w h i c h f u n c t i o n as a d j e c t i v e s , it makes sense to assume that most l e x i c a l items w h i c h are marked as prepositions in the l e x i c o n are also marked as a d j e c t i v e s . If we consider the stress pattern in phrases l i k e in after years or his under jau, it w i l l not be doubted that these are AN phrases, i.e. that after and under .are a d j e c t i v e s . PN compounds are thus in r e a l i t y AN compounds. °

3 . 4 . 5 . NV Compounds The f i r s t t h i n g to be noted about t h i s type of compound is that it is very u n p r o d u c t i v e . In £ac±, S e l k i r k (1982: 15) m a i n t a i n s that they do not exist at a l l . We s h a l l m a i n t a i n that they do e x i s t , as the f o l l o w i n g examples show: (3-70)

hand-wash, Jcnife-cut,

pencil-mark

and that their r e l a t i v e l y unproductive nature can be explained in part by the principles of percolation given above. F i r s t l y , let us examine a hypothetical compound (to} bat-fly where bat is the l o g i c a l subject of fly. I n t u i t i v e l y t h i s compound w o u l d have no d i s t r i b u t i o n , as the subject is already present in the compound. T h i s f a c t can be explained if we remember that the arguments t r u c t u r e of the head-verb in the compound percolates to the w h o l e compound. In a sentence in w h i c h the compound appears, t h e r e f o r e , the subject-role must be assigned e x t e r n a l l y . But as fly can only assign one subject-role (see Chomsky (1981: 3 6 ) ) , and t h i s haa already been assigned to bat, the compound is not acceptable. 45 This r e s t r i c t i o n does n o t , however, seem to apply in German, where it is the case that there not only are l e x i c a l i s e d compounds such as Zuischenboden 'between f l o o r ' , Zwischenstufe 'between s t a g e ' , " i n t e r m e d i a t e stage", but where the pattern is also productive of ad hoc compounds. 46 This observation only applies to compounds of the type PN and not to the type [ P N ] N , such as after-dinner hour or German LfnterbettschranJs 'under-bed cupboard'. Such compounds a c t u a l l y do contain a p r e p o s i t i o n , w h i c h assigns the object role to the f o l l o w i n g noun. As t h i s is not the head of the construction, there are no problems w i t h percolation and the r e a d i n g obtained, namely "hour a f t e r dinner" or "cupboard beneath the bed" is the correct one. 47 See also Marchand ( 1 9 6 9 : 9 6 ) . 48 T h i s does not need to be s t i p u l a t e d in terms of, for e x a m p l e , a "First-Sister P r i n c i p l e " (see Roeper E. Siegel ( 1 9 7 8 ) ) . 74

S i m i l a r l y , a h y p o t h e t i c a l NV compound in w h i c h the noun is the object of the verb w i l l also be i m p o s s i b l e , as the argumentstructure must percolate because of the h e a d - p r i n c i p l e and must thus be r e a l i s e d e x t e r n a l l y . Since it is already realised i n t e r n a l l y , t h i s is not p o s s i b l e . Thus a compound such as caJce Jbake w i l l be u n g r a m m a t i c a l . These considerations show that compounds of the type NA and NV w i l l always be unacceptable when the head assigns a -role to the non-head. This leaves the p o s s i b i l i t y that an NV compound can occur w i t h an e l e m e n t in the n o n - h e a d - p o s i t i o n w h i c h is not part of the a r g u m e n t - s t r u c t u r e of the V. In f a c t , all the examples given above of NV compounds seem to be of this type. However, it appears that not all possible s e m a n t i c arguments of the V can appear in the N p o s i t i o n . Consider the f o l l o w i n g compounds: (3-71)

queue-stand,

summer-sing,

morning-uash

Such compounds c l e a r l y do not occur, * we are, however, at the moment not in a position to g i v e a p r i n c i p l e d e x p l a n a t i o n for t h i s observation. The correct g e n e r a l i s a t i o n seems to be that verbs can o n l y be m o d i f i e d in compounds by i n s t r u m e n t a l semantic arguments, but t h i s is m e r e l y a d e s c r i p t i v e g e n e r a l i s a t i o n and has no e x p l a n a t o r y v a l u e . As it does not seem that any of the p r i n c i p l e s discussed thus far could possibly account for this d i s t r i b u t i o n of the data, we must assume that it has something to do w i t h s p e c i f i c restrictions on w o r d - f o r m a t i o n .

3 . 4 . 6 . VV Compounds Compounds of t h i s type are rare and are considered by S e l k i r k ( 1 9 8 2 : 15) in f a c t to be n o n - e x i s t e n t . In Boase-Beier & Toman (in p r i n t ) it is shown that VV compounds m i g h t t h e o r e t i c a l l y be thought possible, assuming that one verb takes the other as an argument. But we have a l r e a d y seen that verbs cannot be arguments. In a phrase such as (to} attempt to leave it m i g h t 49 See also Lieber (1983: 2 6 2 f . ) 50 Note that these compounds are not ruled out per se but only their occurrence in s y n t a c t i c structures is ruled out. We should thus expect compounds in w h i c h such verbs form the f i r s t element to be acceptable. In E n g l i s h t h i s type is rather unproductive. In German, however, compounds such as Briefschreibkunst " l e t t e r - w r i t e art", i . e . ' t h e art of w r i t i n g l e t t e r s ' , Autowaschanlage " c a r - w a s h - f a c i l i t y " , TeppichMlopfgerät " "carpet-beat-instrument" i.e. ' c a r p e t - b e a t e r ' , belong to a very p r o d u c t i v e type. 51 Lieber ( 1 9 8 3 : 2 6 4 ) has an example stream-vade in w h i c h , she m a i n t a i n s , stream is the local s e m a n t i c argument of vade. We m a i n t a i n that such a compound would be u n g r a m m a t i c a l , and that it does n o t , in f a c t , e x i s t . 75

seem as though just t h i s is the case, i . e . , as though leave is the argument of attempt, but we assume that the s i t u a t i o n in t h i s case is d i f f e r e n t because the 0-role is not assigned to the verb but to a syntactic phrase c o n t a i n i n g a COMP-node. As this must be regarded as the head of the phrase, the 0-role is thus assigned to the COMP-node, w h i c h is [ + N ] . There remains, however, the case in w h i c h no 0-role assignment has taken place, that is, in which one verb is a semantic argument of the other. But t h i s case is e q u a l l y unacceptable, as verbs cannot be semantic arguments, e i t h e r . The only possible VV compounds w h i c h could occur seem thus to be co-ordinated ones. Both types of co-ordination discussed in section 3.3.5 seem in p r i n c i p l e to be possible, as the f o l l o w i n g examples show: ( 3 - 7 2 ) a . thinJc-speaJc meaning " t h i n k and speak" smile-hesitate meaning " s m i l e and hesitate" b. run-ualk meaning "to walk such that one is (in r e a l i t y ) running" sinff-speaJf meaning "to speak such that one is singing" Although these compounds are t h e o r e t i c a l l y possible, the type is very unproductive. One reason for this is the restriction that only words of the same category and w i t h the same characteristics can be co-ordinated. Thus VV compounds cannot be formed f r o m verbs w i t h d i f f e r e n t argument-structures, i . e . , compounds l i k e the f o l l o w i n g w i l l not be possible: (3-73)

*sit-sell (a car),

* enter-say

It is clearly the case that the compounds in (3-73) are much less acceptable than those in ( 3 - 7 2 ) . As there are no other relations w h i c h could hold between two verbs, in the case where 0 - m a r k i n g had not t a k e n p l a c e , we shall assume that VV compounds other than co-ordinated ones are impossible.

3 . 4 . 7 . AV Compounds As neither verbs nor adjectives can be either proper arguments or semantic arguments, compounds of this type w i l l obviously be restricted as far as f u l f i l m e n t of argument-structure is concerned. A d j e c t i v e s w h i c h can take verbs as arguments, e . g . able, might be expected to give rise to compounds such as: 76

(3-74)

able-leave m e a n i n g "to leave leave"

such that one is able to

However, as we have already noted for VV compounds, these compounds w o u l d also be ruled out owing to the r e s t r i c t i o n that only n o m i n a l e l e m e n t s can be arguments. The occurrence of structures l i k e able to go in sentence syntax can in p r i n c i p l e be explained in the same way as attempt to leave. Other compounds, in w h i c h the a r g u m e n t - s t r u c t u r e of the V is not f u l f i l l e d , should be possible if the non-head is an adverb. This in fact seems to be the case, as the examples g i v e n by Lieber ( 1983: 2 6 4 ) show: (3-75)

fast-dance,

slou-dry

3 . 4 . 8 . PV Compounds It seems, on the f a c e of it, c l e a r enough that PV compounds e x i s t ; compare t h e f o l l o w i n g e x a m p l e s : (3-76)

overdo, underestimate, outbid

N e i t h e r of the two e l e m e n t s of the compound can assign the other a β-role, as n e i t h e r p r e p o s i t i o n s nor verbs can be arguments. As they cannot be semantic arguments e i t h e r , readings based upon t h i s p o s s i b i l i t y are also u n a c c e p t a b l e . In a construction such as to underestimate the effect, the argument-structure of the headverb is realised outside the compound; t h i s conforms to the p r i n c i p l e s of p e r c o l a t i o n and the r e a l i s i n g of argumentstructures in sentences. But the non-head must in t h i s case have a d i f f e r e n t status f r o m that of semantic a r g u m e n t . As an aid to establishing what e x a c t l y its status is, it is of interest to observe that compounds such as the f o l l o w i n g are u n g r a m m a t i c a l : (3-77)

u

betueen-go, (someone)

u

betueen-read

(a

book),

u

behind-value

If we assume here again that p r e p o s i t i o n s such as between and behind do not have an ^ a d d i t i o n a l entjry in the l e x i c o n as adjectives (cf. also betueenmost, behindmost) , then the assumption seems reasonable that just these prepositions cannot occur as adverbs. The f a c t t h a t , p a r a l l e l to compounds l i k e overred, under-ripe, in w h i c h the non-head appears to be an adverb, there are no compounds l i k e between-red, behind-ripe, appears to support t h i s a s s u m p t i o n . It thus seems to be the case that compounds w h i c h at f i r s t sight m i g h t be considered as b e l o n g i n g to the type PV are in 77

actual f a c t of the type A d v V . Thus underestimate does not (and cannot) mean something l i k e "to estimate under" but rather "to estimate too l i t t l e " and overdo means "to do e x a g g e r a t e d l y " . In other words, most prepositions are also listed in the l e x i c o n as adverbs, w i t h t h e corresponding m e a n i n g s , i . e . "too l i t t l e " f o r under, "exaggeratedly" for over, and so on.

