FROM CASE TO ADPOSITION AMSTERDAM STUDIES IN THE THEORY AND HISTORY OF LINGUISTIC SCIENCE Series IV - CURRENT ISSUES IN LINGUISTIC THEORY FROM CASE TO ADPOSITION THE DEVELOPMENT OF CONFIGURATIONAL SYNTAX IN INDO-EUROPEAN LANGUAGES TM 1hc paper used in this publication meets the minimum requirements of American National Ο© Standard for Information Sciences — Permanence of Paper for Printed Library Materials, ANSI Z39.48-1984. Library of Congress Control Number: 2006047967 ISBN 978 90 272 4795 7 (Hb; alk. paper) fc) 2006 - John Benj am in s B. V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, with- out written permission from the publisher. John Benjamins Publishing Co. · P.O.Box 36224 · 1020 ML Amsterdam · 'lhe Netherlands John Benjamins North America · P.O.Box 27519 · Philadelphia PA 19118-0519 · USA AUTHORSf PREFACE 1.0 Case Systems In our earlier monograph Tense and Aspect in Indo-European Languages, (Hewson & Bubenik 1997) we traced the verbal systems of Indo-European languages, and the different trajectories taken from what must have existed, from the evidence of the morphological reflexes in the daughter languages, at the level of the protolanguage to give all the different results, both ancient and modern. The analysis and reconstruction of the morphology and paradigms (the expression systems in Hjelmslev's terms) had at that point already been worked on for over a century; what we did was to analyse the evolution of each different system of representation (the content systems in Hjelmslev's terms) and how these meaningful systems had evolved out of the meaningful contrasts that could be traced to the protolanguage. The result was an emphasis on meaning (rather than morphosyntax), and on systemic contrasts and systemic evolution (rather than on atomistic contrasts and the evolution of single items). This is not a criticism of our predecessors; what we did could not have been done without the ground that they had already prepared in two centuries of historical and comparative work on Indo-European languages. Only from an understanding of the morphology can one come to an appreciation of the meaningful contrasts that the morphology marks, and ultimately of the systemic relationships built on those meaningful contrasts, the content systems, both simple and complex, whose cognitive structure is marked by paradigms, by morphological sets. In the present work we have turned from an examination of the meaning of the verbal morphology to scrutinize that of the nominal morphology. Here we had a model in the work of two distinguished predecessors: Louis Hjelmslev's (1935) two-part book La catégorie des cas of some 270 pages, and Roman Jakobson's (1936) article "Beitrag zur allgemeinen Kasuslehre: Gesamt- bedeutungen der russischen Kasus" of 46 pages: both authors had attempted to analyse the underlying meaning of nominal case systems. While not totally successful in the enterprise, they had established principles and ways of looking vi FROM CASE TO ADPOSITION IN IE LANGUAGES at the data that allowed us, over a period of years, to reflect on ways of improving on what they had already achieved. Hj elms lev, for example, had already established the terminology of content and expression (1935:xii), and the relationship of every expression system (i.e. grammatical paradigm) to a corresponding content system (a coherent set of contrastive meanings). Following the dictum established by Saussure, he noted that there was a priority in the meaningful contrasts (Saussurian valeurs, or values), the morphology of the corresponding paradigm being secondary: its only purpose is to mark the meanings and make them observable. Hjelmslev noted, in fact, that this priority of content is explicitly stated in the 19th century work of Franz Wüllner, a disciple of Franz Bopp. 2.0 The Priority of Content This priority of content is also carefully argued by Jakobson in the opening pages of his 1936 article, praising Hjelmslev and quoting with approval his resume of Wüllner's position (Hjelmslev 1935:84), which is translated in the reprint of the Kasuslehre article in Jakobson's On Language (1991:337): Ά grammar is a theory of basic meanings or values and of the system formed by means of them, and it must proceed empirically in accomplishing its task'. This point of view is also asserted by another distinguished post- Saussurian, Gustave Guillaume, in two laws, frequently discussed in his work (e.g. 1971:70-71; 1984:73-74; 2004:25, to give just a sample): (i) the Law of Coherence, which states that the content system {système de valeurs), the system of contrastive meanings, is fully coherent, and (ii) the Law of Simple Sufficiency, which states that the corresponding morphological paradigm is only as coherent as it needs to be: sufficiently coherent to mark the systemic meanings. Saussure's analogy of the game of chess drives the point