[helena_kurzová]_from_indo-european_to_latin_the(bookzz.org)

277

Upload: -

Post on 09-Dec-2015

32 views

Category:

Documents


5 download

DESCRIPTION

Лингвистика

TRANSCRIPT

Page 1: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)
Page 2: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

FROM INDO-EUROPEAN TO LATIN

Page 3: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

AMSTERDAM STUDIES IN THE THEORY AND

HISTORY OF LINGUISTIC SCIENCE

General Editor E.F. KONRAD KOERNER

(University of Ottawa)

Series IV - CURRENT ISSUES IN LINGUISTIC THEORY

Advisory Editorial Board

Henning Andersen (Los Angeles); Raimo Anttila (Los Angeles) Thomas V. Gamkrelidze (Tbilisi); John E. Joseph (Hong Kong) Hans-Heinrich Lieb (Berlin); Ernst Pulgram (Ann Arbor, Mich.) E. Wyn Roberts (Vancouver, B.C.); Danny Steinberg (Tokyo)

Volume 104

Helena Kurzová

From Indo-European to Latin The evolution of a morphosyntactic type

Page 4: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

FROM INDO-EUROPEAN TO LATIN

T H E EVOLUTION OF A MORPHOSYNTACTIC TYPE

HELENA KURZOVA Czech Academy of Sciences

JOHN BENJAMINS PUBLISHING COMPANY AMSTERDAM/PHILADELPHIA

1993

Page 5: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

Library of Congress Cataloging-in-Publication Data

Kurzová, Helena. From Indo-European to Latin : the evolution of a morphosyntactic type / Helena Kur­

zová. p. cm. - (Amsterdam studies in the theory and history of linguistic science.

Series IV, Current issues in linguistic theory, ISSN 0304-0763; v. 104) Includes bibliographical references (p.) and index. 1. Latin language-Morphology. 2. Indo-European languages-Influence on Latin. 3. Latin language-Syntax. I. Title. II. Series. PA2137.K87 1993 475-dc20 93-1456 ISBN 90 272 3606 2 (Eur.) / 1-55619-558-3 (US) (alk. paper) CIP

© Copyright 1993 - John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher.

John Benjamins Publishing Co. • P.O. Box 75577 • 1070 AN Amsterdam • Netherlands John Benjamins North America • 821 Bethlehem Pike • Philadelphia, PA 19118 • USA

Page 6: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

Acknowledgements

I would like to express my thanks to all those who have offered advice and suggestions concerning matters of content as well as to those who helped me with preparing the final form of the manuscript. In this I would like to mention first of all my colleagues at the Institute of Classical Studies of the Czechoslovak Academy of Sciences.

This study was largely written during extended stays at the Linguistic Institute of the University of Bonn which were financed by a scholarship granted by the Heinrich Hertz Foundation, to whose President, Dr. Lingens, I owe a special debt of gratitude. My particular thanks are due to Professor Karl Horst Schmidt, Director of the Institute, for his continuing interest in the progress of my work. His stimulating proposals concerning the typology of Indo-European were among the fundamental ideas guiding my research. I am also very much indebted to other colleagues from the Institutes of Linguistics in Bonn and Cologne, especially to Reinhard Stempel and to Hans-Jürgen Sasse, whose contributions to the discussion of concept of 'IE subject' and related phenomena as well to language typology in general have been of great importance for my conclusions. The opportunity for the presentation of preliminary findings and for subsequent discussion at the Universities of Bologna, Vienna, and Bonn too provided me with additional stimulation for my work.

Finally, I would like to express my thanks to all those who helped me with the English style of the present volume in its various stages of completion, especially to Karin Hlaváček (Bonn), Ludmila Kollmanová (Prague), Timothy Bowen (East Sussex), Chris Searles (Cologne), and the anonymous readers who commented on the pre-final revision. The list of my benefactors is far from complete, I'm afraid. How­ever, I feel I must make mention of the kind cooperation and support received from Konrad Koerner, editor of this series, throughout the different phases of readying the text for publication. Remaining shortcomings remain my sole responsibility.

Prague, September 1992 H.K.

Page 7: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)
Page 8: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

Contents

PART 1 THE ORIGINS AND EVOLUTION OF THE INDO-EUROPEAN FLECTIONAL TYPE THE DERIVATIVE-FLECTIONAL TYPE OF INDO-EUROPEAN AND THE PARADIGMATIC-FLECTIONAL TYPE OF LATIN . 1

1.1 Introduction 3 1.1.1 Typology and historico-comparative research: Typological

framework 5 1.1.2 Evolutionary perspective: 'from non-differentiated/synthetic

to differentiated/analytic' 7 1.1.3 Approach to flectional morphology: Word modification vs.

morpheme addition: Decomposition via paradigmatization . 8 1.1.4 Semantic interpretation of the flectional structure: Enlarging

the notion of linguistic form 9 1.1.5 From derivation to inflection: Preliminaries on the recon­

struction of the derivative-flectional type 10 1.1.5.1 The principle of the grammatical relevance of lexical

categories: Morphological classes 11 1.1.5.2 The distribution and semantics of grammatical formants . 11 1.1.6 Derivative-flectional vs. paradigmatic-flectional types:

an approach to the typological characterization of the morphosyntactic structure of Indo-European and its development: Introductory remarks 13

1.2 The Nominative-Accusative Structure of Indo-European . . . . 15 1.2.1 Nominal ergativity/activity 16

Page 9: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

FROM INDO-EUROPEAN TO LATIN

1.2.1.1 Ergative/active case and the animacy hierarchy 17 1.2.1.2 Other structural evidence contra ergative/active hypothesis 18 1.2.2 Verbal activity 19

Main Characteristics of the Derivative-Flectional and Paradigmatic-Flectional Types 23

1.3.1 Morphological processes and characteristics 23 1.3.1.1 Vocalic alternation and suffixation: Internal and external

inflection 24 1.3.1.2 Implicit semantic distinctions: Inherent characteristics of

roots/stems . . . . . . . . . . . 26 1.3.2 Characterization of IE flectional morphology: Preliminaries 27 1.3.2.1 Cumulative/fusional and complex character of IE mor­

phology: Allomorphy and form variation 27 1.3.2.2 IE flection and natural morphology 29 1.3.3 The morphemic structure of the IE word 30 1.3.3.1 The IE word as a hierarchically structured unit:

Lexico-derivative and grammatical components: Paradigmatization 30

1.3.3.2 The structure of the flexive ending in the paradig-matic-flectional language: Evidence from Latin 31

1.3.3.3 The structure of the flexive ending in the deriva-tive-flectional stage: Evidence from Greek 35

1.3.4 Character of the morphological semantics: Category-im­manent vs. relational semantics 37

1.3.4.1 The IE word as a semantic complex 38 1.3.4.2 Category-immanent vs. relational semantics 39 1.3.5 Word categories: Noun and verb distinction. Position of the

adjective 41 1.3.6 Structure of the Indo-European sentence: Word autonomy in

the non-formalized sentence structure . 42 1.3.6.1 Word autonomy and agreement 42 1.3.6.2 Government: The place of the verb in the IE sentence . . 43 1.3.6.3 Word order 44 1.3.7 Analytic and incorporation-like features of Indo-European . 45 1.3.7.1 Analytic morphemes 45 1.3.7.2 IE composites as pseudo-incorporative elements 46

Page 10: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CONTENTS ix

PART 2 THE BASIC PRINCIPLES AND ORIGINS OF THE NOMINAL SYSTEM AND INFLECTIONS 47

2.1 Introduction Principles of Nominal Inflection: Derivative-Flectional Morphology of Nouns 49 2.1.1 The flectional principle of expressing number and case . . . 49 2.1.2 Types of allomorphy in Latin noun inflection: Allomorphy in

Latin and Indo-European 51 2.1.3 The morphological character of basic case suffixes: In­

do-European allomorphy motivated by the accent-ablaut alternations 54

2.1.3.1 Hysterodynamic, proterodynamic and amphidynamic infle­ctions of consonantal and -i-/-u- stems 55

2.1.3.2 The -o- stems inflection 56 2.1.4 Noun and adjective: Word autonomy and the semantics of

nominal categories 57 2.1.5 The Latin and Indo-European nominal systems: Preliminaries 58 2.1.6 Derivative flectional morphology of nouns . 59

2.2 Animate and Inanimate Noun Classes The Category of Gender 61 2.2.1 Morphological differences between animates and inanimates

and their semantic motivation 61 2.2.2 Gender and agreement 63 2.2.2.1 IE agreement in gender and agreement in nouns class of

Bantu languages 65 2.2.2.2 Agreement and word autonomy: Semantic,

morphological/paradigmatic and syntactic functions of gender 66

2.2.3 Conclusions 67

2.3 The Category of Number 69 2.3.1 Singular — Dual — Plural 69 2.3.2 The nominative and accusative plural: The nucleus of the

plural paradigm 71

Page 11: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

FROM INDO-EUROPEAN TO LATIN

2.3.2.1 The nominative plural: Intraparadigmatic analogical innovations 71

2.3.2.2 Agglutination-like form of the accusative plural 72 2.3.2.3 Transfer of the -s marker from the nominative plural to

the accusative plural 73 2.3.2.4 The element -s as a representative plural marker 75 2.3.2.5 Remarks on plural nominatives without -s 75 2.3.3 Conclusions 76

The Case System 77 2.4.1 The basic case system of animate nouns: Preliminaries . . . 77 2.4.1.1 The basic case forms 77 2.4.1.2 The semantic analysis of the basic case system 78 2.4.2 Case suffixes -s, -m vs. ~es/-os, -om: Nominative, accusative

vs. genitive 80 2.4.2.1 Derivational semantics of the suffixes -s and -m: Types of

participation 80 2.4.2.2 Paradigmatization of case oppositions: Preliminaries . . . 81 2.4.2.3 Genitive marking by -s and -m suffixes: Possessivity and

partitivity. Genitive - ablative 82 2.4.2.4 The -s marking and animateness: Restriction in -s

nominative 83 2.4.3 Nominative and subject constitution 83 2.4.3.1 Verb orientation in the IE sentence 84 2.4.3.2 Thetic and categorical statements 84 2.4.3.3 The semantic/prototypical and syntactic/generalized

subject 86 2.4.4 The paradigmatization of nominative - accusative - genitive

oppositions 86 2.4.4.1 Remarks on the position of nominative: Nominative -

accusative opposition 87 2.4.4.2 Remarks on the genitive problem 88 2.4.5 Case suffixes -i vs. -ei: Locative vs. dative 89 2.4.6 Other components of the IE and Latin case paradigms . . . . 89 2.4.6.1 Casus indefinitus or primitivus 90 2.4.6.2 Case forms with vocalic suffix: Instrumental 90 2.4.6.3 Cases with marked postinflectional suffixes 91 2.4.6.4 The Latin ablative and its Indo-European components . . 91

Page 12: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CONTENTS xi

2.4.7 The plural paradigm 92 2.4.7.1 The Latin dative-ablative plural in -is and its IE origins . 92 2.4.7.1.1 The IE locative plural suffix -si/su 93 2.4.7.1.2 The IE plural instrumental in -oi-s 93

'2.4.7.1.3 The postinflectional character of -su/-si and -s 93 2.4.7.2 The Latin dative-ablative in -bus 93 2.4.8 Innovative case suffixes of -o- stems: The pronominal

subsystem 94 2.4.8.1 Genitive singular in -os-io/-os-o . . . . . . . . . . . . . . . 94 2.4.8.2 Nominative plural in -oi 94 2.4.9 The PIE case paradigm and its development 95 2.4.9.1 The evidence of Hittite 95 2.4.9.2 The Greek paradigm with alleged syncretism 96 2.4.9.3 The late IE isoglosses in case formation 96

2.5 Remarks on Latin Paradigmatization 97 2.5.1 The morphological and the inflectional paradigms 97 2.5.2 Some features and tendencies of Latin paradigmatization . . 98 2.5.2.1 Semantic and formal markedness (singular vs. plural) . . 99 2.5.2.2 Homonymy and its structural disambiguation and dehomo-

nymization: The position of the nominative singular . . . 100

2.6 Conclusions 103

PART 3 THE INDO-EUROPEAN ORIGINS OF THE LATIN VERBAL SYSTEM 105

3.1 Introduction Derivative-Flectional Morphology of the Indo-European Verb: Paradigmatization 107 3.1.1 Derivative-flectional and paradigmatic-flectional morphology

of the Indo-European verb: Reconstruction of the two-class system 107

3.1.2 Methodological remarks 109 3.1.2.1 Approach to the development of the Indo-European verbal

system 109

Page 13: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

Xli FROM INDO-EUROPEAN TO LATIN

3.1.2.2 The Greek/Aryan vs. the Latin verbal system as two alternative realizations of the paradigmatization process . 109

3.1.3 Morphological and semantic-syntactic categories I l l 3.1.4 Category-immanent semantics of the Indo-European verb

morphology: Injunctive-based system 112

3.2 The Active and Inactive Verb Classes The Development from an Injunctive-Based Two-Class System into an Integral Present-Based Verbal System 115 3.2.1 Characterization of active and inactive classes 116 3.2.1.1 The active vs. inactive verbs are morphological classes

(see 1.1.4), i.e. non-oppositional lexical classes with mor-phosyntactic relevance . . . 116

3.2.1.2 Diathetic classes and aspectual subclasses . . 118 3.2.1.3 Injunctive-based and present-based verbal system . . . . . 119 3.2.1.4 The meaning and function of the classification active vs.

inactive: Its diathetic character . 120 3.2.1.5 The non-oriented, diathetically vague inactive verbs . . . 121 3.2.1.6 The active vs. inactive verb classes and animate vs.

inanimate noun classes: The nominative-accusative morphosyntax of Indo-European 124

3.2.2 Preliminaries to the reconstruction of the active and inactive verb classes 127

3.2.2.1 Hypothesis on the morphonological difference between active and inactive roots 127

3.2.2.2 Representative examples of the active and inactive verbs 129 3.2.3 Morphology of the active and inactive verb classes: Injunc­

tive-based system developing into present-based system . . . 131 3.2.3.1 Personal endings of the active verb . 131 3.2.3.2 'Primary' forms with the postinflectional element -i:

The IE injunctive 132 3.2.3.3 Simple athematic and thematic forms 132 3.2.3.4 The thematic present 134 3.2.3.5 Personal endings of the inactive verb 135 3.2.3.5.1 The laryngeal interpretation 137 3.2.3.5.2 The -d2 and -fy variants of personal endings 137 3.2.3.5.3 The -el-o variants of the 3rd person 138 3.2.3.5.4 The primitive two member system of inactive endings . . 138

Page 14: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CONTENTS xiii

3.2.3.6 Other morphological processes: Stem alternation. Suf­fixation 139

3.2.3.6.1 Stem alternation: Internal inflection 139 3.2.3.6.2 Suffixation 140

Inactive and Perfect The Latin perfect 143 3.3.1 Greek and Aryan opposition of perfect to aorist vs. their

identification in Latin and other IE languages 143 3.3.2 Semantic and morphological basis of the identification of

aorist with the perfect in Latin: The central position of the present 145

3.3.3 The Latin perfect endings 147 3.3.4 The Latin perfect stems . . 149 3.3.4.1 Sigmatic perfects 150 3.3.4.2 Perfects in -vi/-ui .150 3.3.4.3 Reduplicated perfects . 151 3.3.4.4 Perfects with lengthened stems 151 3.3.4.5 Unmarked perfects 152 3.3.5 The Latin perfect system vs. present system 152 3.3.6 Inactive and perfect: Conclusions 155

Inactive and Middle-Passive The Latin deponent/passive 157 3.4.1 Inactive origin of the middle-passive: Preliminaries 157 3.4.2 Discussion of alternative explanations 158 3.4.3 'From inactive to middle-passive': Semantic analysis . . . . 159 3.4.3.1 The Greek middle-passive vs. the Latin deponent/passive:

Syntactic diathesis and semantic version 160 3.4.3.2 The origin of the oppositional and marked middle-passive

in the non-oppositional and non-marked inactive 162 3.4.3.3 The innovative character of the Greek/Aryan oppositional

middle 163 3.4.3.4 Impersonal passive: Latin backgrounding passive vs.

Greek foregrounding passive 164 3.4.4 Formal characteristics of the Greek and Latin middle-passive 165 3.4.4.1 The middle-passive endings of the -toft) type 166 3.4.4.2 The -r endings 167

Page 15: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

xiv FROM INDO-EUROPEAN TO LATIN

3.4.5 Foregrounding -to(i) middle-passive vs. backgrounding -r middle-passive: Dialectal difference in IE area 168

3.4.6 The endings of the Latin deponent/passive: A revised analysis 170

3.4.7 Inactive and middle-passive: Conclusions 171

3.5 Thematic -e/o- suffix and Long Vocalic Suffixes Latin subjunctives and futures: Latin present conjugations 173 3.5.1 Functional parallelism of -e/o- injunctive and long vocalic

-ă-, -ë- injunctives 174 3.5.2 The preterital and modal value of IE injunctive 176 3.5.3 The semantics of the vocalic suffixes 176 3.5.4 Latin subjunctives and futures with -ă- and -ë- suffixes . . . 177 3.5.5 Greek thematic imperfects and subjunctives 179 3.5.6 Conclusions: Latin present conjugations. Basic tendencies of

Latin paradigmatization 180

3.6 Sigmatic Forms of the Latin Present and Perfect Systems The Indo-European -s- Suffix 181 3.6.1 The aoristic and modal -s- suffix 181 3.6.2 The morphology of formations with consonantal suffixes:

Latin sigmatic forms . 182 3.6.2.1 The derivational basis of Latin sigmatic forms 183 3.6.2.2 The vocalic endings of the Latin sigmatic forms 183 3.6.3 Imperfect subjunctive: The -se- formation 184 3.6.4 Futures and subjunctives of the type faxo, faxim (-se / so- and

-si- formations) 185 3.6.5 The perfect forms with sigmatic suffix . 185

3.7 Latin Imperfect in -bam and Future in -bo The Indo-European -dh- suffix 187 3.7.1 The analytic explanation of the Latin indicative imperfect . . 187 3.7.2 The flectional explanation of the indicative imperfect . . . . 187 3.7.3 The -dh- suffix in Greek, Germanic and Latin preterite: The

Latin -bo future 188

3.8 Conclusions 191

Page 16: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CONTENTS xv

Summary 191

Notes 195

References 215

Index 233

Page 17: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)
Page 18: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

PART 1

THE ORIGINS AND EVOLUTION OF THE

INDO-EUROPEAN FLECTIONAL TYPE

THE DERIVATIVE-FLECTIONAL TYPE OF

INDO-EUROPEAN AND THE

PARADIGMATIC-FLECTIONAL TYPE OF LATIN

Page 19: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)
Page 20: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

1.1 Introduction

1.1.0 In this study an attempt will be made to describe the typological characteristics of the original Indo-European structure, called the deri­vative-flectional (d-f)1 stage or (sub) type, and to trace its development to the paradigmatically organized structure of the individual Indo-European languages, called the paradigmatic-flectional (p-f) stage or (sub)type. This development is demonstrated in Latin, a language which is characterized by a highly developed inflection and at the same time attests, especially by its ver­bal system, an alternative way of paradigmatizing the original structure, differing from languages on which traditional reconstruction was based, i.e. Old Indian and Greek. In this way, the present study with its primary typological orientation attempts to contribute to the topical problems of the reconstruction of Proto-Indo-European and traces the developmental history of the IE morphosyntactic type by showing alternative variants of this development.

By means of the notion of derivative-flectional type I will attempt to penetrate to the original form and historical sources of the IE flectional type without presupposing radical typological change between Proto-IE and IE. This differs from the traditional theory of prehistoric change of IE structure from isolation to flection via agglutination,2 the original isolating structure being considered as a structure with analytic expressions of grammatical categories and with fixed word order. In our conception, the origins of flection lie in lexico-derivative categorization. Instead of non-motivated transfer of the attested flectional structure to a hypothetical structure based on different typological principles, I attempt to exhaust the interpretational possibilities of the attested structure (see Kurzová 1973, 1986).

In its current form the typology distinguishing flection, agglutination and isolation as main typological principles (= FAI typology, with further possible subdivisions) is not concerned with semantics. In the present study,

Page 21: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

4 FROM INDO-EUROPEAN TO LATIN

however, I will try to interpret IE flectional structure semantically, and with regard to the complementary relationship between word and sentence structure. Though the morphological system will remain the centre of our attention, the morphological processes and categories will be considered with respect to their semantic-syntactic relevance. Systematic treatment of the semantics enables us to see the basic agreements and characteristic differences of both stages of flection and to understand the internal connection between the more remote and more recent stages of IE reconstruction. This furnishes a basis for placing the reconstructed phenomena in relative chronological order or in alternative relationship. I shall try to suggest a relative chronological or alternative interpretation of the morphosyntactic phenomena and systems examined. I will not, however, try to draw any general conclusions with respect to a spatiotemporal differentiation of IE. A more subtle differentiation of the stages of development, as suggested especially by Meid (1975), is a matter for future systematic research which is beyond the possibilities and scope of this study.

The first part of this book describes the general characteristics of IE morphosyntactic structure and defines the main differences between the two stages (subtypes) of IE flection. In this context the assumption of either the active or ergative structure of Proto-IE is critically discussed. Both the nominal (Part 2) and the verbal (Part 3) systems of IE and Latin are analyzed on the basis of the presupposed typological development from the derivative-flectional type of IE to the paradigmatic-flectional type of Latin. The different conditions in regard to typologico-historical analysis encountered in the nominal and verbal systems lead to certain differences in our approach to the noun and verb. We shall explain the basic principles of flectional type in connection with nominal categories, the nominal case-number morpheme serving as a prototypical cumulative morpheme, and gender as a grammati-calized lexical class. While the paradigmatization of basic cases was already completed in the Common IE period, the paradigmatization of the verbal system exhibits more alternative variants in the IE languages. Here the typological analysis can contribute in an important way to the interpretation of the data offered by IE research. The analysis of the verbal system presented in Part 3 verifies the explanatory ability of the typological approach chosen for this study.

Page 22: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INTRODUCTION 5

1.1.1 Typology and historico-comparative research: Typological framework

Typological evaluation of the data given by the historico-comparative analysis is a necessary precondition of the reconstruction of IE structure and its development. Typology serves as a basis for so-called internal recon­struction, which allows the analyst to go beyond the language structure that emerges via primary reconstruction. However, primary reconstruction also cannot be based on the data offered by the historico-comparative method only, i.e. by so-called external reconstruction. Typology provides a necessary con­trol device of the results of external reconstruction, i.e. of the compatibility and synchronizability of the given data. The decision whether two or more variants resulting from external reconstruction are to be put in the relative chronology A => B or considered as alternatives going back to stage C is possible only on the basis of what we "know" about the character of the structure and of the evolutional process we are reconstructing.

Two extreme views of and attitudes towards the material as offered by external reconstruction — on the one hand the naive realistic identification of these data as elements of the parent language, on the other hand the lack of interest in interpreting the conglomerate of the heterogenous and incoherent elements (which can then be freely manipulated and enlarged through re­constructive finesse) as potential elements of language structure — are to be rejected via the typological evaluation of this material.

Yet, for these purposes we need a typology which attempts to give a characterization of the entire language structure by determining its dominating principles.3 Only this kind of typology guarantees the adequate use of the results of the general comparative research for the reconstruction of IE. As we shall see, inadequate extrapolations from non-Indo-European languages in many cases have deformed the picture of IE structure. We must always have in mind that the questions which are posed by one language structure in relation to the other must be answered with respect to the whole character of the compared structures.

Classical FAI-typology, if enlarged in scope, is able to define the language types according to the principle of how the word and the sentence are structured with respect to the expression of grammatical categories. In the flectional structure the word represents a complex hierarchically structured unit which expresses grammatical categories by means of the integral

Page 23: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

6 FROM INDO-EUROPEAN TO LATIN

modification of its structure,4 not by the addition of a formal element serving as a sign of the grammatical category. The sentence structure is non-formal­ized, allowing the free combination of words (the principle of 'word autonomy').

If in the further development of IE languages flexive morphology is more or less suppressed by agglutinative or analytic morphology, we must shift the basis of typization. It is the formalization of sentence structure which can be considered as the general tendency of the development of European languages (see Kurzová 1974, 1981). This basic process can again be typified by defining the dominating principles of its realization, and in this framework the differences in the use of flectional agglutinative and analytic processes5

will be motivated. An essential feature of the typology aimed at the characterization of the

entire language structure is its ability to furnish a perspective, where not only the phenomena prototypically corresponding to the defined principles are taken into account, but all language phenomena relevant to the selected base of typization. We shall see that the analytic or postinflectional and the incor­poration-like elements of IE structure have a character which is determined by the dominating flexive principle (1.3.7). And the same applies to agglutina­tion-like structures (2.3.2.2). Syntactic processes such as agreement (1.3.6.1; 2.2.2) and government (1.3.6.2) have a different character in IE than in other types of structure, and these differences can be explained by typological characteristics of the respective structures. These examples reflect my attitude toward typology. Seemingly identical phenomena and properties have different characteristics in the different language types, occupying different positions in the structural hierarchy. Therefore the differences between language types cannot be appropriately defined by simple clusters of properties. They are determined by underlying principles responsible for the character of the whole structure. Typology must describe and explain the internal coherence of the language structure, in agreement with the intention of classical typology.

The typology determining the principles of word and sentence structuring with respect to the expression of grammatical categories also includes in its characterization word order rules (1.3.6.3) and the 'syntax of fundamental relations' (ergative/active vs. nominative structure - 1.2), which structural properties can be typologically evaluated only in this wider frame-work.

The enlarged conception of flectional type as defined above (and 1.1.4) furnishes a perspective which, by adopting further, more specific criteria, can

Page 24: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INTRODUCTION 7

provide a subtle characterization of the individual IE languages. I have applied this method in a comparative characterization of Latin and Greek (see especially Kurzová 1970, 1983, 1987a), which in the present study will be considered only for those features with wider relevance for IE languages.

1.1:2 Evolutionary perspective: 'from non-differentiated/synthetic to differentiated/analytic'

For our explanation of IE development the connection of typological characterization with an evolutionary perspective is essential. The selection of FAI-typology as the basis for our analysis makes no unequivocal presupposi­tions as to the evolutionary perspective of interpreting the development of IE structure. The most common assumption made in the framework of FAI-typo­logy is the change from an analytic/isolating to agglutinative and final flexive structure in the prehistorical development of IE. This conception, according to which the fundamental language types alternated in language development, is unsatisfactory and certainly not the only possible evolutional perspective. It seems more likely that the main language types, all well represented by the oldest attested languages of individual language families, cannot be derived from each other in a particular chronological sequence, but represent alternative ways of structuring the sentence with more lexemes. None of the attested language types can be equated with a primitive language assumed to contain simple one-word or bipartite statements. The transition from this primitive stage to complex sentence structuring gives rise to the alternative types of dividing the expression of grammatical categories and relations between word and sentence. In the flectional type, grammatical categories are expressed by internal modification of the word, in the agglutinative type by morpheme addition, and in the isolating type by specific words or by word ordering. Therefore, according to our conception, the main principles of language structure which are defined in the various language types of FAI-typology represent different original principles of structuring statements con­taining more than one lexeme, i.e. structures where word = sentence.

The typological evolution of Indo-European from derivative-flectional to paradigmatic-flectional stages did not occur on a compositional basis, i.e. one assumed for an evolutional process going from isolation to flection via agglutination, but, on the contrary, on a decompositional basis. This evolu­tional process can be characterized as the development from non-differentiated complexes to more differentiated, analyzable, and explicit complexes. In both

Page 25: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

8 FROM INDO-EUROPEAN TO LATIN

evolutional stages the word is structured according to its function in the sentence. It represents a complex unit joining lexical and grammatical information in its internal structure. The difference between both stages consists in the development of an explicit and discrete expression of grammatical semantics thus leading to its relative separation from lexical-derivative semantics. The IE word as a formally and semantically non-differ­entiated complex was differentiated and decomposed in the process of paradigmatization by means of paradigmatized oppositions into formal and semantic components (1.1.3). It seems that processes based on decomposition are very typical for the biological and anthropological world. In biology, the organisms and organs with non-differentiated structures are genetically prior. The relationship of mythological to logical thought also follows this perspective. Mythological thought evokes a complex and deep understanding of the world, which is then made explicit and differentiated into individual components by logical thought. These components are, however, somehow present or already implied by mythological ideas. Therefore, we presuppose for the IE word as a typologically relevant unit the evolution from a non-differentiated, non-analyzable whole to a more highly differentiated, relatively analyzable (decomposable) whole, which, however, also differs in its analyzable stage from the units of more analytic, i.e. isolating and agglutinative, structures. With respect to the agglutinative structure we mark this difference with the help of the terms additive vs. non-additive. (2.1.1). It is crucial for our understanding of the original IE structure to do away with the tendency which forms the basis of 'analytism', to explain the flectional morphemes as originally analytic elements (grammatical words). This tendency is based on the prejudice that the primary meaning-constitutive activity should concern words only and not word modifications. This assumption, however, lacks justification. We may assume that word modifications also had primary significative value.

1.1.3 Approach to flectional morphology: Word modification vs. morpheme addition: Decomposition via paradigmatization

The definition of flectional expression of grammatical categories as an integral modification of the word (or of the word structure), which I have accepted (1.1.1, note 4), determines my approach to flectional morphology in the sense that I shall not attempt to decompose the word into morphemes6

with respect to individual grammatical categories. This is possible only for

Page 26: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INTRODUCTION 9

language structures based on morpheme addition. The evolutionary tendency to develop more explicit and analyzable/

decomposable structures is governed and controlled in a flectional language by the integrating tendency which dominates the word structure. The mutual conflict between these two tendencies is solved by the principle of paradigmatization. The decomposition and making explicit of the form -function relations is realized by the paradigmatic oppositions that exist between word forms which have a certain common formal and semantic component while differing in others. On the basis of paradigmatic oppositions these components are decomposable. The flexibility of the paradigmatic interrelations accounts for the great decompositional capacity of such an organized structure: the same formal components have different values in different paradigmatic relationships (e.g. the verbal ending -o in dic-o signals the indicative with respect to die-am and the 1st person with respect to die-is, etc.). We shall examine the phenomenon of paradigmatization, a recent centre of attention for linguistic research, especially from the historical point of view, i.e. as the basis for the renewal and enrichment of the morphosyntax.