3 . 4 . 9 . NA Compounds Independently of w h e t h e r or not p r e d i c a t i o n is regarded as an instance of θ-role a s s i g n m e n t , it seems clear that there is one main r e s t r i c t i o n on NA compounds: n a m e l y , that if the non-head of the compound is that of w h i c h the a d j e c t i v e is p r e d i c a t e d , the compound w i l l b e u n g r a m m a t i c a l . T h i s f a c t i s consistent w i t h t h e observation made, about NV compounds: the assignment of P - r o l e s by t h e head t o t h e non-head w i l l a l w a y s r e s u l t i n u n g r a m m a t i c a l compounds of the c a t e g o r i e s A and V. In t h i s case, as in the case of NV compounds, t h i s is a r e f l e c t i o n of the f a c t that the a r g u m e n t - s t r u c t u r e of the head w o u l d have to p e r c o l a t e to the w h o l e compound, so that it w o u l d then have to be r e a l i s e d both e x t e r n a l l y and i n t e r n a l l y , w h i c h is not p o s s i b l e . If there is no assignment of θ-roles, the noun in the non-head could t h e o r e t i c a l l y be a s e m a n t i c argument to the a d j e c t i v e . In practice, however, it is d i f f i c u l t to imagine how an adjective could be m o d i f i e d l o c a l l y , t e m p o r a l l y or i n s t r u m e n t a l l y by a n o u n ; such compounds do not in f a c t e x i s t . H o w e v e r , there seems in E n g l i s h to be a m a r g i n a l p o s s i b i l i t y for f o r m i n g compounds of t h i s t y p e , whereby the n o n - h e a d is a s e m a n t i c argument of a type w h i c h does not g e n e r a l l y b e l o n g to the set of possible s e m a n t i c arguments, n a m e l y a basis for c o m p a r i s o n . T h i s p o s s i b i l i t y leads to the p r o d u c t i o n of e x a m p l e s such as: (3-78)

blood-red,

crystal-clear, iron-hard, ice-blue

We have said that such compounds b e l o n g to a m a r g i n a l t y p e . By t h i s we do not mean that they are q u a n t i t a t i v e l y rare but that t h i s type, a l t h o u g h c o n t a i n i n g a large number of l e x i c a l i s e d compounds, is rather r e s t r i c t e d in its p r o d u c t i v i t y for ad hoc compounds. T h e f o l l o w i n g coined e x a m p l e s i l l u s t r a t e t h i s : (3-79)

inJc-red, glass-clear, steel-hard, vater-blue

There seems no apparent reason for the f a c t that these compounds are less l i k e l y to occur in standard English t h a n those in (37 8 ) . All c o n t a i n a substantive as f i r s t m e m b e r , w h i c h is s t e r e o t y p i c a l l y associated w i t h a s p e c i f i c c h a r a c t e r i s t i c . T h i s observation is substantiated by the fact that compounds corresponding to the latter three in ( 3 - 7 9 ) exist in G e r m a n , n a m e l y glasklar, eisenhart, eisgrau. But in E n g l i s h they sound

78

odd. s ? We n o t i c e , h o w e v e r , that they cannot be r u l e d out by p r e e m p t i o n , because, at least in the case of ink-red and ice-grey, no compound e x i s t s i n t h e l e x i c o n w h i c h denotes e x a c t l y t h i s colour. We s h a l l thus m a i n t a i n that t h i s is not a p r o d u c t i v e type of c o m p o u n d , and that e x i s t i n g compounds are l i m i t e d to l e x i c a l i s e d cases, w h e r e possible e x t e n s i o n is by means of a p e r i p h e r a l , not a c e n t r a l , process. An important question in c o n n e c t i o n w i t h NA compounds is the· q u e s t i o n of the e x i s t e n c e of a n u l l - m o r p h e m e ( 0 - m o r p h e m e ) . In order to i l l u s t r a t e what is m e a n t by t h i s , let us c o n s i d e r s t r u c t u r e s of the f o l l o w i n g t y p e : (3-80)

star-struck, sun-blinded, uave-vashed

At f i r s t s i g h t , there seem to be f o u r possible s t r u c t u r e s such words. These are i l l u s t r a t e d as f o l l o w s :

for

(3-81)a.

52 It is a d m i t t e d l y not easy to decide w h e t h e r t h i s oddness rests in the f a c t that the type is u n p r o d u c t i v e , whereby l e x i c a l i s e d e x a m p l e s f o r m a n e x c e p t i o n , o r w h e t h e r i t results s i m p l y f r o m the f a c t that these compounds are not l e x i c a l i s e d whereas those in ( 3 - 7 8 ) a r e . H o w e v e r , if we compare other p a i r s of l e x i c a l i s e d and n o n - l e x i c a l i s e d e x a m p l e s , such as bookseller and dress-seller or bookshop and tent-shop, we do not f e e l the l a t t e r e x a m p l e in each case to be odd or in any way " p o e t i c " . T h i s s u b s t a n t i a t e s our v i e w that the p r o b l e m in ( 3 - 7 9 ) is not s i m p l y that these compounds are ad hoc, but l i e s in the f a c t that t h e y are f o r m e d a c c o r d i n g to a n o n - p r o d u c t i v e p a t t e r n . A l t h o u g h L i p k a ( 1 9 6 6 ) m a i n t a i n s that t h i s type i s very p r o d u c t i v e i n E n g l i s h , i t i s o u r v i e w that t h i s j u d g e m e n t arises f r o m a d e f i n i t i o n o f p r o d u c t i v i t y w h i c h f a i l s t o d i s t i n g u i s h between lexicalised and n o n - l e x i c a l i s e d compounds. 53 T h i s is assumed in many t r a d i t i o n a l s t u d i e s to be the process i n v o l v e d in the f o r m a t i o n , for e x a m p l e , of verbs such as father. See Marchand ( 1 9 6 9 ; 3 ) . 79

b.

The f i r s t two structures (3-.81a) and (3-81b) i n v o l v e a 0-morpheme w h i c h changes the compound verb or past p a r t i c i p l e into an a d j e c t i v e . Cases (3-81c) and (3-81d) i n v o l v e the assumption that the s u f f i x -ed belongs both to category A and to category V. Now if the complex words in ( 3 - 8 0 ) had the structure in (3-81b) or ( 3 - 8 1 d ) , that is, if they were NA compounds, then one would expect them to have a marginal i n t e r p r e t a t i o n of the type "blinded l i k e the sun", p a r a l l e l to cases such as blood-red, leaf-green etc., but t h i s meaning does not seem even to be m a r g i n a l l y possible in these cases. This leaves structures (381a) and (3-81c). We shall assume, for reasons s i m i l a r to those given in Lieber ( 1 9 8 0 ) , but w h i c h we s h a l l not discuss here, that what is involved is a 0-morpheme, as in (3-81a), One o b j e c t i o n to this structure might at f i r s t sight appear to reside in the f a c t that in t h i s structure an NV compound sun-blind is first generated, and, as we r e c a l l from the discussion above, NV

80

compounds in w h i c h the N is the subject of the V are impossible. However, it is important to realise that they are not per se impossible but o n l y in the sense that they cannot appear in sentences. We should expect them to exist w o r d - i n t e r n a l l y , where such questions do not arise. That there must be the addition of a 0-morpheme, i . e . , that the complex word cannot be a verb, is clear because, if this were the case, the agent w o u l d also be able to appear in a sentence in w h i c h it occurred. Thus, although blinded can be both an a d j e c t i v e (as in the blinded man) and a verb (as in he uas being- blinded (by the sun), the word sunblinded can o n l y be an a d j e c t i v e . And the expression he uas being sun-blinded is, in f a c t , u n g r a m m a t i c a l . A d j e c t i v e s such as: (3-82)

eye-blinding, sand-beating

are analysed a n a l o g o u s l y . These cases, too, derived a d j e c t i v e s , rather than compounds.

are

assumed to be

3 . 4 . 1 0 . VA Compounds Compounds of t h i s t y p e , m a i n t a i n s S e l k i r k ( 1 9 8 2 ) , do not e x i s t ; Lieber ( 1 9 8 3 ) sees the type as f a i r l y u n p r o d u c t i v e . If we assume that verbs and a d j e c t i v e s can be neither arguments nor semantic arguments, then there w i l l be no possible interpretation for such compounds. Lieber (1983: 2 6 1 f . ) suggests that there are compounds such as appear-productive, in w h i c h the verb appear takes productive as its a r g u m e n t . This w o u l d be impossible in our system, and, in f a c t , our i n t u i t i o n suggests that such compounds are u n g r a m m a t i c a l .

3 . 4 . 1 1 . AA Compounds

Because a d j e c t i v e s cannot be assigned 8-roles, we should not expect any AA compounds to exist except co-ordinated ones, that is, compounds such as the f o l l o w i n g : ( 3 - 8 3 ) purple -blue, darJc-green These compounds u s u a l l y have the meanings "a shade of blue w h i c h is p u r p l e " , i . e . " p u r p l i s h - b l u e " , and "a shade of green w h i c h is dark" (see 3 . 3 . 5 ) . This compound-type is not very p r o d u c t i v e in SL, being l i m i t e d in general to colour-compounds. Such compounds might be interpreted as h a v i n g an adverbial m o d i f i e r in the non-head but as 81

t h i s w i l l n o t a f f e c t t h e i r possible f u r t h e r consider t h i s q u e s t i o n .

readings,

we shall

not

3 . 4 . 1 2 . PA Compounds T h e o r e t i c a l l y , PA compounds could be expected to be very restricted. This c i r c u m s t a n c e is explained by the f a c t s about percolation and the realisation of argument-structures. P r e p o s i t i o n s and a d j e c t i v e s cannot be either proper or semantic a r g u m e n t s . We must t h e r e f o r e assume that compounds such as overripe are to be analysed in some other w a y . The o n l y f e a s i b l e analysis seems to be t h a t , p a r a l l e l to PV compounds, these compounds consist of an adverb and an a d j e c t i v e . This assumption is supported both by the semantics of these compounds - overripe means the same as overly ripe, i . e . "too r i p e " - and by the f a c t that l e x i c a l items w h i c h o n l y appear as p r e p o s i t i o n s , and do not have a separate entry as adverbs, cannot f o r m such compounds: (3-84)

betueen-ripe,

behind-stated

On the other h a n d , l e x i c a l items w h i c h are l i s t e d both as p r e p o s i t i o n s and as adverbs, such as over and under, o c c u r very f r e q u e n t l y in compounds of t h i s t y p e .

3 . 4 . 1 3 . NP Compounds NP compounds are p r e d i c t e d by our system to be i m p o s s i b l e . I n t u i t i o n t e l l s us that compounds l i k e treeunder or chairabove do not e x i s t . L i e b e r ( 1 9 8 3 ) accounts for t h i s f a c t by saying t h a t all p r e p o s i t i o n - f i n a l compounds are r u l e d out by an independent l e x i c a l c o n s t r a i n t on the class of p r e p o s i t i o n s . We are of the o p i n i o n , h o w e v e r , that such a c o n s t r a i n t is not r e q u i r e d , as the facts about argument-structure and semantic facts about p r e p o s i t i o n s w i l l s u f f i c i e n t l y account f o r t h e f a c t that these compounds are r u l e d o u t . In a h y p o t h e t i c a l NP compound such as treeunder, we see that under w o u l d pass its argument-structure to the c o m p o u n d , and the compound w o u l d t h u s , in a sentence, have ^o r e a l i s e its o b j e c t - r o l e e x t e r n a l l y . Thus, expressions such as treeunder the table w o u l d then a r i s e . But what is then the nonhead-noun? The p r o b l e m w h i c h arises is the same one we have met already in the case of NV compounds: the head can o n l y 0 - m a r k one e l e m e n t for each role a n d , because it must do so e x t e r n a l l y , the non-head cannot r e c e i v e a θ - r o l e . It can thus o n l y be a s e m a n t i c a r g u m e n t . But because a p r e p o s i t i o n cannot be m o d i f i e d , at least 54 T h i s e x p l a n a t i o n is also put f o r w a r d by Lieber ( 1 9 8 3 : 2 6 1 ) . H o w e v e r , she suggests no p a r a l l e l e x p l a n a t i o n for PN and PV compounds. We see no reason to treat the non-heads of these compound-types d i f f e r e n t l y .

82

not by a n o u n ,

NP compounds w i l l be ruled out.

3 . 4 . 1 4 . VP Compounds N e i t h e r the verb nor the p r e p o s i t i o n can be a proper or a s e m a n t i c a r g u m e n t , and such compounds are thus not g r a m m a t i c a l . As we have a l r e a d y said above (see 3 . 3 . 1 ) , we s h a l l not regard f o r m s such as sit-in, eat out as VP compounds.

3 . 4 . 1 5 . AP Compounds A g a i n , as n e i t h e r a d j e c t i v e s nor prepositions can be semantic or proper a r g u m e n t s , t h i s type should be u n g r a m m a t i c a l . E m p i r i c a l e v i d e n c e substantiates t h i s : (3-85)

nice-out,

green-under

H o w e v e r , if the f i r s t e l e m e n t is an a d v e r b , rather than an adjective, it could m o d i f y the preposition, resulting in compounds such as far-out e t c . T h i s is to be seen as a m a r g i n a l p o s s i b i l i t y f o r f o r m i n g AdvP compounds ( i f , i n f a c t , they a r e c o m p o u n d s , w h i c h is not r e a l l y c l e a r ) , s u b j e c t to the necessary semantic restrictions.