home: the system lies in the moves each piece can make (its meaningful role). It does not lie in the pieces: if a rook is lost, for example, it can be replaced by a spice bottle, or, as Hjelmslev points out (1959:28), by "any conceivable object of a suitable size", without altering in any way the system of the game. The pieces are merely markers, just like the linguistic morphology that marks meanings and functions. Consequently, it is not unusual to have irregularities in the morphological paradigms. One notes, for example, that the meanings of the English plurals geese, mice, men represent the same kind of plurality as do the regular forms cats, dogs, horses. In the grammatical system of number in English there is a binary contrast in which (i) a unitary singular is established; and (ii) transcending that singular establishes a plural, based on the singular. One cannot have plurality that is not based on a singular, in the same way that photocopies can only be produced if one has an original document. English noun plurals are normally marked AUTHORS' PREFACE vii regularly, sometimes irregularly, for various reasons. The details of this simple content system with its variable morphology were worked out in Hirtle's (1982) monograph Number and Inner Space, and Wickens subsequent (1992) detailed overview of the whole range of singular/plural usage in English, entitled Grammatical Number in English Nouns. This way of thinking, though manifestly correct, was not popular during the 20th century, when the anti-mentalist bias of the positivist philosophers led linguists to search for system in the directly observable morphosyntax. As a result Bernard Bloch, because the morphology of the English strong verbs was not regular, gave to the irregular past forms, such as took, a zero suffix (to parallel the -ed suffix of the weak verbs) and indicated that the ablaut difference between take and took was meaningless: took was simply the allomorph of take to which the 'past inflection' was added, 'the only difference being that after this particular base the preterit suffix has the phonemic shape zero, as it has also after the base pu f (Bloch 1947/1957:245); this was done because real systems have to be coherent. Certainly, took and talked do have something in common, but it is not a now-you-see-it-now-you-don't suffix, but a meaning, past tense, representing time co-eval with the memory, Memorial Time. And took is a cumulative morph, a single morph which marks two or more different meanings together, as was pointed out by C.E. Bazeli in a little gem of a book entitled Linguistic Form, published in 1953, with numerous examples of cumulative morphs (portmanteau morphs in the Bloomfieldian tradition) from Indo-European languages. Since meaning was dismissed by the Bloomfieldians from its rightful position at the core of linguistics, it was shifted to the periphery, and some linguists took up the Truth Function Semantics of the philosophers, which stems ultimately from the 19th century philosopher Frege, and was only taken up seriously by linguists after many philosophers had already discarded it. If truth function is supposed to determine linguistic meaning, there is a major problem: how can truth function determine meaning, when the meaning of the sentence has to be known before its truth function can be established? If linguistic meaning necessarily determines truth function (you cannot tell whether a sentence is true or false if you don't understand it!), what sense is there in proclaiming that truth function determines meaning? It is time that the anomalies of Truth Function Semantics were recognized, and the whole anti-mentalist bias that underlies it finally discarded. In recent decades, in fact, there has been considerable reaction on several fronts against the view that grammar is syntax, that syntax is meaningless, and meaning is to be examined by virtue of its truth function. The work in Cognitive Linguistics, for example, is based on a principle that the basis of grammatical viii FROM CASE TO ADPOSITION IN IE LANGUAGES structure is meaning (e.g. Langacker 1987), and although there is no clear idea of system, work in this paradigm obviously has similarities to our own. Systems exist in the permanent storage of the subconscious mind (Saussure's langue) and, as Jakobson and Guillaume have shown, often have a hierarchical form (e.g. a phonological system > a vowel system > a system of nasal vowels) and are consequently not accessible to investigations that do not make some kind of Saussurian langue I parole distinction. But this distinction is missing from work in Cognitive Linguistics; instead one finds a concentration on lexical meaning, with a presupposition that lexical meaning is not categorically different from grammatical meaning, with the result that paradigmatic elements have been largely ignored. This was typical also of the generativist tradition, where there were disagreements as to whether paradigms belonged to the lexicon or were a sub-category of syntax, with neither alternative reflecting the observable reality. The simple paradigm of number in the English noun (cat versus cats), for example, is neither lexical not syntactic: I see the cat/s has the same syntax and lexicon for both singular and plural versions. 