I shall try to describe the complex expression of the respective grammatical category by determining the paradigmatic opposition within which this expression is identifiable. E.g. the present indicative dic-o, die-is is marked by the present stem and endings with respect to the perfect indicative dix-i, dix-isti, whereas the same category is marked by modal vocalic exponents (partially fusional with the ending) dic-o, dic-i-s with respect to the present subjunctive dic-a-m, dic-a-s. The decomposable formal elements will be called 'formants', or alternatively 'markers', 'exponents' if their relation to a certain semantic feature is particularly clear. However, taking into account the tendency toward explicitness and the formation of external inflection (1.3.1.1), I shall not avoid the terms morpheme and morph when speaking about the morphemic structure of the word and the interrelation of lexical/derivational and inflectional (grammatical) components/morphemes (1.3.3.2).

1.1.4 Semantic interpretation of the flectional structure: Enlarging the notion of linguistic form

In this study I will work with an enlarged conception of flectional type which goes beyond the simple assertion that the grammatical categories are expressed not by the addition of separate morphemes of agglutinative or

Page 27: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

10 FROM INDO-EUROPEAN TO LATIN

isolating type, but by the integral modification of word structure. Two mutually connected questions are decisive for the deeper characterization of the flectional structure: 1) what type of grammatical semantics is bound to flectional morphological processes; cf. 1.3.4. 2) how the morphological processes represented by suffixation and stem alternation relate to other formal expressions of semantic-syntactic categories and relations.

It is an enlarged notion of linguistic form which must be applied in this characterization (see Kurzová 1987:203). The grammatically relevant semantics is expressed not only by morphological flectional processes, but also by syntactic devices such as government and agreement, by lexical subcategorization,7 or by substitution possibilities. These various devices can be covered by the useful term 'morphosyntactic form'. These other formal means contribute at the same time to the specification and disambiguation of the meaning of the polyfunctional and vague morphological formants of flectional structure.

The enlarged notion of linguistic form is a most important result of modern linguistics, common to all various linguistic schools and trends. Yet, it should be taken into account that the hierarchy of formal devices is different in various types of language structure and that these devices differ also in their internal characteristics. The fact that, in general, the same morphosyntactic form bears a different character in the individual structures will be examined especially with respect to agreement and government (1.3.6). Also the principle of grammatical relevance of lexical categories is a common and universal feature of language structure, but its position in the hierarchy of morphosyntactic devices is widely different. The dominant position of this feature characterizes the d-f structure.

1.1.5 From derivation to inflection: Preliminaries on the reconstruction of the derivative-flectional type

The reconstruction of the original form of IE flection is based on the observations and interpretation of certain morphological and morphosyntactic features of ancient IE languages.8 These observations concern two pheno­mena, both showing the primary character of lexical and derivational catego­rization which was the basis for inflection proper: 1) the grammatical rele­vance of lexical morphemes (1.1.5.1); 2) the distribution and semantics of grammatical formants (1.1.5.2).

Page 28: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INTRODUCTION 11

1.1.5.1 The principle of the grammatical relevance of lexical categories: Morphological classes

We can observe in the ancient IE languages that the morphological categories are not valid for all nouns and verbs, i.e. for noun and verb as such, but for certain noun and verb classes only. We have so-called deponent verbs in Latin such as hortor or orior, which are not marked for the active vs. passive opposition. And we also have perfects such as memini or odi, so called presento-perfects, which do not have the corresponding present form. In the nominal system the neuters are not marked for the nominative vs. accusative opposition, but possess an indifferent case only, which can be called absolutive (see 2.2.1). Therefore, in the nominal system in direct form and in the verbal system in residual form, large lexical classes are attested, which were relevant grammatically. It can be proved — as we shall show further on — that the morphological differences between classes were motivated semantically in the original structure, so that the grammatical relevance of the lexical classes was not only a formal and morphological but also a semantic-syntactic phenomenon. The grammatical relevance of lexical categorization was a dominant feature of IE flectional structure, which affected basic semantic-syntactic categories and relations (case-relations, diathesis, and aspecto-temporal distinctions).

1.1.5.2 The distribution and semantics of grammatical formants

We can further observe that the same category does not have the same expression in all nouns and verbs, but is expressed by different formants, either by suffix or by alternation (ablaut). It can be proven in many cases that these alternative (allomorphic) formants originally differed in their meaning. They exhibited specific features and were only by progressive abstraction united secondarily under the same grammatical category. This took place on the basis of a certain common semantic feature which became relevant for the given category. In the nominal system we have the example in the plural formant -a of the neuters (verb-a, nomin-a) which is semantically identified with the plural of the feminines and masculines in Latin and considered as a pure allomorphic expression of the same category. Originally, however, this formant had the specific meaning of expressing collectivity (2.2.1).

To a much greater extent the original distributional and semantic properties of formants are observable in the verbal system. The Greek aorist

Page 29: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

12 FROM INDO-EUROPEAN TO LATIN

and present formations present a good example of this. The opposition present vs. aorist was grammaticalized in Greek as

aspectual opposition. Various formations serve as expression of aorist and presents in Greek. We have athematic root aorists such as 9 thematic aorists such as , reduplicated (Homeric) aorists such a s , sigmatic aorists such as , aorists with suffix such as and

such as Correspondingly we have athematic purely thematic reduplicated and suffixal presents

etc.). Different present formations can correspond to the same type of aorist: the thematic aorist has simple thematic present with vowel gradation whereas the thematic aorists are opposed to the suffixal presents

The specific original meanings of various aorist and present formations can be partially reconstructed (see especially Strunk 1967, Kølln 1969). They were subjected to the grammatically relevant aspectual meaning perfective/ aoristic vs. imperfective/presentic, and then suppressed in favour of this meaning.

The Latin present vs. perfect opposition, which semantically represents opposition of an another aspecto-temporal type, is realized by similar diversity of formations (3.3.4) without, however, any clear remnant of original specific meanings. Therefore, the formants which serve to express the morphological categories in IE languages were originally restricted to certain nouns and verbs only, and had specific meanings. This, however, is a trait typical of derivation and not of inflection proper. The original inflective formants are close to the derivational formants in their distributional properties. The same also applies to their morphological properties and to their type of morphological semantics. The inflectional categories of p-f structure developed by way of progressive grammaticalization, consisting of a) abstraction from the original specific meaning which was abandoned or backgrounded in favour of the basic, grammaticalized meaning, and b) the tendency towards generalization of the category with grammaticalized meaning, i.e. towards its expression in all nouns and verbs.

The different morphological systems of individual IE languages in this historical perspective appear as the result of differently realized processes of grammaticalization and paradigmatization of the original structure. This offers a new possibility for explaining the different morphological systems, especially the verbal systems (3.1.1), as alternative realization of the same historical process.

Page 30: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INTRODUCTION 13

1.1.6 Derivative-flectional vs. paradigmatic-flectional types: an approach to the typological characterization of the morphosyntactic structure of Indo-European and its development: Introductory remarks

In the definition of flectional type the formal-morphological features were the main centre of attention. By reconstructing the older stage of flection I will try to answer the fundamental question of what type of grammatical semantics was inherent in the full semantic word and bound so closely to lexical semantics. The semantic problems which thus became decisive must be treated with respect to the semantic-syntactic structure of the sentence toward which all morphologic and morphosyntactic processes are oriented in their final result. If we speak of the primacy of word structuring to sentence structuring as an invariable of the flectional type, we mean the formalization and hierarchization of the internal structures of these complementary units which participate in the expression of grammatical semantics: the IE word as a relatively autonomous and hierarchically structured complex vs. the IE sentence as a non-formalized structure. However, in any case, grammatical semantics belongs to the semantic-syntactic level, be it expressed morpho­logically by inflection or syntactically.

By including the semantic-syntactic level, the interpretation of the flectional type receives a necessary conceptual syntheticity allowing us to define the principal agreements and characteristic differences between both subtypes. I will attempt to give a synthetic characterization of the flectional type in its two stages in section 1.3. I shall move from the morphological to the semantic-syntactic level, trying to show the typological consistency of structural features belonging to different levels. Before doing this, in section 1.2.I will briefly examine the assumption of an ergative or active structure of (Proto)-Indo-European.10

The assumption of either an ergative or active structure shows us that the transfer of an isolated structural feature from one language structure to another without taking into account the entire character of the compared structures is an inappropriate and misleading method. On the other hand, the comparison with ergative and active languages has been stimulating for IE research and we may make use of it positively, while at the same time rejecting the hypothesis about the alleged ergative or active structure of PIE.

Page 31: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)
Page 32: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

1.2 The Nominative-Accusative Structure of Indo-European

1.2.0 In discussing the character of fundamental syntactic relations in IE we must anticipate many points of our morphosyntactic analysis of IE. The theories of the original ergative or active structure of IE are based on extrapo­lations not respecting the whole character of compared language structures. I shall try to show that the question of nominative-accusative11 vs. ergative/ active structure must not be treated in isolation from other structural features of the languages in question.

The theories of the ergative character of original IE structure are based on interpretations of certain phenomena of nominal morphology, which resemble some features of ergative languages:12 the marked IE -s nominative and its possible connection with the genitive; the 'absolutive' case of neuters (2.2.1). In new research the hypothesis about the non-nominative structure of IE is modified in the sense that this structure is rather of active than of ergative character.13

The active structure differs from the ergative structure in that the opposition agentive vs. non-agentive is realized also in the case of intransitive verbs, which thus are differentiated into agentive/active14 verbs like "go" and non-agentive/inactive verbs like "lie". An essential feature common to both structures is the identity of the patient of transitive active verbs (IE object of the verb "kill" etc.) with the non-agent 'subject' of inactive verbs.

(1) active structure /Agent tr = Agent intr/ ≠ /Patient tr = Non-Agent subj/ + kill + go + kill + lie, know

Page 33: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

16 FROM INDO-EUROPEAN TO LATIN

(2) ergative structure /Agent tr/ ≠ /Patient tr = Non-Agent subj/ + kill + kill + lie, know

+ go Agent tr = agent of active transitive verbs / kill Agent intr = agent of active intransitive verbs / go Patient tr = patient of active transitive verbs Somebody kills me Non-Agent = non-agentive 'subject' (primary actant) of inactive verbs

I lie, I know

The differentiation and identification of these actants is realized by means of nominal case morphemes and/or by means of verbal person affixes. In ergative structures both nominal and verbal expression are equally well attested. 'Classical' ergative languages like Georgian and Basque are languages with cases. The opposition of Agent tr vs. Patient tr, Non-Agent subj is expressed by means of ergative vs. absolutive cases. On the other hand, for active languages, verbal expression by means of active and inactive person affixes is typical; they are languages without cases or with very restricted case marking. Thus for IE noun the atypical expression of active vs. inactive op­position by cases is reconstructed in the active hypothesis. At the same time two verb classes with different personal affixes are reconstructed for IE and defined as active vs. inactive verbs. I shall discuss the nominal ergativity/ activity and the verbal activity separately.

1.2.1 Nominal ergativity/activity

Both the ergative and the active hypotheses reconstruct for IE two basic cases of animate nouns, one expressing agent: ergative or active case, and other expressing non-agent (Patient 4- Non-Agent subj): absolutive or inactive case.

Therefore, the nominal active hypothesis is only a variant of the older ergative hypothesis, both assuming that the marked IE -s nominative of animate nouns was originally restricted to the agentive/active verbs only, i.e. in the ergative variant to transitive active verbs of the type "kill", whereas the active variant has also intransitive active verbs of the type "go". The non-agentive/inactive verbs expressing processes and states such as "lie" (intransitive, one-actant verb) or "know", "see" (two-actant verbs) are supposed to have occurred with non-marked 'absolutive' case.

Page 34: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOMINATIVE-ACCUSATIVE STRUCTURE 17

(3) the assumed active and ergative case marking of IE animate nouns active variant

-s -O Agent tr + kill Patient tr + kill Agent intr + go Non-Agent subj + lie, know

ergative variant -s -0

Agent tr + kill Patient tr + kill Non-Agent subj + lie, know

+ go

We have many reasons for rejecting this assumption. Some of them are based on general comparative evidence concerning the distribution of case marking and animate/personal vs. inanimate/non-personal features (1.2.1.1). Other structural traits of IE also do not support the ergative/active hypothesis (1.2.1.2).

1.2.1.1 Ergative/active case and the animacy hierarchy

In active languages, basic expression means of syntactic relation are personal affixes on verbs. Active languages having restricted case marking show an opposite distribution of cases with respect to the features animate/ personal vs. inanimate/non-personal compared with that of the alleged active structure of IE. A more animate/personal class is marked for Non-Agent (Patient, Non-Agent subj):

(4) active languages with case suffixes (after Mallinson & Blake 1981:52, McLendon 1978)

Guarani -pe for Patient tr of animate nouns Eastern Porno -al, -yiy for Patient tr Non-Agent subj of pronouns,

kinship terms and proper names

Similarly, the following rule which is stated with respect to the ergative structures (and holding also for Eastern Porno where less personal nouns are marked for agent) contradicts the situation reconstructed for IE (see Moravcsik 1978, Plank 1979:4-5; for Australian languages Silverstein 1976): If the noun classes distinguished by the features animate/personal vs. inanimate/ non-personal behave differently with respect to the ergative (agent) marking,

Page 35: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

18 FROM INDO-EUROPEAN TO LATIN

then it is always the less animate/personal class which is marked for ergative (agent). This regularity holds also for the classical ergative languages such as Georgian, where the personal pronouns are excluded from ergative marking.

The very fact that the alleged ergative or active Proto-Indo-European should be excepted from the regularity valid for attested ergative/active structures makes the ergative/active hypothesis improbable. As we shall see later, this assumption does not fit in with other structural features of IE.

1.2.1.2 Other structural evidence contra ergative/active hypothesis

The inadequacy of the nominal ergative/active hypothesis and its incompatibility with the structural data of IE can be summarized as follows: 1) The alleged situation in which the -s nominative would be restricted to agentive verbs of the type "kill" (and "go" in the active variant) whereas the verbs of the type "lie", "know" (and "go" in the ergative variant) would have non-marked absolutive 'subjects', is not attested in IE. The distribution of marked -s nominative and absolutive follows the class distinction animate vs. inanimate. The verbs "know" and "go" have regularly animate subjects which in all attested IE languages are expressed by marked -s nominative or its reflex. As follows from 1.2.1.1 this subject case restricted to animate nouns cannot be classified as either 'ergative' or 'active'. 2) The transition from the alleged situation with marked -s subject restricted to verbs of the type "kill" ("go") only to the historically attested situation is not motivated in IE, where no ergativity/activity split according to aspecto-temporal distinctions is reconstructed. In Megrelian which has been adduced as a parallel for the assumed IE development (see K. H. Schmidt 1979) the starting situation is that of split ergativity and the whole process was motivated by the tendency toward the uniform expression of syntactic rela­tions. The extension of ergative into the function of nominative is here a secondary 'nominativization', which is preconditioned by the existence of the nominative in the present and other related tenses. 3) The alleged situation in which different verbs or verb classes would require a different case marking does not correspond to the free, non-compact struc­ture of the IE sentence, which is well attested in ancient IE languages. This sentence structure (1.3.6) is characterized by strong autonomy and mutual independence of nominal and verbal constituents, the centralizing function of the verb is weak. A whole series of verbs of both active and inactive meaning can refer to the same subject expressed only once for the whole series: the

Page 36: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOMINATIVE-ACCUSATIVE STRUCTURE 19

active and inactive verbs are regularly coordinated. 4) Not only substantival subjects of verbal predicates Idominus venit/, but also subjects of nominal predicates Idominus bonus (est)/ and adjectives in predicative and attributive function Idominus bonus/ have marked -s nomina­tive form. This differs from the Adygean ergative, with which Ivanov (1981) wants to equate the Indo-European -s form. The Adygean ergative is restricted to agent function with transitive verb and is expressed only once in the noun adjective phrase (group inflection).

(5) comparison of Latin and Adygean noun phrase Latin amicus bonus

Subst-NoM Adj-NoM Adygean

Subst-0 Adj-ERG

The Adygean group inflecting ergative, serving as agent expression in verbal sentences, has an analytic origin.15 IE word inflection is to be explained on the basis of the derivational form and semantics of cases. It is in this type of semantics that the marked form of the IE nominative finds its motivation (2.4.4).

1.2.2 Verbal activity

Typical active languages have bipersonal verb with active and inactive personal affixes expressing the difference of Agent ≠ Non-Agent and the identity of Non-Agent subj = Patient tr.

(6) Verbal prefixes in Dakota (after Klimov 1977:34) Agent intr wa-t'i

lSG.ACT-dwell "I dwell" Non-Agent subj ma-śica

lsG.INACT-be bad "I am bad" Patient tr Agent tr ma-ya-k'te

lsG.INACT-2sG.ACT-kill "You kill me"

The active prefixes express Agent (tr intr), the inactive prefixes Non-Agent subj and Patient tr.

A characteristic feature of these languages is a large category of

Page 37: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

20 FROM INDO-EUROPEAN TO LATIN

inactive verbs expressing predicates of the type "be bad", "be great" etc. which regularly occur with the non-personal actant (subject). These active verbs correspond to adjectival predicates (with or without copula) of IE languages.

In both characteristics mentioned above the IE structure differs from that of active languages. The IE verb is unipersonal, hence no combination of active and inactive affixes similar to above mentioned Patient tr + Agent tr construction occurs and the personal affix never refers to the second actant of a transitive active verb of the type "kill" (Patient tr, object). IE lacks the category of adjectival verbs. IE adjectives are of nominal character (1.3.5), they take nominal case and number morphemes. The capacity of also functioning as nominal predicates is a syntactic feature which nouns and adjectives have in common.

Also the atypical active language with unipersonal verb such as Guarani exhibits both characteristics of active languages mentioned above: the large category of adjectival verbs and the ability to identify Non-Agent subj with Patient tr by means of verbal affixes. The form with inactive prefix is a means of expressing the fact of me-beating:

(7) xe-pete lsG.INACT-beat "somebody (you, he, she) beats me/ I am

beaten by somebody" (after Bossong 1980:376)

The formation of the sentence in Guarani is governed by the hierarchy of persons, according to which the first person is always privileged and expressed explicitly by a personal prefix. On the other hand in IE the active transitive verb of the type "kill", "beat", was oriented unequivocally to agent which (before the formation of the secondary passive diathesis) could only be referred to by the personal suffix.

We reconstruct for IE two verb classes which can be classified as active and inactive, because they differ according to features concerning agentivity (3.2.1.3). However, the inactive suffix does not refer to Patient tr of the two-actant verb. The opinion of Gamkrelidze & Ivanov (1984:289), who argue that the inactive suffix of two-actant verbs refers to Patient tr (object) in a construction like "the man lays the stone", is unmotivated. These inactive transitive verbs are non-oriented as for actor - undergoer opposition (3.2.1.4). In any case we have no support for the assumption that the Patient tr of active verbs of the type "kill", "beat" etc. could be expressed by an inactive suffix. This is impossible because of the unipersonal character of IE

Page 38: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOMINATIVE-ACCUSATIVE STRUCTURE 21

verb which expresses only Agent in the case of active verbs. The identification of Non-Agent subj = Patient tr could not be expressed in constructions with primary distribution of case marking and personal affixes. Only in the later development is the marked passive voice used to signal the secondary iden­tification of Non-Agent subj with Patient tr in passive constructions.

Therefore, the existence of verb classes differing in agentivity in IE does not imply the existence of the active structure as a type of fundamental syntactic relations. The definition of this structure includes Non-Agent subj = Patient tr identification. The principle of grammatical relevance of the noun classes animate vs. inanimate and verb classes active vs. inactive has a broader validity, not restricted to the active type of syntactic relations. In IE this principle is a component of the derivative-flectional type where the lexical and derivational categorization has a dominant position in morphosyntactic structure.

Page 39: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)
Page 40: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

1.3 Main Characteristics of the Derivative-Flectional and Paradigmatic-Flectional Types

1.3.0 In this section I will try to give a synthetic characterization of the IE flectional type and its development from the reconstructed d-f structure of IE to the attested p-f structure of Latin and other IE languages. I shall proceed from morphology to syntax trying to define the invariant characteristics of both stages/subtypes of IE flectional type and the main differences between them. At the end of this exposition the marginal analytic and pseudo-incorpo-rative features of IE will be examined. As for the morphology, this is a survey of the main issues for the readers interested more in general linguistics and typology than in IE research proper. Consequently, rather frequent references are made to the part of the book where a particular problem is treated in a more detailed way.

1.3.1 Morphological processes and characteristics

We have defined the flectional expression of the grammatical categories as internal modifications of the word (1.1.3). The whole formal/derivative process holding between two word forms accounts for their functional/ semantic relationship. We shall now concretize these statements and bring them into connection with the dominating principle of the original d-f structure of IE, i.e. the principle of primary grammatical relevance of lexico-deriva-tional categories. According to this principle the lexical stem should determine the morphological processes to be applied and the morphosyntactic character­istics of the word form. The main noun and verb classes differ in the application of case morphemes and personal endings (1.1.5.1). Further, more specific lexico-derivational subcategorization is grammatically relevant, deter­mining the application of inflectional formants (1.1.5.2). This determination

Page 41: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

24 FROM INDO-EUROPEAN TO LATIN

going from stem to endings is an invariable of both d-f and p-f structures. However, in p-f structure it is the inflectional stem which replaced the lexico--derivational stem in this respect.

1.3.1.1 Vocalic alternation and suffixation: Internal and external inflection

In both stages of the IE flectional type the morphological categories are expressed by the combination of internal and external inflection, i.e. by stem modification and affixation. The main morphological processes in IE are vocalic alternation and suffixation. Though reduplication is also a quite widespread morphological process, its application is more restricted and it is semantically marked. Its function as a regular perfect marker is an OI and Greek innovation. Prefixation developed as a marginal and secondary phenomenon, its only inflectional application being the so-called augment, which goes back to the original particle (3.1.2). The nasal infixation involved in the formation of the present stem is also a secondary development. However, the distinction between internal and external inflection is only relatively valid, both vocalic alternation and suffixation being components of a derivative process consisting of the internal modification of the entire word. Vocalic alternation apply not only to lexical roots/stems, but also to suffixes. Suffixation has non-additive character and is also used as a characterization of derivational and inflectional stems and not only of inflectional endings proper. The lexical/derivational and inflectional stems has determining value for the application of the inflectional ending.

The inflectional stems already containing suffixes are a relatively internal part of inflection to which the ending proper applies as an external part of inflection. The inflectional stem characterizes the inflectional declension class in nouns and the aspecto-temporal stems and inflectional conjugation class within these stems in verbs. So the -ê- suffix is a part of external inflection in Latin future dic-es and the same suffix is a part of the inflectional stem in Latin present vide-s or imperfect dice-ba-s (3.7.2). As we can see, the suffixes can be applied cyclically to the stems already containing a suffix of the same formal type, as long vowel suffixes in our examples.

In d-f structure vocalic alternations are represented by the following set of grades: full grade with -e- vocalism {-ter-) and -o- vocalism ('-o- grade': -tor-), zero grade with zero or reduced vocalism (-tr-, -tr0-), long grade (-ter-, -tor-). The distribution of full grade and zero grade is supposed to be related to the alternation of accent, which itself, however, has a distinguishing

Page 42: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

DERIVATIVE AND PARADIGMATIC FLECTION 25

function in IE16. We are interested in the morphological/functional coopera­tion of accent and ablaut, not in presupposed morphonological motivation of ablaut by accent, which can be partial only. This approach to ablaut-alterna­tions allows us to take into account the morphological naturalness of form -function relationships (1.3.2.2).

The fact that not only the lexical stem, but also suffixes, are subjected to alternation give rise to the differentiation of case suffixes into two separate forms with distinct morphological functions (2.4.1). The morphologically irrelevant alternation of lexical noun stems has already been abandoned in the prehistorical development of IE. Instead, the stem in verb forms receive a morphological function (*uoid-: *ueid-: *uid-, etc. — 3.2.3.6.1). In Latin, vocalic alternation also takes place (in altered form and distribution) as part of the perfect stem vs. present stem distinction (vīd-i17 vs. vid-eo). In Germanic, under specific conditions, this principle is applied more systematically and constitutes a very important part of verb morphology.

In the development into p-f structure the system of vocalic alternations reconstructed for IE d-f structure is as a whole disturbed. The principle of internal stem modification is, however, preserved in another form. In Latin, the diversity of perfect vs. present stem marking makes it possible to classify this marking in its entirety, including the suffixal formations with an -s- or -re­formant, as internal modification of the verb stem (1.3.3). This refers to inflectional stems constituting the most internalized part of external inflection.

Internal and external inflection appear as a continuum in IE structure. It is possible to cyclically apply suffixes to bases which arise by decomposi­tion in paradigmatic relationships and their application is connected with various additional adaptations (insertion of the hiatus filling consonants, binding vowels, etc.) and analogical contaminations (2.3.2.3) E.g. the -ă-suffix with modal or preterital interpretation is applied to the root in the archaic subjunctive form adven-a-t, to the present stem in classic adveni-a-t (in paradigmatic relationship to the present indicative adveni-t) and, as part of the suffix -bă-, to the derived base advenie-bat (3.5.1).

The definition of flectional morphological expression as an internal modification of the word holds for both stages of IE flectional structure. The difference between the d-f and p-f stages lies in the varying extent and relevance of the lexico-derivational and inflectional part of the word. In the more highly explicit p-f structure the grammatical morpheme became larger and more complex, and the cyclical paradigmatization processes enriched the inflectional morphology. The stem itself became in part an inflectional stem,

Page 43: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

26 FROM INDO-EUROPEAN TO LATIN

participating in inflection not by its lexico-derivational semantics, but by its formal morphological characteristics.

The lexical stem in d-f structure is itself the result of cyclical suffixation, which, however, took place at the level of lexical derivation; the reconstructed minimal lexical morphemes are in part roots and in part stems with no clear demarcation between them. The system of vocalic alternations is also part of the formal complexity of the lexical stem. The greater variety of accent-ablaut alternations which are examined in new IE research18 is in keeping with the closer internal connection between lexico-derivational and grammatical parts of the word in d-f structure. These, in fact, are formal manifestations of and correlate with the basic principle of d-f structure, i.e. the principle of grammatical relevance of lexical, word inherent semantics. In the next subsection we shall deal with other possible formal manifestations of this underlying principle.

1.3.1.2 Implicit semantic distinctions: Inherent characteristics of roots/stems

As explained in the preceding section, the principle of grammatical relevance of the lexical and derivational categories as the basic principle of d-f structure accounts for the character of IE morphemic processes and structures. Word inherent semantics manifests itself in the derivational and inflectional characteristics of the respective word category. It can, however, also remain implicit and be manifested in the larger morphosyntactic structure with the help of another word form, as in the case of masculine vs. feminine gender which manifests itself only in adjectives. Also implicit in lexical roots and implied in the derivational category of 'Aktionsart' is the aspectual distinction 'perfective vs. imperfective' which becomes grammaticalized on the basis of aspect in Greek, whereas in Latin it becomes a part of tense distinction of present vs. perfect stem (3.3).

The basic noun and verb classes with the grammatical relevance (animate vs. inanimate; active vs. inactive) display morphological and morphosyntactic characteristics. Their differing morphology is semantically motivated in the original d-f structure. In the p-f structure the morphological differences became mere allomorphic differences in the case of noun classes (absolutive and plural of neuters — 2.2). In verbs, the different endings of active and inactive verbs are either allomorphized (thematic present — 3.2.3.4, perfect of Latin type — 3.3) or are involved in marking inflectional oppositions (perfect of Greek type — 3.3, middle-passive — 3.4).

For the d-f structure we may consider the possibility that the class

Page 44: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

DERIVATIVE AND PARADIGMATIC FLECTION 27

semantics which manifests itself morphologically in case-number morphemes and personal endings appears also in the shape of root and stem. This phenomenon, which is compatible with the principle of the grammatical relevance of lexical root/stem in the d-f structure, will have to be examined with respect to verb roots (3.2.2.1), where the morphonological difference between active and inactive roots seems to be reconstructible. In the case of noun morphology, the characteristics of neuters can be found in the derivational capacities of stems, exhibiting also such characteristic processes as heteroclisis19 and having a proper position with respect to the system of stem alternations (2.2.1).

1.3.2 Characterization of IE flectional morphology: Preliminaries

1.3.2.1 Cumulative/fusional and complex character of IE morphology: Allomorphy and form variation

Inherent class semantics accounts for a part of the IE cumulative expression of categorial distinctions. The basic case morphs express the distinction animate vs. inanimate, so that, together with the cumulative expression of number (2.1.1), the respective morphemes contain three accu­mulated categorial distinctions: nom.sg. animate; acc.sg. animate; absol.sg. inanimate, etc. Similarly, the personal verb endings are cumulative expres­sions of person (+number) and active vs. inactive class, and thus we have morphemes with the cumulative structure lsg. active, lsg.inactive, etc. The cumulative character of flectional morphology is not only a formal but a semantically motivated phenomenon. With the development of the paradigma-tized flectional oppositions, flectional forms are subjected only to relative decomposition. A strong tendency to word integration (1.1.3), which is semantically motivated by the determining role of the semantics of category inherent type, governs flectional structure and controls the tendency to explicitness.

Characteristic for the IE word is its internal hierarchical complexity with respect to form - meaning correlations. In order to comprehend the various, though mutually related, aspects of IE flectional morphology I will use the following terms. 'Cumulative' in the strict sense are expressions of those categories which were never marked separately, as in the case of class distinctions animate, active, etc. contained in case morphemes and personal endings. The term 'fusion, fusional' is used to cover those complex form -

Page 45: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

28 FROM INDO-EUROPEAN TO LATIN

meaning correlations which are expressed by markers united in only relatively decomposable units. The number expression with its complicated derivational history (2.3) or the similarly complex derivational history of verb suffixal formations both show that the difference between cumulation and fusion is only a gradient one. We are mainly concerned with cumulative/fusional expressions.