3 . 4 . 1 6 . PP Compounds L i e b e r ( 1 9 8 3 ) gives e x a m p l e s such as into and onto as PP compounds, but later m a i n t a i n s that such compounds, l i k e all compounds with a prepositional head, are ruled out for independent reasons. In our system, compounds of t h i s type need not be r u l e d out by such r e s t r i c t i o n s , a l t h o u g h there is good reason to suppose t h a t r e s t r i c t i o n s of t h i s nature do, in f a c t , e x i s t . As p r e p o s i t i o n s can be n e i t h e r proper nor semantic a r g u m e n t s , PP compounds w o u l d receive no i n t e r p r e t a t i o n . A l t h o u g h c o - o r d i n a t i o n is in theory p o s s i b l e for such c o m p o u n d s , t h i s does not seem to a p p l y to words l i k e into. Into does not mean "a sort of to, w h i c h is also i n " . We shall t h e r e f o r e assume that the f a c t that prepositions may not be m o d i f i e d also e x p l a i n s why they cannot occur as c o - o r d i n a t i v e structures. In a d d i t i o n , the f a c t that words l i k e into do not b e l o n g to a p r o d u c t i v e p a t t e r n ( c f . in-uncler, on-under) substantiates t h i s a s s u m p t i o n .

*## 55 It is not q u i t e clear to us w h e t h e r or not it makes sense to speak of a p r e p o s i t i o n ' s b e i n g m o d i f i e d in a phrase such as nearJy under the tattJe. For a d i s c u s s i o n of t h i s q u e s t i o n , see Boase-Beier & Toman (in p r i n t ) .

83

We have seen, then, that there are several p r i n c i p l e s at w o r k , which together account for possible and rule out impossible compounds. S p e c i f i c a l l y , we have been d e a l i n g w i t h the headp r i n c i p l e and the p r i n c i p l e s of θ-assignment and p e r c o l a t i o n , as w e l l as restrictions on the nature of arguments and on the m o d i f i c a t i o n of a d j e c t i v e s and prepositions. Our brief survey of the types of compound e x i s t i n g in E n g l i s h has, we hope, served to show in general w h i c h compounds are possible and w h i c h are not, though we have, of course, l e f t out of consideration some q u e s t i o n s of pragmatic l i k e l i h o o d . It remains to be seen in the course of the next chapter, whether all these predictions apply to poetic compounds, too. We expect to f i n d that they do not.

84

PRINCIPLES OF POETIC LANGUAGE

4.1.

The Generation of Poetic Compounds

In Chapter 3 we established t h a t , in order to study poetic compounds, we must be concerned w i t h a three-way d i v i s i o n of compounds into (i) those possible in SL and PL, ( i i ) those impossible or m a r g i n a l in SL but possible and productive in PL and ( i i i ) those not n o r m a l l y possible in either SL or PL. Compounds of the f i r s t type are accounted for by the system of compounding discussed in Chapter 3. The third type of compound can, for obvious reasons, not be the subject of a systematic study and is thus of l i t t l e interest to us here. It is w i t h the second type of compound, which constitutes an area of what we have called PL, that t h i s chapter is concerned. Such compounds are not generated by the system presented in Chapter 3, w h i c h is of course a desirable r e s u l t , for they do not u s u a l l y occur in standard language. Consider the f o l l o w i n g compounds, all of w h i c h were f o u n d in the poetry under c o n s i d e r a t i o n : (4-1)a.

hover-still [ V A ] ( + R 4 2 ) , +ducJc-eggshell emptiness [ N N ] (+SS 6 8 ) , grin-long [ N A ] ( R 1 0 4 ) , flover-lovely-false [ [ N A ] A ] ( + N M 2 1 ) , deep-root-ueather [ [ A N ] V ] ( + M T 6 i > ) , shadou-blacJc sharp [ [ N A ] A ] ( + M 5 9 ) , hover-search [ V V ] ( + SS 2 7 ) , mistlost [ N A ] ( + M 3 4 ) , summer-hang [ N V ] ( + I 1 8 ) , horse-uorship [ N V ] ( + W N P 5 7 ) , drought-auaited [NA] ( + I 1 2 ) , fruit-offer [NV] (+NM 3 0 ) ;

b.

bread-crumb sill [ N N ] C+BS 4 8 ) , blue-dark [AA] ("""HT 3 1 ) , rainlight [ N N ] ( N P 8 5 3 ) , uorld uonders [ N N ] (DH 3 3 ) , angel tree [ N N ] ( + T O 3 2 ) .

We shall m a i n t a i n that these compounds are not compounds of SL, in the sense that those in ( 4 - l a ) are unacceptable and those in ( 4 - l b ) , a l t h o u g h not per se u n a c c e p t a b l e , may only occur under certain c o n d i t i o n s (such as c e r t a i n types of context-dependency), w h i c h are not necessary to t h e i r occurrence in poems. It is important to stress t h a t , as we have stated above, r e f e r r i n g to certain compound types as "impossible in SL" is not to be taken as meaning that such compounds may never occur in SL because 1

One might argue about the assignment of i n d i v i d u a l compounds to one group or the o t h e r . This does not m a t e r i a l l y a f f e c t what we have to say.

85

o b v i o u s l y , under special c i r c u m s t a n c e s , they m a y . H o w e v e r , t h e q u a l i f i c a t i o n "under special circumstances" is important because we w i s h to m a i n t a i n that these circumstances amount to a transference of PL r u l e s into the area of SL. If these compounds are not generated by our compounding system, h o w , t h e n , are they generated? The answer to t h i s question was g i v e n in a broad sense in Chapter 1 , where we stated that we regard these compounds as the result of an interaction between the p r i n c i p l e s of the grammar of standard E n g l i s h and certain p r i n c i p l e s of poetic language. In the f o l l o w i n g pages we shall discuss in d e t a i l those poetic compoundtypes w h i c h are p r o d u c t i v e in the texts under c o n s i d e r a t i o n and say, f i r s t l y , why they are i m p o s s i b l e or of m a r g i n a l status in SL, and, s e c o n d l y , why they are possible in PL. In answering this latter question we s h a l l try to characterise poetic language p r i n c i p l e s , and to say what consequences the assumption of their existence h a s , both for the area of c o m p o u n d i n g and for other aspects of poetic l a n g u a g e . We shall f u r t h e r try to establish just what f o r m the i n t e r a c t i o n between p r i n c i p l e s of poetic language and p r i n c i p l e s of the grammar of SL takes. It does not make sense to assume that the poetic p r i n c i p l e s in question are s p e c i f i c to the area of compounding. Such an assumption w o u l d i m p l y that all areas of the grammar are coupled w i t h s p e c i f i c p o e t i c p r i n c i p l e s ; t h i s w o u l d result in an e n o r m o u s l y c o m p l e x system of poetic language w h i c h w o u l d both lack e x p l a n a t o r y power and w o u l d f a i l to a c c o u n t , for e x a m p l e , for the i n t u i t i o n that a compound such as grin-long and a syntactic construction such as as long as a grin, have something more than just t h e i r l e x i c a l items in c o m m o n . F u r t h e r m o r e , as we have e x p l a i n e d in Chapter 3.1, it seems reasonable to assume ( f o r reasons indepedent of the nature of PL) that neither compounding nor the larger area of w o r d - f o r m a t i o n are to be regarded as autonomous components of the grammar. If this be so, then it seems u n l i k e l y that the p o e t i c p r i n c i p l e s w h i c h interact w i t h compound p r i n c i p l e s w o u l d b e c o m p o u n d - s p e c i f i c . We shall t r y , t h e n , to e x a m i n e in what relation the assumption of the general nature of p o e t i c p r i n c i p l e s stands to f a c t s about p o e t i c language.

4 . 2 . Metaphor, Metonymy, Synecdoche and Ellipsis Consider the f o l l o w i n g poetry under discussion:

examples

of

complex

words

from

the

(4-2)a. [NV] moth-walk ( + A B 2 6 ) , deep-root-ueather ( + M T 6 +6 ) , fruitoffer (nakedness) ί + Ν Μ 3 0 ) , gold-Jbeat (hair) ( SP 2 ) ;

b.

[[NN]ed]A child-minded (UM 4 4 ) , fly-footed (DV 1 1 ) , dragon-toothed ( + N M 3 1 ) ;

86

( + IF 6 7 ) ,

angel-eyed

c. [ N A ] goose-vhite (Fu 3 5 ) ; map-logical ( + Μ 6 0 ) ; afternoonsubtle ( T O 1 6 ) , drovned hair silent [ C A N ] A ] ( R 4 6 ) ; d.

IVA]

hover-still ( + R 4 2 ) , hover-poised (.K 9 6 ) , sag-heavy (+MT 5 5 ) , hum-rigid ( M T 6 7 ) ;

e. [ N N ] rusk-vhisper (IF 7 4 ) , ducJf-eggshel 1 emptiness ( + SS 6 8 ) , cat-ecstasy (RE 1 1 4 ) , root-hunger ("""IF 3 2 ) ; £ .[ N N ]

shillinq-hair ( + J A 1 5 ) ; leopard-leaves ( + U W 3 8 ) , ocean (*DH 4 9 ) , hird-of-paradise cloak (SM 2 7 ) ;

hyphen

g. [ N A ] heron-pale ( + M T 1 0 1 ) , oxygen-red ( + A 9 7 ) , ( + DV 4), prayer-young ( + A 2 3 8 ) .

mirror-close

All these compounds w o u l d be at best of m a r g i n a l status in SL; many w o u l d be considered u n a c c e p t a b l e . To see why t h i s is so, it is necessary to r e c a l l some of what was said in Chapter 3. To start w i t h a verbal compound of group ( 4 - 2 a ) , for example fruitoffer, it is apparent, even w i t h o u t consideration of the context in w h i c h this compound appears, that it cannot be a possible compound of SL. It was established in Chapter 3 that the f i r s t element of an NV compound can n e i t h e r be the subject nor the o b j e c t of the verb in the h e a d - p o s i t i o n . So to fruit-offer cannot mean "to o f f e r f r u i t " nor can it mean " f r u i t o f f e r s " , where fruit is the subject ( t h i s latter w o u l d in any case v i o l a t e s e l e c t i o n a l r e s t r i c t i o n s on the verb offer, w h i c h g e n e r a l l y requires an a n i m a t e s u b j e c t ) . In other words, the compound cannot somehow " m e t a p h o r i c a l l y " mean "to o f f e r f r u i t " as the compound cannot exist w i t h t h i s m e a n i n g . The o n l y r e m a i n i n g p o s s i b i l i t y is to assume that θ-assignment does not take place in the compound and that fruit is a s e m a n t i c argument of offer. From the set { l o c a t i o n , t i m e , i n s t r u m e n t ) , o n l y t h e latter is possible f o r fruit. But the reading "to o f f e r something by means of f r u i t " does not appear to m a k e sense. If we look at the poem in w h i c h the compound o c c u r s , we f i n d as its i m m e d i a t e context: ( 4 - 3)

his partner [ . . . ] fruit-offering in vain her nakedness ( + N M 3 0 )

O b v i o u s l y , t h e n , the is her nakedness and seem to be "to o f f e r to be on the f a c e of

subject of offer is his partner, the o b j e c t the correct r e a d i n g of the compound w o u l d l i k e f r u i t " . That is, the word fruit seems it a s e m a n t i c argument expressing the basis

87

of comparison. B u t , as we said in Chapter 3, we assume that (in SL, at least) semantic arguments constitute a closed class. If, in PL, the argument "basis for comparison" were to belong to t h i s group, then it w o u l d seem that we have to do here w i t h a poetic p r i n c i p l e . So w h a t , should we assume, is the correct f o r m u l a t i o n of such a p r i n c i p l e ? We c o u l d , of course, account descriptively along these lines for compounds such as fruit-offer (nakedness), gold-beat (hair) and so on by postulating a poetic p r i n c i p l e such as the f o l l o w i n g : (4-4)

For NV compounds the set of semantic arguments of w h i c h N can be a member consists of {location, t i m e , instrument, basis for comparison}.