3.0 Dependency versus Constituency This book, however, is not simply about case meanings, and the cognitive contrasts that underlie nominal paradigms of case: it also has a diachronic dimension, an analysis of the evolution of nominal forms that largely parallels the diachronic survey of verbal systems in the earlier book. One of the notable diachronic shifts has often been commented on: the replacement of certain of the case forms by adpositional phrases in many of the languages. In these instances a single word with reasonably complex morphology, and without any necessarily fixed place in the sentence, is replaced by a phrase with a fixed order of words, a new departure that is to become the initial impulse for a new typology that has overtaken the whole phylum in varying degrees. All modern IE languages have some degree of configurational syntax, which was virtually non-existent in the earliest extant texts, and some modern languages have gone to the extent of complete loss of case in the noun, with the result that most syntactic strategies in these languages are configurational. To deal with this typological shift we have used a simple dependency syntax that originally goes back to Aristotle and the Greek grammatikē. This kind of dependency syntax was rejected by Leonard Bloomfield early in the twentieth century because dependencies, unlike immediate constituents, are not directly observable. Bloomfield consequently invented a syntax that was based on Immediate Constituents, syntactically related items that are observably contiguous (in English!). Chomskyan syntax followed Bloomfield, but it was soon discovered AUTHORS' PREFACE ix that, even in English, Constituency Syntax did not work very well. Ironically, one of the first adjustments was X-bar syntax, which was a way of reintroducing Dependency Syntax by the back door. Meanwhile the tradition of Dependency Syntax had not been totally eclipsed, and we have used and expanded this 2500 year old tradition which represents a simplicity of profound common sense. It is also a meaningful syntax, since when one item is predicated of another (= becomes dependent on another) the dependent element typically brings its meaning to the support to which it is predicated: in Jespersen's (1924:96) example of extremely hot weather, the adjective hot is said of weather, and the adverb extremely is said of hot. This is not a syntax which is meaningless. 4.0 Adpositional (prepositional and postpositional) Systems Dealing with the Adpositional Phrase also entails two other requirements: (i) the necessity of dealing with adpositional meaning, which has been a much discussed topic in the last twenty years, and (ii) relating adpositional meaning to the meanings of case systems, where there is obviously a degree of overlap, leading to cases being replaced by adpositions. In dealing with the first requirement, David Bennett states in the early pages of his seminal study of English prepositions (1975:5): "One of the most difficult problems in an analysis such as the present one is that of deciding how many senses to ascribe to a given lexeme". In many modern studies the underlying meaning of a preposition such as over is seen as a group of related meanings, a sort of molecule. Others see it as a single meaning, a monoseme, or ideal meaning, Jakobsonas Gesamtbedeutung, and there are often strong differences of opinion on these two possibilities for the underlying, permanent storage: is it single or multiple by nature. Bennett, in fact, states (1975:10) that his own "rather general senses of English prepositions ... are essentially similar to the Gesamtbedeutungen of Jakobson 1932 and 1936", and he proceeds to compare the variant meanings (allosemes) with allophones which are determined contextually, and the underlying meaning (sememe) to the single phoneme which determines the allophones, a comparison we have often made in our own work. Where such strong differences of opinion exist, however, it is normal that both parties are at least partially right, and there are ways of accommodating both views. In post-Saussurian linguistics the term norm was used by Hjelmslev and further developed by Coseriu (1952) to indicate that two different communities can use the same element with the same underlying meaning in slightly different ways, in the way that British usage allows Have you a pencil handy? whereas American usage normally requires the DO auxiliary here. From this perspective