The term 'complex' is used, at a more general level, when speaking about the association of complex semantics (sets of categorial distinctions) with the form of the flectional word. In both the d-f and p-f stage the relationship between word semantics and word form was complex. The differ­ence consists in the increased explicitness of p-f structure with more differ­entiated complexity of flectional endings. The integral non-differentiated complexity of word form is characteristic of d-f morphology with implicit and stem internal expression of semantic distinctions. The complex form is not presented by a particular segment, but is identifiable in the relationship to the derivational base or to other forms of the same paradigm (1.1.3).

In the narrower sense I use the term 'complex' to cover the complex expression of the same category, which is particularly characteristic of verb morphology (the Latin present or perfect being expressed by a combination of stem and ending, etc.).

Another characteristic feature of flectional morphology is allomorphy. Although the marker of a given category may be relatively distinct, as with the subjunctive marker -ă- in Latin dicas vs. dicis, it does not serve as a marker of the same category in all verbs and nouns, but only in certain inflectional (p-f) or derivational (d-f) classes of nouns and verbs. Allomorphy in a strict sense, i.e. based on complete semantic identity of the allomorphs, increases with the development of p-f structure (2.1.2). The form variation typical of d-f structure is, however, more or less restricted in developed paradigmatic struc­tures. What is meant by the term form variation as used by Lewy (see 1961:205) is, in part, semantically motivated variation in the original structure. The derivational formants with distinct semantics are by the process of paradigmatization identified as allomorphs on the basis of their grammaticalized meaning (1.1.5.2; 3.1.3). This holds especially for verb mor­phology. In noun morphology paradigmatization consists of the reduction of the number of derivational classes (various vocalic and consonantal stems) which already have inflectional, non-semantic character in the d-f structure. On the other hand, the fusion of the derivational and case suffixes (cf. Latin datives in -ae, -ö, -f, etc.) give rise to systematical allomorphy.

Page 46: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

DERIVATIVE AND PARADIGMATIC FLECTION 29

1.3.2.2 IE flection and natural morphology

As has been explained in previous sections the morphological processes do not interest us in their casual morphonological motivation and interdependence, but in their functional/semantic relevance. This holds also for the relationship of accent and ablaut. It is their functional cooperation, not the partial dependence of zero vs. full grade alternation on accent, which is morphologically relevant. Also, other word modifications such as the applica­tion of the -o- grade and long grade alternants or of the thematic -e/o- suffix (3.2.3.3) are considered as primary morphological processes with semantic relevance.

The predominant interest of IE research in morphonological processes (accentual word types, insertions of vowels and consonant to prevent clusters and hiatus) which could motivate certain morphological processes diverted attention away from the systematic motivation of morphological phenomena by the principle of naturalness or iconicity determining the non-arbitrary character of significative processes and structures.20

The idea that there is a scale of naturalness in the organization of morphological semantics and that there exist certain natural correlations between form and meaning is developed in our exposition in various aspects, and has proved its explanatory power. We consider the correlation between formal and semantic markedness, which perhaps has universal validity, dif­fering in concrete realizations cross-linguistically. The semantically marked forms are regularly more extended and tend to be expressed by more marked phonological realizations (see on case endings 2.4.6.3; 2.5.2.2). The formally non-marked vocalic suffixes have vague semantics (3.2.3; 3.5.3). The change from a semantically non-marked to marked character is followed by a corresponding change in form (see the discussion of the inactive and middle-passive 3.4.4 as well as the perfect 3.3.6).

Naturalness, however, also has typologically relevant aspects and manifestations. Some phenomena are natural within a certain linguistic type, being preferred as prototypically corresponding to underlying typological principles. In our characterization of two stages of IE flectional type the correlations between formal/distributional and semantic phenomena are systematically examined. The principle of typological naturalness helps us interpret some historical changes by the tendency toward processes which are typical of and natural for the given structure. The fact that some morpho­logical forms were originally more analyzable or decomposable, whereas the

Page 47: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

30 FROM INDO-EUROPEAN TO LATIN

formal fusion took place in some later development, is interpreted by the followers of isolation → agglutination → flection hypothesis (1.1.2) as evidence of original analytic structure. Examples are the Latin datives feminae, domino, pedi where the original identity of dative suffix -ei is no more detectable. However, the fusion of the inflectional class marker with the case marker is in keeping with the principle of the flectional expression of grammatical categories by the integral formal difference of opposite forms, by integral modification of the word, and not by any additively ordered sequences of morphemes. This principle having its realization in inherent characteristics of roots/stems, in cumulative morphology, and in internal inflection (alternation) is original for IE structure; the fusion is only another type of realization of this natural and dominating tendency. Similarly, when in the development of IE languages new stem alternations appear, particularly in Germanic languages (but also in Latin — 1.3.1.1), this cannot be interpreted in such a way that one sees the stem alternation as a later phenomenon. It is again a manifestation of how the language repeatedly tends toward certain processes, which are natural and characteristic for the given structure.

1.3.3 The morphemic structure of the IE word

1.3.3.1 The IE word as a hierarchically structured unit: Lexico-derivative and grammatical components: Paradigmatization

We have already defined the non-additive and internal character of the morphological expression valid for both stages (subtypes) of the IE flectional type. It is not the identity of the formal element added, but the identifiability of morphological opposition which is relevant in both stages; cf. the discussion about nominative vs. genitive opposition in 2.1.3.1. This, however, does not mean that the IE word is an internally unstructured whole. On the contrary, it exhibits hierarchical internal relationships ranging between lexical, derivation­al, and inflectional components. The root determines the choice of derivations, and the derivational stem determines the choice of flectional formants. If all three components are segmentally expressed, they follow the sequence lexical - derivational - inflectional, but with no strict formal demarcation between the three components and the individual formants. The preposed formants (reduplication, augment) are restricted and serve mostly as subsidiary expression means in the cooperation with endings.

We have already suggested the main differences between d-f and p-f

Page 48: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

DERIVATIVE AND PARADIGMATIC FLECTION 31

morphology, which lie 1) in the close affinity between derivational and grammatical formants of the d-f structure, manifested in both formal/ distributional and semantic features; 2) in the predominantly semantic motivation of the grammatical component by the lexico-derivational component in the d-f structure, whereas in the p-f structure the flectional ending is only formally determined by the inflectional stem; 3) in the structure of the flexive ending proper.

All these differences are reducible to the notion of paradigmaticity, central to the p-f structure. Their existence originates in the process of paradigmatization which takes place between the d-f and p-f stages. In the d-f stage, the derivations of the same base do not constitute any regular system of oppositions whatsoever. The dérivâtes formed by the same process (by the application of the same suffix) are mutually correlated on the basis of certain common semantic characteristics. These correlations are, however, of unsystematic and irregular character. In the p-f structure, the derivational pairs are integrated into the paradigm. We speak about a paradigm of developed flection in two different ways and at two different levels: the obligatory system of morphological categories and distinctions expressed in a given word category (morphological paradigm); and the system of inflectional forms displaying the same morphological processes (inflectional paradigm).

The fact that the application of formants is semantically motivated by the lexico-derivative stem in d-f structure and only formally motivated by the inflectional stem in p-f structure, is obvious proof of the original character of the d-f structure. However, the nature of the distributional and structural dif­ferences, which will now be examined with respect to the entire morphemic structure of the word, is also such that the reverse historical relationship of the two subtypes seems to be impossible.

1.3.3.2 The structure of the flexive ending in the paradigmatic-flectional language: Evidence from Latin

The flexive ending serves to express several morphological categories: number, case, gender in nouns; person/number, tense/aspect, mood and voice in verbs. In spite of the non-additive and integral character of flectional morphology, the segmentation into relatively separable units must be examined as an important part of morphological characterization. We will therefore use the term morpheme to designate a form-functional unit which can be defined by the combination of formal and semantic criteria. The separability of morphemes is relative and gradual in IE, and there are categorial distinctions

Page 49: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

32 FROM INDO-EUROPEAN TO LATIN

to which no separate morpheme can be ascribed. Therefore, we will not try to segment the flexive ending into its potential minimal morphemes, each of them corresponding to one semantic categorial distinction, but we define as morphemes only those combinations of categorial distinctions which have a relatively separable segmental representantion.

Due to the systematic inseparability of case and number markers (it is impossible to say which part of the suffix -is in pedis or -i in domini, -ovum in dominorum expresses the genitive and which part the number), it is appropriate to treat the categories of number and case as constituting one complex morpheme: nominative-singular, genitive-plural, etc. On the other hand, the two components are discernible in both formal and semantic fea­tures, whereby an iconic, natural isomorphy exists between the formal and semantic complexity of case-number morphemes, the plural morphemes being more formally marked. There is also a tendency to develop a representative marker of plurality (2.3.2.4). Semantically, the meaning of number distinction is the same in all cases. In the p-f structure this also holds for the absolutive of neuters, which in the d-f structure still has a specific meaning, determined by gender distinction.

Equally inseparable and cumulative are the expressions of person and number in verbal endings. Moreover, they are also semantically inseparable in the combination 'person + number' because of the semantic non-equivalence of the plural of different persons.21 In the following diagram we will describe the verbal persons not as linear combinations of person and plurality, but as simultaneously expressed categories of '1st person singular' (= lsg.), etc. As for the other verbal categories, it seems reasonable to describe the verbal ending as a sequence of two complex morphemes: tense-mood and person-diathesis, i.e. present-indicative, lsg.-active, etc. This corresponds to the semantic coherence and formal cumulative/fusional integration of the categories tense - mood and person - diathesis. Semantically, tense and mood distinctions relate to events, person and diathesis to actant - event correlations.

We can show the structures of Latin finite verbal endings in the following list, taking the 3sg. as an example:

(8) Latin verbal endings present indicative 3sg. active dic-i-t present indicative 3sg. passive dic-i-tur present subjunctive 3sg. active dic-a-t voc-e-t present subjunctive 3sg. passive dic-a-tur voc-e-tur future indicative 3sg. active dic-e-t voc-a-bit

Page 50: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

DERIVATIVE AND PARADIGMATIC FLECTION 33

future indicative 3sg. passive dic-e-tur voca-bi-tur imperfect indicative 3sg. active dice-ba-t imperfect indicative 3sg. passive dice-ba-tur imperfect subjunctive 3sg. active dice-re-t imperfect subjunctive 3sg. passive dice-re-tur perfect indicative 3sg. active dix-i-t perfect subjunctive 3sg. active dix-er-i-t pluperfect indicative 3sg. active dix-er-a-t pluperfect subjunctive 3sg. active dix-is-se-t

Dividing the Latin verbal ending into two basic complex morphemes has only relative value. It helps us to notice the correlations between forms containing a marked mood (-ă-, -ě- subjunctives) and those containing a marked tense (-ê-, -bi- futures, -bă- imperfect) or a marked tense + mood {-re- imperfect subjunctive). The flectional expression of the morphological distinction is, however, complex and integral. The basic tense/aspect distinc­tion is, in fact, expressed by the inflectional stem of the present and the perfect, and at the same time by the type of ending (present and perfect endings which except for the 3sg. and lpl. are distinguished for all persons), whereby the type of ending is determined by the inflectional stem in agreement with the structural hierarchy between the components of the IE word. Also, the pluperfect has a proper inflectional stem, marked by a sepa­rate -er/is- formant. Further, there is also an additional stem modifying element in the imperfect: -e- in dic-e-bam, dic-e-rem. These are manifestations of a complex expression of categorial distinctions, with hypercharacterization and redundance as typical features as well as cooperation of internal and external inflections.

The separability of morphemes does not hold for the entire inflectional paradigm and all inflectional classes to the same extent. We can correlate the marker -i- with indicative-present in the 2 and 3 sg. of the 3rd conjugation (dicis, dicit) vs. -ă- as a marker of the subjunctive-present, but the same is impossible for the lsg. dico, where the marker of the indicative-present has been fused with the person marker. The marker of the inflectional conjugation class is involved in the expression of the indicative vs. subjunctive opposition in part of the morphological forms, in others it functions as a separate marker. In the form dic-i-t vs. dic-a-t we take the vowel -i- as a modal marker, it can, however, at the same time be seen as a marker of the inflectional class; or it may be assumed that this class has a 0 marker. The difference between both conjugations with clear-cut inflectional markers, the 2nd and 1st conjugations,

Page 51: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

34 FROM INDO-EUROPEAN TO LATIN

is symptomatic: cf. dele-t vs. dele-at, with an identical stem for indicative and subjunctive, as opposed to voca-s vs. voc-es with the fusion of the inflectional and modal marker in both forms. There is, therefore, no sharp demarcation between the inflectional stem and the present-indicative morpheme. Further­more, there is also a continuum between the lexical root/stem (da-t, fa-tuf) and the inflectional stem with separate marker (voc-a-t, voc-a-tur).

From these observations, however, it does not follow that the flexive ending is an unordered structure. On the contrary, the structural position of markers is fixed, and the markers are hierarchically ordered. In the above list of Latin verbal endings the morpheme or formant furthest to the left, nearer to the lexical stem, always determines the following one and the lexical verbal stem determines the structure and form of the flexive ending as a whole.

We may also observe some natural correlations between sound reali­zation and semantic distinction. The modal (indicative vs. subjunctive) markers are vocalic, whereas the marked tenses are expressed by consonantal markers. The future reflects its intermediary position between mood and tense, having the 'modal' vocalic markers in the 3rd and 4th conjugations (dic-e-s: dic-i-s, dic-a-s) and consonantal 'tense' markers in the 1st and 2nd conjugations (voca-b-o: voca-b-as, voca-v-i).

These observations reveal that Latin morphemic structure agrees with the general characteristics of flectional morphology, where categorial distinctions are in principle identified in oppositions and not expressed by separate segments. The flectional ending of the finite verb represents an integral expression of grammatical semantics (in the sense of '3sg. active of the present indicative'), with the preference of certain segmental structures and form - meaning correlations.

On the other hand, paradigmatization is progressive in Latin and mani­fests itself at both levels of the paradigm. In Latin we have a set of categorial distinctions, which for the finite verb is regular (morphological paradigm). The inflectional class (inflectional paradigm) is represented in Latin in a highly elaborate form. The Latin present classes are representative examples of a paradigmatized structure at the inflectional level. The original form variety is reduced to 4 well defined classes, which are characterized by inflectional markers in the basic present-indicative form and by the type of subjunctive and future. The indicative vs. subjunctive opposition is formalized in Latin, with no homonymy of formants within the same inflectional class. The preterite of the present system, i.e. the imperfect, has a distinct form, common to all classes. These features can be duly appreciated when compared

Page 52: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

DERIVATIVE AND PARADIGMATIC FLECTION 35

to languages which preserve more of the original form variety of the d-f structure (Greek — see below) or which do not have such a rationalized distribution of inflections (see 3.5.6 on Tocharian).

1.3.3.3 The structure of the flexive ending in the derivative-flectional stage: Evidence from Greek

The flexive ending in the d-f stage is strongly determined in its form and structure by the lexico-derivational stem. There are two main differences in comparison with the p-f structure: restricted explicitness and large form variety.

The structure of the d-f ending is both less explicit and less complex due to the fact that the paradigmatic oppositions of the forms belonging to the same lexeme were restricted. The personal ending of the verb is determined by the lexical class, which is originally non-oppositional and whose features active and inactive were inherent in the verbal lexeme. This differs from the marked diathesis of the p-f structure, i.e. the middle-passive, which is explicitly expressed by means of the middle-passive endings as opposed to the active endings of the same verb. The nominal ending is also less complex, due to the fact that 1) the lexeme is vague with respect to number, and 2) the mar­ked number (non-singular) originally had restricted case distinctions. Also in the fundamental object relation either the accusative or the plural was originally expressed, but not both marked categories simultaneously (2.3.2.3).

The form and structure of the d-f ending varies in the different lexico-derivative noun and verb classes. If we compare the Latin with the Greek verb in this respect, the difference is very clear-cut, the Greek verb reflecting the original 'form variety' typical for the d-f structure.

It would be impossible to set up a similarly simple inventory of ending structures for Greek, which (with some restrictions) would be valid for all verbs or for the verb as such, as we did for Latin. In Latin, the flectional ending expresses a simple inventory of categorial distinctions which appear in regular combinations and restrictions (no subjunctive with future, no passive with perfect and pluperfect). The endings of the various verbs differ with respect to allomorphy, expressing, however, the same categorial distinctions and having the same linear and hierarchical structure — with some inflection-ally well established variants, as described above. This is not the case with the Greek verb. The individual aspecto-temporal stems do not combine with the identical categorial distinctions of diathetic nature. The endings of Greek verbs differ as to the amount of the expressed categorial distinctions and also as to

Page 53: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

36 FROM INDO-EUROPEAN TO LATIN

their positional and hierarchical ordering. There is a categorial feature 'intransitive/passive', combinable with the aorist (aorist in -ŋv and - θŋv) — and, less systematically, also with the future (passive future in -θŋσoµαι) — and standing here in opposition to the medium which in the present is part of the medio-passive/middle-passive category . There are also other types of the Greek aorist, which can be partially characterized semantically as a combina­tion of the aspectual feature with certain semantically vague features of a diathetic nature: sigmatic with reduplicated aorist (attested especially in Homeric Greek) on the one hand, and thematic -e/o- aorist with athematic root aorist on the other hand. These aorist pairs differ in their diathetic semantics, which is manifested by their distribution with respect to verb classes, the thematic and root aorists occurring in inactive (see 3.2), mostly intransitive verbs. The sigmatic aorist as a productive aorist formation is the unmarked member of this aoristic subsystem. Though no regular oppositional forms of the same verb are formed, several verbs have such oppositions. The sigmatic aorist with transitive/causative meaning is opposed to the root aorist in e.g.

"I set up" vs. "I stood" — see Kølln 1969:15f. with more examples. There are other differences in the system of morphological forms of the various verbs (i.e. differences in the morphological paradigm), especially in the perfect and middle categories. The distribution of the active and middle-passive forms is irregular. There are groups of verbs with the active present and the middle future as or of verbs with the middle present and the active perfect as This differs from the situation in Latin, where the deponents constitute a semi-semantic/semi-inflectional class of verbs which have the passive/inactive -r series of endings in all forms of the present system, i.e. regularly in all synthetic finite forms.22

These are only representative examples of variety which the Greek verbs exhibit in their systems (paradigms) of morphological oppositions. The purely inflectional variety of the Greek verb, much larger than in Latin, can again be only suggested through simple reference to the variety of personal en­dings and to the variety of the inflectional stem characteristics. As for the personal endings, the inherited series, i.e. athematic vs. thematic and primary vs. secondary in the present-aorist system, the perfect and middle series, are enlarged by Greek innovations, i.e. series of sigmatic aorist, primary vs. secondary middle-passive series.

With the diversity of morphological and inflectional paradigms greater variety in the positional hierarchy of markers and greater complexity of

Page 54: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

DERIVATIVE AND PARADIGMATIC FLECTION 37

expression arises. The same categorial distinction is marked by a combination of different formants, cf. the complex and integral expression of the aorist: augment, stem characteristics, type of endings. And these markers have another positional hierarchy in the various verb groups, cf. stem alternation or reduplication in the thematic and reduplicated aorists, suffix as part of the flexive ending in the sigmatic and intransitive/passive aorists, etc. Again, Greek innovations such as the systematic use of reduplication (in perfect) increase the complexity of expression.

The Greek verbal system typologically corresponds more to the older d-f stage of flection than does the Latin one. This, however, does not mean the Greek system as a whole can be projected onto IE. I will try to show in Part 3 that the Latin and Greek systems are, in principle, alternative results of the paradigmatization of the original d-f structure. However, the paradigmati-zation in Greek was realized in such a way that the original form variety and other structural features of the IE d-f verb have been preserved and, in certain cases, increased by the help of innovations. The system of morphological oppositions is maximized in the Greek verb and this represents an innovation. For the realization of this tendency, however, the original lexical/non-op-positional and derivational formations are used and, in another form, preserved. We must not only distinguish the archaisms and innovations, but we have to take into account the typological relevance of archaisms and innovations. As we have suggested, certain Greek innovations have the effect of increasing certain structural features typical of the d-f type. Therefore, we shall distinguish between type-changing and type-preserving innovations.

1.3.4 Character of the morphological semantics: Category-immanent vs. relational semantics

The close connection between lexical and grammatical components in the word complex as well as determination of the grammatical component by the lexico-derivational component have been shown to be characteristic of the flectional type. On the basis of the facts attested in ancient IE languages we are trying to reconstruct the original d-f structure, where the distributional and structural features of flectional morphology find their semantic motivation. It is necessary to pose the fundamental question about which type of morphological semantics is to be presupposed for these word complexes with predominance of the lexico-derivational component.

We shall discuss the original character of the morphological semantics

Page 55: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

38 FROM INDO-EUROPEAN TO LATIN

in the IE flectional structure by examining the semantic complexity of the word in d-f and p-f stages (1.3.4.1) and by showing the primacy of category-immanent vs. relational semantics as characteristic of the original structure (1.3.4.2).

1.3.4.1 The IE word as a semantic complex

In conformity with the complex/integral and non-additive expression of grammatical categories via the internal modification of the word, IE word structure is to be semantically defined not as an addition of the grammatical information to the lexical component, i.e. as 'verb lexeme + active + lsg. ' , etc., but as a semantic complex, i.e. 'active verb in the lsg. ' , or similarly 'plural animate noun in nominative', etc. This is the primary definition of the word structure as a whole. In the paradigmatic relationships which tend to constitute the system of oppositions in the p-f stage, individual semantic dis­tinctions are identified and associated with certain formal distinctions. The one-to-one attribution of a certain formal feature to a certain semantic feature is, however, not achieved in the p-f structure either.

In the original d-f structure the grammatically relevant semantics is implied in or determined by the lexico-derivational stem. Some grammatically relevant features (animate vs. inanimate, active vs. inactive, perfective vs. imperfective) are originally implied in the lexical stem as distinct class charac­teristics; others, such as number distinctions, are only vaguely present (2.3.1) and are only specified facultatively by distinct formal markers. The grammatical categories not directly inherent in the lexical stem (such as person category) are governed by the features implied in the lexical stem (active vs. inactive).

In the d-f stage the unsystematic paradigmatic relationships are realized by derivational pairs (founding and founded) which do not constitute any fixed paradigm. The relationship between the lexical base and the derivation, which is semantically motivated, is essential. This relationship is also valid for grammatically relevant derivations. The determination of the grammatical component by the lexico-derivational component has its structural basis in this relationship. On the other hand, in the p-f structure the forms derived from the same lexeme are organized into paradigms of inflectional oppositions, i.e. they receive a clearly inflectional character. The semantic relations within these paradigms become essential. At the same time, the semantically motivated determination of the grammatical component by the lexico-derivational stem develops into a determination of formal, allomorphic character between the in-

Page 56: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

DERIVATIVE AND PARADIGMATIC FLECTION 39

flectional ending and the inflectional stem. It is the semantically motivated relationship between grammatical and lexico-derivational word components of the d-f structure which is important for the question of the original type of morphological semantics in IE.

1.3.4.2 Category-immanent vs. relational semantics

The grammatical semantics bound to the lexico-derivational stem is not primarily oriented towards the classification of verbo-nominal relations, but to classification within word categories. It is the word-category-inherent (immanent) classification of nouns and verbs into more specific semantic clas­ses which constitutes the basis of morphological semantics, thus appearing as semantics of a category inherent type.

On the basis of category inherent semantics, nouns and verbs are in principle independently classified into animate vs. inanimate and active vs. inactive. By means of this classification the semantic-syntactic value of respective noun and verb classes in the sentence is determined. Hence, the semantic-syntactic relationships are not directly expressed by morphological devices, but through the intermediary of category inherent classifications. This orientation of morphological semantics is also maintained in another form in the p-f structure. In paradigmatically organized flection, morphological distinc­tions assert themselves within the paradigmatic opposition. The semantic-syntactic value is once again predetermined by intracategorial oppositions, not directly orientated to the intercategorial syntagmatic relationships.

On the other hand, semantic-syntactic aspect is, from the very beginning, included in this type of morphological semantics. The fundamental (animate vs. inanimate, active vs. inactive) and the more specific classifica­tions (individual - non-individual, partial - total, etc. in nouns; perfective - im-perfective, progressive - non-progressive, etc. in verbs) are not effected before and outside of, but within and with respect to, the sentence forming acts. The individuals and events are classified as parts of situations denoted by the sentences (2.4.1.2). In the flectional IE structure, however, these denotational correlates of the sentences are primarily structured with respect to the intra­categorial relationships and category inherent features, not with respect to intercategorial and relational ones.

As we shall explain in Part 2 and 3, the semantics of the nominal categories, not only that of gender, but also that of number and case, is originally semantics of the category-immanent type. And in the verb, not only aspecto-temporal and modal distinctions referring to events, but also the

Page 57: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

40 FROM INDO-EUROPEAN TO LATIN

diathesis as a category relating to actant - event correlations is conceived as a category inherent distinction, and also the category of person is determined by categorial semantics. This is possible due to the complex character of the finite verb, including the subject person as an internal part of its structure.

Again, Greek preserves the original category-immanent type of morphological semantics better than Latin. The tendency of Latin morphologi­cal semantics towards a relational interpretation, however, exists not because of progressive paradigmaticity itself, but because of the tendency towards sentence formalization (1.3.6). This difference between Greek and Latin in the orientation of morphological semantics (see esp. Kurzová 1987a) towards semantic subcategorization within a word category (category-immanent semantics) on the one hand and towards syntactic and relational distinctions (relational semantics) on the other is comparatively large. With respect to the main nominal and verbal categories, the different orientation of Greek and Latin morphology is especially apparent in the semantics of the accusative and genitive, in the primacy of aspect in Greek vs. primacy of tempus in Latin (3.3.5), in the semantic vs. syntactic definition of diathesis (3.4.3.1).

As we shall not be examining this question in our study of case semantics in Part 2, we must remember here the main differences between the Greek and Latin accusatives and genitives. In Latin, the tendency to subordinate case to a certain syntactic relation is conspicuous. The Latin accusative can be defined (within the simple sentence) as the case of direct object, the Latin genitive as the predominantly adnominal case. On the other hand, in Greek the ungoverned accusative of respect (cf. e.g.

X.Mem, 1,6.6., etc. "I suffer with respect to my legs") has a comparatively wide use. The accusative in object function enters into semantic opposition with the genitive on the basis of features 'total vs. partial5 (with verbs like , etc.). Therefore, the paradigmatic semantics of cases also asserts itself in the function of primary object. The adverbal object genitive is common in Greek in comparison with the predominantly adnominal genitive in Latin. Again, some Greek innovations such as the wide use of the accusative of respect (so-called accusativus Graecus) work towards reinforcing the original typological feature, i.e. the semantic/category-inherent orientation of d-f morphology vs. the syntactic/relational orientation of Latin morphology, as connected with the sentence formalization.

Page 58: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

DERIVATIVE AND PARADIGMATIC FLECTION 41

1.3.5 Word categories: Noun and verb distinction. Position of the adjective

It is a familiar fact that the noun and verb distinction is maximal in the IE flectional structure23 and manifests itself in lexical and morphological derivations and morphosyntactic specifications. I shall not discuss this question separately because it follows from the entire characterization of the IE nominal and verbal system given in this book. What I shall briefly discuss here is the semantic-syntactic relevance of the noun - verb distinction and the position of adjectives.

Nouns are primarily referential expressions denoting individuals/ objects, verbs are primarily predicational expressions denoting 'states of affairs'/events.24 The semantic-syntactic position of the verb is partially determined by its inherent relationality, according to which the verb opens slots for its complements. Yet, there is another important feature which con­cerns the relationship between noun and verb in the sentence and which results from their primary referentiality vs. predicativity. As referential expressions, nouns open slots for or are oriented towards predicates. Being subject to predication is for nouns the main and necessary process of being introduced into the sentence, due to their restricted possibility of functioning as predicates (in a naming and identifying/including function). As non-referential predicates, verbs open slots for or are oriented towards referential elements. These two kinds of slot-opening by verbs (on the basis of their relationality/valency and on the basis of their predicativity and non-referentiality) accounts for the difference between semantic roles (agent, experiencer, patient, etc.) and the referential role of subject, as we shall see in our discussion of subject con­stitution in IE (2.4.3). Here, the definition of nouns as referential expressions vs. verbs as non-referential/predicational expressions serves as a basis for determining the position of adjectives in IE.

IE adjectives are, in fact, referential expressions constituting a subclass of nouns in a larger sense. They refer to the individual less directly, by denoting non-essential properties of the individual, whereas substantives refer to the individual directly by naming its essential property. Yet, the difference between constitutive, (essential) and non-essential property25 is only relative, and the adjective can function as noun head without derivational modifications in IE. The referential and nominal character of the IE adjective is important for our understanding of the original character of agreement in IE (2.2.2).

Page 59: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

42 FROM INDO-EUROPEAN TO LATIN

1.3.6 Structure of the Indo-European sentence: Word autonomy in the non-formalized sentence structure

Hierarchically structured word complexes have their counterpart in the linear, non-formalized sentence structure. This character of the sentence is fully motivated in the reconstructed d-f stage, where lexical-derivational semantics is dominant, and morphological semantics is of the category-imma­nent type. The paradigmatization of morphological oppositions which is reali­zed in the p-f structure has no direct effect on the essential relationship of word and sentence structure. The word in the developed p-f structure remains a complex unit, to which the grammatically relevant semantics is bound as an internal component. The autonomy of the word (i.e. the minimal constituent of the sentence) has, as its necessary counterpart, a free, non-formalized (or according to Meillet 1934 appositional) structure of the sentence. This also holds for the p-f stage. The formalization of sentence structure which we observe in the development of European languages is not a mere consequence of the changes in morphological systems (paradigmatization). We must assume that sentence formalization, which is accompanied by a change in the type of morphological semantics (1.3.4.2) and in the morphosyntactic expression of grammatical categories in general (increased role of syntactic devices), was motivated from 'above', by the development of thought and communi­cation.26

The theory about word autonomy in the non-formalized sentence, outlined by Meillet (1934) and further elaborated by Havránek (see especially 1968, 1971, 1973; further Kurzová 1973, 1981), is an essential contribution to characterizing the IE morphosyntactic type. We have concentrated on formal and semantic word structure in this study, but paying systematic respect to the semantic-syntactic functions. The IE non-formalized sentence structure will only be briefly characterized by examining the character of main syntactic devices such as agreement, government, and word order.