But, apart f r o m i n t u i t i o n s one might have about the nature of p r i n c i p l e s , there is clearly at least one important t h i n g wrong w i t h ( 4 - 4 ) : it is so s p e c i f i c in nature that it e n t i r e l y fails to account for the speaker's i n t u i t i o n that there are many other aspects of PL, i n c l u d i n g those represented in the other compounds in ( 4 - 2 ) , w h i c h have something in common w i t h t h i s phenomenon. Let us look at this o b j e c t i o n more closely, by considering some other examples f r o m ( 4 - 2 ) , namely the group of a d j e c t i v e s map-logical, goose-vhite, and so on (group ( 4 - 2 c ) ) . These compound a d j e c t i v e s are at f i r s t sight interpretable and might thus seem not to be e s p e c i a l l y "poetic" at a l l . But in f a c t this is not the case. Imagine I were giving a description of a man to the p o l i c e . I might say he has snou-uhite hair but I should certainly not say he has goose-uhite hair. Then recall what we said in Chapter 3 about NA compounds: a thematic reading is not possible, and a non-thematic reading is u s u a l l y not possible because it is g e n e r a l l y d i f f i c u l t to m o d i f y an a d j e c t i v e l o c a l l y , temporally or i n s t r u m e n t a l l y . A n d , as we have said above, the set of semantic arguments does not contain "basis for comparison" in any case, which f u r t h e r accounts for the fact that the NA compounds of this type are not productive; the only ones f o u n d are lexicalised compounds based on stereotypes of the noun in the non-head-position. In the examples in ( 4 - 2 c ) above, goose-white and map-logical might also be said to be based on stereotypes of goose and map, although they are not l e x i c a l i s e d . Drouned hair silent and afternoon-subtle, on the other h a n d , are not o n l y nonl e x i c a l i s e d compounds, but they are not based on stereotypes of the nouns. It seems that for these NA compounds there is a p r i n c i p l e at work s i m i l a r to that at work in the NV compounds above. In other words, the descriptive " p r i n c i p l e " given in ( 4 - 4 ) would have to be f o r m u l a t e d to apply to these cases too. Let us consider a type of compound in w h i c h "principle" ( 4 - 4 ) would not even be d e s c r i p t i v e l y adequate - that of the NN compounds given in ( 4 - 2 f ) above. In order to see why shilling hair or leopard leaves are not possible compounds in SL, we must r e c a l l what was said in Chapter 3 about the interpretation of NN compounds. There it was established that (apart f r o m contextdependent compounds) an NN compound must be interpretable either by means of an inherent r e l a t i o n , as in choir-leader, a 88

s t e r e o t y p e - r e l a t i o n , as in shoe factory, or a basic r e l a t i o n of co-ordination as in composer-director (or possibly by means of a local r e l a t i o n ) . Thus, compounds such as those in ( 4 - 2 f ) are only possible in SL if the r e l a t i o n between t h e i r elements is stated in the context. However, this is not the case w i t h an example such as shilling hair, above, w h i c h means "hair as bright as a s h i l l i n g " , "silver h a i r " . In order to account for the presence of such compounds, we might then postulate an a d d i t i o n a l r e l a t i o n of s i m i l a r i t y w h i c h w o u l d have in common w i t h the basic r e l a t i o n of c o - o r d i n a t i o n and the marginal basic r e l a t i o n of l o c a t i o n that it would be context-independent but not inherent in one of the elements of the compound. Thus, in order to account for the compounds in ( 4 - 2 f ) we should have to propose an a d d i t i o n a l " p r i n c i p l e " such as: (4-5)

In PL, the set of basic r e l a t i o n s {location, co-ordination, similarity}

consists

of:

But this would be subject to the same c r i t i c i s m as that directed at ( 4 - 4 ) above, as it misses the s i m i l a r i t y between examples in ( 4 - 2 f ) and those in the other groups. It also f a i l s to account correctly for the other KN group ( 4 - 2 e ) , for w h i c h the same type of interpretation w i l l not do. W h i l e the interpretation of shilling-hair w i l l be some v a r i a t i o n on the basic interpretation "hair w h i c h is l i k e a s h i l l i n g " , obviously rusJc-uhisper is not "whisper w h i c h is l i k e a rusk" but rather "whisper l i k e the sound a rusk makes". We thus require yet other means to interpret the compounds in ( 4 - 2 e ) , and to e x p l a i n their presence in PL. As for group ( 4 - 2 d ) , w h i c h consists of VA compounds, these were found in Chapter 3 not to e x i s t . A n d , assuming that in PL adjectives could be m o d i f i e d by a semantic argument and that, by "principle" ( 4 - 4 ) above, these, in PL i n c l u d e "basis for comparison", we should perhaps be a b l e to e x p l a i n the compounds in ( 4 - 2 c ) , but we should s t i l l have no e x p l a n a t i o n for the compounds in ( 4 - 2 g ) , as they w o u l d now receive the r e a d i n g s "pale as a heron", "red as oxygen", e t c . , when in f a c t they mean "pale as a h e r o n ' s eyes", "red as blood c o n t a i n i n g o x y g e n " , something w h i c h obviously involves more complicated interpretation procedures than " p r i n c i p l e " ( 4 - 4 ) w o u l d a l l o w . S i m i l a r l y , f o r group ( 4 - 2 b ) , " p r i n c i p l e " ( 4 - 5 ) w o u l d give us the reading "having teeth l i k e dragons", "having feet l i k e f l i e s " , and so on, when in fact the compounds mean "having teeth l i k e dragon's teeth" (- but note the a m b i g u i t y of this compound in its actual context; see A p p e n d i x - ) , "having f e e t l i k e f l y ' s f e e t " , etc. The f a c t that these compounds are not p o s s i b l e , or o n l y of very m a r g i n a l status, in SL, suggests that we have to do w i t h a s t r i c t l y p o e t i c p r i n c i p l e or p r i n c i p l e s of a general nature. It is evident that in all the cases above, there are elements of metaphor, t a k i n g "metaphor" here in its i n t u i t i v e sense, n a m e l y to mean " c o m p a r i s o n " . N o w , it may seem that the presence of 1 e x i c a u s e d compounds such as hlood-red, snou-uhite, snou-goose, 89

to name just a f e w , indicates that the m e t a p h o r i c a l process also plays a role in compounding in SL. We shall argue, however, that this process occupies only a very marginal position in SL. There are five main reasons for this assumption. Firstly, as established above, almost all compounds occurring in SL w h i c h have any sort of metaphorical interpretation are l e x i c a l i s e d . Even compounds in w h i c h the comparison is based on stereotypes, and w h i c h would thus be easy to understand, o n l y occur in SL as l e x i c a l i s e d compounds. Secondly, w h i l e metaphoric compounds of the types KN and NA do occur as lexicalised words in SL, other compound-types, such as NV and V A , do not. There are no, or very f e w , lexicalised examples p a r a l l e l to moth-ualJs or hoverstill. In other words, the m e t a p h o r i c a l interpretation is l i m i t e d to certain types in SL but not in PL. T h i r d l y , there is a distinct tendency to avoid metaphor - except in certain circumstances - in SL, whereas there is a tendency to employ it in PL. Having said that context-independent examples do not occur in SL, it might be thought that context-dependent ones do. This is not so, though. Consider the f o l l o w i n g passage of SL: (4-6)

The juba plant, of which some leaves are dark green and others mottled yellow, grows on the banJcs of the Amazon, Local men wear the leopard leaves as hair ornaments.

The metaphorical interpretation of the compound leopard-leaves is clear from the context. However, it is nonetheless unacceptable, or at least very odd. This tendency to avoid metaphor in SL is also seen in the f a c t that s i m i l e s , in which the relation of comparison is clearly stated, are much more acceptable than strict metaphors, in w h i c h it is not, though s i m i l e s , too, are subject to more constraint in SL than in PL. Consider the f o l l o w i n g poetic l i n e : (4-7)

The bowl of sugar is a cup of lighted bulbs (AB 67)

This is not a possible sentence of a non-poetic f o l l o w i n g , if a l i t t l e f a n c i f u l , is: (4-8)

text,

but the

The bowl of sugar looJcs like a cup of lighted bulbs.

This is exactly what we should expect if there is some sort of metaphorical p r i n c i p l e in poetic language; in SL, where such p r i n c i p l e s do not e x i s t , for any constructions using c o m p a r i s o n , t h i s w i l l have to be c l e a r l y stated. 2

The distinction we are m a k i n g between metaphor and s i m i l e , namely that in the former the r e l a t i o n of comparison is i m p l i c i t , whereas in the latter it is e x p l i c i t , goes back at least as far as Q u i n t i l i a n . See McCall ( 1 9 6 9 : 2 2 9 ) .

90

Fourthly, the few lexicalised examples which do exist in SL are based on stereotypical q u a l i t i e s of the noun w h i c h serves as basis for comparison, c f . snov-goose, strawberry-mark, blood-red, jade-green. This applies not o n l y to m e t a p h o r i c a l compounds but also to many s i m i l e s : as red as blood, as playful as a kitten. That this is not the case in PL is shown by (non-contextd e p e n d e n t ) compounds such as drouned hair silent, afternoonsubtle, rusk-whisper, duck-eggshell emptiness, and also by s y n t a c t i c phrases such as bright as bread (JA 4 8 ) , as white as Sunday (UW 3 9 ) , She [ . . . ] stared like a strange blossom or a window-pane (M 3 2 ) . The f a c t that many poetic compounds are in fact based on stereotypes has, we shall assume, to do with ease of interpretation and has nothing to do w i t h the nature of metaphor i t s e l f . F i f t h l y , where metaphor does exist in SL it is usually employed to compare things w h i c h are inherently s i m i l a r . A g a i n , t h i s has nothing to do w i t h the nature of metaphor itself but is a constraint on metaphor which ensures that it be interpretable in SL. It is thus not possible in SL to say something tastes dark or the noise was like a flower, though there are of course many l e x i c a l i s e d instances of this type such as a loud ( i . e . brightlycoloured, conspicuous) jacket, a sharp taste. In PL, of course, comparisons of thia type are very common, as the f o l l o w i n g examples i l l u s t r a t e : (4-9)

sickle-shout C N N ] (RE 1 2 1 ) , apple-slicing chill [NN] (S 20),pearly mutter [ A N ] (WQ 6 0 ) , voice of rough silver (SG 15)

There is, t h e n , clearly a d i f f e r e n c e between the marginal nature of metaphor in SL which is only to be found in unproductive types, lexicalised examples, or in a very limited way in e x p l i c i t l y stated comparison, and the nature of poetic metaphor w h i c h is creative, r e l a t i v e l y unconstrained and central to poetic language. There seems, though, to be some c o n f l i c t here w i t h the c l a i m s of many researchers that metaphor is present everywhere in our language. We wish to m a i n t a i n , however, that there is in fact no c o n f l i c t involved in these views, but only a d i f f e r e n t use of the term "metaphor". But before we discuss this question it is necessary to consider another apparent c o n f l i c t between two d i f f e r e n t v i e w s of metaphor. As pointed out by Max Black ( 1 9 6 2 ) and the s i t u a t i o n has not r a d i c a l l y changed since - the m a j o r i t y of the enormous w e a l t h of studies on metaphor are examples either of the view that metaphor principally involves s u b s t i t u t i o n , or of the v i e w that it is based on comparison. The former view (such as that put forward by Jakobson (1971)) is in 3 T h i s is a type of catachresis, on one d e f i n i t i o n of the term (see Lanham (1968: 2 1 ) ) . 4 See, for e x a m p l e , L a k o f f & Johnson (1980) and also Bühler ( 1 9 3 4 : 3 4 3 ) ; the latter quoted in Brekle ( 1 9 8 4 : 8). 5 See those listed in Chapter 1 . 2 , for example. 91

p r i n c i p l e the assumption t h a t , if we say (4-10)

The sun is a diamond.

we mean something l i k e "the sun is b r i g h t " , that is, we do not use diamond such that its intension a p p l i e s , but such that it stands for whatever aspects (associations, stereotypes, connotations) are compatible w i t h sun. The latter v i e w , suggested by Abraham (1975) among others, holds that the phrase in ( 4 - 1 0 ) above is taken to mean "the sun is l i k e a d i a m o n d " , that is, the meanings of sun and diamond are constant, but is is somehow to be interpreted as "is l i k e " . W i t h o u t at this point going into this c o n f l i c t in more detail let us consider a f u r t h e r problem w h i c h must be dealt w i t h in a study of metaphor, namely that of the variousness of the phenomena which have been explicated under the heading "metaphor". These have t r a d i t i o n a l l y included both expressions l i k e those in ( 4 - 2 ) and ( 4 - 1 0 ) and also phrases l i k e the leaves laughed. We shall argue that we have here to do w i t h more than one poetic phenomenon and thus w i t h more than one p r i n c i p l e . An e x a m i n a t i o n of the data in ( 4 - 2 ) above substantiates t h i s . All the compounds are in some way (on a broad understanding of the t e r m ) metaphorical, yet even among compounds w i t h the same combination of categories, such as those in ( 4 - 2 c ) and those in ( 4 - 2 g ) , or in ( 4 - 2 e ) and ( 4 - 2 f ) , there seem to be at least two d i f f e r e n t types. If we were not to assume that more than one p r i n c i p l e is at work here, we should have to f o r m u l a t e an extremely complex metaphoric p r i n c i p l e to deal with such inhomogeneous compound-types. In v i e w of the d i f f i c u l t i e s involved in attempting to d e f i n e an area to w h i c h the term "metaphor" can be a p p l i e d , it seems that it m i g h t be h e l p f u l to d i s t i n g u i s h between four meanings of the term: (i)

a general psychological process of comparison;

(ii)

the type of expression in SL such as as black as pitch or it puts my JtjacJc up;

It w i l l be n o t i c e d that t h i s v i e w of metaphor w i l l be inherent in any v i e w of word meaning as changing w i t h context (see, for e x a m p l e , Bosch 1 9 8 4 : 1 8 f f . ) . The advantage of t h i s v i e w is that it a l l o w s a u n i f i e d treatment of such phenomena as m e t a p h o r , vagueness, a m b i g u i t y , and so o n , by regarding them as i n t e r p r e t a b l e using the normal i n t e r p r e t a t i v e means of l a n g u a g e ; c f . Bosch ( 1 9 8 3 ) . I t s disadvantage i s that i t cannot p r o v i d e any account of poetic m e t a p h o r .

92

(iii)

a general poetic f i g u r e , i n c l u d i n g expressions l i k e the sun is a diamond, the leaves laughed, as uhite as Sunday (UW 39), etc.; and

(iv)

a s p e c i f i c f i g u r e of PL, standing in opposition to other p o e t i c f i g u r e s such as m e t o n y m y , hyperbole and so on.

D i s t i n g u i s h i n g these f o u r meanings should enable us to account for the three apparent c o n f l i c t s just m e n t i o n e d . Let us look at each of the f o u r meanings in t u r n . In (i) the term "metaphor" is used to r e f e r to a general process of comparison, w h i c h may m a n i f e s t i t s e l f as metaphor proper, as m e t o n y m y , s i m i l e , "dead" metaphor, m o d e l , hyperbole, euphemism or i r o n y . It is also t h i s sense of the term "metaphor" w h i c h is used in discussions of the role of metaphor in l i n g u i s t i c e v o l u t i o n , language change, or the structure of thought. It is in this sense that P a u l ' s statement that metaphor is "something w h i c h of necessity f l o w s from the human character and makes i t s e l f f e l t not only in poetic but above all in the everyday language of common people" ( ( 1 9 0 9 : 9 4 ) ; quoted also in Bühler ( 1 9 3 4 : 3 4 4 ) ) is to be t a k e n . The cases i n v o l v e d in the second sense of the term "metaphor" are a subset of those i n v o l v e d in the f i r s t sense, in that the second is a s p e c i f i c m a n i f e s t a t i o n of the general metaphoric p r i n c i p l e . In t h i s second sense, the term r e f e r s to the use of s i m i l e , "dead" metaphor and expressions w h i c h are based on the secondary meanings of words. In our o p i n i o n , these are not true cases of metaphor, as phenomena of this type are u n p r o d u c t i v e . Even in cases in w h i c h metaphor might seem to be productive in SL, it is clear that we have to do w i t h analogy to e x i s t i n g secondary m e a n i n g s . That is, to take an example f r o m L a k o f f & Johnson (1980: 4), if the verb attack is used to mean a verbal a t t a c k , we regard this not as a metaphorical f i g u r e of speech but as a secondary ( l e x i c a l l y l i s t e d ) meaning of the verb attack. In 7

8 9 10 11 12 13

Most scholars seem to make l i t t l e d i s t i n c t i o n at all between these types. See, for e x a m p l e . W h e e l w r i g h t ( 1 9 6 2 ) or Ortony ( 1 9 7 5 ) . However, B a r f i e l d ( 1 9 4 7 ) does d i s t i n g u i s h between senses ( i ) a n d ( i v ) . This is the sense in w h i c h the term is used by, for e x a m p l e , Abraham ( 1 9 7 5 ) . For a d i s c u s s i o n of "dead" metaphor see, for example. L e v i n (1977: 9 f ) . That irony and hyperbole can be regarded as types of metaphor is suggested by M o o i j (1976: 6f. and passim). See also Lanham (1968: 61). See, for e x a m p l e , Wegener ( 1 8 8 5 ) , discussed in Langer ( 1 9 6 9 ) , q.v. See discussions in A u s t i n ( 1 9 6 1 ) and U l l m a n n ( 1 9 6 7 ) . See, for e x a m p l e , Beardsley ( 1 9 5 0 ) and L a k o f f i. Johnson (1980). 93

analogy to this and the secondary meaning of other words such as uin (an argument) and so on, one might speak of losing a battle in the sense of "losing an argument". T h i s is n o t , we shall assume, the same as creative poetic metaphor, but is m e r e l y analogy to e x i s t i n g (secondary) meanings of words. Whether or not the secondary meanings in the f i r s t place came about by metaphor is not our concern here. In its third sense, the term "metaphor" denotes a group of productive poetic f i g u r e s . It c l e a r l y describes phenomena d i f f e r e n t f r o m those covered by the second sense because, in PL, simile, metonymy and all other metaphorical figures are productive and unconstrained and do not need to be formed in analogy to e x i s t i n g expressions. When we speak of metaphor in the f o l l o w i n g we shall either be using it in t h i s general poetic sense or in its f o u r t h and stricter m e a n i n g . Here the term "metaphor" means a trope of poetic language by w h i c h something is stated to be identical to something else, w h e n , in f a c t , what is meant is that the two things are s i m i l a r . This is e s s e n t i a l l y the "comparison" v i e w of metaphor. In t h i s sense it does not i n c l u d e metonymy and s i m i l e but stands in opposition to them as a d i f f e r e n t trope. The second view of metaphor discussed above, namely that it represents a s u b s t i t u t i o n , has been proposed on the assumption that dead metaphor and creative metaphor are to be interpreted a l i k e . But they are n o t . S u b s t i t u t i o n m i g h t provide a reasonable interpretation of a dead metaphor such as Fred is an old vornan, to take an example f r o m Bosch ( 1 9 8 4 : 1 8 ) , but it will not ( n o r was it in this instance intended to) e x p l a i n poetic metaphor such as fabric CisJ a tunic of kisses (AB 6 0 ) . The important t h i n g about this latter case is that the reader may not know exactly what aspects of the two t h i n g s are being compared, and thus cannot p e r f o r m a s u b s t i t u t i o n . What he does know is that they are, in some w a y , being compared, i . e . , that the is in a phrase such as the above must be interpreted as "is l i k e " . T h i s is the essence, though by no means the w h o l e , of poetic metaphor. We thus regard metaphor in the f o u r t h , strictest p o e t i c a l sense, as a s p e c i f i c poetic m a n i f e s t a t i o n of metaphor in the f i r s t sense; it is a f i g u r e of comparison. As a f i g u r e of comparison, we should expect it to be closely connected w i t h both metonymy and synecdoche. To see w h y , compare the t r a d i t i o n a l d e f i n i t i o n s of m e t o n o m y : 14 The author, t o o , may not k n o w . Consider the following statement by a psychologist: "I am convinced by p s y c h o l o g i c a l investigation that it is impossible for the maker of some metaphor ever to report w i t h certainty those f e a t u r e s he intended to point to by his metaphor." (Deese ( 1 9 7 4 : 2 1 2 ) Compare s i m i l a r statements by poets, when they are asked about their metaphors and symbols. Ted Hughes, for example, questioned about central recurring images in Crou r e p l i e d " I ' m not quite sure what they s i g n i f y " (Faas ( 1 9 8 0 : 2 0 7 ) ) . 15 In this connection see the work of the Groupe μ ( 1 9 7 0 ) , who have stressed the importance of synecdoche and viewed both metaphor and metronymy as c o m b i n a t i o n s of synecdoche, a view 94

Substitution of cause for e f f e c t or e f f e c t for cause, proper name for one of its q u a l i t i e s or v i c e versa. (Lanham 1968: 67) and synecdoche: Substitution vice versa.

of part for w h o l e ,

genus for species, or (Lanham 1968: 67)

It is clear that the r e l a t i o n "is l i k e " , taken to be the r e l a t i o n concerned in m e t a p h o r i c a l expressions is by nature such that if two e n t i t i e s X and Υ are s i m i l a r w i t h respect to characteristics χ of X and y of Y, whereby χ and y are s i m i l a r , I can say X is l i k e Y, a n d , if I say X is l i k e Y, I can assume they possess some s i m i l a r characteristics χ and y. That is, the very nature of comparison is such that it presupposes a m e t o n y m i c a l or synecdochial r e l a t i o n s h i p between the t h i n g s compared and the characteristics w h i c h are a c t u a l l y s i m i l a r . There can thus be no such t h i n g as "pure" metaphor, in w h i c h neither metonymy nor synecdoche is i n v o l v e d . To see t h i s , consider the f o l l o w i n g e x a m p l e s of the type of c o n s t r u c t i o n g e n e r a l l y regarded as metaphorical: ( 4 - 1 Da.

The sap is the bJood of the plant.

b. The sun is a diamond. c. His suord is his

life.

In example ( 4 - l l a ) , it m i g h t seem that the o n l y m e t o n y m i c a l r e l a t i o n s h i p s present are between sap and plant on the one hand and blood and a non-expressed person on the o t h e r , and thus that the phrase, in comparing the aspects (sap and blood) themselves, is "pure" metaphor. But t h i s is not so, of course. Sap and blood are o n l y compared w i t h respect to certain q u a l i t i e s w h i c h each possesses ( " l i q u i d " , "essential" e t c . ) . S i m i l a r l y , i n ( 4 - l l b ) , i t is the brightness ( e t c . ) of the sun w h i c h is compared to that of a diamond and in ( 4 - l l c ) the i m p o r t a n c e of the sword is compared to the importance ( f o r the person in q u e s t i o n ) of his l i f e . Thus metaphor can be seen not to exist independently of m e t o n y m y . However, though metaphor always involves metonymy or synecdoche, the reverse does not h o l d : synecdoche or metonymy can exist independently of m e t a p h o r , as the t r a d i t i o n a l examples of the f i g u r e s - r e f e r r i n g to a book as print or coal as heat for of metynomy w h i c h is not c o m p a t i b l e w i t h ours. Other scholars have regarded metaphor as based on two m e t o n y m i e s - a v i e w w h i c h has something in common w i t h ours. See, for e x a m p l e , Henry (1971: 6 6 ) , quoted in W e s t s t e i j n (1983: 1 0 0 ) . 95

m e t o n y m y , and to a car as vheel f o r synecdoche - i n d i c a t e . Another c o n s i d e r a t i o n w h i c h is important in d e t e r m i n i n g the exact nature of p o e t i c metaphor is the r e l a t i o n s h i p of metaphor to s i m i l e . A s i m i l e w o u l d be the sun is JiJce a diamond, a corresponding metaphor w o u l d be the sun is a diamond. Yet b o t h , i n t u i t i v e l y , are interpreted to mean the same. The above considerations lead us to the conclusion that we have to do w i t h at least two separate p r i n c i p l e s ; a P r i n c i p l e of Metaphor, w h i c h a l l o w s constructions of the f o r m "X is Y" to be interpreted as "X is l i k e Y" and a P r i n c i p l e of M e t o n y m y , w h i c h substitutes an aspect of a p a r t i c u l a r t h i n g for the t h i n g i t s e l f , or v i c e versa. We shall f o r m u l a t e the f i r s t as f o l l o w s : (4-12)

The P r i n c i p l e of Metaphor Assign a construction having the f o r m "X is Y" a s e m a n t i c representation of the f o r m "X is l i k e Y " .