1.3.6.1 Word autonomy and agreement

IE agreement between noun and adjective and between noun subject and verbal predicate is a consequence of word autonomy and word inflection. In attributive and predicative syntagms each full semantic word, i.e. minimal constituent, is autonomously marked for the grammatical categories which are

Page 60: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

DERIVATIVE AND PARADIGMATIC FLECTION 43

common to the whole syntagma. Adjectives agree with nouns in the categories of case, number, and gender, the verb agrees with the noun subject in the category of subject, which is expressed by the nominative in the noun, and by the subject-person marker in the verb.

The syntactic function of agreement as a device serving to identify co-constituents of the same syntagma (allowing their distance position and free word order) is a secondary consequence of the primarily semantic function of agreement. Primarily, IE agreement had a semantic character, each word in­dependently expressing the same semantic distinction, or as Hjelmslev (1971) puts it, the same content category. Therefore, IE agreement is a manifestation of word autonomy and represents an integral component of the d-f structure with word inherent expression of grammatical semantics.27

1.3.6.2 Government: The place of the verb in the IE sentence

The non-formalized character of verb government is a phenomenon which is very symptomatic for 'appositional'28 sentence structure, where according to Meillet (1934:359): "Chaque mot a la forme que demande le sens à exprimer, non une forme commandée par un autre mot de la phrase; il n y avait pas en indo-européen de rection d'un mot par un autre [...]; l'autonomie du mot est le principe qui commande la structure de la phrase indo-européen­ne". With respect to non-formalized government, Meillet (1934:358) argues that "le cas auquel sont mis les compléments ne dépend pas du verbe, mais seulement du sens à exprimer".

The non-formalized character of government is easily observable in Greek, as explained and documented in more detail in Kurzová (1985), (1987a) and (1988). There are alternative expressions of object, although the lexical meaning of the verb is the same. The accusative and genitive in the object function enforce their meaning of total and partial affectedness re­spectively (see above), and similarly, though not in such a systematic way, other forms too can be used as alternative object expressions (accusative -dative, prepositions) with the same verb as semantic unit (i.e. with the same verbal 'sememe'). Further, the position of the object is left unoccupied under very free conditions, so that the verb implies the object by its semantic valency only. This is a manifestation of the selfsufficient and complex character of the finite verb (verb autonomy). It is, however, the autonomy of the noun which must be especially stressed in comparison with the modern European sentence, the latter having the verb in a central and centralizing position. The non-formalized character of government,, shown by the fact that

Page 61: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

44 FROM INDO-EUROPEAN TO LATIN

the occurrence of a particular verb does not necessarily cause the occurrence of a nominal in a particular form, is the clearest evidence of the weak predicability of the sentence form on the basis of the verb. Another manifestation of the weak formalizing and centralizing function of the verb is the large occurrence of nouns and nominals in attributive-predicative and appositional functions, which represent constituents not subjected to verb government and to the dependency-frame of the verb. In some cases, they are equivalents of verb dependent expressions (complements, adverbials), used in modern languages in similar contexts, "I went to sleep in all content" Lys. 1.1.13.

The verb is not an obligatory sign of sentencehood in IE. Nominals (nouns and adjectives) have a predicative function without requiring the copula as an overt sign of sentencehood. This lack of overt segmental sign and the lack of formalized word order are related to the non-formalized character of the sentence as a linguistic unit at the level of segmental representation. The identity of the sentence at a more abstract level, where it is defined as a 'closed field of the syntactic relations' (Trost 1962), is, of course, not affected by this lack of overt form.

1.3.6.3 Word order

Word order is not grammaticalized in IE as a device for signalizing syntactic relations. In all basic syntagms, both relative positions are allowed and the co-constituents can stand in distant position. The only word order rules or regularities which can be observed in the ancient IE languages are directed towards signalization of the sentence frame, i.e. towards the marking of the beginning and the end of each sentence. In Latin, this marking is comparatively elaborated: the verb is usually in the final position, the initial position is occupied by conjunctions, pronouns and other connecting elements. Free non-grammaticalized word order is supposed to follow the pragmatic, communicative organization of the sentence. This, however, is not to be considered as a function of word order as a linguistic sign. We can hardly say that word order is a linguistic sign of pragmatic distinctions. More appropriately, we can say that pragmatic distinctions manifest themselves in word order, and in an unsystematic way at that. The word order in this respect has the role of symptom and not of sign (see Kurzová 1988:158). Attempts to distinguish marked and non-marked word order types29 are based on transfer of descriptional procedures from grammar proper to the level of pragmatic organization. Typologically, free word order and word autonomy are solidary

Page 62: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

DERIVATIVE AND PARADIGMATIC FLECTION 45

phenomena. The words/constituents are defined by autonomous characteristics inherent in word categories and not by their relative position or by syntactic devices marking their relations.

1.3.7 Analytic and incorporation-like features of Indo-European

1.3.7.1 Analytic morphemes

As already suggested (1.1.1), the IE analytic elements have only marginal character and functions not disturbing the basic flectional principles. The prepositions specify the semantic relations holding between co-constitu­ents, whose basic semantic-syntactic relationship is, however, sufficiently indicated by case. Also, in other instances the analytic elements (pronouns, articles, modal and sentential particles) do not change the categorial character of the segment with which they are related. The only exceptions are the subordinating conjunctions, whose application is connected with the change of the sentence into a sentence constituent, hence with the change of its categorial character. The subordinate conjunctional clauses belong, however, to a later development of IE. They develop from appositional relative clauses introduced by relative adverbs. In ancient IE languages clauses with an intermediary character between relative appositional clauses and conjunctional clauses are attested. They are not directly embedded in the matrix sentence, but by means of a pronominal or nominal constituent to which they are appositionally added; cf. the Greek type of conjoined clauses with or , etc. and a demonstrative adverb or , etc. in the partner sentence.30 The proper, embedded conjunctional clause is, with its changing effect on categorial character, an element disturbing the flexive principle. This relates, however, to the syntax of the complex sentence and not to the basic syntax.

Individual IE languages differ in their use of analytic elements and this difference is not given by the degree of paradigmatization or formalization of sentence structure. On the contrary, a comparatively intensive use of analytic elements in Greek tends towards greater autonomy of word/constituent and individual clause. Articles increase the autonomy and expansional possibilities of the nominal phrase, which can be enlarged to the detriment of the verbal phrase, whose expansion and hierarchization is characteristic for the development of formalized sentence structure (see above, 1.3.6); also cf. the difference between Greek and Latin in the constructions vs. pugna apud lacum Regillum facta (binding function of the article in Greek

Page 63: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

46 FROM INDO-EUROPEAN TO LATIN

vs. centralizing function of the verb in Latin; see Kurzová 1985:12). The attempts to explain flexive morphemes as originally analytic words

fail to recognize the integral/complex and cumulative/fusional character of flexive morphology. Basic IE flexive formants clearly show their derivational origin. Their word-inflectional distribution is opposed to the distribution of analytic elements, which are used only once for the whole group (see the discussion on Ivanov's analytic explanation of -s nominative in 1.2.1.2, note 15). The fact that the analytic explanation of flexive formants requires proof of their original group inflection is duly considered by Haudry (1980). He shows the original group inflectional distribution for 01 instrumental -ă, hence for the marginal case suffix (2.4.6.3). Similarly, verbal formants derivable from particles originally have marginal specifying functions and their use is facultative only (3.1. 4). The basic aspecto-temporal distinctions are signalled by flexive formants of derivational character in the IE structure. The augment e- or the postinflectional particle -i serves to further specification. Subsequent integration of the particle -i into the structure of the personal ending can be interpreted in the sense of typological naturalness (1.3.2.2).

1.3.7.2 IE composites as pseudo-incorporative elements

The IE composites attested in ancient IE languages are only formally reminiscent of the noun phrases of incorporative languages. Functionally, they belong to the domain of the lexicon and not that of grammar.

There is also no support for assuming (with Jacobi 1897, W. P. Lehmann 1969, 1974:76, Gamkrelidze & Ivanov 1984:324-325) that compo­sites originate in the original syntagms and that they attest original rigid word order. In general, the conception of composites as original syntagms is not justifiable (see Bühlcr 1934:320ff.). In IE, fixed linear sequence of the components in IE composites agrees with their functional character as lexical units. Fixed structure is typical for the IE word, but not for the IE syntagma. All languages possess naming units (lexical units) containing several lexemes. And according to the free, autonomous position of constituents, these naming units are expressed by composition and not by the syntagma in IE.

Page 64: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

PART 2

THE BASIC PRINCIPLES AND ORIGINS OF THE

NOMINAL SYSTEM AND INFLECTIONS

Page 65: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)
Page 66: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

2.1 Introduction: Principles of Nominal Inflection: Derivative-Flectional Morphology of Nouns

2.1.1 The flectional principle of expressing number and case

The mutual relation of the number and case categories is an especially good example of the difference between flectional and agglutinative structure. The expression of number is united in one non-analyzable morpheme with the expression of case. This is a prototypical example of the cumulative character of flectional morphemes. The complex nominal morpheme includes the cat­egory of number and case without it being possible to determine the formal element expressing number and case respectively. Pater, amicus are complex forms of the nom. sg.; patres, amici are complex forms of the nom. pl., and similarly patrem, amicum; patres, amicos with respect to the acc. sg. and pl. Also in marginal cases the ablatives patre, amico and patribus, amicis are complex non-analyzable forms of the abl. sg. and pl., with the non-analyzable morphemes -e, -ö, -ibus, -Ts. There is no formal element identifiable which would represent a single categorial distinction of number (sg., pl.) and case (nom., acc, abl.) in all forms where it is present. This is a manifestation of the non-additive, non-separative character of the flectional expression of grammatical categories. The formal fusion of the case morph proper with the stem vowel is complete in the case of the second declension, representing the IE -o- stems (-0 < -o-ei, -is < -ois). Also, the complexes 'acc. sg.', 'ace. pl.' are expressed differently in the different declension classes. This is a typical flectional feature which is called allomorphy.

These principles show their peculiarity when compared with regular agglutinative morphology. We shall illustrate the difference between the two structures by comparing the Latin expression of the nom. sg. pl. and acc. sg.

Page 67: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

50 FROM INDO-EUROPEAN TO LATIN

pl. with the Hungarian:

(1) sg. pl. nom. atya pater atyá-k patr-ês

harát amicus barát-ok amic-î ace. atyá-t patr-em atyá-k-at patr-es

barát-ot amic-um barát-ok-at amic-os

The agglutinative nominal inflection is illustrated by the nouns atya "father", barát "friend". The marked number (pl.) and marked 'case' (acc.) have autonomous, separative formal exponents common to all nouns: -k for pl., -t for acc. In the form marked for both categorial distinctions (acc. pl.) these two suffixes combine in the fixed number-case order maintaining their distinctness. These are structural features contrasting with the IE flectional principle. The integral non-additive morphology of IE and Latin stands in contrast to the additive and separative morphology of agglutinative languages. The Hungarian nominal inflection, as indicated above, has neither cumulation and fusion (1.1.2) nor allomorphy.

We have dealt with the character of IE flectional morphology in the first part, and we shall illustrate the complex and cumulative/fusional expres­sion by concrete examples in the later exposition of noun morphology. We have shown that this principle is an invariant for both derivative-flectional and paradigmatic-flectional subtypes, though its concrete manifestation differs. In the following sections allomorphy will be examined via the process of comparing the reconstructed IE and Latin morphology and discussing the question of whether and to what extent this feature is also an invariant manifestation of both subtypes.

Our comparison of Hung. and Lat. noun inflection should be concluded with the following remark. We identified, as usual, the Hung. -t suffix with the IE acc. suffix. However, the Hung. suffix -t is functionally unambiguously defined as a syntactic device expressing the direct object relation, whereas in the case of the IE accusative this definition does not exhaust the functional application of the acc. marker (2.4.1.1). This ambiguity vs. unambiguity of form - function relations is another feature which can be included under the contrastive properties of both language types and under the common label of complex, non-additive (non-separative) structuring and its opposite.

Page 68: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOMINAL SYSTEM 51

2.1.2 Types of allomorphy in Latin noun inflection: Allomorphy in Latin and Indo-European

First of all we must distinguish between allomorphy of the whole case number morpheme (= C+N allomorphy) and the allomorphy of separate case or number exponents (= C V N allomorphy). Strictly speaking, because of the non-analyzable character of the case-number morph, only C+N allomorphy can be considered as allomorphy in the proper sense. However, here also the separative vs. non-separative distinction is not absolute (cf. the -s element as representative plural exponent 2.3.2.4). CVN allomorphy is one of the intermediaries between complexity-cumulativity and allomorphy, and at the same time one of the preconditions of allomorphy proper. Therefore this phenomenon belongs in the present connection, irrespective of whether it is reasonable to use the same term 'allomorphy' for both instances.

The allomorphic differences we find in Latin have various foundations: 1) They are conditioned by morphological classes of animate and inanimate nouns, i.e. lexical classes with grammatical relevance. The differences between animate and inanimate nouns in the expression of fundamental cases are allomorphic in Latin, representing different expressions of the categorial distinctions acc. sg. pl., whereas in the original structure these differences are semantic, marking 'abs. vs. nom. acc.' and 'collective pl. vs. individual pl. ' . In Latin the original specific meanings are suppressed or backgrounded and the identical morphological value prevailed. The non-differentiated nom.acc. of neuters is considered as synonymous with nom. and acc. of M/F, the pl. of neuters as synonymous with the pl. of M/F. The allomorphy holds, however, only with respect to case and number. The nom. and acc. morphemes function as exponents of the gender category. 2) Other instances of C+N allomorphy are conditioned by inflectional class. They differ with respect to motivation and age. A) Some of them arose by phonological and morphonological changes. Allomorphic differences of this kind result especially from the fusion of the case suffix proper with the declension marker (stem vowel) into one case ending. Cf. the acc. and dat. markers of three Latin declensions: domin-um, domin-o; femin-am, femin-ae; urb-em, urb-i. The inflectional declension marker and case marker proper are no longer separable and identifiable. The final form of this fusion develops in Latin. It is, however, predetermined by the structures, where the application of two forms of the consonantal suffixes,

Page 69: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

52 FROM INDO-EUROPEAN TO LATIN

with or without a preceding vowel, is morphonologically motivated. Some fusions of stem vowel and case suffix vowel had already occurred in IE (2.1.3.1). B) Other allomorphs are morphologically motivated. The respective formants originally represent different morphological elements and/or are formed by morphological analogy or morphological modification.

Some allomorphic differences of this type are due to the fact that different derivational formants asserted themselves in different inflectional classes. This kind of allomorphy resulting from the fact that formants with specific meaning are identified on the basis of the dominating grammaticalized meaning, is more notable in the verbal system, but also occurs in the noun inflection. The marginal cases in particular are formed by different formants whose distribution differs in individual languages and in the various inflec­tional classes. In Latin abl. sg. and dat. abl. pl. are allomorphic formations of this type. Some allomorphic differences in the formation of the nom. sg. probably belong to the morphologically motivated allomorphy reflecting the original difference between nom. formed by -s suffixation and nom. formed by lengthening (2.4.2.4).

The formants of nom. pl., gen. sg. and gen. pl. of the second and first declension also belong to the morphologically motivated type of allomorphy. They are first transferred from the pronominal declension to one of these declensions (-oi > -f of the nom. pl. and -f of the gen. sg. to first decl., -arum of the gen. pl. to second decl.) and then analogical forms develop in the other declension. The analogy based on the mutual influence of the inflectional classes/declensions operates in many Latin case forms. Some of these analogical formations have the restriction of possible allomorphs as their effect, such as the endings -es and -ibus of the consonantal 3rd declension (type pater) which are transferred from -i- stems (type civis) by analogy: nom. pl. -es < -ei-es, dat.- abl. -i-bus.

The relative antiquity of the morphologically motivated allomorphic differences varies. The difference between the suffixed and the lengthened nom. sg. must be projected back to IE (2.4.2.3).

The tendency to develop allomorphs of the gen. sg. of -o- stems is observable in all ancient IE languages with exception of Hittite. The modified forms of the type -os-o, -os-io are common to the main groups of IE languages. The non-sigmatic nom. pl. of -o- stems represent a dialectical difference projectable also into late IE. On the other hand the gen. pl. forms transferred from pronominal stems are restricted to Latin and Greek, and the

Page 70: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOMINAL SYSTEM 53

gen. sg. in -ï is a specific innovation of Latin and Celtic. 3) The third type of allomorphy relates to the C V N allomorphy. Case and number (cumulative) exponents exhibit in their mutual relationship allomorphy conditioned by the morphological category valid for all number and case distinctions and all declensions. The categorial distinctions sg. vs. pl. represent a condition for the choice of respective allomorphs of the given case: e.g. sg. → ace. -um, pl. → acc. -ös in the second decl. The case distinctions represent conditions for the choice of the sg. or pl. exponents: e.g. acc. → sg. -um, → pl. -ös in the second decl. This type of allomorphy is an original part of the cumulative structure of IE nominal morphemes.

To sum up our observations concerning Latin and IE allomorphy, we must conclude that allomorphy increases during the development from the derivative d-f flectional to the p-f paradigmatized structure. The paradigmati-zation gives rise to allomorphy in more than one way. The morphological changes in the fusion of stem vowel and suffix vowel result in the systematic allomorphy of declension classes. The originally distinct derivational formants (see above under 2 B) are allomorphized by the process of grammaticalization. The original differences between lexical noun classes (see above under 1) are also allomorphized. The original forms are changed by morphological processes (analogy and other morphological modifications) which are motivated by paradigmatic structuring (distinctness of paradigmatic oppo­sitions, intraclassal and interclassai relations and hierarchy).

On the other hand the C v N allomorphy is present in the original cumulative structure and the C+N allomorphy is conditioned by the IE structure in more than one respect. The semantically based allomorphy has its origin in the semantically vague and unsystematic character of derivative-flectional morphology and some allomorphs, such as the suffixed and lengthened nominative, are already present in IE structure. The 'therapeutic' morphological modifications are of IE origin. The gen. sg. in -os-o/-os-io can be traced back to the later IE development. Already in IE, the pronominal subsystem is allomorphic as a whole and represents a source of the nominal allomorphic formations (see esp. 2.4.8). Morphologically motivated allomorphy also has an IE origin which we shall deal with in the following section, in connection with the morphological character of suffixation. What follows is a survey of the allomorphy types in Latin with representative examples.

Page 71: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

54 FROM INDO-EUROPEAN TO LATIN

(2) Types of allomorphy in Latin 1 C + N allomorphy conditioned by morphological class

nom. acc. of M/F nom./acc. of N 2 C + N allomorphy conditioned by inflectional class A morphonologically motivated allomorphy

nom. sg. amic-us, mon-s, ov-is acc. sg. amic-um, mont-em, ov-em

B morphologically motivated allomorphy derivationally based: abl.pl. -fs < -oi-s;-i-bus analogically based: gen.pl. -orum, -arum vs. -um

transferred from pronominal subsystem nom.pl. -es patres after cives, etc. based on interclassai relationships

3 C V N allomorphy conditioned by morphological category nom acc

sg. -us -um valid for all N and C distinctions pl. -f -ós and all declensions

2.1.3 The morphological character of basic case suffixes: In­do-European allomorphy motivated by the accent-ablaut alternations

The basic case suffixes -s, -m(n) and -i are subjected to the same morphological rules as the derivational suffixes. They occur in two variants conditioned by the accent-ablaut alternations between root or stem morpheme and suffix: -s vs. -es/-os; -m vs. -om; -i vs. -ei. As we shall explain in section 2.4 these two variants are functionally differentiated and used as expressions of the case distinctions: nom. vs. gen. (-s vs. -es/-os), acc. vs. gen. (-m vs. -om), loc. vs. dat. (-i vs. -ei).

The monomorphemic root nouns such as pes < *ped-s or monomor-phized stems like dens < *dent-s exhibit an alternation between root and case suffix. The so-called strong cases are characterized by accented full grade root and the reduced grade of the suffix, i.e. without a preceding vowel, whereas the so-called weak cases are characterized by unaccented reduced root and accented full grade suffix, i.e. with a preceding vowel. We shall illustrate this rule with the nom.- gen. and loc- dat. pairs:

Page 72: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOMINAL SYSTEM 55

nom. -s *péd-s31 > pes *h1dént-s > dens gen. -es/-os *ped-és > pedis *h1dnt-és > dentis loc. -i *péd-i > pede *h1dént-i > dente dat. -ei *ped-éi > pedī *h1dnt-éi > dentī

(3)

However, when the case suffixes are applied to nouns already containing a stem suffix, more possibilities arise. The types of alternation between root, stem suffix and case suffix have been studied in recent research following Kuipers stimulating observations (see 1.3.1.1, note 18). These types are considered to be a result of the different accentuation classes of IE words.

2.1.3.1 Hysterodynamic, proterodynamic and amphidynamic inflections of consonantal and -i-/-u- stems

For our purposes the well established types of OI and Gr. -i- and -u-stems and their comparison with the basic alternational structures of the stems with the consonantal suffix -ter/tor- present instructive evidence.

The stems with the suffix -ter/tor- exhibit a common type of alternation between derivational suffix and case suffix, which corresponds to the type attested by root nouns: the case suffix has two forms according to the grade of the preceding morph. In the proposed classification of alternating word this type is called hysterodynamic.32

(4) nom. Gr. gen. Gr. loc. * - — i OI dat. OI

In the -i- and -u- stem declensions two types of endings are attested, reflecting two different distributions of ablaut grades which, originally, are determined by the animate vs. inanimate categorization33. The inanimate *ouis type shows a non-alternating -i- (and similarly -u-) stem vowel and alternating case suffix, while the *men-ti-s type (lat. mens, 0I matih) has fusi-onal alternation of the whole ending (stem vowel + case suffix)34. It is the ment-is type which represents the dominating declension of -i- stems and -u-stems.

Page 73: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

56 FROM INDO-EUROPEAN TO LATIN

(5) ouis type nom. -is -us *ouis 0I avih gen. -i-os -u-os *ou-i-os 0I avyah Lat. senat-u-os

(6) mentis type nom. -is -us *ment-is 0I matih gen. -ei-s -ous *mnt-eis 0I mateh Lat. senat-ous

In Latin, the difference between both declension types is attested only for the -u- stems by archaic forms like senatu-os beside senat-ous which has its continuation in the classic normal form senatus.35

With respect to the proposed classifications, the *ouis type represents the proterodynamic type, with alternations between root36 and case suffix, whereas the *mentis type is classified as amphidynamic, with alternations between root and stem suffix. These types of morphologically based allomorphy can be projected into IE. In the paradigmatized structure of Latin the original diversity is reduced. The fusional genitive ending -eis is characteristic of the flectional structure, in which it is not the identity of the exponent but the distinctness of the opposition (here nom. vs. gen.) to the relevant morphological forms that is important.

The nominative vs. genitive opposition is, therefore, expressed not by a simple (agglutination-like) addition of the case suffix, but by the integral modification of the word, typical of flectional structure. The genitive ending itself has also two allomorphs -es and -os in the consonantal declension, as well as a fusional form -eis in the mnteis type, and most probably also an allomorph -s in the proterodynamic inflection with ablauting root and non-ablauting case suffix (type nokwt-s vs. nekwts — see note 32 above).

2.1.3.2 The -o- stems inflection

In the case of -o- stems the fusion of the thematic -e/o- suffix with the case suffix must be considered as IE. Incapable of zero grade - full grade alternations, the thematic vowel appears in -e- and -o- grades with unclear distribution and motivation. The loc. and dat. forms differ by the short and long quality of diphthong: loc. -oi vs. dat. -öi < -o-ei. For the gen. only Hit­tite attests the form in -as which can be interpreted as -ös < -o-es. However, the -ǎs interpretation is also possible and it is this form which is attested in other IE languages in the genitives -osio, -eso, -oso where the stem vowel in -o- or -e- grade is followed by the -s suffix. The formant -os/-es (which would

Page 74: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOMINAL SYSTEM 57

be homonymous with the nom. in its -o- realization) is enlarged by the element -io or -o, probably representing a connective anaphoric/relative particle (Gelenkpartikel).37 The use of the connective particle was possible, however, only in certain genitive functions, i.e. in the case of possessive or partitive adnominal genitives. This explains why the case system of -o-declension has a more notable tendency to integrate the marginal form marked by the dental suffix than the other declensions. This form expressed the ablative function which belonged to the function of the IE genitive. The form in -osio is now attested also in archaic Latin (see note 71).

Therefore, the morphological defectivity of the gen. form of -o- stems causes its morphological modification and the adaptation of other derivational formations to the case function. This development corresponds to the principles of the older derivative-flectional structure, where the inflectional systems of various noun classes can differ with respect to the number and character of semantic distinctions. These differences are eliminated in Latin by the process of paradigmatization (see 2.5.1).

Both the derivative-flectional morphology of IE and the paradigmatic-flectional morphology of Latin exhibit diversity in the inflectional systems of different noun clauses. But while in the original structure the differences are motivated morphonologically or semantically, they represent synchronically non-motivated allomorphic differences in Latin.

2.1.4 Noun and adjective: Word autonomy and the semantics of nominal categories

In IE structure adjectives also have complex case number morpheme (which in the nom. acc. forms serves also to express gender oppositions). They are, therefore, autonomously marked for the syntactic relation into which the whole nominal syntagma (noun phrase) enters. This principle will again show its peculiarity when compared with the contrastive structure of an ag­glutinative language with so-called group inflection. We shall continue our comparison of Latin with Hungarian.

(7) a jó atya pater bonus a jó atyák patres boni a jó atyát patrem bonum a jó atyákat patres bonos

Page 75: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

58 FROM INDO-EUROPEAN TO LATIN

In Hungarian the adjective (jó "good") and the article (a) are not inflected for number and case distinctions; only the noun expresses these categories. In comparison with IE word inflection38, this type of expression of the 'case' relations is called group inflection. The noun follows the adjective in the fixed word order typical of group inflection.39 In Latin, the adjective also inflects for number, case and gender, which are united in a complex morpheme, and selects the allomorph which is determined by its own inflectional class, independent of the class of the noun head. This formal diversity is a characteristic feature of IE agreement which will be compared with the formalized agreement of Bantu languages in the section on gender.

As was explained in the introduction, case, number and gender agree­ment in IE is to be interpreted as a manifestation of word autonomy (1.3.6.1). We recall this structural feature here because of its implications for the semantics of nominal categories, especially for the semantics of case (2.4.1.2) which also belongs to semantics of word-category-inherent type (1.3.4.2). The conception of agreement as manifestation of word autonomy is also relevant for the character of gender agreement (2.2.2).

2.1.5 The Latin and Indo-European nominal systems: Preliminaries

The nominal system of Latin is formed by the tripartite category of gender, the bipartite category of number and the case paradigm with six members, including the vocative. The relation to the richer system as recon­structed for IE (with dual member and 8 cases) will be discussed in connection with the paradigmatization process. These categories do not represent systems independent of each other. They appear in characteristic formal and semantic interrelations, typical of flectional structure. These relationships change during the development from d-f to p-f structure without the mutual interdependence and cumulative expression of the nominal categories being lost.

The system of inflectional categories and distinctions relates also to other parts of the nominal morphology: the system of word-formation and the system of declension classes. The relationship of these three components of noun morphology is different in the two types of flectional structure. They have relatively clear-cut semantic demarcation in Latin, representing inflec­tional categories with grammatical meaning, derivational categories with lexical meaning and flectional classes without meaning. The flexive endings which result from the fusion of the specific class marker, i.e. stem vowel, with the case suffixes represent formal markers of inflectional class. This fu-

Page 76: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOMINAL SYSTEM 59

sion is a common tendency of all IE languages, being more radically realized in Latin than in Greek or OI. In Latin, the system of morphological categorial distinctions is the same in all inflectional classes. This is one of the differences between the paradigmatized structure of Latin and the reconstructed d-f struc­ture of IE. In the later structure the derivational and inflectional morphology are closer to each other as we have already suggested in section 1.3. We shall briefly specify the concept of the d-f structure with respect to nominal morphology, trying to show the main differences between noun and verb.

2.1.6 Derivative flectional morphology of nouns

The relationship between p-f and d-f flectional subtypes is different in more than one respect in the nominal and verbal systems. The possibilities of reconstructing d-f structure from the situation attested in IE languages are also different. 1) The noun classes with grammatical relevance preserve the character of lexical, non-oppositional classes: animate/masculine,feminine vs. inanimate/ neuter. Gender as an oppositional inflectional category occurs in adjectives. On the other hand, the verb classes active vs. inactive are not attested in IE languages, they are only reconstructed on the basis of inflectional categories showing traces of the original non-oppositional classes with specific formants. This difference between noun and verb is motivated by the different relevance of noun and verb categorization for the lexical and semantic-syntactic levels (see below under 2 and section 3.1.1). The main difference between d-f and p-f subtype lies in the above mentioned allomorphization of the originally semantic differences between animates and inanimates in number and case. 2) The derivational character of the flexive formants is attested by the distribution and semantics of the verb formants in IE languages (see esp. 1.1.5.2). On the other hand only the general morphological character (2.1.3) and the type of semantics (2.4) suggest the affinity of derivational and case suffixes. The distribution of basic cases is fully paradigmatized in the attested structures. What can be observed is a certain graduality between derivational and inflectional morphology in the application of the marginal case suffixes, where the difference between an integrated case form and an adverb is gra­dient. However, there is a clear difference between a derivational category like the category of the agent noun expressed by the -tor/ter- suffix and the inflectional category of case. On the other hand there is no such distinct difference between the derivational and the inflectional verb category. The

Page 77: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

60 FROM INDO-EUROPEAN TO LATIN

derivational suffixes are the main source of the formation of inflectional verb categories in IE languages, as we show in Part 3. This difference is connected with the primary referential, non-predicational character of the noun as opposed to the predicational character of the verb. The distinctions expressed by the derivative noun formation assert themselves fully on the level of designation, whereas the distinctions expressed by the verb suffixes have systematical relevance for semantic-syntactic categories such as temporality/ aspectuality, modality and diathesis. 3) The differences in the morphemic structure of noun and verb are also connected with the different relationship between the derivational and inflec­tional categories of noun and verb. In the case of the noun the non-separative character of flectional morphology is most clearly realized in the cumulative case-number-gender morpheme of IE and Latin; the tendency toward fusion of the inflectional class marker with the case suffix is particularly strong in Latin. In verb morphology where cumulation and fusion are not so extreme, the verbal ending is analyzable into distinct morphs.