We have f o r m u l a t e d the p r i n c i p l e as a s s i g n i n g a p a r t i c u l a r s e m a n t i c representation to a p a r t i c u l a r surface f o r m . H o w e v e r , a closer look at the type of structure we s h a l l , in our v i e w of metaphor as comparison, consider m e t a p h o r i c , suggests that the p r i n c i p l e is too s p e c i f i c . It seems clear that metaphor is an interpretative process w h i c h is restricted to certain s u r f a c e f o r m s , but not only to one s p e c i f i c one. We should c e r t a i n l y wish to say that all three of the f o l l o w i n g expressions are examples of metaphor: ( 4 - 1 3 ) a . The sun is a diamond. b. the diamond-sun c. the diamond of the sun If we m a i n t a i n a m e t a p h o r i c interpretation is only possible for a s p e c i f i c surface f o r m , it w i l l be d i f f i c u l t to e x p l a i n t h i s . What ( 4 - 1 3 a , b, c) have in common is not s u r f a c e f o r m , but semantic representation. If we do not take t h i s f a c t into account, we can neither e x p l a i n how it is possible for both syntactic phrases and 16 Here and in the following, we have f o r m u l a t e d poetic p r i n c i p l e s either as interpretation or production p r i n c i p l e s , according to their level of interaction. In f a c t , all production principles will have a counterpart in interpretation, and all interpretation p r i n c i p l e s w i l l have a counterpart at the level of p r o d u c t i o n , because only interpretable structures are w e l l - f o r m e d and only w e l l - f o r m e d structures are a c c e p t a b l e . 96

compounds to be m e t a p h o r i c a l nor how it is that both seem to embody metaphors w h i c h are s i m i l a r (though not quite i d e n t i c a l ) in m e a n i n g . These considerations lead us to the conclusion that the P r i n c i p l e of Metaphor does not work at the level of s u r f a c e f o r m but at the level of s e m a n t i c representations. We must thus revise the p r i n c i p l e as f o l l o w s : (4-14)

The P r i n c i p l e of Metaphor ( r e v i s e d ) Assign a structure w h i c h in SL would be given the semantic representation "X ( w h i c h ) is Y" a semantic representation of the f o r m "X ( w h i c h ) is l i k e Y".

This principle, l i k e all the poetic principles which f o l l o w , is assumed to be f r e e in a p p l i c a t i o n . Whether or not it applies w i l l depend on f a c t o r s e x t e r n a l to the p r i n c i p l e i t s e l f : the type of t e x t , the presence or absence of c o n f l i c t s its a p p l i c a t i o n w o u l d give rise to, the p o s s i b i l i t y of i n t e r p r e t a t i o n , and so on. Of course, we could have f o r m u l a t e d the p r i n c i p l e to work d i r e c t l y upon the semantic r e p r e s e n t a t i o n . However, the process of interpretation w o u l d then lose some of its c l a i m to " p s y c h o l o g i c a l r e a l i t y " , as it w o u l d presuppose stages in the interpretation of a metaphor w h i c h i n t u i t i v e l y would seem not actually to take place, such as the assignment of a representation "X is Y" as a f i r s t stage. By f o r m u l a t i n g the p r i n c i p l e for a structure " w h i c h in SS would be given the semantic representation ' X is Y ' " , we leave open the question as to whether such a stage a c t u a l l y appears in the poetic i n t e r p r e t a t i o n . The k n o w l e d g e of the f a c t that the interpretation "X is Y" would be assigned in SL, but is never a c t u a l l y assigned in PL (or at least, only at an early stage of an i n t e r p r e t a t i o n ) leads to the often-discussed "tension" in metaphoric constructions between a l i t e r a l and a m e t a p h o r i c r e a d i n g . The three examples given above in interpreted as;

(4-13),

would,

in SL, be

( 4 - 1 5 ) a . "the sun is a diamond" b. "the sun w h i c h is a d i a m o n d " c. "the sun w h i c h is a d i a m o n d " 17 It is sometimes m a i n t a i n e d ( s e e , for e x a m p l e , L e v i ( 1 9 7 8 : 2 8 2 ) ) , that the m e t a p h o r i c a l i n t e r p r e t a t i o n of a compound such as diamond-sun is "sun w h i c h is a d i a m o n d " . It is our o p i n i o n that t h i s v i e w rests on a c o n f u s i o n b e t w e e n l i t e r a l and metaphorical interpretation. 97

In PL they diamond."

are

interpreted

as

"the

sun

(which)

is

like

a

In its a p p l i c a t i o n to compounds, the p r i n c i p l e has the e f f e c t that all compounds w h i c h in SL could be interpreted as "Y w h i c h is X" (that is, all compounds r e c e i v i n g the second type of coordinative i n t e r p r e t a t i o n ) , can, in PL, be interpreted as m e t a p h o r i c a l compounds w i t h the m e a n i n g "Y w h i c h is l i k e X " . This assumption w i l l have the consequence that o n l y compounds whose two elements are of the same category can be interpreted m e t a p h o r i c a l l y . This is both i n t u i t i v e l y correct and capable of e m p i r i c a l s u b s t a n t i a t i o n . It w i l l also have the consequence that metaphorical compounds w i l l always have the interpretation "Y w h i c h is l i k e X" and n o t , for example, "X w h i c h is l i k e Y." To return to the q u e s t i o n of m e t o n y m y , we assume that only i n d i v i d u a l l e x i c a l items can be i n v o l v e d . We shall f o r m u l a t e a p r i n c i p l e as f o l l o w s : (4-16)

The P r i n c i p l e of M e t o n y m y Assign an element X a semantic representation corresponding to an element x, so that the r e l a t i o n s h i p between X and χ is one of the f o l l o w i n g : {part/whole, causality, predication, similarity, ...}.

It w i l l be n o t i c e d that we have f o r m u l a t e d the P r i n c i p l e of Metonymy to i n c l u d e what is t r a d i t i o n a l l y called synecdoche ( i . e . , the p a r t / w h o l e r e l a t i o n s h i p ) . W h i l e it is no doubt of v a l u e in works of r h e t o r i c or l i t e r a r y c r i t i c i s m to separate these two phenomena, we shall assume that for our purposes it is quite s u f f i c i e n t to regard them as aspects of the same phenomenon. The dots in the set of r e l a t i o n s g i v e n above are to be t a k e n as m e a n i n g that any r e l a t i o n by w h i c h χ and X are associated is in p r i n c i p l e an acceptable one for f i g u r e s of m e t o n y m y . S i m i l a r i t y is not on the w h o l e cons„Ldered to be a m e t o n y m i c r e l a t i o n s h i p in t r a d i t i o n a l studies, but examples such as the f o l l o w i n g suggest that it should be: (4-17)

The leaves Je ugh ed.

Such expressions are t r a d i t i o n a l l y considered instances of m e t a p h o r . H o w e v e r , there seems no way in w h i c h ( u n l e s s we are 18 The close c o n n e c t i o n between c o - o r d i n a t i v e and m e t a p h o r i c a l readings of compounds has been noted by Hatcher (1951: 4). 19 In f a c t , many scholars have regarded them as aspects of the same trope, cf. Jakobson ( 1 9 7 1 ) and Le Guern ( 1 9 7 3 ) . 20 But compare the e n t r y for metonymy in H o l m a n ( 1 9 7 2 ) , w h i c h i n c l u d e s an example based on s i m i l a r i t y .

98

content to be e x t r e m e l y vague) we can e x p l a i n as metaphor both the sentence in ( 4 - 1 7 ) and the sentence: (4-18)

The sun is a diamond.

We should c e r t a i n l y wish to say that the latter is metaphor. But then the f o r m e r cannot be (in the strict sense). We shall m a i n t a i n that the f o r m e r phrase is given the reading (4-19)

"the leaves trembled laughing)"

(so that

it

looked/sounded

like

That is, the m e a n i n g of trembled is substituted for laughed, by a m e t o n y m i c a l process. E x a m p l e ( 4 - 1 7 ) is thus not a metaphor. In f a c t , we b e l i e v e that this corresponds e x a c t l y to our i n t u i t i o n s about the two phrases: the f i r s t i n v o l v e s a s u b s t i t u t i o n , w h i c h is the essence of m e t o n y m y , and the latter i n v o l v e s a comparison, w h i c h is the essence of metaphor. Metaphor and metonymy are thus two separate phenomena w h i c h i n t e r a c t . They should not be treated as aspects of the same phenomenon. Another look at the data in ( 4 - 2 ) suggests, h o w e v e r , that the P r i n c i p l e s of Metonymy and of Metaphor alone w i l l not be s u f f i c i e n t to account for all the compound-types g i v e n . For e x a m p l e , the compound moth-walk, representative of group ( 4 - 2 a ) , means "to w a l k j as a moth w a l k s 2 " (the indices have been added for c l a r i t y ) , that is, it is not the case that the elements moth and valJf^ are being compared, but that ualJf^ is being compared w i t h a structure " w a l k 2 ( m o t h ) " w h i c h is represented in the compound by moth. T h i s , i n c i d e n t a l l y , provides f u r t h e r support for the v i e w that it cannot s u f f i c e , in order to e x p l a i n such compounds, to extend the group of semantic arguments to include "basis for comparison". O b v i o u s l y moth cannot be seen as a s e m a n t i c argument of comparison on a par w i t h , say, a s e m a n t i c argument of i n s t r u m e n t a l i t y in a compound hand-wash, because moth is a c t u a l l y the s u b j e c t of a second r e l a t i o n , n a m e l y ualJf^. Thus it can be seen that we have here to do w i t h a type of structure w h i c h does not even have a p a r a l l e l in SL. The question is, how does moth in the compound moth-ualJc come to receive the s e m a n t i c representation " w a l k 2 ( m o t h ) " ? We s h a l l m a i n t a i n that this is possible by v i r t u e of e l l i p s i s . We shall d e f i n e a poetic P r i n c i p l e of E l l i p s i s as f o l l o w s : (4-20)

The P r i n c i p l e g_£ E l l i p s i s At the level of semantic representation, replace a s i n g l e element - e i t h e r an argument or a r e l a t i o n - by an argument-structure c o n t a i n i n g this e l e m e n t .

Of

course,

ellipsis

is

a

traditional 99

poetic

figure,

usually

d e f i n e d as "omission of a word easily understood" (Lanham 1968: 4 0 ) . It is, in a sense, the p a r a l l e l of metonymy ( o r , more traditionally, synecdoche) at the level of semantic representation. Where synecdoche in one of its most typical m a n i f e s t a t i o n s uses part of the thing r e f e r r e d to to represent the w h o l e , e l l i p s i s uses part of an argument-structure to represent the whole argument-structure. Note t h a t , though some of the e l l i p t i c a l constructions in PL can c e r t a i n l y be accounted for by the gapping p r i n c i p l e of SL, and are thus not d e v i a n t , others are not possible in SL: ( 4 - 2 1 ) a . That night, such storms. (RG 19) b.

the lover in his Jiving sleeve is giving more, and will receive, than in the naked land of sight. (SM 2 5 )

The f i r s t example ( 4 - 2 1 a ) is not a case of gapping at a l l , as it does not involve a co-ordinated structure, and the second example ( 4 - 2 1 b ) , though it contains a c o - o r d i n a t i o n , w o u l d not be a l l o w e d by gapping in SL, because the deleted element does not c o n t a i n a verb, ar)4-> gapping of direct objects is in any case not possible. E l l i p s i s must t h e r e f o r e be seen as a poetic interpretative p r i n c3 i p l e , w h i c h a l l o w s d e l e t i o n of elements not deletable in ^· It does seem, however, to be subject to a number of constraints, that is, not just a n y t h i n g can be o m i t t e d , as L a n h a m ' s d e f i n i t i o n indicates. W h i l e it is not quite clear just what "easily understood" in this d e f i n i t i o n means, the d e f i n i t i o n does seem to correspond to many of the f a c t s about e l l i p s i s . One f a c t w h i c h is of p a r t i c u l a r interest to us is that it seems, in 21 Various types of d e l e t i o n ( u s u a l l y i n v o l v i n g puns) are r e f e r r e d to in the l i t e r a t u r e as " z e u g m a " . The term is, h o w e v e r , used w i t h a great deal of inconsistency. See Lanham (1968: 1 0 4 f . ) a n d B a n f i e l d (1973a: 5 3 f f . ) . I n addition t o i t s importance in PL, e l l i p s i s (or z e u g m a ) also plays an important role in the language of j o u r n a l i s m , p a r t i c u l a r l y in newspaper headlines. See Sandig ( 1 9 7 1 ) . 22 See S t i l l i n g s ( 1 9 7 5 ) for a discussion of gapping, i . e . , of constructions .of the type John sau Mary and Bill Ann. 23 A proper substantiation of this statement w o u l d require a discussion of what is possible as e l l i p s i s and what is n o t . Furthermore, we are assuming that e l l i p s i s is a peripheral l i n g u i s t i c phenomenon, and that it makes sense to speak of e l l i p t i c a l expressions ( w i t h some e x c e p t i o n s ) as g r a m m a t i c a l l y d e v i a n t . This p o s i t i o n is by no means s e l f - e v i d e n t . The reader is r e f e r r e d to discussions of e l l i p s i s in Meyer-Hermann & Rieser ( 1 9 8 5 ) . 100

compounds, that the head cannot be an e l l i p t i c a l element. It is clearly the case that the head of a compound must be a word, and that phrases are ojrly possible (though as a marked case) in the m o d i f i e r position. But this should not r u l e out the possibility of ellipsis in the head, as an e l l i p t i c a l element is a phrase only at the level of its semantic representation. We must therefore assume that there are semantic reasons for this, to do with ease of interpretation. Whatever the reason, this fact is important for the interpretation of metaphorical and co-ordinated compounds because it means that if, for example, an NA compound is to be assigned the relation of co-ordination, something which is impossible without the assumption of e l l i p s i s , because both elements of the compound must belong to the same category, this can only proceed by means of interpreting the N as e l l i p t i c a l , not the A. That is, the compound w i l l be given a representation in which it corresponds to an AA, not an NN compound. A further restriction, w h i c h is i n t u i t i v e l y covered by Lanham's d e f i n i t i o n , is that the deleted element must occur elsewhere; presumably it is otherwise not recoverable. That is, in a compound moth-ua2k, what is deleted in the non-head is the walJc of the argument-structure "walk ( m o t h ) " . It cannot, apparently, be any other verb. The compound cannot, for e x a m p l e , mean "to walk l i k e a moth f l i e s " , but only "to walk like a moth w a l k s " . The former is a conceivable comparison, but it cannot be expressed by this compound. This is, therefore, clearly a constraint on e l l i p s i s , not a constraint on metaphor. In f a c t , metaphorical compounds do occur in w h i c h two argument-structures, each with a d i f f e r e n t r e l a t i o n , are compared: (4-22)

[ [ N A ] A ] flover-lovely-false + ( + N M 21) heart-hollov drab ( N P 137) shadow-blacJc-sharp ( + NM 59)

These compounds mean, respectively: "false as a f l o w e r is l o v e l y " , "drab as the heart is h o l l o w " , "sharp as a shadow is b l a c k " . They are only interpretable as such because both relations are expressed. Let us now return to the examples of compounds given above in ( 4 - 2 ) - a rather inhomogeneous group, as we have seen - and consider in w h i c h ways the Principles of Metonymy, Metaphor and E l l i p s i s interact w i t h the grammar to render the compounds w e l l 24 See Toman (1983: 4 6 f . ) for a discussion of t h i s point. 25 That it is a conceivable comparison is shown by the existence of s i m i l e s l i k e happy as the day is long (cf Lord 1975: 4 f . ) . The compound day-happy w o u l d , however, not mean t h i s , but "happy as the day is happy". 101

formed in PL, w h i c h , as we have seen, they are not in SL. We have shown that the compound (to) fruit-offer (something) means "to offer ( s o m e t h i n g ) as one o f f e r s fruit". This interpretation can be represented in three stages (whereby we do not necessarily mean to correlate the stages t e m p o r a l l y w i t h actual mental processes). N o r m a l l y , the interpretation of a compound in w h i c h the n o n - r e l a t i o n a l element is n e i t h e r an argument nor a semantic argument of the r e l a t i o n a l o n e , as it cannot be here, proceeds just l i k e the interpretation of an UN compound. That is, the compound is assigned a new r e l a t i o n , of which both its elements are arguments. Now we have seen in Chapter 3 that there is only one context-independent relation which is not based on stereotypes of £he compound-elements, and this is the r e l a t i o n of co-ordination. In PS, this relation can be either co-ordination or the metaphoric r e l a t i o n of comparison, depending on whether the P r i n c i p l e of Metaphor is involved in the interpretation or not. However, there seems to be no way in w h i c h we could assign an NV compound a metaphorical r e l a t i o n as its two elements belong to d i f f e r e n t categories. (It w i l l be remembered that a prerequisite for c o - o r d i n a t i o n (and thus also for comparison in PL) is i d e n t i t y of categories to be c o - o r d i n a t e d . ) So f i r s t a representation has to be found in w h i c h both elements do belong to the same category. This is the task of the P r i n c i p l e of E l l i p s i s , which a l l o w s the f i r s t stage of interpretation of the compound. In the semantic representation of the compound, the non-head-element fruit can, by v i r t u e of the P r i n c i p l e of E l l i p s i s , be replaced by an argument-structure " o f f e r ( f r u i t ) " . Because the structure now consists of two elements of category V, it could be assigned the co-ordinative reading, w h i c h , as we saw in Chapter 3 for a compound AB is "B w h i c h is A". This possible assignment of a co-ordinative r e l a t i o n is a s u f f i c i e n t condition for the P r i n c i p l e of Metaphor to a p p l y ; i . e . for the compound to be interpreted as "B w h i c h is l i k e A". Before we embark upon a ( p a r t i a l ) reconstruction of the interpretation of the compound fruit-offer, it is necessary to make clear some assumptions w h i c h we are m a k i n g about the interpretation of l i n g u i s t i c structures. As stated in Chapter 3, we assume that the interpretation of a structure involves the construction of a semantic representation. The semantic representation is mapped on to uninterpreted s u r f a c e structures (or some representation of t h e m ) on the one hand and on to an interpretation on the other. Both these mappings are governed by semantic p r i n c i p l e s . We shall not here discuss those p r i n c i p l e s which govern the mapping of the correct interpretation on to the semantic representation. Our concern is w i t h the f o r m a t i o n of the latter as t h i s is the level at w h i c h the poetic p r i n c i p l e s of metonymy, ellipsis and metaphor are effective. We have represented the output of each poetic p r i n c i p l e as a new stage in the semantic representation, because o n l y in t h i s way w i l l the e f f e c t of the p r i n c i p l e s be c l e a r l y seen. These intermediate 26 We are ignoring here the possible m a r g i n a l location as its existence is u n c e r t a i n . 102

relation

of

"semantic representations", w h i l e they may have some sort of "psychological r e a l i t y " in that they are r e c o n s t r u c t i b l e as elements i n v o l v e d in the i n t e r p r e t a t i o n of a p o e t i c f o r m , and may play a role in the f i n a l total poetic e f f e c t , do not themselves receive any i n t e r p r e t a t i o n . Only the final stage of the representation is interpreted. F u r t h e r m o r e , we have w r i t t e n semantic representations as though they consisted of the l e x i c a l items themselves rather than representations of these l e x i c a l items. The two stages in the interpretation of the compound fruitoffer as representative of group ( 4 - 2 a ) can thus be represented as f o l l o w s : (4-23)

fruit-offer 1) e l l i p s i s : replace fruit by " o f f e r

(fruit)"

semantic representation ( S R ) : " ( o f f e r

(fruit))-offer"

2) metaphor: assign the1 structure the representation "B w h i c h is l i k e \" ' SR: " ( o f f e r ) which is like ( o f f e r interpretation:

(fruit))"

"to o f f e r l i k e one offers f r u i t "

In other words, a f t e r e l l i p s i s the compound, if it were possible in SL ( w h i c h it is n o t ) , would be assigned a co-ordinative relation as the o n l y possible context-independent r e l a t i o n . In PL such compounds are d i r e c t l y assigned a metaphorical reading. Now it m i g h t be thought that there is a f u r t h e r possible e x p l a n a t i o n for the compound fruit-offer. It could be suggested that the metaphoric p r i n c i p l e makes it possible, in a syntactic phrase to offer nakedness, to c o n j o i n J.ike fruit to the object nakedness, so that the verb offer marks the whole phrase nakedness like fruit w i t h the role of o b j e c t ; in other words, only the objects fruit and nakedness are compared, not the relations of o f f e r i n g . This would be parallel to the situation in a phrase he eats bread and meat in w h i c h bread and meat as a phrase is assigned the role of o b j e c t . T h e n , we might t h i n k , t h i s c o n j o i n e d object can in a compound be " s p l i t " , so that one element is placed before the verb and one a f t e r and both are marked as objects thus:

27 What is meant here is of course the P r i n c i p l e of Metaphor in its i n t e r a c t i o n w i t h the semantics of compounds (see a b o v e ) .

103

( o b j . )-

(4-24)

πι ι Ν

nakedness (obj.)· The problem here, however, is that the θ-role of object w o u l d have to be assigned to the w h o l e phrase "nakedness l i k e f r u i t " , if t h i s were the correct structure (a question w h i c h f r o m a semantic point of v i e w is in any case d o u b t f u l ) and that the two roles could not be assigned separately. E x a c t l y the same state of a f f a i r s is f o u n d in the case of c o - o r d i n a t i o n : he ate bread and meat cannot be expressed as he bread-ate meat where bread and meat both receive the θ-role of object f r o m the verb. In other words, co-ordinated elements do not a f f e c t the θ - c r i t e r i o n , w h i c h is exactly what we should expect. Or, to put it another w a y , the β - c r i t e r i o n prevents the separating of elements of a co-ordinated argument and thus precludes cases such as that just g i v e n . The same would then apply to elements c o n j o i n e d by the r e l a t i o n " l i k e " . However, this is the case in SL and we might argue that in PL the case is d i f f e r e n t , n a m e l y that the 0 - c r i t e r i o n does not apply and that co-ordinated arguments can be split. This does not seem, at least for arguments c o n j o i n e d w i t h "and", to be the case. No such examples appear in the poems under question. Though this of course does not mean they are i m p o s s i b l e , i n t u i t i v e l y they do not seem l i k e l y to occur. We shall thus assume that t h i s e x p l a n a t i o n is not a v i a b l e one and that in f a c t the e x p l a n a t i o n given above is the correct one. This assumption is f u r t h e r supported by what appears to be the correct interpretation of the compound, namely a comparison of two types of o f f e r i n g . Let us now consider the e x a m p l e angel-eyed, representative of group ( 4 - 2 b ) . This is a complex a d j e c t i v e consisting of the compound angel -eye and the a d j e c t i v a l e n d i n g - ed and means "having an angel-eye". But what is an angel -eye? Not an a n g e l ' s eye, p r e s u m a b l y , but "eye2 l i k e the eye^ of an a n g e l " . If t h i s is what it means, then it must contain two argument-structures, n a m e l y " l i k e ( e y e j , e y e 2 ) " a n d "have ( e y e ^ a n g e l ) " . T h e compound contains as f i r s t e l e m e n t angel f r o m the second argumentstructure and as second e l e m e n t ey&2 £ r o m t h e f i r s t argumentstructure . The compounds of group ( 4 - 2 g ) have s i m i l a r structures. In the case of the compound oxygen-red, w h i c h means "red as blood c o n t a i n i n g o x y g e n " , t h e t w o e l e m e n t s belong t o t w o d i f f e r e n t argument-structures w h i c h can be paraphrased as "red as blood" and "blood c o n t a i n s o x y g e n " . In prayer-young, prayer is taken f r o m an argument-structure w h i c h can r o u g h l y be paraphrased as "X 104

says a prayer for Y" and young f r o m a structure "young as Y " . That is, the compound means "young as something one w o u l d say a prayer f o r " . At t h i s p o i n t it w i l l be remembered that we have g i v e n a s e m a n t i c p r i n c i p l e , ( 3 - 3 1 ) , in Chapter 3, governing the r e a l i s a t i o n of a r g u m e n t - s t r u c t u r e s in compounds. We repeat this p r i n c i p l e ( a s ( 4 - 2 5 ) ) here f o r c o n v e n i e n c e : (4-25)

The No-Mixing Principle In the s e m a n t i c representation " R ( A , B ) " of a compound AB, A and B must be assigned arguments of the same relation.