The inflectional class marker can serve at the same time as a marker of semantic distinctions of aspectual, modal or 'Aktionsart' character and pre­serves, therefore, its distinctness from the person marker proper. Yet, another characteristic trait of flectional morphology is more prominent in verb morphology. It is the complex character of morphological expression (1.3.2; 3.1.1).

In the following analysis of the nominal categories of IE and Latin we shall show the relevance of the concept of d-f structure and its paradig-matization for the explanation of the Latin nominal system. The reconstruction of the original structure helps us find the motivation of the synchronically unmotivated structural features, and, at the same time, understand the development and its main tendencies.

Page 78: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

2.2 Animate and Inanimate Noun Classes: The Category of Gender

2.2.0 The tripartite category of gender which we find in the majority of ancient IE languages (0I, Avest., Gr., Lat., Goth., OCS) is composed of two oppositions: 1) animate vs. inanimate/neuter; 2) masculine vs. feminine. These two oppositions differ with respect to their structural position and most probably also with respect to their relative age, though both of them are projectable into common IE before or after the separation of Hittite.

The fact that Hittite does not distinguish masculine and feminine is to be interpreted either as an archaism, Hittite reflecting the presupposed original stage, or as a secondary return to the presupposed original stage.40 In the latter case the opposition of masculine to feminine develops (to a certain de­gree) in Hittite and is then lost.

The solution of this Hittitological problem, however, is not decisive for the assumption that the opposition animate vs. inanimate is older than the opposition masculine vs. feminine.41 What is important is the different structural position of both oppositions. Only the opposition animate vs. inanimate has morphosyntactic relevance with respect to fundamental nominal categories and relations. The close connection of this classification with the morphosyntactic structure of the basic sentence is to be interpreted in the frame of original derivative-flectional structure where the semantic-syntactic relevance of word categories is the dominating principle.

2.2.1 Morphological differences between animates and inanimates and their semantic motivation

Animate (masculine and feminine) and inanimate (neuter) noun classes differ in the expression and semantics of the fundamental cases and of the

Page 79: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

62 FROM INDO-EUROPEAN TO LATIN

plural. The inanimate nouns, the neuters, do not distinguish nominative and

accusative, but have only undifferentiated case, which can be called 'absolutive'. However, in using this term we must bear in mind that this absolutive cannot be equated with the absolutive of ergative languages like Georgian or Basque. In these languages the absolutive is opposed to the ergative of the same noun, the ergative expressing an agent and the absolutive a patient (or 'non-agent' — 1.2.1). In the case of IE neuters, on the other hand, nominative and accusative are not distinguished. When the neuter is used with a transitive verb as in tempus mutat mores "time changes customs", it stands in the same indifferent absolutive form.

(8) MF nom. amicus bonus/utilis N abs. verbum bonum/utile

acc. amicum bonum/utilem

This difference in the system of cases is not only of a formal character, but has a semantic basis as well. As we shall explain in the section on cases (2.4) the nominative marks the category of animate nouns with the distinctive features + autonomous/+individual. Only this category can stand in the privi­leged syntactic position of the subject conceived as privileged participant to which the sentence is oriented.

The opposition animate vs. inanimate which is basic in IE nouns, however, does not correspond to the logical or scientifical notion of animateness. This ethno-culturally bound notional distinction is connected with the process of animation, by means of which logically inanimate entities are conceived as animal beings which are compatible with the features +autonomous/individual. The same entity can be conceived as animate or inanimate viewed from different aspects and with respect to its different manifestations as is documented especially by words designating "fire" and "water": cf. Lat. (OI etc.) masc. ignis vs. Gr. neuter , Lat. fem. aqua vs. Gr. neuter .42 There is a close connection between the process of animation and the constitution of the subject which is important for IE morphosyntax.

The type of the plural is also different in animate and inanimate nouns. The neuters have their own plural formant -a/-a which is supposed to have originally had a collective meaning. On the same basis the identity of this suffix with the feminine suffix -ā/-ă is presupposed (see Hardarson 1987, esp.p.87 concerning the -eh2 collectivum and feminine). The following

Page 80: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CATEGORY OF GENDER 63

evidence indicates the characterization of the suffix -a as a marker of collectivity. 1) the rule 'neutrum plurale habet verbum singulare' attested in Greek, OI, Avestan, Hittite, but not in Latin: type Gr. "the animals run".43

2) the plural in -a of some masculine nouns (especially locus "joke", locus "place", sibilus "hissing") attested beside the plural in -f.

The semantic difference between both forms is observable in the case of locus: loca expresses the non-individualized notion of space, region, whereas loci expresses single places, e.g. topics or passages in a book. Again, the difference in the expression of the M/F pl. vs. N pl. is semantically motivated and connected with the distinctive feature 'autonomous/individual' which is decisive for the difference in case marking of animate and inanimate nouns.

The morphosyntactic relevance of animate vs. inanimate distinction is a component of the structure with dominating position of category-inherent semantics (1.1.4). We can pose the question of whether the animate and inanimate nouns differed also by their formal morphological characteristics as it is suggested for verb classes active vs. inactive (1.3.1.2; 3.2.2). There seems to be no similar systematic difference in the shape of roots/stems as it is reconstructible for verbs. However, the specific morphonological rules of neuters are attested by heteroclisis (1.3.1.2, note 19) and by the inflection with stable root accent. The latter phenomenon is studied in recent research in progress whose results cannot be fully reported here.

2.2.2 Gender and agreement

Unlike the opposition animate vs. inanimate, the opposition masculine vs. feminine is not connected with the morphosyntactic difference in case and number categories. Morphosyntactically, this categorization serves only as an expression of agreement between substantive and adjective.

Whereas the distinction M/F vs. N is explicitly marked by the charac­teristic case and plural marking, the distinction M vs. F is only implicit in the noun stem; it is explicitly marked only in adjectives — including the adjectival pronouns. Also in the case of adjectives the M vs. F marking is morphologi­cally more restricted (-o- vs. -ă- stems, bonus vs. bona) than the M/F vs. M marking (bonus/bona vs. bonum; utilis vs. utile).

The classification masculine vs. feminine presupposes in IE the process

Page 81: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

64 FROM INDO-EUROPEAN TO LATIN

of animation which results in the distinction animate vs. inanimate.44 We can only guess at the values attributed to the basically sexual opposition which is extended to the whole sphere of animates. With respect to the restricted expli-citness and irregularity of gender marking, this extension must be considered as semantically motivated. Extension on a formal morphological basis can be considered for -ă- stems only.45

We can observe in IE languages some systematic rules in the assign­ment of a certain gender to a certain natural class of entities. Trees are feminine, whereas rivers are masculine in Latin (cf. Wackernagel 1928:30, 32). The appellative arbor is also feminine and the names of the trees are feminine even when they belong to the predominantly masculine -o- declen­sion: quercus, "oak", populus "poplar". In these cases the mythology with personified deities contributes to the perseverance of this systemic attribution. In general, outside the sphere of persons and animals gender is unmotivated in attested IE languages (see Bechert 1982:24; Meillet 1921:228f.).

Gender is a non-oppositional, lexical category of substantives46

becoming an oppositional morphological category by adjective marking which belongs to the domain of IE agreement. However, for gender as an agreement category as well, the semantic/referential, not the syntactic function is primary (1.3.6.1). Its function as a device expressing the syntactic connection between a noun and an adjective belonging to the same syntagma (noun phrase) is secondary, though morphosyntactically the only relevant function.

In the situation where gender distinctions are motivated semantically, as it is in d-f structure, nouns and adjectives can be understood as referring independently to the entity conceived as M, F or N. This corresponds also to the non-hierarchically structured noun phrase of IE with autonomy of adjec­tive.

Not only the semantic feature realized in the substantive is binding for the choice of the adjective form but also the feature realized in the adjective is binding for the choice of the substantive, so that the entity designated as quercus should be qualified as alt-a, but also the entity qualified as alt-a has to be referred to by a feminine noun. There is a correlation between the semantic-referential character of IE agreement and the formal expression of the agreement. This will be clarified by comparison with the different situation as found in the African languages with noun classes.

Page 82: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CATEGORY OF GENDER 65

2.2.2.1 IE agreement in gender and agreement in nouns class of Bantu languages

The IE genders have been compared with the noun classes of the Bantu languages (cf.especially Meinhof 1936). There are, however, distinct semantic-conceptual and structural differences between IE genders and the noun classes of Bantu languages. Conceptually these classes are based on more concrete classificatory criteria, which are not so strongly determined by the process of animation as the two gender distinctions in IE. The class of persons is one of several noun classes.

What is of special interest to us, is the character of agreement in gender or class between noun and adjective, which seems to be a common trait of IE and Bantu languages. I will give a short comparison of IE and Bantu to show the main differences in the character of agreement.47

In the Bantu languages we find noun classes with eight or ten members in the main, the markers of which are expressed primarily in the substantive itself. The prefix characterizing the substantive class js then repeated in the same form in the co-occurrent adjective which follows the substantive in the grammaticalized linear order. Cf. the following examples from Swahili:

(9) the agreement between noun and adjective in Swahili (Bantu) m- prefix of persons m-tu m-zuri

homo pulcher hi- prefix of 'instruments', etc. ki-su ki-kali

culter acutus

The prefix of persons, m- and the prefix of 'instruments' ki- are expressed in the noun and repeated in the adjective. On the contrary in Latin and IE there is no special gender marker which would be expressed in the noun and repeated in the adjective. The expression of gender is part of the flexive cumulative/fusional morphology.

These differences in formal structure suggest that both types of agreement differ functionally as well. It can be assumed that the repetition of the same marker in the adjective serves as a device expressing the syntactic relation of the adjective to the noun, so that the syntactic function of agreement would be primary in Swahili. Only in the noun, would class distinctions have semantic/referential function in Bantu structure. This cor­responds to the position of adjective in these languages. Adjectives form only

Page 83: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

66 FROM INDO-EUROPEAN TO LATIN

a restricted group of expressions — with 10-50 members according to Dixon (1977). No class affinity of adjective and noun as we encounter it in IE (1.3.5) holds for Bantu languages, in which most adjectives are derived from verbs.

Other structural differences between the IE and Swahili noun phrase confirm the different definition of agreement. In Swahili, agreement is connected with fixed word order, both phenomena representing syntactic devices serving to bind the syntagma and to give it an explicit and obligatory form. On the other hand the position of the IE adjective is not fixed. As is well documented in Latin, the adjective can precede or follow the nouns and it can be in non-contact position with the noun. Agreement in IE is to be considered as a manifestation of word autonomy and non-formalized syntax. The syntactic function of agreement which serves to identify the constituents showing the same categorial distinction as part of the same noun phrase (cf. Anderson 1985:167) is a secondary consequence of the primary semantic/ referential function of agreement.

2.2.2.2 Agreement and word autonomy: Semantic, morphological/ paradigmatic and syntactic functions of gender

As we have already explained in section 1.3 word autonomy and IE agreement are correlated phenomena. Hjelmslev's (1971) conception of agreement according to which the adjective expresses content categories which are ascribed to the entity the noun phrase refers to is valid for gender agreement as well. However, in the case of gender agreement these categories are in part (with respect to M vs. F distinction) merely implied in the lexical stem of the substantive; they are morphologically expressed only in the adjective.

Therefore, agreement originally has a semantic/referential character in IE. In the original derivative-flectional structure the categorization based on the features 'animate vs. inanimate, masculine vs. feminine' is motivated by actual conceptual structures from the realms of religion and mythology. In the further development of IE languages these distinctions are grammaticalized, i.e. morphologized in the sense that they lose their conceptual meaning and remain semantically relevant in a very restricted domain only.

In the paradigmatized structure of Latin, the gender distinction has primarily morphological-paradigmatic function, dividing nouns into morphological classes. Though semantically empty, the gender distinctions still represent modifications of nouns and adjectives which assert themselves primarily in the semantic/referential relation. The syntactic function of

Page 84: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CATEGORY OF GENDER 67

agreement as a device binding the adjective to the substantive is secondary. The substantive and adjective, which can stand in non-contact position, refer to the same entity by ascribing the same feature to it; their syntactic binding is a consequence of their identical referential intention. In this sense agreement is closely connected with word autonomy in the non-formalized sentence (1.3.6.1).

In the process of formalization of the sentence structure gender agreement either disappear, being replaced by other devices binding together the members of the formalized syntagma or it changes its character, becoming itself a syntactic device. The use of agreement as formalized syntactic device is typical of the Balkan languages (see Kurzová 1974). Here the double article and the agreement in gender cooperate in expressing a formalized syntagma of noun and adjective. Cf. the following Rumanian and Bulgarian examples:

(10) prietenul cel bun

In Rumanian the double article -/ + cel expressing defmiteness and gender serves as a binding device. In Bulgarian, the syntagma has only one article (-t), but the noun is marked for gender by the so called 'gender declension' where the declensional class serves as marker of gender.

2.2.3 Conclusions

The IE category of gender forms a tripartite system composed of two hierarchized oppositions:

animate inanimate masculine feminine neuter Both distinctions differ with respect to their character and antiquity. 'Animate vs. inanimate' represents a distinction which was connected with the different semantic-syntactic value of animates and inanimates in the original structure. The distinction 'masculine vs. feminine' is not connected with the different position of the nouns in the sentence, but represents only a categorization which has its morphosyntactic manifestation in the agreement between noun and adjective.

In the paradigmatic structure of Latin the morphosyntactic differences in case and number between animate and inanimate noun become purely mor­phological differences with allomorphic character.

Page 85: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

68 FROM INDO-EUROPEAN TO LATIN

The agreement in gender is, like the agreement in case and number, originally of semantic/referential character. The syntactic function of agree­ment is secondary. The non-formalized character of gender agreement is apparent (cf. Swahili). Agreement and word autonomy are interrelated phenomena common to both d-f and p-f structure. In the typological comparison of languages it must be taken into account that superficially simi­lar phenomena differ as to structural position and functional hierarchy in different languages, as has been suggested by comparison of IE and Latin agreement with agreement in Swahili and in Balkan languages, where agreement serves as a binding device in the formalized noun - adjective syntagma.

We have dealt rather extensively with the gender category which could be considered as occupying a rather marginal position in the morphosyntactic structure. However, it appears that the categorization 'animate vs. inanimate' is connected from the beginning with the fundamental sentence structuring. Also, agreement is a typical morphosyntactic phenomenon of IE structure.

The development of the gender category reflects closely the change in overall structure. In the development from d-f to p-f type it changed from a semantic to predominantly inflectional category, in the development from non-formalized to formalized sentence structure it developed into a syntactic device. In our treatment, restricted to the relation of Latin and IE, we cannot follow the disappearance of the neuters in the development from Latin to the Romance languages, which was characteristically simultaneous with the disappearance of the opposition nominative vs. accusative.

This structural position of the gender category is a clear manifestation of the grammatical relevance of lexical categories which we defined as a basic and original principle of IE structure.

Page 86: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

2.3 The Category of Number

2.3.0 We have already dealt with two problems concerning the category of number in IE: 1) the flectional principle of the treatment of number and case as categories accumulated in one complex morpheme; 2) the special plural marker of the neuters, originally with collective value (2.2.1).

Because of the cumulative character of the expression of number and case we shall treat the plural paradigm as a part of the case system and its development, as is the common practice. However, some general problems concerning the mutual relation of dual and plural and their expression in the fundamental cases, i.e. the nominative and the accusative, are important for the origin and development of the paradigmatic flection and should be treated separately before the exposition of the case system.

2.3.1 Singular — Dual — Plural

Latin lacks the dual as a grammatical category or categorial distinction. Only the isolated numerals duo and ambo reflect the original dual forms. Neither the opposition of dual vs. plural in nouns denoting persons and animals nor the dual as the characteristic non-singular form for entities which occur in pairs (eyes: Gr. ) are present in Latin. Cf. the Greek examples:

(11) singular the-NoM man-NoM dual the-NoM/Acc man-NoM/Acc plural the-NoM man-NoM singular the-NoM old man-NoM dual the-NoM/Acc old man-NoM/Acc plural the-NoM old man-NoM

Page 87: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

70 FROM INDO-EUROPEAN TO LATIN

(12) Greek du. vs. Latin pi. oculi like - like amici

The IE dual is attested outside of Greek in Aryan, Baltic, Slavic, Tocharian and Old Irish. The dual case systems are restricted to two or three forms. Besides the nom.acc. (-e) Greek has only one common form for gen. dat. We shall restrict our exposition to the nom.acc. form, where the dual form is relevant for the historical explanation of plural forms. For Latin as well, the non-attested categorial distinction helps to explain the attested plural forms.

The nom.acc. dual marker can probably, with support from Gr. evi­dence, be reconstructed as -e in its basic, postconsonantal form. Its realization in vocalic stems is irregular. This element, which can be reconstructed as -e/-h1, causes lengthening of the stem vowel in the case of -i-, -u- and, in part, -o- stems (Gr. ) whereas in -ă- (OI -e-, OCS -ë) and, in part, -o-stems (0I -au) it represents the second element of the diphthong. These irregularities can be explained as sandhi variants of the suffix -e/-h1.

The relation of dual and plural forms in Greek consonantal declension can probably be projected into IE:

(13) du. ' pl.

The mutual relation of these non-singular forms invites us to explain the plural forms as dual forms enlarged by -s and leads to the conclusion that the dual form in -e was the original expression of non-singular in IE (see Specht 1944:366, who considers both -e and -s as deictic elements). This presupposes that the -e form had originally a wider meaning, designating not only two entities, but restricted, countable quantity in contrast to the non-countable collective notion typical of the inanimates. Note that the expression of non-singularity can be facultative only in the original structure. The 'singular' as primitive or basic form is neutral or, better, vague (see Wierzbicka 1980:60-61) with respect to number. Only after the new plural of animates is formed, marked by -s, is the old non-singular restricted in its use to the dual proper, designating two closely united entities.

Page 88: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CATEGORY OF NUMBER 71

2.3.2 The nominative and accusative plural: The nucleus of the plural paradigm

2.3.2.1 The nominative plural: Intraparadigmatic analogical innovations

We have reconstructed the old form of the non-singular of the type *pod-e/*ped-e to which the marker -s is added and we assume that this -s represents the same suffix -s which signals the nominative of animate nouns (see 2.4.2.1 on the original value of this marker). From the nominative singular this marker is transferred to the non-singular to signal the features +individual, +autonomous:

(14) analogical transfer of -s to the non-singular form sg. *pod-s → pl. *pode-s non-sg. *pode

Note that this transfer belongs to the changes which can be classified as analogical in the broader sense of being based on the paradigmatic relationships between morphological forms. However, they are not based on the analogical proportions typical for the interclassai (interparadigmatic) analogies between declensions. Here the suffix which was primarily applied to the singular stem was secondarily applied also to the non-singular form with the effect of providing the non-singular ( → plural) with the features +individual/+autonomous. Morphological analogy is only one part of this formative process, since the application of the case suffix is semantico-syntac-tically motivated by the need to mark the non-singular form for subject. Next we shall discuss another more complicated, intraparadigmatic analogical transfer which we assume for the accusative plural. We think it probable that the irregular analogical formations connected with contaminations are typical of d-f morphology. In the opposition between the old non-sg. *pode and the plural *po-de-s48 the element -s appears as a marker of plurality. This could provide a basis for the extension of the marker also to other members of the pl. paradigm. This tendency, however, is realized only partially in IE languages.

Page 89: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

72 FROM INDO-EUROPEAN TO LATIN

2.3.2.2 Agglutination-like form of the accusative plural

Unlike other plural cases which are formed by a suffix different from the. singular forms, the accusative of singular and plural are formed by the same basic suffix -m/-n (further on only -m).49 However, in the plural the formant -s is added to the accusative form. Cf. the reconstructed IE accusative forms of the consonantal stems, with Latin and Greek regular correspond­ences:

(15) acc.sg. acc.pl. *ped- Lat. ped-em *ped- s Lat. ped-es < *ped-ems *pod- Gr. *pod- s Gr.

This superficially resembles the situation found in an agglutinative language like Hungarian, so that the form *ped- -s, if considered in isolation from the other forms of the plural paradigm, seems to represent a sequence which corresponds in reverse order to the Hung. acc.pl. (see Szemerenyi 1985:518):

(16) Hung. láb-ak-at IE *ped- s leg-PL-ACC leg-ACC-PL

However, there are obvious synchronic and diachronic differences between both structures, which contradict the equation of the IE acc.pl. to the corresponding agglutinative form. The element -s does not carry the meaning 'plurality' as its primary and original meaning, its primary and original meaning being + individual/+autonomous and its function being nominative marking. The opposition nom.sg. vs. nom.pl. does not follow the same agglutination-like structure. Unlike Hung. nom.pl. láb-ak which is marked by the addition of pl. marker to the nom.sg. form, the IE nom.sg. and nom.pl. are marked by the cumulative exponents -s (sg.) and -es (pl.). Therefore the agglutination-like structure valid for the acc.pl. does not apply to the nom.:

(17) acc.pl. = acc.sg. + s *ped s = *ped + s nom.pl. ≠ nom.sg. + s *pedes ≠ *peds + s

And this interpretation is not valid for other IE plural case forms either, though some of them contain a final -s which tends to be interpreted as

Page 90: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CATEGORY OF NUMBER 73

representative plural exponent (2.3.2.4). Therefore, the structure of IE paradigm does not allow to take agglutination-like interpretation of the acc.pl. for more than isolated and accidental phenomenon which has no support in the relations between case forms within the paradigm. Historically, the acc.pl. is not formed by simple addition of the plural marker to the sg. form but this formation has more complicated historical content, typical for the d-f struc­ture, as we shall see in next paragraph.

Note also that the IE acc.pl. in -m-s does not follow the rule according to which the marker of plural as more internal, non-relational category occurs closer to the noun base than the marker of case as more external, relational category (see Bybee: 34, and here note 5). This failure to obey the iconicity principle valid for agglutinative markers reveals a basically different, i.e. decompositional character and origin of the ~s element as tentative plural marker.

2.3.2.3 Transfer of the -s marker from the nominative plural to the accu­sative plural

If we return to the nominative of non-singular and plural as examined above (see under 13 and 14) we can see that in the relationship to the old non-singular form (*pede) which continued to be used in the function of dual the element -s (*pede-s) can be considered as marker of plural vs. dual. This can provide a basis for the extension of this marker to the accusative form. It follows, however, from the above observations that this transfer cannot be realized on the basis of the proportional division 'singular case + plural marker', because this division is not valid for the nominative plural, i.e. the founding form (cf. 17).

In light of this, how can we explain the transfer of -s from the nom.pl. to the acc.pl.? As we have said, the element -s which was originally a marker of animate non-singular nominatives (*pede-s) could be interpreted as a marker of plural in the opposition to the dual form (*pede). The nominative plural is a complex form expressing nom. and pl. at the same time, nom. with respect to other cases, pl. with respect to sg. and dual. This corresponds to the flectional principle according to which the value of the cumulative morpheme is realized in the oppositions. As for the accusative, the following picture can be assumed: In the singular the acc. with the nasal suffix -m was opposed to the nom. with the suffix -s. In the dual, however, the single form *pede stood for both nom. and acc. As stated above, the 'singular' form was unmarked

Page 91: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

74 FROM INDO-EUROPEAN TO LATIN

and originally vague with respect to number. Accordingly, there was a pos­sibility of choosing between two forms to express the non-sg.acc: the form *ped- , if the case relation mattered, or the form *pede, if the non-singularity were to be accentuated. When the element -s appearing in the nom.pl. as a marker of individualized non-singularity (plurality) vs. simple non-singularity (duality) starts to be employed also in the accusative, as it is assumed, it is not added to the dual form *pede (neutral as to the nom. vs. acc. opposition) but to the form with nasal *ped- , marked unambiguously for the acc. The result of this process is the form *ped-e-m-s with the value of acc.pl., both explicitly marked in the complex but analyzable morpheme. The form of acc.pl. develops, therefore, by analogical transfer accompanied by the functional contamination of two possible forms for the acc. of non-singular, *ped- and *ped-e:

(18) acc. sg. *ped-pl. *ped- s

non-sg. *ped-e

We can pose a question whether the functional contamination of the forms *ped- and *ped-e leading to the acc.pl. *ped- -s is in part accompa­nied also by the formal contamination of both forms.

We observe some discrepancy in the form of acc.pl. of -o- stems which can be perhaps explained in this way. The forms of the acc.pl. of some IE languages, i.e. 0I , Lith. and probably also Latin, presupposes the long vowel -öms (see Szemerenyi 1980:170, Stang 1966:186). The long vowel can be explained as the result of the contamination: long -ô of the dual + -m of the accusative + -s:

(19) the possible formal contamination in acc.pl. of -o- stems: acc.sg. *amico-m

acc.pl. *amicö-m-s nom.acc.du. *amicō

We can imagine that there was an oscillation between -öms and -öms, i.e. non-contaminated and contaminated form, whereby diverse languages chose one of the variants.

2.3.2.4 The element -s as a representative plural marker

Page 92: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CATEGORY OF NUMBER 75

The interpretation of -s as plural marker which we presuppose for nom.pl. vs. nom.du. and for the acc.pl. vs. acc. sg. and du. suggests the question whether and to what extent the -s element can be considered as the representative pl. marker in IE languages. In Latin, all pl. cases with the exception of nom.pl. of -o- and -ă- stems and gen.pl. have final -s. In most other IE languages this tendency is more restricted. To a great extent it is realized in Hittite (cf. also 2.4.9.1 below). With the exception of the abl. and instr., which do not have pl. forms and of the -an variant of the gen.pl., all Hittite pl. cases are marked by final -s, whereby not only acc.pl. but also directive/ terminative pl. shows an agglutination-like form (sg. -a vs. pl. -as). (The final -s of directive/terminative can go in part back to the local suffix -si/-su).

The tentative association of plurality and final -s cannot be denied. On the other hand not only the non-complete realization of this tendency, but also other structural features, especially the systematic non-identity of the sg. case morph and the part of pl. case ending preceding the final -s, do not allow us to consider -s as a separate plural morph. It is also only a part of the complex morpheme, but has representative character with respect to the plurality. The tendency to extend the final -s could also be motivated by formal structure, this being a useful device for making the sound shape of pl. forms more volu­minous and providing it with consonantal auslaut (iconicity between sg. and pl. paradigm 2.5.2.2).

2.3.2.5 Remarks on plural nominatives without -s

Since the suffix -s applies originally to the basic sg. form only and is transferred to the non-sg. form secondarily, the non-sg. nominatives without -s are original. Besides the non-sigmatic forms with dual meaning we find forms without the final -s also in plural function. They occur in pronominal declension and in a group of languages including Italic, Celtic, Greek and Baltoslavic also in the declension of -o- stems (and analogically also in -ă-stems of Greek and Italic). The plural ending in -oi is thus opposed to the dual form in -ö: Gr. nom. du. vs. nom.pl. The plural in -oi belongs to the IE pronominal declension and in the above mentioned group of languages is extended to the nominal -o- stems.50 The dual -ö forms and plural -oi forms can perhaps be explained as semantically differentiated sandhi variants of the same form.

Page 93: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

76 FROM INDO-EUROPEAN TO LATIN

2.3.3 Conclusions

We have examined the expression of marked number in fundamental case relations. The formation of nom.pl. (-es) and acc. pl. (-ms) helps to concretize our insights into the d-f structure. The plural forms are not formed by the simple addition of the separative pl. marker, but develop by processes based on the relative analysability of forms which are connected in deriva­tional opposition: transfer of the nom. marker -s from *pede to *pede-s, interpretation of -s as a plural marker in the opposition *pede-s to *pede, and its transfer to the *ped-m based on the functional contamination of *pede + *pedrm as selective forms for non-sg. acc. The resulting forms are the complex morphemes -es, -ms with only relative distinctness of -s as the representative pl. marker.

The original forms are more simple as to the categorial distinctions with overt marking. They are marked only for case as *ped-s and *ped-m or for number only as *ped-e. Semantically, however, the forms are complex and imply the non-overt category: the basic 'sg.' case form is vague with respect to number, the *pede form is neutral with respect to the nom.- acc. opposi­tion. The vague complexity is made explicit not by the additive elements, but by the internal modification of forms. The representative pl. exponent has only relative validity and stability, reflecting the tendency of flectional structure to­ward more analytic expression. In principle, the forms with more categorial distinctions are formed not by the combination of morphs expressing the single individual distinctions, but by the constitution of complex cumulative/fusional morphemes typlcal of flectional type.

We shall deal with the expression of plural in more marginal cases in section 2.4. The forms with basic suffixes are originally vague with respect to number, the forms with marked suffixes are adapted for the expression of the plural cases.

Page 94: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

2.4 The Case System

2.4.0 The case paradigm we find in Latin and in other IE languages appears to be fully unmotivated as to the mutual relation of case forms and their relation to other forms of the structure of the language. This leads to the search for the historical motivations of this synchronically unmotivated system. I have already analyzed the attempts made to explain the case suffixes as the original analytic elements (pronouns or postpositions). Having rejected the assumption which contradicts the word autonomy of IE (word inflection vs. group inflection in the case of analytic elements — 1.2.1, note 15; 1.3.7.1), I would like to attempt to explain the cases within a derivative-flectional structure as reconstructed for PIE. I assume that derivation is an original formative process in IE and that there is no essential contrast but only a gradual difference between the derivation expressing lexical categories and the derivation expressing grammatically relevant categories, i.e. inflection. From this point of view I shall now analyse the original form and meaning of cases and the development of the case system. As already explained, in broader connections the animate and inanimate classes differ by their case morphology and we can assume that the basic cases develop primarily in animate nouns.

2.4.1 The basic case system of animate nouns: Preliminaries

2.4.1.1 The basic case forms

The basic cases are formed by means of three suffixes, whose application is connected with the internal modification of the word called apophony or accent/ablaut alternation. In this respect as has already been explained in 2.1.3 the case suffixes behave like derivative suffixes.

Page 95: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

78 FROM INDO-EUROPEAN TO LATIN

In the prototyplcal domain each suffix has two forms: -0C vs. -VC, i.e. -s vs. -es/-os; -m vs. -om; -i vs. -ei.51 With regard both to the formal and functional hierarchy we can say that the forms without vowel representing nominative, accusative and locative are founding, whereas the forms with vowel representing genitive and dative are founded, i.e. derived.

(20) the IE case system of animate nouns FG + -s RG + -Vs nom.sg. gen.sg. *dént-s *dnt-ós FG + -m RG 4- -Vm acc.sg. gen.pl. *dént-m *dnt-óm FG + -i RG + -Vi loc.sg. dat.sg. *matér-i *matr-éi

This distribution holds for the prototyplcal domain (2.1.3.2) with full grade (FG) of the root or stem syllable preceding the case suffix in the so-called strong cases and reduced grade (RG) of this syllable in the so-called weak cases. From the singular cases attested in IE languages only instrumental and ablative are not included in this system (on vocative see 2.4.7.1). As we shall see further on, the structural position of these two cases is very different, or rather contrastive: the simple residual form in the case of instrumental vs. the marginal marked form in the case of ablative. Both of them, however, are only partially integrated into the p-f structures of IE languages as distinct case forms.

The basic cases are fully represented in the respective cases of Latin consonantal third declension. The locative is presupposed to represent a component of the Latin ablative (see 2.4.7.4).

(21) basic cases in Latin singular plural

nom. reg-s gen. reg-is < -es reg-um < -om dat. reg-i < -ei acc. reg-em < -m abl. reg-e < -i = locative

< -e = instrumental

2.4.1.2 The semantic analysis of the basic case system

The widely accepted division of the IE case system into the grammatical/abstract and local/concrete cases does not allow us to see the parallel formation of the case oppositions nom./acc. vs. gen. (-s, -m vs. -es

Page 96: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CASE SYSTEM 79

(-os), -om) and loc. vs. dat. (-i vs. -ei).52 The concept of the d-f structure provides us with a unifying basis for the interpretation of the case semantics within which the distinction of abstract and concrete case appears to be only gradual. The synchronically non-transparent relationship between nom./acc. vs. gen. and loc. vs. dat. is the result of the grammaticalization of the original structure where the mutual relations of cases are transparent and semantically motivated.

The cases of the IE languages are syntactically and semantically poly-functional. This has led to the attempts to interpret the function of cases with the help of a special code-system structuring the paradigmatic relations of cases, without defining the case directly by its relation to the syntactic categories of subject, object, etc., or to the semantic categories as agent, patient, etc. The various structural descriptions as proposed by Jakobson (1936), de Groot (1939), (1956a), (1956b), Hjelmslev (1935),(1937), Kuryłowicz (1949), etc,53 give a more or less adequate account of the mutual case relations in the developed paradigm. These relations can be described by the opposition — like independent vs. dependent, total vs. partial, central vs. marginal. However, if we attempt to reconstruct the original meaning of the cases, a certain notional framework is necessary to account for the intentional, significative relation of cases to the extralinguistic correlate in the represented 'reality'. We use the usual terminological apparatus54 with some adaptations to the needs of the structure where the nouns have a relative semantic-syntactic autonomy and carry the relevant grammatical meaning. We designate the correlate of the sentence 'situation', and the semantic focus of the situation expressed by predicate is called 'state of affairs'. The correlate of nominal expression is called 'participant in the situation'.

Semantic roles like 'agent', 'patient', 'experiencer', 'benefactive', etc. are defined with respect to the verb categorization, i.e. the participation in the situation is viewed from the point of the verbal predicate. The IE cases, however, express the types of participation which is viewed from the point of the nouns as primarily referential expressions (1.3.5). Whereas the types of participation viewed from the verb side belong to the relational semantics, the morphological semantics of cases is the semantics of a referential type,55

immanent to the noun category. The idea of d-f structure provides the structural-typological basis where this type of semantics finds its broader moti­vation and is explained by the derivational origin of the inflectional case category.

We consider the categorial/immanent semantics as typlcal of the d-f

Page 97: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

80 FROM INDO-EUROPEAN TO LATIN

stage, and it corresponds to other constitutive features of this type of language structure as defined in sections 1.1 and 1.3: word-internal morphological modifications; word as a complex and hierarchically structured morphemic unit (with determining position of the lexico-derivational component); word autonomy in the non-formalized sentence. The paradigmatization consists of progressive abstraction of a grammatically relevant component from the origi­nal, more concrete meaning, and of the formation of the paradigm as a system of oppositions obligatory for the respective word category. This general idea will be applied to the case category and its paradigmatization. We shall start our discussion with the fundamental nominative and accusative cases and their relation to the genitive as derived, founded form.

2.4.2 Case suffixes -s, -m vs. ~es/-os, -om: Nominative, accusative vs. genitive

2.4.2.1 Derivational semantics of the suffixes -s and -m: Types of partici­pation

As has already been stated, the meaning of the IE fundamental cases, i.e. the nominative and the accusative, cannot be defined either by the syntactic functions of subject and object or by the semantic functions of agent and patient. In the developed p-f structure, the subject and object functions can be ascribed to the nominative and accusative respectively as their basic syntactic functions. However, the use of both cases is not exhausted by these functions. Moreover, both functions are expressed not only by means of cases, but also with the help of other devices, such as agreement in the case of subject and government in the case of direct object. As for the basic semantic functions, the function of agent is expressed by means of diathesis, i.e. by verbal voice in p-f structure and by verb class in the reconstructed d-f structure (3.2.1.3). The function of patient is again only one of the accusative functions which has its basis in verb categorization.

The noun categorization which is expressed by the case suffixes -s and -m reflects the types of participation of the participant in the situation. By means of the basic suffixes -s and -m the types of participation in the situation are expressed not with respect to the category of predicates (as is the case of semantic participant roles such as agent, patient, etc.), but on the basis of the classificational criteria immanent to the noun category and connected with its referentiality. The suffixes with mutually contrastive meanings express two

Page 98: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CASE SYSTEM 81

fundamental types of participation of the animate nouns. The nominative represents a participant who has an independent and privileged role in the designated situation, forming the centre to which the situation is oriented. The accusative represents a participant whose position is dependent on the activity of another participant and who is affected by the designated situation. We presuppose that this contrastive marking of the oppositional types of participation developed simultaneously. However, logically, the nominative marking is primary. The accusative marking of the contrastive type of participation is based on the suppression of the characteristic properties which were ascribed to an animate participant by nominative marking (features + autonomous/+individual).

If we want to answer the question of what was the meaning of the derivational suffixes adapted to the expression of fundamental cases, we can only approximatively interpret a vague subcategorization whose semantic focus cannot be determined exactly. We would prefer to determine this semantic distinction by means of a set of features, considering this to be a good way to account for the (objective) vagueness of this categorization and (subjective) uncertainty of its definition. We presuppose the contrastive features: +autonomous/individual, -hindependent for the nominative -s suffix vs. —autonomous, + dependent, 4-affected (by the activity of another participant) for the accusative -m suffix.

The same suffixes and semantic distinctions, if applied in the derived form (-es, -om) to the possessive/partitive structures56, served the genitive marking, i.e. the marking of the participant participating in the situation not totally, but only partially, and with the intermediary of the possessum. This semantic-syntactic differentiation of founding and founded form of the same suffixal formations is the result of paradigmatization of the basic case system which is already effected in PIE. The basic case oppositions are common to all IE languages including Hittite.

2.4.2.2 Paradigmatization of case oppositions: Preliminaries

Paradigmatization of -s and -m suffixal formations consists in the gram-maticalization of their meaning and the formation of paradigmatic oppositions. The derivational formations develop into the paradigmatically organized and obligatory case oppositions. Within the paradigm, the accusative, which in op­position to the nominative is interpreted as —autonomous, + dependent, +af-fected, is in opposition to other dependent cases interpreted as directly and

Page 99: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

82 FROM INDO-EUROPEAN TO LATIN

totally affected. The dependent cases, therefore, form a system of distinctions differing

by direct and total (accusative) vs. partial (genitive) and indirect (locative/ dative) participation.

In connection with the process of grammaticalization there is a tendency to define the fundamental cases with respect to their basic syntactic function, which, however, is never completely realized. On the one hand, the case semantics overlaps the syntactic functions and, on the other, it is only an incomplete means of the expression of subject and object, as the syntactic devices of agreement and government are necessary for the expression of subject and object relation. We will further examine the nominative -accusative - genitive paradigm (2.4.4) after the discussion of subject constitution (2.4.3), which is a phenomenon of primary importance for the IE morphosyntax. Here some remarks will be made about the use of the -s and -m suffixes, i.e. about their distribution in genitive function (2.4. 2.3) and about certain restrictions in the use of -s nominative marker (2.4.2.4).

2.4.2.3 Genitive marking by -s and -m suffixes: Possessivity and partitivity. Genitive - ablative

In the paradigmatized structure of IE languages the -s variant is specified to the gen.sg., whereas the -m variant is specified to the gen.pl. This differentiation can probably be explained by the prevailing association of possessivity with an autonomous animate participant (domus patris) and with singularity, and the association of partitivity with non-autonomous participant (pars domus) and with plurality (unus militum). The possessor has an independent and privileged position with respect to the possessum so that a close connection between nominative subject and genitive possessor is understandable. With the gradation of the partitive notion (pars domus, particeps sceleris) the position of the genitival participant becomes more 'non-autonomous'.

The partitive component of the IE genitive is very strong, since the genitive also expresses the ablative function. This is a very clear-cut example of the union of more abstract and more concrete distinctions in the meaning of cases.

In Latin, the genitive is predominantly an adnominal case (including its occurrence with adjectives). Adverbal genitive occurs only with several verbs like accuso, impleo, misereor. The ablative is in Latin united in one case not

Page 100: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CASE SYSTEM 83

with genitive, but with instrumental and locative meanings.

2.4.2.4 The -s marking and animateness: Restriction in -s nominative

The suffixai formations with -s and -m suffixes are constituted primarily from animate nouns. By the process of grammaticalization the genitive marking is applied also to the neuters occurring in the possessive and partitive structures. However, the application of the -s and -m suffixes in nominative and accusative functions remains restricted to animate nouns. This restriction is motivated semantically in the original d-f structure. After the absolutive -0 marker of neuters became an allomorphized expression of nominative and accusative in p-f structure, it remains as a device expressing gender and agreement.

There is another restriction in the use of the -s suffix for nominative marking. Stem alternation without -s suffix is certainly the original marker of animate nominative vs. inanimate absolutive in the case of -ă- stem feminines; the -eh2 variant is chosen for nom.fem. vs. -h2 variant for nom.pl. of neuters. The alternative expression of the privileged participation by stem alternation (lengthening) can be projected into IE irrespective of the possible explanations of the lengthened nominatives of the type Lat. vict-or < -or(s), hom-o < -on(s) by secondary elimination of the -s suffix.57

2.4.3 Nominative and subject constitution

I have tried to show in the preceding paragraphs that the semantics expressed by basic case suffixes was of the category immanent type. The morphological category of case, which one would expect to be a relational category with respect to its syntactic effect is a noun-inherent category in IE, expressing the semantic features of the nominal participant. The nominative suffix of animate nouns marks the privileged (prominent) participant position in the situation denoted by the sentence. This marking is made with respect to the semantic/referential relation to the denoted participant, not with respect to the noun - verb relation, i.e. it is not determined by the verb categorization. It consists in the further subcategorization of the animate gender category by the features +autonomous/4-individual, etc.

In order to answer the question of how nominative marking contributed to the constitution of the subject - predicate relation58 two other aspects shall be considered: 1) the contribution of the verb to the constitution of the subject

Page 101: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

84 FROM INDO-EUROPEAN TO LATIN

- predicate relation in the verbal sentence; 2) the general condition for subject constitution in the categorical statement.

2.4.3.1 Verb orientation in the IE sentence

The unipersonal IE verb is marked for only one of the possible participants involved in its relational frame. This provides a precondition for the establishment of its relation to the privileged participant. Important for subject constitution is the fact that the verb is marked not for the most involved and internal participant, but for the most autonomous and external one: the agent of an active verb, the non-agentive actor of an inactive verb such as "I turn something" (verto), the animate experiencer of an inactive verb such as "I see, I know" (video). The patient of the active verb is excluded from the person marking. The same applies to the inanimate undergoer of the inactive verb, because of the non-personal and non-oriented third person of the inactive verb (3.2.1.4). Therefore, the unipersonal marking of the IE verb, being solidified with the nominative marking, constitutes a contribution to the establishment of the subject - verb relation from the verb side. The subject -verb relation has its morphosyntactic expression in the agreement between subject nominative and subject person, which is extended also to the 3rd person of inactive verbs. Prior to the transfer of the consonantal active marker to the inactive verb (see 3.2.3.4, 3.4.4.1), agreement in inactive verbs may have been expressed by means of an -o/-a variant of the 3sg. ending (3.2.3.5).

(22) agreement in subject marking amicus veni-t → amicus vidi-t friend-NOM.SG. come-3sG friend-NOM.SG. see-3sG "the friend came" "the friend saw"

2.4.3.2 Thetic and categorical statements

A necessary precondition for the constitution of subject as sentence constituent standing outside of the predicate (as predication base) and outside of the complement frame of the verb (not governed by the verb) is the existence of the type of statement which allows or supports this extraction. The difference between thetic and categorical statements, whose linguistic relevance was proved by Sasse (1987), concerns the relationship between the referential act of the noun and the predicational act as the sentence-constituting

Page 102: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CASE SYSTEM 85

act. In thetic statements the reference of the nouns is completed together with the reference of the whole sentence, i.e. with the accomplishment of the pre-dicational act. Yet the nouns can also refer to 'absolutes' in the sense of Wilensky (1987), i.e. to the elements of the actual knowledge shared by the speaker and the hearer. This makes a statement of the categorical type possible. The noun in its basic naming form enters into the sentence as a referentially closed expression whose referential intention is already satisfied before the act of predication. It constitutes a predication base to which the predicate ascribes some property or state of affairs. For our purposes, the following constructed Latin sentences can illustrate the difference between both types of statements. It is presupposed that the difference between uni-partite and bipartite structure manifests itself in intonation patterns similar to those known in modern European languages (the whole accentuated word is underlined):

(23) thetic statement of the nominal ('entity central') type ecce, amicus optimus "Look, [this is my] best friend"

(24) thetic statement of the verbal ('event central') type amicus venit "The friend is coming"

(25) categorical statement amicus hodie venire non potest "The friend cannot come today"

The subject - predicate relation is grammaticalized in the IE sentence, so that the thetic statement amicus venit also has subject predicate structure expressed by the verb agreement. This holds for the formal-syntactic level of sentence constituents. Both types of statements have, however, different semantic-syntactic structure and interpretation. As in many other languages, the thetic interpretation of the sentence can be indicated by subject inversion in the Latin sentence:

Page 103: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

86 FROM INDO-EUROPEAN TO LATIN

(26) ardebant oculi [... ] Expectabant omnes quo tandem progressurus [...] esset f.. ] Clamabat ille miser [...] Caedebatur virgis in medio foro Messaenae civis Romanus Cic.Verr.2,6,161 "his eyes blazing [...] Everyone was wondering how far he would go [...] The unhappy man cried out that he was a Roman citizer [...] There in the open market-placc of Messana a Roman citizer was beaten with rods"

Note that this essential difference between the two relationships of noun reference and predication, which accounts for the bicentral character of the sentence structure based on categorical statement, must not be confused with the distinctions such as theme - rheme or toplc - focus which are made with respect to the distribution of new and old information and of the emphasis in the discourse. These discourse categories are assumed to concern all constitu­ents of the sentence and have no direct relevance for the internal semantic-syntactic structure of the sentence; it is only casual coincidence of the theme with the predication base of the categorical statement which may suggest it.

2.4.3.3 The semantic/prototyplcal and syntactic/generalized subject

The formation of the grammaticalized subject - predicate relation expressed by the combination of nominative marking and agreement is a characteristic feature of p-f structure. The original prototyplcal IE subject is 1) semantically defined by the nominative marking with the features + animate, +autonomous, etc. and 2) limited to categorical statements. In these state­ments the referentially satisfied noun enters into the agreement relationship with the verb.

The development of the grammaticalized subject involves more struc­tural changes, which are dealt with only in part in this study: the allomorphi-zation of inanimate absolutive which is semantically identified with animate nominative and accusative respectively, the development of agreement in inactive verbs, the rise of the passive, etc..

2.4.4 The paradigmatization of nominative - accusative - genitive oppositions

2.4.4.1 Remarks on the position of nominative: Nominative - accusative opposition

Page 104: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CASE SYSTEM 87

We define the nominative marking in d-f structure as marking of ani­mate nouns by the features + autonomous/individual (2.4.2.1). This category-inherent meaning of the -s marker serves to further specify the animate class, which, however, has semantic-syntactic relevance. By this specification the privileged role of animate participant in the sentence is signalled. The marked form of the nominative is satisfactorily interpreted by this category-inherent specification,59 so that the ergative/active hypothesis cannot have its support in the marked form of the IE nominative. This isolated agreement between er-gative and IE nominative is only a superficial phenomenon, while the structur­al and functional differences between both cases are essential, as explained in section 1.2.

In the developed p-f structure the nominative enters into the opposition with other cases and at the same time it serves as a sign of animate gender. The structural opposition of the nominative marked by the -s suffix is typlcal of the cumulative semantics of the IE flectional morphemes. According to the different structural relations and oppositions of this form, the -s marker is 1) a marker of the animate class; 2) a marker of case with respect to the oppo­sition to other cases of animate nouns, especially to the contrastive accusative case; 3) a marker of singular with respect to dual/non-singular (2.3.1).

As to the character of the opposition nominative - accusative, we do not find it necessary and possible to consider all morphological oppositions as privative oppositions, with one member marked positively, whereas the other member is neutral with respect to the given categorial distinction. We will take into account the equipollent oppositions, with both members marked positively, i.e. +autonomous, (—dependent), for the nominative vs. (—autonomous), + dependent (affected) for the accusative. The equipollent semantic relation is probably typlcal of derivative suffixes which are not fully grammaticalized. Only during the process of grammaticalization do the priva­tive oppositions become prevalent. The grammaticalized opposition of nomina­tive - accusative can, with the help of markedness theory, be described so, that the -s nominative was (by the feature + autonomous) positively marked for the gender, whereas, within the opposition of nominative - accusative cases, the accusative represents the member marked as + dependent, - affected.

(27) the structural position of nominative -s vs. -0 + animate vs. —animate -s vs. -m —dependent vs. + dependent

Page 105: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

88 FROM INDO-EUROPEAN TO LATIN

2.4.4.2 Remarks on the genitive problem

We explained the genitive singular as a suffixal formation whose relation to the nominative form is the relation of a derived form to the fundamental one, based on the distinction of the strong and weak cases (stem alternations). This explanation has clear advantages compared with those going back to Wijk (1902). According to Wijk the nominative and the genitive have an identical form, i.e. with full -es or -os suffix. In the gen. + nom. syntagma the dependent form, i.e. the genitive should precede the nominative in IE, and the strong accent, relating to the whole group, is placcd on the ending -és of the genitive. As a result of this accentuation, the non-accentuated -es ending of the nominative is syncopated.

(28) development of the gen. + nom. group assumed by Wijk: *hominés pedes > *homines peds > hominis pes

This explanation is based on the assumptions which do not correspond to the IE structure: 1) the contact position and fixed word order in the attributive syntagma of genitive + nominative; 2) the restriction of the genitive to the attributive function. As to point 2, the analysis given by Watkins (1967), on the contrary, shows that the predicative function in nominal sentences is the primary function of the genitive: pes (est) hominis. This corresponds to the general rule stating that the primary function of the morphological forms is that function which is exercised in the simplest structures. In IE structure nouns can exercise the predicative function without any support in the verb (the copula), and the predicative structure is more elementary than the attributive one. The conception of genitive as a derived form (with respect to nominative and accusative) is compatible with its primary use in the predicative structures with possessive or partitive meaning. Also the attributively posed genitive can regularly follow the nominative in IE languages. In Latin both positions are well attested (see esp. Marouzeau 1922, de Jong 1983).

As already mentioned in our discussion in section 1.2 the formal identity of the nominative and genitive is one of the supports of the ergative hypothesis. Again this correspondence is only isolated and superficial. The ergative has often a more marginal case function, besides the agentive one, expressing especially the instrument. This functional unity is based on the semantic affinity between agent and instrument as two participant roles

Page 106: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CASE SYSTEM 89

determined by the verbal relationality. The structural relationship between nominative and genitive is different. Syntactically, they co-occur as members of predicative or attributive syntagms, and semantically they represent a fun­damental and derived variant of the same referential case role.

2.4.5 Case suffixes -i vs. -ei: Locative vs. dative

The suffix expressing location is also differentiated into the fun­damental and derived forms: -i vs. -ei.60 The former is represented by the locative expressing the pure and actual location and the latter by the dative expressing the virtual location, marked + directed, + dynamic. Cf. OI matár-i vs. matré < *ma-tr-éi. The location can be viewed as a rather general and abstract notion compatible with animate nouns, around which the case system was centred: 'in the sphere of s.o., in the interest of s.o.'; in the meaning of +dynamic member (dative) also the alienable possession is included.

The opposition locative vs. dative (—dynamic vs. +dynamic), however, is not so commonly grammaticalized in the IE languages as the oppositions nominative vs. accusative vs. genitive. In most IE languages, including Greek and Latin, loc. and dat. are not integrated in the case paradigm as distinct and regular oppositional forms. Only certain substantives have specific locative forms, as adverbial expressions of placc, e.g. ruri, domi. According to the common explanation, the locatives of type ruri are shaped analogically to the type domi, where the short ending -i contracts with the preceding thematic vowel: *domo-i. Of course, it can also be assumed that the more marked dative -ei ending prevails with consonantal stems as the expression of location ±dynamic. Only with thematic stems was the semiadverbial locative of the type domi < *domo-i differentiated from the dative *domo-ei. Consider the fact that the dative function is realized predominantly with personal/animate nouns. We cannot definitely decide whether the locative ending -i forms also a component of Latin ablative ending dente, trabe, rure, etc. or whether these forms represent only the original instrumental in -e (2.4.6.2).

2.4.6 Other components of the IE and Latin case paradigms

Besides the cases formed directly by the suffixes -s, -m, and -i and the forms of nom.acc.pl. with indirect application of these suffixes (2.3.2), the IE paradigm has the following components: 1) stem form with -0 suffix, the so-

Page 107: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

90 FROM INDO-EUROPEAN TO LATIN

called casus indefinitus — 2.4.6.1; 2) instrumental form with vocalic suffix -e/-h1 — 2.4.6.2; 3) forms with marked postinflectional suffixes: ablative and plural cases — 2.4.6.3.

2.4.6.1 Casus indefinitus or primitivus

The stem form with -0 suffix has a different role in the declension of animates and inanimates (neuters): A) In the animate declension the non-marked form occurs in the vocative function. The vocatives are distinguished from both the sigmatic and non-sigmatic (lengthened) nominatives by their short vowel stem: cf. Gr.

vs. , OCS ženo (a) vs. žena (ă). In the declension of -o- stems the vocative is characterized by the -e- grade of thematic vowel: Gr. vs.

, Lat. lup-e vs. lup-us < -os. In Latin, with the exception of -o- stems, the nominative and vocative

have the same form. This can be explained as a result of morphonological or analogical shortening of lengthened nominatives (pater < *pater, femina < *dhemenă).61 In the case of sigmatic nominatives, however, the extension of the nominative into the vocative function evidently takes place. With the short­ening and other changes (ager < *agros)62 a rather large occurrence of nominative without -s marker arises. These secondary -0 nominatives must be interpreted as components of the characteristic position of the nominative in the Latin paradigm as an extra- or supra-paradigmatic case.

In some IE languages the use of the casus indefinitus in the non-vocative function as a quotation form, the so-called casus commemora-tivus, is attested. The -e vocative of -o- stems is, however, not attested in this function, so that we have no possibility of following this use in Latin too. B) In the declension of neuters the absolutive is represented by the pure stem: genus, mare < *mari, etc. Only -o- stem neuters are marked with -m suffix: verbum.

Other use of the casus indefinitus of inanimate nouns is that of the locative in the broader sense, which is attested in -r-, -n-, -s-, and -t- stems.63

2.4.6.2 Case forms with vocalic suffix: Instrumental

The case with vocalic suffix -e/-h1 (homonymic with the dual/non-singular suffix — see 2.3.1) is classified as instrumental in the treatment of IE case paradigms.64 We have this form in Lat. abl. ped-e and as underlying

Page 108: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CASE SYSTEM 91

(and partially co-occurring see 2.4.6.4) form in the ablative of vocalic stems: dom-ö < -oH(d), femin-â < -aH(d), etc.

This little marked form had weak position, and was subjected to reinforcements and postinflectional enlargements, in agreement with seman-tically marked value of instrumental case (iconicity principle). In 0I the -o­stern form v k-ă (besides of x k-ena, from v kah "wolf") was extended to consonantal stems: pad-ă (instr. from pad "leg").65 In Mycenaean Greek the form e-re-pa-te with instrumental meaning is attested. The suffix is, however, probably that of dat. -ei (= not ).66 In any case the instrumental was not integrated into the Greek paradigm (2.4.9.2). Only -o-stems with this vocalic suffix are attested in Lith. instr. vilk-u < -ö (so also OHG wolfu), and perhaps in Goth. dat. wulf-a < -ě (see Szemerényi 1980:169). Otherwise, in Baltoslavic the -m forms were also introduced in the singular paradigm. The -a case of Hittite goes probably back to the 'instrumentalle' -e/-h1 suffix.

It seems probable that the instrumental originally represented (together with the absolutive) a case form of inanimate nouns. If this is true, then the IE case paradigm would arise by the integration of two formational systems of nominal classes, just as the verbal system developed by the integration of active and inactive systems.

2.4.6.3 Cases with marked postinflectional suffixes

They are represented by A dental suffix -t/-d 67 of the ablative singular — 2.4.6.4 B marked suffixes of -CV type: -si/su, -bhi/-bho, -mi/-mo which were paradigmàtized as plural cases — 2.4.7

The marked postinflectional suffixes68 are used as signs of marginal case distinctions, in accordance with the position of postinflectional elements in the IE structure (1.3.7.1). The presumably postinflectional character of these suffixes does not imply that they should have been applied also as autonomous particles. Certainly, they were integrated into the word form and were applied word inflectionally, not group inflectionally, as were suffixes of derivational origin.

2.4.6.4 The Latin ablative and its Indo-European components

The Latin ablative of the consonantal declension can go back to two IE cases, i.e. to the locative in -i and to the instrumental in -e. The most probable

Page 109: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

92 FROM INDO-EUROPEAN TO LATIN

assumption that the dental suffix we find in Aryan and Latin ablative of vocalic stems69 (on dental cases of Hittite see 2.4.9.1) is a postinflectional suffix applied to the underlying instrumental form would suggest that the instrumental is a basic component of the Latin ablative.

In 0I the ablative with a dental suffix is attested in the -o- declension and this situation has been projected into EE, whereas the more extensive use of the dental ablative in Avestan and Latin, where it occurs also in other vocalic stems, was considered as secondary innovation based on the analogy with -o- stems. However, these differences can be results of alternative grammaticalizations of the originally optional postinflectional element. The fact that the dental ablative asserted itself especially in -o- declension can have its motivation in the presup­posed semantic restriction of the innovative -oso, -osio genitive (2.1.3) which did not serve the expression of the ablative function covered by the genitive in other declensions. In Latin, according to the analysis by Prat (1975), the co-occurrence of -ö, -ă and -õd, -ăd forms is attested in the case of -o- and -ă- stems.70

2.4.7 The plural paradigm

The plural paradigm in Latin and similarly in other IE languages which are representative of the IE cases, contains forms of different structure and origin. We can distinguish three types of case forms: A) The nominative and accusative plural have their origin in the dual/non-sg. form and in the corresponding singular cases, as explained in section 2.3. B) The genitive plural originated directly as one of the forms of the basic system, formed by the suffix -om. C) The datives and ablatives, which have common forms in all Latin declensions, are formed by the marked suffixes, just like the corresponding plural forms, i.e. datives, locatives and instrumentals of other IE languages.

2.4.7.1 The Latin dative-ablative plural in -is and its IE origins

The plural dative-ablative forms are made up of suffixes of the type CV, and in some instances by the addition of the representative -s marker of plurality. Both processes have correspondences in the respective case forms of other IE languages.

Page 110: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CASE SYSTEM 93

2.4.7.1.1 The IE locative plural suffix si/su This suffix is attested in Aryan (01 dat-su, mat -su, v k-e-su) and

Baltoslavic (Lith. dantyse, OCS materech , vi cech ). In the variant si this suffix is attested in Greek ( ). As can be seen especially clearly in the Greek , this ending su/si is adjoined to the dative/locative form in -oi. Also in Latin the ending -is of the -o- stems of the 2nd declension can go back to -oisu. However, the non-attestation of -s < si/su in other Latin declensions speaks for the derivation -is < -ois (2.4.7.1.2).

2.4.7.1.2 The IE plural instrumental in -oi-s

The plural instrumental had a distinct form only in the case of -o- stems, where we find the ending -ois attested as an autonomous instrumental ending in 0I and Baltic. In Greek, this form is a component of the dat.pl. besides the form in si. Similarly in Latin, this form is reflected by the dat./abl.form in -is <-ois and analogical dat./abl. of -a- stems.

2.4.7.1.3 The postinflectional character of -su/si and -s

In both variants of the plural cases of -o- stems, i.e. -oi-su/si and -oi-s the final formant -su/si or -s is applied postinflectionally to the dative/locative form in -oi which was originally vague as to the number. This is clear evidence of the postinflectional character of the final formant which is not apparent in the consonantal declensions (cf. Gr. ); see above on the casus indefinitus with locative function. Both formants, however, differ in their character and origin. The formant su/si represents a suffix with concrete local meaning, the formant -s represents an -s plural marker transferred to the underlying sg. form by analogy with the nom. and acc.pl. (2.3.2.4).