C l e a r l y , all the examples discussed ( a n d in f a c t all the examples given above in ( 4 - 2 ) except for those in group ( 4 - 2 f ) ) v i o l a t e t h i s p r i n c i p l e . It seems, in f a c t , that t h i s p r i n c i p l e is r e g u l a r l y v i o l a t e d in PL. In SL this is not the case. There are two types of e x p l a n a t i o n w h i c h could be given for this circumstance. The f i r s t would be to postulate a poetic "principle" like the f o l l o w i n g : (4-26)

The N o - M i x i n g P r i n c i p l e does not a p p l y in PL.

H o w e v e r , the same o b j e c t i o n s can be made to t h i s as were discussed above. It is too s p e c i f i c , it does not capture important g e n e r a l i s a t i o n s and so on. Worse than t h i s , it is i n c o r r e c t . A compound rose forest cannot mean "forest in w h i c h a m a g i c i a n lives who eats roses" in PL any more than it can in SL. That is, we need p r i n c i p l e ( 4 - 2 5 ) to e x p l a i n other phenomena of PL and cannot thus s i m p l y abandon i t . We shall argue that the P r i n c i p l e of E l l i p s i s and the P r i n c i p l e of Metonymy both interact w i t h the i n t e r p r e t a t i o n of compounds in such a way that the NoM i x i n g P r i n c i p l e does not r u l e them o u t . To return to the a d j e c t i v e angel-eyed, we can disregard the -ed e n d i n g , as it can be assigned a semantic interpretation ( " h a v i n g x " ) d i r e c t l y . O n l y the compound angeJ-eye w h i c h the complex a d j e c t i v e contains requires an interpretation in several stages. T h i s can be represented as f o l l o w s : (4-27)

angel-eye 1) interpretation as r e l a t i o n a l compound SR: " a n g e l ' s eye" 2) m e t o n y m y : replace " a n g e l ' s eye" by "x w h i c h is an a n g e l ' s e y e " i n t e r p r e t a t i o n ( i n c l u d i n g -ed e n d i n g ) : "having x l i k e an a n g e l ' s e y e ( s ) "

105

like

The i n t e r p r e t a t i o n no longer v i o l a t e s the N o - M i x i n g P r i n c i p l e , because the r e l a t i o n " l i k e " does not obtain between eye and angel ( w h i c h is i t s e l f the argument of another r e l a t i o n ) but between eye and " a n g e l ' s e y e " . In other words, at a f i r s t level of i n t e r p r e t a t i o n , angel-eyed means s i m p l y "having an a n g e l ' s eyes". It is only because of the i n t e r a c t i o n of the P r i n c i p l e of Metonymy that the compound does not denote the eyes of an angel but something ( u s u a l l y eyes) w h i c h are regarded as " a n g e l ' s eyes" because they are s i m i l a r . Thus all examples in ( 4 - 2 b ) can be shown to be possible words of PL, by v i r t u e of the P r i n c i p l e of Metonymy. The correct e x p l a n a t i o n is thus not that the N o - M i x i n g P r i n c i p l e f a i l s to apply h e r e , but that these examples no longer constitute problems for it. As we should e x p e c t , examples l i k e the above, w h i c h are e x t r e m e l y productive in PL, are not productive in SL, although there is a l e x i c a l i s e d class of complex a d j e c t i v e s such as duck-billed, dog-faced, ivy-leaved, w h i c h m a i n l y occur as plant- and a n i m a l - n a m e s . Note that the χ of the f i n a l interpretation can in p r i n c i p l e stand for any noun. For e x a m p l e , in the case of an a d j e c t i v e dragon-toothed (- n o t e , a g a i n , the a m b i g u i t y of t h i s compound in its actual context -), if it occurs in the phrase dragon-toothed man, then presumably χ w i l l stand for "teeth", that is, the phrase w i l l mean "man w i t h teeth l i k e a d r a g o n ' s teeth". If it occurs in the phrase dragontoothed plant it m i g h t stand for "leaves", i . e . the phrase w i l l mean "plant having leaves l i k e a dragon's t e e t h " . In other words, what χ w i l l be depends on the context. For the compounds in groups (c) and (d) above, the process w i l l be s i m i l a r to that discussed for type ( a ) . Taking as representatives of these types the compounds goose-uhite and hover-still, it w i l l be clear that no r e l a t i o n can be assigned i m m e d i a t e l y , as the elements of the compounds are of d i f f e r e n t categories; in these cases, t o o , it seems that one element is e l l i p t i c a l . The two stages in the interpretation of the compound goose-uhite are as f o l l o w s : (4-28)

goose-uhite 1) e l l i p s i s : replace goose by " w h i t e (goose)" SR: " ( w h i t e ( g o o s e ) ) - w h i t e " 2) metaphor: assign the structure the representation "B w h i c h is l i k e A" SR: " ( w h i t e ) w h i c h i s l i k e ( w h i t e ( g o o s e ) ) " interpretation:

"white l i k e a goose is w h i t e "

For the compound hover-still, the i n t e r p r e t a t i o n is as f o l l o w s :

106

(4-29)

hover-still 1) e l l i p s i s : r e p l a c e hover by " s t i l l (x) and hover ( x ) " SR: " ( s t i l l ( x ) a n d hover ( x ) ) - s t i l l " 2) m e t a p h o r : assign the structure the representation "B w h i c h is l i k e A" SR: " ( s t i l l ) w h i c h i s (x))" interpretation: hovering"

"still

like like

(still

( x ) a n d hover

(one is)

still

when

The i n t e r p r e t a t i o n s w h i c h result f r o m these two stages a r e , then, e x a c t l y the i n t e r p r e t a t i o n s w h i c h these compounds and the others i n ( 4 - 2 c ) a n d ( 4 - 2 d ) r e q u i r e . W e n o w turn t o group ( 4 - 2 e ) , represented by the compound rusJs-uh isper, w h i c h again i n v o l v e s metonymy and is interpreted in a way p a r a l l e l to the i n t e r p r e t a t i o n of compounds in ( 4 - 2 b ) . There is no d i f f i c u l t y in assigning t h i s compound the m e t a p h o r i c a l representation, as both of its e l e m e n t s are n o u n s : (4-30)

rusJf - vh isper 1) m e t a p h o r : assign the structure the representation "B w h i c h is l i k e A" SR: " ( w h i s p e r ) w h i c h i s l i k e ( r u s k ) " 2) m e t o n y m y : r e p l a c e "rusk" by "sound of a rusk" SR: " ( w h i s p e r ) w h i c h is l i k e (sound of a r u s k ) " i n t e r p r e t a t i o n : "whisper w h i c h is a rusk"

l i k e the sound of

A g a i n , in the case of compounds in groups ( 4 - 2 f ) and ( 4 - 2 g ) , the P r i n c i p l e of M e t o n y m y plays a r o l e . Using the example leopard-leaves for group ( 4 - 2 f ) , it is clear t h a t , as both e l e m e n t s b e l o n g to the same c a t e g o r y , there is no d i f f i c u l t y in assigning to the compound the m e t a p h o r i c a l representation. We s h a l l assume the f o l l o w i n g two stages a p p l y : (4-31)

leopard-leaves 1) m e t a p h o r : assign the structure the representation "B w h i c h is l i k e A" SR: " ( l e a v e s ) w h i c h a r e l i k e ( l e o p a r d s ) "

107

2) m e t o n y m y : replace "leaves" by " c o l o u r i n g etc. of leaves" and "leopards" by " c o l o u r i n g etc. of leopards" SR: " ( c o l o u r i n g etc. of leaves) w h i c h ( c o l o u r i n g etc. of leopards)"

is

i n t e r p r e t a t i o n : "leaves w h i c h (in respect c o l o u r i n g e t c . ) are l i k e leopards"

like of

their

In the case of compounds in group ( 4 - 2 g ) for w h i c h we take heronpale as an e x a m p l e , a relation cannot be assigned u n t i l the two elements, by v i r t u e of the P r i n c i p l e of E l l i p s i s , can be regarded as of the same category: (4-32)

heron-pale 1) e l l i p s i s : r e p l a c e heron by "pale ( h e r o n ) " SR: " ( p a l e ( h e r o n ) ) - p a l e " 2) metaphor: assign the structure the representation "B w h i c h is l i k e A" SR: " ( p a l e ) w h i c h i s l i k e (pale ( h e r o n ) ) " 3) metonymy: replace "heron" by "heron's eyes" SR: " ( p a l e ) w h i c h i s l i k e ( p a l e ( h e r o n ' s e y e s ) ) " interpretation: "pale as a heron's eyes (are p a l e ) "

In these cases, metonymy w i l l be seen as a principle replacing either whole by part ("heron" by " h e r o n ' s e y e s " ) , object by q u a l i t y (.Jeafesby "colouring etc. of leaves") or object by t h i n g it is s i m i l a r to ( " a n g e l ' s eye" by "x l i k e a n g e l ' s e y e " ) . W h i c h of these substitutions is a c t u a l l y the appropriate one w i l l be decided by context, as in the case of heron-pale, w h i c h m o d i f i e s the noun eyes, or by stereotypical i n f o r m a t i o n , as in the case of leopard-leaves where it is clear that the characteristics compared w i l l be those s t e r e o t y p i c a l l y associated w i t h both words. It w i l l be noticed that in all cases e l l i p s i s takes place before a metaphoric r e l a t i o n is assigned to the compound, whereas metonymy takes place a f t e r . This r e f l e c t s our i n t u i t i o n that e l l i p s i s is concerned with parts of argument-structures w h i c h stand for whole structures and thus must operate at the level of u n i n t e r p r e t e d ( a n d u n i n t e r p r e t a b l e ) representations whereas metonymy is concerned w i t h the denotations of elements in 28 Compare also Gläser (1971: 2 7 4 ) : "the comparison proper is established on grounds of more d e l i c a t e semantic features f r o m which the tertium comparationis is d e r i v e d . " (Gläser works in an extended K a t z - P o s t a l - l i k e semantic f r a m e w o r k ; see K a t z S, Postal ( 1 9 6 5 ) ) . 108

an i n t e r p r e t a t i o n - it assigns an "extended" i n t e r p r e t a t i o n to an already i n t e r p r e t a b l e s t r u c t u r e . It can thus be seen that compounds which might intuitively be referred to as " m e t a p h o r i c a l " in f a c t i n v o l v e a variety of d i f f e r e n t structures and r e s u l t f r o m t h e i n t e r a c t i o n o f several d i f f e r e n t p o e t i c p r i n c i p l e s at d i f f e r e n t stages in t h e i r i n t e r p r e t a t i o n . The same type of inhomogeneity will of course also characterise any group of l i n g u i s t i c structures regarded as e x a m p l e s o f m e t a p h o r . T h e above-mentioned c o n f l i c t w h i c h B l a c k discusses between v i e w s of m e t a p h o r as s u b s t i t u t i o n or c o m p a r i s o n is thus not s t r i c t l y a c o n f l i c t but a r e f l e c t i o n of attempts to e x p l a i n d i f f e r e n t phenomena under the same heading. From t h e above d i s c u s s i o n i t w i l l have become clear that o b v i o u s s e m a n t i c i n c o m p a t i b i l i t i y is n o t , as has o f t e n been a s s u m e d , necessary to m e t a p h o r . A compound diamond-stone is not internally i n c o m p a t i b l e - it can be assigned a c o - o r d i n a t e interpretation - but it m i g h t be a m e t a p h o r . The correct a s s u m p t i o n seems t h u s t h a t metaphor can be r e c o g n i s e d as such e i t h e r by c o n t e x t - i n f o r m a t i o n or by s e m a n t i c i n c o m p a t i b i l i t y . In the l a t t e r case, as in the compound diamond-sun, a c o - o r d i n a t * reading w i l l be semantically deviant, which should be s u f f i c i e n t to ensure that a m e t a p h o r i c i n t e r p r e t a t i o n is assigned to the c o m p o u n d . If there is no s e m a n t i c i n c o m p a t i b i l i t y and th