2.4.7.2 The Latin dative-ablative in -bus

In other declensions, as well as in archaic Latin in the case of -ă stems, the dative-ablative suffix -bus is used, which has a closer correspondence in other IE plural case suffixes containing -bh- and more distant correspondences in similar suffixes containing -m- (see Porzig 1954:90 on this dialectal difference). The final -s represents again the tentative plural marker.

Page 111: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

94 FROM INDO-EUROPEAN TO LATIN

2.4.8 Innovative case suffixes of -o- stems: The pronominal subsystem

Innovative case formations which we find in Latin and other IE languages are connected with the elaboration and therapeutical transformation of the inflectional paradigms of -o- stems, followed partially by the analogical transformation of -ă- stems. The point of departure for these innovations was the subsystem of adjectival gender distinguishing pronouns formally corresponding to -o- and -ă- stems, and occurring systematically in syntagmatic relations with them. In Latin, we find the innovative forms in gen.sg., nom.pl. and gen.pl.

Whereas the gen.sg. in -i (see Blümel 1972, Devine 1970) and the gen.pl. in -drum, -arum are Latin innovations with only restricted correspondences in other IE languages, the gen.sg. in -os-io/-os-o and the nom.pl. in -oi represent late Indo-European innovations.

2.4.8.1 Genitive singular in -os-io/-os-o

This ending is attested in the form -osio in archaic Latin,71 so that Latin shares the innovation which seems to include a large part of IE area. Only Hittite directly attests the -os ending, which can be considered as an original product of the application of -es/-os suffix to -o- stems (2.1.3). The -osio/-oso forms are directly attested in Aryan and Greek. The areal distribution of this form could, however, be larger, except that its integration into the paradigm did not take placc, as was also the case with Latin. In Baltoslavic the genitive of the -o- stems goes back to the form -öt which, ablative in origin, was generalized in genitive use. The form in -so is, however, attested in the pronominal declension, cf. OCS česo.

2.4.8.2 Nominative plural in -oi

The non-sigmatic nominative pl. in -oi instead of -os of Aryan, Gothic and Osco-Umbrian (and Olr vocative — see 2.3.2.5, note 50) is attested in Latin, Greek, Baltoslavic, and OIr (where we have nom. pl. OIr fir < *uiroi vs. voc.pl. firu < *uirös). This form is also supposed to be transferred from the pronominal subsystem. Unlike the -osio/-oso form which was innovative also within the pronominal declension, the non-sigmatic nom.pl. goes back to the original non-singular forms without the -s suffix (2.3.2.5).

Page 112: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

CASE SYSTEM 95

2.4.9 The PIE case paradigm and its development

The basic system formed by the suffixes -s, -m, -i applied primarily to animate nouns is of PIE origin. It is commonly attested in all IE languages which are representative with respect to the case morphology. The IE languages accord also in the weak and non-stable differentiation of the locative -i vs. dative -ei forms.

The differences between case paradigms of the IE languages consist in 1) the different integration of the residual non-marked form -e/-H (instrumental) and 2) the different integration of marked suffixal {-CV) formations. The latter difference relates especially to the plural paradigm. Here the cases formed by direct (gen.pl.) and indirect (nom.pl., acc.pl.) application of basic suffixes can be projected into IE, whereas there are expressive differences in the use of marked forms.

2.4.9.1 The evidence of Hittite

With respect to the important position of Hittite given by its early attestation on the one hand and early (though perhaps only relative — see Meid 1975, 1979) isolation and non-participation in late IE developments on the other, the analysis of the IE case paradigm and its development must respect the Hittite evidence. The Hittite case system (cf. especially Neu 1979, Starke 1977, and Schmid 1973) confirms the conclusion that the basic case system formed with help of the suffixes -s/-es,-os; -ml-om; -i/-ei is of PIE origin. The lack of the grammaticalized difference between locative (-i) and dative (-ei) is also confirmed by the Hittite evidence.

We have two forms with a final dental element in Hittite: the ablative -az, -z and the instrumental -it, -t. These suffixes are probably applied postin-flectionally to the underlying -a case, which represents the IE instrumental with -el-H suffix, and to the underlying dative/locative in -i.72 The dental cases, ablative and instrumental, do not have a corresponding plural form (and according to Starke 1977:126 are not used for referring to plural entity). The dat./loc.sg. in -i and the directive/terminative sg. in -a have the identical corresponding form in -as in the plural, which represents the -a form with added -s marker of plural (which can partially go back to -si/-su suffix). After the generalization of -as genitive also for the plural, all plural forms of Hittite end in final -s.73

Page 113: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

96 FROM INDO-EUROPEAN TO LATIN

2.4.9.2 The Greek paradigm with alleged syncretism

The basic case system is also well represented in Greek, with non-integrated semi-adverbal locative of the type We cannot speak about the syncretism of IE cases either in the dative/locative, whose differentiation is only tentative in IE,74 or in the genitive/ablative representing an original semantic unit of IE. Only in the case of -o- stems was this unit disjoint by mutually connected innovations: the formation of the genitive in -osio followed by the integration of postinflectional formation in -t/-d into the paradigm. The really characteristic difference between the Greek paradigm and the paradigms of other representative languages consists in the elimination of the instrumental (which also explains the non-existence of an -öt/-öd form based on the instrumental) and its merging with or substitution by the dative. This is of course only an alternative and radical realization of the common tendency to avoid the little marked -e/-H form, which was in Greek not compensated for by the paradigmatization of marked suffixes (the instrumental -oi remained only semi-adverbal in Greek).75 The possibility of a merger of the dative and instrumental is given by the disambiguating effect of animate vs. inanimate (as a notional category, not as a gender category) categorization, which determines the dative vs. instrumental (and locative) interpretation (see on this Kurzová 1988:151).

2.4.9.3 The late IE isoglosses in case formation

We have discussed or mentioned three innovative forms which were produced in the development of IE after the separation of Hittite: 1) gen. in -oso/-osio of -o- stems; 2) nom.pl. in -oi of -o- stems; 3) -bh- and -m-dat./instr.pl.76 The first two changes do not have special interest for the dialectal differentiation of IE. A large and not exactly determinable area of IE was affected by the -oso/-osio formation. The -oi form has more restricted and limitable occurrence. The change consisted, however, in the transfer of the archaic, non-innovative form from one subsystem to other, so that parallel development is possible. Only the difference between -bh- and -m- pl. eases can be considered really relevant for the dialectal differentiation of IE.

Page 114: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

2.5 Remarks on Latin Paradigmatization

2.5.1 The morphological and the inflectional paradigms

The developed paradigmatic-flectional structure of Latin shows a high degree of paradigmatization of both the morphological and inflectional paradigms. As explained in 1.3.3.1 the morphological paradigm is a system of morphological distinctions valid for the given word category, whereas the inflectional paradigm is a realization of these distinctions valid for the given inflectional class, in this case the declension class.

The Latin nominal paradigm represents a system of distinctions obligatory for all nouns. For the derivative-flectional noun morphology of IE, a certain diversity in the system of morphological distinctions can be assumed, though it is rather restricted in comparison to the verb morphology. The basic case paradigm was already formed in PIE, the morphological diversity of the noun systems of IE languages can be observed only in marginal case func­tions. It was the ablative function which had double range in IE, being a component of the genitive in the basic -es form of athematic singular declensions, whereas in the -bh/m- forms (applied especially in plural paradigms) and in the thematic -öt/-öd form, it was connected with the instrumental in one case form. In Latin, this ambiguity was eliminated, the ablative/genitive union is non-attested (2.4.2.3). Similarly as in other IE languages, the originally semantic difference of animate nominative and accusative vs. inanimate absolutive appears as only an allomorphic, inflectional difference.

Thus, all inflectional declension classes display the same system of morphological distinctions in their singular and plural paradigms, only with a higher degree of homonymy in the plural (total homonymy of dat.-abl., partial homonymy of nom.- acc). Also, neuters are equated with animate masc. and fem, genders in their morphological system. Since the original absolutive

Page 115: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

98 FROM INDO-EUROPEAN TO LATIN

became an allomorphized expression of nom. and acc, the identity of nom. and acc. of neuters becomes part of intraclassal homonymy (see below).

I will not attempt to give either a systematical analysis of the intraclassal and interclassai relationships within the Latin inflectional system77

or a historico-comparative analysis of Latin declensions.78 Only some aspects of the paradigmatization process will be discussed which are relevant from the point of view of natural morphology, since they reveal certain relationships between the morphological categories (singular vs. plural, case relations).

2.5.2 Some features and tendencies of Latin paradigmatization

In Latin, we can observe a progressive formal fusion of stem marking vowel with case suffix into one case ending. We have already mentioned this tendency, which is characteristic of the nominal inflection when we compared the morphemic structure of noun and verb (2.1.6). In the verb, the stem marking vowels preserve their distinctness from the person markers proper and function not only as conjugation markers, but also as mood/tense markers (1.3.3.2). The following series of endings characterizes the declensions of classic Latin:

(29) case forms of classic Latin singular

decl. nom. gen. dat. acc. abl. 1 -a -ae -ae -am -a 2 -US/-0 -ī -ö -um -ö 3 -S/-0 -is -i -em,- 79

-un -e,-i 4 -us -ūs -ui-u -um -Ü 5 -ēs -ei

plural -ei -em -ë

decl. nom. gen. dat./abl. acc. 1 -ae -arum -ís -as 2 -i -örum -ís -ōs 3 -ēs -um,-ium -ibus -ēs 4 -us -uum -ibus -ūs 5 -es -êrum -ebus -ēs

Page 116: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

LATIN PARADIGMATIZATION 99

2.5.2.1 Semantic and formal markedness (singular vs. plural)

The volume of plural cases exceeds that of the singular cases in Latin paradigm or is the same in gen.sg. -is vs gen.pl. -um (decl.3) and nom.sg. -ēs vs. nom.pl. -ēs (deci.5). This general rule reflects the iconicity principle in the distribution of singular vs. plural forms. Beyond that, more specific regulari­ties can be observed in the form of singular and plural cases. The singular endings are unisyllabic (exceptions: dat. of 4th declension -uī, gen. and dat. of 5th declension -eī), the combination 'long vowel + consonant' (VC) occurs only in nom.sg. -es and gen.sg. -s, again in 4th and 5th decl. With exception of nom. pl. -ae, and -f of 1st and 2nd declensions all plural endings end with a final consonant and are bisyllabic or bimoric (— = --), with exception of the gen. pl. in -um of 3rd declension.

(30) the shape of singular and plural cases singular plural

_ -ārum, -örum, -ěrum, -ěbus — ~ -ui, -eī -ium, -uum, -ibus — VC -ēs, -üs -ās, -ös, -üs, -s, -is — V -ae, -ö, -ā, -i -ü, -ë -ae, -i _ VC -us, -is, -am, -um, -em, -im -um _ V -e

Some preferences observed in analogical changes were probably governed by the tendency to make the plural endings more voluminous and marked than the singular ones. The generalization of the abl. pl. -ibus for the whole 3rd declension (by the extension from the -i- stems to the consonantal stems) can thus be explained, as well as the fact that the gen.pl. ending -ium is preferred, i.e. it is applied to a greater extent to the consonantal stems than the other -i- stem variants.80 By this process the exceptional unimoric plural ending -um became restricted in its use.81 This iconicity between formal and semantic markedness can also be observed in the innovative gen.pL -drum, -ārum of -o- and -ā- stems.

On the other hand, the bisyllabic character of the gen.sg. in -osio could be one of the causes why this form was abandoned. Also the favoring of the abl.sg. ending -e vs. -f (in contrast to the favoring of the gen.pl. -ium vs. -um) results in the preferred unimoric character of singular forms.82

The exceptional bisyllabic singular endings -uf (dat. of 4th declension)

Page 117: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

100 FROM INDO-EUROPEAN TO LATIN

and -et (gen. and dat. of 5th declension) may suggest that there are certain differences between declensions with a view to the grade of the fusion of the declension and case marker. Probably the distinct markers -u and -e of the 4th and 5th declensions respectively are not so clearly treated as part of the case endings as they are in the other declensions.

2.5.2.2 Homonymy and its structural disambiguation and dehomony-mization: The position of the nominative singular

The homonymy of case endings is allowed in classic Latin with respect to both intraclassal and interclassai relationships.

In classic Latin we have the following homonymous case endings:

(31) homonymous case endings -ae 1st decl.: gen.dat.sg., nom.pl. -f 2nd decl.: gen.sg. and nom.pl.

3rd decl.: dat.sg. and abl.sg. (-i stem variant) -ei 5th decl.: gen. and dat. sg. -ö 2nd decl.: dat. and abl. sg. -is 1st and 2nd decl.: dat. and abl. pl. -ibus 3rd deel.: dat. and abl. pl.

4th decl.: dat. and abl. pl. -ebus 5th decl.: dat. and abl. pl. -es 5th deel.: nom.sg.

3rd and 5 decl.: nom. and acc. pl. -üs 4th deel.: gen.sg., nom. and acc. pl. -ü 4th deel.: dat. and abl. sg.,nom./acc. sg. of neuters83

In its final form this homonymy of case endings arose by the morphonological (fusion of case suffix with class marker) and phonological changes realized in the development of Latin; homonymy is a counterpart of increased allomorphy (2.1.2). Homonymy of flectional formants was also, however, a typlcal feature in the original IE structure, where the basic grammatically relevant formants had restricted phonological realizations. Cf. this original homonymy in instances such as -s = animate nominative marker, 2sg. active marker; -e/-H = dual marker, instrumental marker, vocative marker of -o- stems, 3sg. inactive marker, etc. Furthermore, in the paradigmatization process the formants were provided with new functions. Thus the -s which in nominal morphology was animate gender and nominative

Page 118: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

LATIN PARADIGMATIZATION 101

singular marker also became a representative plural marker. The interclassai or even intercategorial (-s as a nominal and verbal

suffix) homonymy of formants did not bring about the homonymy of entire word forms. Also, the intraclassal homonymy of formants, such as -e as dual and instrumental marker or -s as nominative marker and representative plural marker, was dehomonymized in the structure where the morphological distinctions were expressed not merely by the addition of a segment but by the entire modification of the word forms in mutual morphological relationships (stem characteristics preceding the application of a final segment).

In the case where the homonymy of formants brought about the homonymy of the entire word forms, the disambiguation was provided by syntactic/constructional and semantic means. Within this domain of homonymy we can distinguish the instances where the homonymy seems to be connected with the affinity of cases on a more abstract functional level from those instances where the homonymy seems to be an accidental phenomenon only, tolerable with respect to the existing disambiguational devices.

Both nom.-acc. (pl. of 3rd - 5th decl.) and dat.-abl. (sg. of 2nd decl., pl. of all decl.) homonymy relates to two cases which can constitute a unit on a more abstract level. In the case of nom.-acc. the newly developed homo­nymy of nom.-acc.pl. (which in the case of nom.pl. of 3rd decl. arose from the extension of -ēs < -ei-es of -i- stems to consonantal stems) had its forerunner in the absolutive of neuters which in the Latin p-f structure no longer had any semantic motivation. The paradigmatic affinity of both cases lies in the fact that both nom. and acc. were fundamental cases expressing the basic actants (actor and undergoer) or the basic syntactic constituents (subject and direct object) in their main paradigmatized functions. The paradigmatic affinity was at the same time connected with the disambiguation of both cases by syntactic devices of agreement and government respectively. In the case of dat.-abl. homonymy, the higher functional unit is that of the adverbal case of indirect and marginal participation. Dative subcomponent expresses the in­direct participation of the primarily animate participant, 'ablative' subcomponent joins in itself more types of participation of the primarily inanimate participants. So not only verbal valency, but also the categorization of nouns in animates and inanimates, function as disambiguating devices. This homonymy was also preconditioned by the earlier IE structure, with its lack of the obligatory dative vs. locative differentiation. On the other hand, other attested homonymies do not point to any functional affinity. Their tolerability is given by the evident constructional disambiguation. This is the case of dat.

Page 119: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

102 FROM INDO-EUROPEAN TO LATIN

and gen. homonymy (sg. of 1st and 5th decl.) which in Latin is made tolerable by the predominantly adnominal use of the genitive. Unlike homonymy within one number, singular or plural, cross-number homonymy, i.e. the gen.sg.-nom.pl. homonymy in the 1st and 2nd decl. and the gen.sg.-nom.pl.-acc.pl. homonymy in the 4th decl., occurs between the cases which are not related to each other by 'paradigmatic affinity'; disambiguation is provided by agreement in the case of nom.pl.84

There is, however, in the IE and Latin development a tendency to avoid the homonymy of the nominative sg. vs. genitive sg. opposition. Resistance against this homonymy is understandable with respect to the regular co-occurrence of both cases as direct co-constituents. As already mentioned, the gen.sg. of -o- stems which was not clearly distinct from the nom.sg. was replacccd by innovative forms. Respecting this, we must give special attention to the Latin homonymy of these case forms.

We find this homonymy in the subclass of the third declension including the -i- stems which were not subjected to syncopation: type civis = nom. and gen.sg. This homonymy was not registered in the above survey, not so much due to its partial character as affecting only subpart of the declension, but rather due to the following considerations. If we examine closely the position of the nominative in the 3rd declension we can see that this homo­nymy is structurally dehomonymized already on the morphological level:

(32) nom.- gen. sg. homonymy and its structural dehomonymization: nom.sg. civis dehomonymized as civi-s gen.sg. civis civ-is

This dehomonymization has support in the structural position of the nominative singular within the 3rd declension. In the formation process of nom.sg. the first step is the formation of the stem which is distinct from the oblique stem in a large part of the 3rd declension. This process is than either accompanied by the application of the -s suffix (type urb-s vs. ur-b-is, pes vs. ped-is) or no suffixation takes placcc (-0 suffix in the type homo vs. homin-is, victǒr vs. victor-is). In the gen.sg. the case suffix is -is in all these instances. Hence, the same relationship is to be considered for civis vs. civ-is homonyms as well. Also here the genitive case suffix is identified as -is and the preceding segment civ- represents the oblique stem to which the nominative stem civi- is opposed, while the final element is identified as nominative -s suffix in the analogy with urb-s, etc.

Page 120: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

2.6 Conclusions

In this part of our study we have further pursued our characterization of the IE morphosyntactic type in its two stages (subtypes), treating from this point of view the development of the nominal system from the d-f structure of IE to the p-f structure of Latin. In the first section the description of IE noun morphology whose complex and cumulative/fusional character had been shown in 1.3, was completed by the analysis of allomorphy and its types and of morphological character of case-number suffixes. In dealing with the separate nominal categories (2.2 gender, 2.3 number and 2.4 case) we have examined their mutual interdependence. The mutual relationship between the nominal categories in d-f and p-f structures presents very clear-cut evidence of the evolutionary perspective which goes from the internally non-differentiated but complex whole to the progressive differentiation of the single form-functional distinctions (1.1). The categories of number and case were determined in d-f structure by the lexical class of gender and they were themselves word inherent semantic/referential categories. This characterization is especially important for our understanding of case semantics. The basic IE cases expressed semantic distinctions of the category-inherent (immanent) type, which was conceived as a further subcategorization of the animate class with more specific features. The paradigmatization of cases was connected with the elaboration and internal differentiation of the semantic-syntactic structure of the sentence, which resulted from the fact that the semantic-syntactic aspect had been included in the case semantics from the beginning. As part of the paradigmatization process, number distinctions became explicit, while the nominal lexemes and also the basic suffixal derivations were originally vague with respect to number.

By the differentiation of case (and number) paradigm into several strata, i.e. 1) the primitive form, the so-called casus indefinitus; 2) the form with non-marked vocalic suffix -e/-H; 3) the basic suffixal derivations with -s,

Page 121: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

104 FROM INDO-EUROPEAN TO LATIN

-m, and -i suffixes in their two variants; 4) the forms with postinflectionally applied marked suffixes -t, -CV, we reached the possibility of explaining the differences between IE languages as differences concerning the integration of the forms of 1, 2, and 4 strata into the case paradigms. The central system with basic suffixes of the 3rd stratum is common to all representative IE languages.

Latin serves as an example of paradigmatized flection, but the detailed synchronic description and characterization of the Latin nominal inflections are not given. In the section 2.5 some basic tendencies of the Latin para-digmatization are outlined.

Page 122: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

PART 3

THE INDO-EUROPEAN ORIGINS OF THE

LATIN VERBAL SYSTEM

Page 123: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)
Page 124: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

3.1 Introduction: Derivative-Flectional Morphology of the Indo-European Verb: Paradigmatization

3.1.0 In the introductory part of the book we explained the main features of verb morphology and morphosyntax with respect to both stages of flection and have characterized the main differences between noun and verb in section 2.1.6. Some of these observations will be resumed here in order to show their relevance for our approach to the development of the IE verbal system (3.1.1 - 3.1.2). Since the exposition in this part of the book does not follow a given system of verb categories, as was the case in Part 2 on the noun, we shall discuss in brief the verb categories, their morphological and morphosyntactic expression, and their semantic-syntactic relevance (3.1.3). Thus we shall obtain the necessary general foundations for the discussion on the morpho­logical semantics of IE d-f verb, including the distinction between an original injunctive-based system vs. a later present-based system (3.1.4).

3.1.1 Derivative-flectional and paradigmatic-flectional morphology of the Indo-European verb: Reconstruction of the two-class system

The morphemic structure of verbs is more complicated than that of nouns because of the greater number of categories expressed in relatively distinct formants: person, diathesis, tense/aspect, mood, and the greater variety of categorial distinctions within these categories. The expression of these categorial distinctions is distributed throughout the entire structure of the word, i.e. the verb form. The tendency towards diversification of flectional ending, with two relatively distinct and separable morphemes (tense-mood and person-diathesis), is to be observed in the paradigmatic-flectional structure

Page 125: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

108 FROM INDO-EUROPEAN TO LATIN

represented by Latin (1.3.3.2). In the Indo-European verb, the expression of morphological distinctions

is less fusional/cumulative than that of the noun, but at the same time more complex. The various implicit and explicit stem characteristics combine with the proper personal ending to express morphological distinctions (cf. perfect stem and perfect endings in Latin, etc.). The combination of internal and external inflection, and their mutually gradient character, are features which hold for both d-f and p-f structure (1.3.1.1).

By recalling the main differences between verb and noun (2.1.6) we shall characterize the basis of our approach to the development of the Indo-European verbal system. 1) In the case of the verb, the original (non-oppositional)85 verb classes, differing as to the personal endings, must be reconstructed. In the attested IE languages they are integrated into a unique verbal system with established oppositional categories, whereby the latter partially reflect the formal and semantic features of the original two-class system. The possibility of recon­structing the original two-class system is the most important result of modern IE research achieved through the analysis of the mutual relationship of perfect and middle (3.2.1). The development of the original system into an integral verbal system was a complicated process, due to the large variety of deriva-tive-flectional formations. 2) A characteristic feature of the IE verb is its large variety of derivative-flectional formations. There is a close structural affinity and historical connection between derivational and inflectional verb categories, semantically motivated by the relevance of the lexical subcategorization of verbs (see 2.1.6 sub point 2 concerning this difference between noun and verb). As a con­sequence, the suffixal derivations become the source of the inflectional categories, the latter developlng via grammaticalization from the derivational categories. The derivational suffixes are then adapted to express aspecto-temporal and modal distinctions. 3) This is reflected in the formal morphemic structure of the verb. Whereas in nouns the inflectional class markers (vocalic suffixes) tend to be fused with the case suffix, in verbs the inflectional class markers simultaneously serve as markers of mood and tense/aspect. Hence they remain distinct with respect to the person marker and can combine with various consonantal formants. As a consequence the verb has a more complicated morphemic structure and is more highly analysable.

Page 126: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

VERBAL SYSTEM 109

3.1.2 Methodological remarks

3.1.2.1 Approach to the development of the Indo-European verbal system

Our study of the development of the IE verbal system is organized in such a way that the reconstruction of the two-class system of active vs. inac­tive verb classes (together with subclasses of aspectual character) will be outlined in section 3.2, along with an examination of their integration into a system of developed inflectional oppositions. This will be followed by a discussion of the main verb categories developed on the basis of aspectual subclasses and diathetic classes, with central attention being paid to the development in Latin (as suggested by the titles of sections 3.3 and 3.4). Then, in the next section (3.5) the modal and future forms with vocalic suffixes will be examined, again with primary emphasis on Latin. Finally, we will look at the two main consonantal suffixes and their grammaticalization in Latin morphology, one of which became a widely used grammatical device in IE languages (3.6), the other remaining more restricted in its application (3.7).

In principle, I have proceeded from basic and central categories to more marked and marginal ones, and at the same time, from elementary to more specific morphological processes (person markers, vocalic suffixes, consonantal suffixes, combined suffixes). Due to the complex and allomorphic character of IE flectional morphology these markedness correlations have only relative validity. The consonantal suffixes are also involved in the expression of the basic opposition between the present and perfect stems in Latin, as in other IE languages. The iconic principle governs, in a more systematic way, the relationships of the morphological forms as a whole. This is evident, for example, in the internal organization of the Latin present and perfect systems. We shall not explicitly examine the morphological naturalness in the syn­chrony and the diachrony of the IE verb, but it is implied in both the way we treat morphological problems and in the way we conceive of the paradigmati-zation process.

3.1.2.2 The Greek/Aryan vs. the Latin verbal system as two alternative realizations of the paradigmatization process

The reconstruction of the original d-f stage, which is characterized by verb classes with different personal endings and by d-f formations with re-

Page 127: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

110 FROM INDO-EUROPEAN TO LATIN

stricted application, enables us to explain the richer system of Greek/Aryan and the more simple system of Latin and other IE languages as two separate alternatives of grammaticalization and paradigmatization. The dilemma put before us by traditional reconstruction, based on the Greek/Aryan system on the one hand and modern revisions on the other, is solved in favour of a system poorer in developed inflectional oppositions, but richer with regard to derivational processes and capacities. This part of the book represents, therefore a contribution to the discussion of the IE verb.86

The assumption about the rich form variety of IE morphology corre­sponds to the traditional view. In recent research, new important contributions to IE form variety in both verbal and nominal morphology have been made, especially in the area of accent-ablaut alternations (see for the verb esp. Narten 1968, Insler 1972). These results are in agreement with the primary importance of internal inflection in the reconstructed d-f structure of IE.

However, the consideration paid to IE languages with more simple systems of inflectional oppositions does not allow us to project the verbal system of Greek/Aryan type onto PIE and makes new interpretation of IE verb morphology necessary. The projection of inflectional oppositions developed in Greek/Aryan onto IE is invalidated by corresponding evidence in Hittite and other east IE languages on the one hand, and some west IE languages on the other. IE languages outside the Greek/Aryan area do not possess a tripartite aspecto-temporal system with present vs. aorist vs. perfect oppositions, but a bipartite temporal system, with present vs. preterite only. Further, there is also no support for the projection of semantic oppositions expressed by the Greek/Aryan middle-passive onto PIE. These basic differences will be discuss­ed in sections 3.2 (in general) and 3.3; 3.4 (with respect to Latin).

On the other hand, the concept of d-f structure with a rich form variety of d-f formations is opposed to the tendency of reducing the number of mor­phological processes in PIE represented by some modern IE research. In splte of the fact that the -ă- formation with subjunctive value is only attested in a few IE languages, it was part of the original d-f structure as a realization of a larger functional value, the IE -ă- 'injunctive'. We may assume that various d-f suffixal formations with a definite (but no longer recognizable) Aktionsart-meaning were used in an 'injunctiva', preterito-modal function, but their application was restricted to certain verbs only, in keeplng with the quasi-derivational distribution of IE formants. These formations were, to varying extents, integrated into the paradigmatized inflections of individual IE languages by the process of primary and innovative grammaticalization. Some

Page 128: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

VERBAL SYSTEM 111

of these developments and differences will be discussed in sections 3.5-3.7. In sum, the allomorphic expression of categories and categorial distinctions found in the individual IE languages may not be reduced to one original expression, but reflects the IE form variety of d-f formations. (Cf. the discussion of -ă- subjunctive and -s- subjunctive in 3.5 and 3.6, respectively).

Furthermore, our assumption that the inflectional categories developed in p-f structure by the grammaticalization of d-f formations with restricted application and specific meanings is also in contradiction with another extreme found in present IE research. In contrast to the theory of reducing the number of IE morphological processes this theory attempts to reconstruct an even greater number of morphological categories than that attested in the Greek/Aryan system (e.g. the category of stative reconstructed by Oettinger 1976 and Rix 1977, 1988 in addition to perfect and middle).

In order to understand the fact that formations with specific Aktionsart-meanings could be adapted to the expression of grammatical distinctions of aspecto-temporal, modal or diathetic character, a brief discussion of the grammatical categories and their morphosyntactic expression is necessary.

3.1.3 Morphological and semantic-syntactic categories

As in nominal morphology, the morphological forms of the verb combine with other linguistic devices to express categories at the semantic-syntactic level. This enables us to speak about the morphosyntactic expression of semantic-syntactic categories. However, the categorization at the semantic-syntactic level is neither independent of nor evidently superior to morphologi­cal categorization, both categorizations being mutually interdependent. Only general categories, such as modality, diathesis and temporality, seem to be common to different language structures, but their subcategorization depends on or reflects (according to the point of view) the morphological categories and categorial distinctions expressed by morphological processes.

Therefore, we must distinguish verbal mood as a morphological catego­ry expressed by modal formants, and modality as a semantic-syntactic category expressed not only by verbal mood but also by other means: modal particles, verbs and predicatives with modal meaning such as oportet, adverbs, verbal adjectives with modal meaning such as the gerundive, and suprasegmental (intonational, constructional) devices (see Kurzová 1987a,b). Yet, the se­mantics of the typologically dominant formants, in our case of the flectional formants, has a prominent position in the semantic organization of modality in

Page 129: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

112 FROM INDO-EUROPEAN TO LATIN

languages such as Latin. The subcategorization of modality as a semantic-syntactic category depends on the morphological meaning of verbal moods. Similarly, we must distinguish verbal voice (or verbal diathesis) as a morphological category, and diathesis as a semantic-syntactic category whose most effective semantic opposition, i.e. active vs. passive, is signalled with the help of syntactic transformations, i.e. the agent-from-subject demotion and the patient-to-subject promotion. These transformations at the same time disam­biguate the semantically ambiguous or polyfunctional verbal voice, such as the middle-passive, which is the basic expression of diathesis in p-f structure of IE languages. In the reconstructed d-f structure of PIE it was not the morpho­logical category of verbal voice, but the morphological class, which served to express diathesis. The reconstructed IE two-class system is constituted by lexical verb classes with the diathetic meaning of 'active vs. inactive'.

The dependence of the semantic-syntactic categorization on the orientation of morphological semantics is most apparent in the case of temporality and its relationship to the morphological categories of aspect and tense. If we want to join the morphological systems with primary aspectual orientation, such as the IE, Greek and Slavic systems, with the primarily temporal morphological systems, such as Latin and the modern IE languages, under the same semantic-syntactic category, the temporality must not be identified with the concept of tense as a 'deictic' category, viewed in its relation to the hic-and-nunc situation. The temporality must include all possible distinctions related to the concept of events perceived as basically temporal phenomena, which themselves can be viewed in their extension, iteration or in some other aspect relevant to the human experience of time. Thus, the different and specific Aktionsart-distinctions expressed by d-f formations can be understood in their relevance to temporality.

3.1.4 Category-immanent semantics of the Indo-European verb morphology: Injunctive-based system

Taken in the wider sense suggested above, temporality concerns basic characteristics of events and belongs to the deepest layer of grammatically relevant semantics connected with verbal predication. The necessary adaptation of the concept of temporality to cover the aspectual and Aktionsart-distinctions leads us again to the basic problem of a general orientation of morphological semantics in IE. As was explained in section 1.3.4 IE morphological semantics was of the category-immanent type with primary orientation to the features immanent (inherent) in a certain word category. Relational semantics connected with the

Page 130: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

VERBAL SYSTEM 113

syntactic relations between the words in the sentence or with the syntactic/ pragmatical relations presupposed by the sentence (tense as deictic category) was secondary.

In the verb, this original character of grammatically relevant semantics is reconstructible for all main semantic-syntactic categories, i.e. diathesis, temporality and modality. We shall anticipate here the major issues which will be discussed in the following sections. a) Diathesis in IE structure was expressed by the lexical verb classes of active and inactive verbs. The different person markers of active and inactive verbs were determined by the meaning of verbal lexemes (3.2.1). b) The basic verb form which can be called 'injunctive' (see 3.2.3.2, and note 101) had a non-actual value, it was not related to the hic-and-nunc situation. This relation could be expressed by the actualizing particle -i joined to the basic form: *bher(e)-t-i (see Brugmann 1916:593, Kiparsky 1968:45). The preterital value could be expressed by another particle which was preposed to the verb form: *e-bher(e)-t, the so called augment (see Brugmann 1916:11, Watkins 1962:113, Kiparsky 1968:45). Probably, both particles were originally applied only as optio­nal elements: as for the augment, its PIE origin is uncertain. Thus the relational semantics of deictic tenses in the original structure was expressed by secondary means, not by the typologically fundamental flectional formants. By the process of paradigmatization the injunctive-based system changed into the present-based system. The new present (with thematic formation as its basic form) included a part of the non-actual value of the original injunctive in its functional content. c) Various suffixes without exactly reconstructible meaning expressed the semantic modifications of verbal lexemes which were grammatically relevant and can be, with a certain reserve, covered by the term 'Aktionsarť 87

('Aktionsart'-formations). The same suffixal form (with thematic -e/o- suffix, with -ë- or -a- suffixes, with -s- suffix etc.) in its primary non-actual (injunctival) use could be interpreted as modal or as temporal/preterital. By the process of paradigmatization both grammatical functions were opposed to present and became expressed by specific forms. For instance, the long vocalic 'injunctives' were limited to the modal (subjunctive and future) function in Latin (whereas they were restricted to the preterital function in Baltic — see 3.5.2), and for the preterital value a new formation with the -dha- suffix was formed (3.7). Thus the tenses/aspects and moods originated in the semi-grammatical (derivational) formations of IE d-f structure, with category-immanent 'Aktionsart' meaning. This will be further explained and specified by examples in the following sections.

Page 131: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)
Page 132: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

3.2 The Active and Inactive Verb Classes: The Development from an Injunctive-Based Two-Class System into an Integral Present-Based Verbal System

3.2.0 Recent research into the IE verb has resulted in the assumption that in the oldest achievable stage of IE the categorial distinction which belongs to the sphere of diathesis was fundamental and superior to the aspectual distinction of present/imperfective vs. aorist/perfective. Two verbal categories with different series of personal endings are reconstructed: aoristo-present with consonantal endings -m, -s, -t in the singular vs. perfecto-medium with vocalic-laryngeal endings -a, -tha, -e/-o in the singular (see 3.2.3). It is, however, not so commonly accepted that this categorization too, like the aspectual one, originally had a lexical, non-oppositional character, so that originally one verbal lexeme allowed only one series of endings. It is my contention, which I share with Perel'muter (1977), K. H. Schmidt (1979, 1986) and Gamkrelidze & Ivanov (1984) that two distinct verb classes, i.e. active and inactive verbs, are to be reconstructed for IE. The inactive class in­cludes not only stative verbs, i.e. the verbs expressing states, but also the verbs expressing processes, and therefore, it represents not only a predecessor of perfect but also of middle (medium).88 In this respect my concept of inactive class corresponds to the perfecto-medium as reconstructed by Neu (1968, 1976), Meid (1975, 1979), Tischler (1982) and others.89 The classification 'active vs. inactive' belongs essentially to diathesis as a semantic-syntactic category, which, however, was morphologically expressed not by an oppositional inflectional category but by a non-oppositional lexical class in IE. In the following section I shall try to specify my conception of active and inactive verbs and to determine their structural position. This problem can be summed up under six headings (3.2.1.1 - 3.2.1.6).

Page 133: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

116 FROM INDO-EUROPEAN TO LATIN

3.2.1 Characterization of active and inactive classes

3.2.1.1 The active vs. inactive verbs are morphological classes (see 1.1.4), i.e. non-oppositional lexical classes with morphosyntactic relevance

In the reconstructed two-class system the active verb had a consonan­tal, aoristo-present series of endings, whereas the inactive verb a vocalic-laryngeal, perfecto-medial series only:

(1) active verbs - aoristo-present - actions OI aorist OI present

lsg. -m a-gă-m han-m-i 2sg. -s a-gă-h < -s ham-s-i 3sg. -t a-gă-t han-t-i

(2) inactive verbs - perfecto-medium - states, processes Gr. perfect Hitt. middle Lat. perfect

lsg. -a < - h 2 e š - J j a - h a r i vid-i < -h2e-î 2sg. -tha < - t h 2 e e š - t a - r i vid-is-ti < -th2e-i 3sg. -e/-o eš-a-(ri) vid-T-î < -e-i-t

The active class is here represented by two verbs, an intransitive and a transitive one: 1) by the verb OI gă- < *gwă- "go": OI aorist (3sg.) a-gă-t < *e-gwă-t (with augment e- as marker of the past indicative)90, cf. Gr. aorist

, Lat. perfect veni, with the nasal transferred from the present stem; 2) by the verb OI han- < *gwhen- "strike", "kill": 0I athematic present (lsg.) han-mi < *gwhen-mi (with -i as marker of the present indicative), cf. the thematic present of this root in Gr. < *gwhen-iö, cf. Lat. verb de-fen-do "defend" with the dental suffix.

Both inactive verbs under (2) Gr. and Hitt. esa are so-called perfects and media tantum, i.e. perfects and middles without the corresponding presents and actives. They have correspondences in other IE languages. The perfect *uoid-a or *uoid-ai (for Lat. and OCS, with present indicative marker -i) is attested in the following representative languages:

(3) lsg.perf. Gr. , 0I veda, Goth, wait, Lat. vidi, OCS "I know, I have seen"

Page 134: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

ACTIVE AND INACTIVE VERB CLASSES 117

As for the verb Hitt. es- and the allied verb Hitt. ki- the Greek - Old Indian - Hittite correspondences are important:

(4) 3sg.med. Gr. OI aste, Hitt. eša(ri) "he sits"

(5) 3sg.med. Gr. 0I śete/śaye, Hitt. kitta-(ri) "he lies"

We have good reason to maintain that these verbs representing perfects and media tantum are relics of the originally lexical inactive class and not isolated members of inflectional oppositions perfect vs. present, middle vs. active. 1) They exhibit archaic semantic and formal features not shared by the representatives of the developed inflectional categories of perfect and middle: a) *Uoida represents an archaic unreduplicated perfect with so called presento-perfect meaning "I know", Gr. has only this meaning, similarly Goth. wait; 0I veda has both meanings, stative "I know" and resultative "I have seen". In Latin, vidi has preterital meaning "I saw, I have seen", but we have another Latin presento-perfects: memini "I remember" with correspond­ence in Gr. presento-perfect "I have in mind, I intend"; odi "I hate". Both verbs are root-perfects of IE roots, b) Both media tantum under (4) and (5) attest the original vocalic ending -o in the 3rd sg. instead of (Hitt. esa) or beside (0I śaye/śete) the more recent 'mixed' ending -to (see 3.4.3) marking middle-passive as an inflectional category. 2) As we shall show especially in sections 3.3 and 3.4, the inflectional catego­ries of perfect vs. aoristo-present and middle-passive vs. active, if developed, are of so varied a character in the individual IE languages that it is impossible to reduce them to an inflectional opposition with common (and allegedly original) semantic definition and formal expression.

The reconstruction of verb classes with active and inactive series of endings is also 'typologically' well founded, i.e. it has its support in the verb classes attested by the general comparative research especially in the American languages (1.2.2). It is advisable, however, to use this typological-comparative evidence only in support of the existence of similar classification in IE in general, and not to force the IE structure into the pattern of some concrete language chosen at random. The definition and characterization of IE verb classes and of their structural position can only be based on the dates achieved by historical-comparative analyses and must be made by such a method of structural-typological (internal) reconstruction as fully respects these dates and

Page 135: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

118 FROM INDO-EUROPEAN TO LATIN

avoids the unjustified extrapolations and inferences 'completing' the concrete dates from different structures of other languages.

3.2.1.2 Diathetic classes and aspectual subclasses

The active verbs expressed imperfective and perfective actions, this subclassification being primary of a lexical character. The imperfective subclass was a predecessor of the inflectional category of present or present-imperfect, the perfective one was a predecessor of the aorist or basic preterite. The inactive verbs expressed not only states as Gr. "I know", but also processes as Gr. "I see, I am flashing" (with Aryan correspond­ences) or Lat. verto/vertor, "I turn" (with Aryan and Baltic correspondences). The inactive class represents therefore not only a predecessor of perfect but also of middle. As stated in 3.2.0, our concept of active vs. inactive distinction corresponds in this respect to the opposition 'aoristo-present vs. perfecto-medium' as reconstructed by Neu, Meid and others. Yet, as explained in 3.2.1.1, the active and inactive classes are considered to be lexical, non-oppositional classes in our conception.

In the following figure the subclasses of active and inactive verbs are enumerated in the second line; in the third line we have inflectional categories which can be considered as the most direct and basic followers of the respective subclasses:

(6) active inactive imperfective perfective process state present aorist91 medium perfect

Yet, this figure gives an instructive, but simplified representation of the mutual relation between the inflectional categories and the original lexical or derivative92 classes. It will be refined and partially relativized, step by step, in discussing the development of the individual inflectional categories and forms of the integral verbal system (3.3-3.8). Here only the general situation will be outlined.

In the development from the two-class into the integral verbal system, not all subclasses were realized as distinct categories in all IE languages; see especially 3.3 on the perfect. An important feature of the two-class system is the fact that the mutual relation of the verb classes active vs. inactive is of a diathetic character, whereas the relation between the two subclasses of both classes is of an aspectual character. This holds for imperfective vs. perfective

Page 136: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

ACTIVE AND INACTIVE VERB CLASSES 119

action as well as for process vs. state, state being defined as non-progressive ("I know, I have seen" for the Gr. and Lat. vidi), process as progressive ("I see, I am in the process of seeing" for the Lat. video, Gr. )93. Thus two possibilities for the state category, i.e. for perfect, are given. It may be either identified with the perfective (i.e. with the aorist) on the basis of non-progressivity (see 3.3 the Latin situation) or opposed to the perfective on the basis of imperfectivity (which is a prototypical interpretation of the stative form) and at the same time to the imperfective on the basis of non-progres­sivity (see 3.3 the Greek/Aryan situation). As for the process category, its development into the marked diathesis on the one hand and into the present formation on the other is given by this double structural relationship: the process form is of a diathetic character in interclassai relation (inactive vs. active), and of an aspectual character in its intraclassal relation to the state form.

With respect to the subclasses of the active verbs which we define as imperfective/durative vs. perfective/non-durative it should be recalled that this subclassification had no distinct marker in IE (see Szemerenyi 1987[1969], 1987b). The aspect opposition we find in Greek was formed by help of different suffixal formations marked for more specific aspectual (Aktionsart) distinctions originally.

Only simple, non-suffixal forms, i.e. root presents, root aorists, and root inactives (perfects and middles), as well as certain suffixal formations such as the sigmatic aorist and long-vowel suffixal formations, can be considered as belonging to one class of the original two-class system. The new common formations of both classes either became identical to some form of the original two-class system such as thematic present (3.2.3.4) which was identified with the athematic root present of active verb, or they developed into a new semantically distinct category such as the middle-passive with respect to the middle (3.4).

3.2.1.3 Injunctive-based and present-based verbal system

The transformation of the two-class system into the integral verbal system was at the same time connected with the transformation of the injunctive-based system into the system with central position of the present, i.e. present-based system. A decisive step in this transformation was the formation of the thematic present as a common formation of both classes (3.2.3.4). The thematic present was functionally identified with the athematic root present of the active verb and this integrated present category gained a

Page 137: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

120 FROM INDO-EUROPEAN TO LATIN

central position in the new integral verbal system. In the following figure the position of 'thematic present' (which in fact

represents the integrated present category) is indicated as the common category of both classes. The basic structural position of present as the centre of the verbal system must be respected in this general outline because it constitutes an important feature of the new integrated verbal system. Figure (7) is specified for Latin, exhibiting a proportional relation between the perfect of aoristic origin (dixi) and of the 'genuine' perfect (vidi) and the central category of present (see 3.3 for detailed exposition):

(7) imperfective perfective process state present aorist medium perfect

thematic present dixi dico

video vidi

3.2.1 .4 The meaning and function of the classification active vs. inactive: Its diathetic character

Having treated the problem of subcategorization of active and inactive classes in the preceding paragraph, we shall now consider the question of the meaning and function of this classification.94 We define the active verbs as verbs expressing intentional actions ascribed to an external agent and oriented to an external goal, whereas inactive verbs expressing processes and states have no such orientation to external actants. With respect to sentences containing animate subjects, we assume that the personal markers of active and inactive verbs serve to specify the semantic role of the IE subject which itself is defined referentially as a privileged participant to which the sentence is oriented. The active verb assigns to the subject the role of agent, the inactive verb that of non-agent. Or, we can also include sentences without animate subjects in our definition and say that the inactive verb signals that there is no agentive subject in the sentence. In other words, the identity of agent and subject does not hold for the inactive verb:

(8) active verb: subject = agent inactive verb: subject = agent

The specification of the semantic role of the subject as agent vs. non-agent belongs to diathesis as a semantic-syntactic category, the opposition

Page 138: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

ACTIVE AND INACTIVE VERB CLASSES 121

agentive vs. non-agentive being the basic diathetic opposition (see more 3.4). In IE structure this basic distinction is not expressed by means of the morpho­logical category of genus verbi, but by means of the morphological class as defined 1.1.5.1. Two series of personal affixes are morphological devices expressing the distinction 'person of active (agentive) verb' vs. 'person of inactive (non-agentive) verb', not the opposition 'agentive vs. non-agentive person' which would be applied to the same verbal lexeme in the sense of inflectional opposition.

All of the active and inactive verbs which have been dealt with so far presuppose personal subjects and assign the semantic role of the agent to their subject in the case of active verbs like 0I hanti (3sg.pres.) "strike" or OI agaţ (3sg.aor.) "go" and the semantic role of experiencer in the case of inactive verbs such as Gr. "I know", Gr. "I see". There are, however, some problems concerning the inactive verbs with non-personal or inanimate primary actants which require special attention (3.2.1.5). They must be treated before we come to a critical discussion of views on the relationship between active vs. inactive verb classes and animate vs. inanimate noun classes, and on the cooperation of person markers and case markers in expressing basic syntactic relations in the sentence (3.2.1.6).

3.2.1.5 The non-oriented, diathetically vague inactive verbs

The classification of verbs as active vs. inactive is linked with intransitivity vs. transitivity. The active intransitive verbs such as already mentioned *gwă- "go" or as *sed- "sit down" (cf. 0I 3sg.aor. asadat) are goal-oriented in the same way as transitive verbs *bher- "bear" (Lat. few, 0I bharămi lsg.pres.) or as *gwhen- "strike" (see above). They are oriented to the external effect, irrespective of whether the action affects the external object - patient or not. The active verb has agent - goal perspective, the subject of the active verb is the agent performing the intentional, goal directed action. On the other hand, the inactive verb lacks this agent - goal perspective, expressing the state or process conceived as going on without intentional goal-orientation. The subject, if present, is defined as non-agent by inactive verb, as already shown in the preceding paragraph with respect to verbs "I know, I see" and their experiencers.

The inactive verb class, however, also includes verbs allowing transitive and intransitive use which from the viewpoint of the active agent-goal oriented verb appear to be diathetically ambiguous, referring to the

Page 139: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

122 FROM INDO-EUROPEAN TO LATIN

undergoer in their intransitive use "somebody/something turns, changes, breaks", but to the actor in their transitive use "somebody turns, changes, breaks something". The undergoer can be and, in case of the verbs such as "to break" etc., regularly is represented by an inanimate noun.

I use the term 'diathetically ambiguous' only with reserve, because in the two-class system the inactive verb did not represent a marked member of the diathetic opposition, but an unmarked member of the semantic classifi­cation; thus the terms 'non-oriented' or 'diathetically vague' verb are more ap­propriate. Notice that following quite common practice the core actants of a more abstract level of sentence structure are labelled 'actor' and 'undergoer', whereas participants on a more concrete semantic level are labelled as agent, patient, experiencer, benefactive etc. This two level model proved useful in the discussion about the passive (see Foley & Van Valin 1984: 28ff.) and when applied to our two-class system it allows us to define the diathetic character of inactive verbs as a device specifying the semantic roles of the subject. The syntactic subject of inactive verbs can be either actor or undergoer, but it is non-agent in both cases.

The non-oriented, diathetically vague or ambiguous verbs are also known in the modern European languages. Being restricted in most of these languages, they form quite a large class in English (see Abraham 1987): to break, to change, to bend, to turn etc. In Latin, only individual verbs are diathetically vague, especially verto "I turn" (intr, tr), muto "I change" (intr, tr), whereby also vertor and mutor occur in intransitive use.

The diathetic vagueness or ambiguity is well attested in Hittite by middle, therefore evidently inactive forms:

(9) NINDA har-ši-in pá[r-a]š-ha-ri bread- thick-ACC break-MED.lsG "I break thick bread"

(10) MUN-aš-ma-kán GIM-an ha-aš-ši-i an-da pár-ši-it-ta-ri salt-NOM but as hearth-LOC in break-med.3sg. "but as salt breaks on the hearth"

In these sentences, quoted after Neu (1968:140), the ambiguous character of the verb pars- "break" is well documented. In the first sentence the verb is transitive "I break the bread", in the second sentence intransitive "the salt breaks". Similar pairs of sentences are attested with some other Hittite middle verbs (though Hittite also can use disambiguating means such as

Page 140: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

ACTIVE AND INACTIVE VERB CLASSES 123

causative affixes as markers of transitivity, or the reflexive particle za as a marker of intransitivity).

These diathetically ambiguous inactive verbs seem to indicate something about the character of the inactive verb in general. Lacking the agent - goal perspective as an internal component of its meaning, the inactive verb expresses the state or process in itself or assigns some stative quality or some process to its actant, without, however, being oriented to a certain par­ticipant role within the sphere of the non-agent, allowing therefore its primary actant to be the actor ("he breaks something"), or undergoer ("it breaks"). An instructive example of another type of polyorientation is the Greek verb

(pres.med.) - (perf.act.), which with a personal subject means "he sees clearly", whereas with an inanimate subject "(the light, the fame, etc.) flashes". In the first, common meaning the subject is a personal experiencer, in the second meaning an inanimate source. The verb by itself expresses the idea of shining, flashing. Cf. an example for the use with inanimate undergoer (source):

(11) "the glory flashes from wide" Pi.0.1,150

This ambiguity or vagueness which was observed in cases of some perfects, justifies views maintaining that the perfects or more widely the inactive verbs are of a similarly neutral character with respect to diathesis, as are verbal nouns. This, however, has no implications for statements about the historical origin of the perfect or its individual formants.

The following observations and ideas relating to the non-oriented inactive verbs are important for IE morphosyntax: 1) The fact that the inanimate undergoer regularly occurs with inactive verbs is motivated by the semantics of inactive verbs expressing processes, which are conceived of as not being intentionally goal-directed. The inanimate undergoer is not conceived of as an autonomous entity representing the goal of action, but rather as an internal component involved in the process of brea­king, therefore "bread-breaking" with respect to the above mentioned example. Recall the fact that the absolutive case of the inanimate noun was originally neutral with respect to the subject vs. object opposition, which was constituted for animate nouns. 2) The non-oriented verb can be paraphrased in the sense "there is breaking with respect to bread"; "there is breaking of bread with respect to me". This holds for its use in so-called event central thetic statements with backgrounded

Page 141: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

124 FROM INDO-EUROPEAN TO LATIN

nominal actants, expressed by an inanimate absolutive or by personal markers only (or by some secondary backgrounding device); see Sasse (1987), cf. 2.4.3.2; 3.2.1.5. 3) However, the whole process of breaking (of bread) can be ascribed to the personal subject expressed by the nominative marker. In these constructions, the personal actant standing in subject position is foregrounded. The function of inactive verbs of the type "to break, to change" is here identical with the function of inactive verbs of the type "to know, to see", etc. They specify the semantic role of the personal subject as non-agent. 4) By the presence of a nominative in the sentence', the second actant is unambiguously signalled as the object. This holds not only for animate nouns marked by accusative marker, but also for the absolutive case of inanimate nouns, which by the presence of the nominative subject is interpreted as object case. Hence, the constructions with inactive verbs, if both actants are explicit­ly represented by nominal constituents, also use the nominative-accusative morphosyntax, as will be briefly explained in the following paragraph.

3.2.1.6 The active vs. inactive verb classes and animate vs. inanimate noun classes: The nominative-accusative morphosyntax of Indo-European

Different oplnions have been expressed on the mutual relations of animate vs. inanimate noun classes and active vs. inactive verbs. Gamkrelidze & Ivanov (1984:267ff.) argue that the noun classes animate vs. inanimate can by themselves be defined as active vs. inactive, whereas the active and inactive verbs are selected by nouns and therefore represent the agreement-classes only. In this conception, verb classification would only duplicate the classification of nouns. Without any support in the evidence, the correlations 'active noun - active verb', 'inactive noun - inactive verb' are assumed95, in which the active and inactive variants of the basically identical verbal meaning are used in agreement with the noun category. Further, the frequent occur­rence of the inanimate undergoer with inactive verbs (of the type "put, change, break" etc.) is accentuated by Gamkrelidze & Ivanov. This, however, can be explained by the lack of intentional goal-orientation in the case of inactive verbs, hence in the framework of verbal agentivity.

According to the model suggested by Klimov (1973 a.o.) the -s nominative expressed only the agentive subject of active verbs, whereas the subject of inactive verbs was assumed to be expressed by the -0 absolutive, just as in the case of animate nouns (see contra-arguments in 1.2) In this

Page 142: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

ACTIVE AND INACTIVE VERB CLASSES 125

conception, the 'active' case (i.e. s nominative) and active verb both serve to mark the agent.

In both of the conceptions just mentioned, which were formulated in the framework of the active/ergative hypothesis, the verb classes only duplicate the semantic and functional distinction expressed by the noun morphology. In my conception, on the other hand, the noun and verb classes and the case and person markers have complementary functions expressing semantic distinctions which are not mutually reducible. The nominative of animate nouns expresses the referential role of the subject as privileged participant, which is then specified as agent vs. non-agent by means of verb classes and their person markers. The verb classes marked as active vs. inactive specify, therefore, the semantic role of the animate subject:

(12) animate subject + verb active agent

verb animate subject inactive non-agent

The inanimate participant was originally conceived of as being involved in the state of affairs expressed by the verbal predicate, and the respective nouns were not marked as to subject vs. object and actor vs. undergoer oppositions.

Therefore, in my conception, case markers and person markers appear to be in complementary relation as to their function in IE. By nominative marking the primary actant of the verb was not defined semantically as the agent, i.e. with respect to its relational semantic role, but referentially and syntactically as the privileged participant/subject. The active and inactive person markers specified the semantic role of this subject.

We have formulated our arguments against the ergative/active hypoth­esis in its nominal and verbal variant in 1.2. The restriction of nominative subject to animate nouns itself speaks against its interpretation as agent. In splte of the presence of the active vs. inactive verb classes, IE is a nominative-accusative language. The active structure in the sense of iden­tification of the subject of inactive verbs with the patient of transitive active verbs is not confirmed by the reconstructed data of IE. The patient of transitive active verbs such as OI hand "strike" or Lat. few "bear" was marked as accusative, whereas the subject of inactive verbs was marked as nominative. Both were neutralized in the case of the inanimate absolutive,

Page 143: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

126 FROM INDO-EUROPEAN TO LATIN

which, however, had no ergative as opposite form. Because of the unipersonal character of the IE verb, there was no means of marking the patient of transitive active verbs by verbal morphology. Only after the formation of the passive could the patient of transitive verbs be transferred from the position of object to the position of subject in derived constructions with marked diathesis. This derived subject was, of course, marked as nominative following the nominative - accusative structure.

Therefore, in our reconstructed two-class system, IE was a language with nominative-accusative morphology and morphosyntax. The constructions not marked for nominative - accusative cannot be interpreted on the basis of active/ergative morphosyntax. They represent, rather, elementary structures common to different structural types of fundamental syntactic relations. The assumption that they represent a common ground of different structuring of fundamental relations in the diachronic sense (see Martinet 1979, Villar 1983:35ff.) seems to have only limited value, in the sphere of so-called thetic statements (2.4.3.2). We have analysed above the constructions with non-ori­ented inactive verbs and backgrounded actants, which represent thetic statements of the event central type. This holds for the use of the inactive two-actant verb with personal marker and inanimate undergoer: "bread breaks", "I break bread" with respect to (9), (10) quoted above. In the case of the inactive constructions, too, the presence of the marked nominative of animates in the sentence resulted in the constitution of the subject - verb relation as described in 2.4.3.

Yet, this relation presupposes the possibility of the interpretation of these structures as categorical statements. Besides the nominative, verbal agreement serves as a device marking the subject. The -o variant of the 3sg. ending of inactive verb (see below 3.2.3.5) can possibly be considered as an agreement device referring to the animate subject. Therefore, in categorical statements, the inactive verbs explicitly follow the nominative - accusative structure.

The correlation between active vs. inactive verb and categorical vs. thetic statement can probably be conceived of in the sense that the active verbs are natural candidates for categorical statements, whereas with inactive verbs the thetic statements seem to be primary. The same preference relation seems to hold for animate vs. inanimate classes in the sense that the animate nouns are natural candidates for the subject of categorical statements.

Page 144: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

ACTIVE AND INACTIVE VERB CLASSES 127

3.2.2 Preliminaries to the reconstruction of the active and inactive verb classes

The systematic reconstruction of the IE active and inactive verbs has not been attempted in previous research and is also beyond the scope of the present study. Our main aim is to show the relevance of this classification to the historical explanation of the IE verbal system, with special attention to Latin development. By discussing this development, many IE verbs will be classified with more or less probability as active or inactive. The present sec­tion will examine some classificational criteria exemplifying them by certain verbs which seem to be especially instructive.

The vague semantic character of active vs. inactive classification does not allow the non-ambiguous inclusion of the verb into one of the classes. We need further semantic, morphological and morphonological criteria to make the decision more plausible. We shall discuss the specific semantic criteria and the morphological criteria by analysing the verbs chosen as instructive exam­ples. First, however, the possibility that the active and inactive verbs differed in the shape of their root will be examined.

3.2.2.1 Hypothesis on the morphonological difference between active and inactive roots

The typlcal active and inactive verbs differ in the shape of their roots or stems. Active roots are CEC roots with a vowel between two consonants. Inactive roots/stems are of the CEIC/CIC, CERC/CRC, CEHC/CHC type, i.e. they have a sonant or a laryngeal as the second radical element. Cf. the following examples of Latin presents and perfects exhibiting the expected vocalism:

(13) active inactive CEC: veko, few, tego CEIC, CERC, CERH, CEHC: vidi, liqui, verti,

genui, pegi

The verb pairs with the same basic lexical content but different semantics96, which can be classified as active vs. inactive, differ in their root shape:

Page 145: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

128 FROM INDO-EUROPEAN TO LATIN

(14) active inactive *sed- CEC sit *ās- EHC *legh- CEC lie, tr.put *kei- CEI

The inactive members exhibit a minimal form of the inactive root. There are, however, verbs which do not correspond to this rule. We have the verbs dico, iungo, hence CEIC, CEUC verbs, which by other criteria are shown to be active (semantics, -s- aorist). This can be explained by assuming that active CEIC, CEUC verbs are derived from the original inactive CEI, CEU roots, In the case of *deik- the connection with the root *dei- "hell glanzen, schimmern, scheinen" (Pokorny 1959:183) is uncertain. The status of Heug- as a root *ieu- "verbinden" enlarged by -g- is acknowledged (Pokorny 1959:508ff.).

The position of long vowel roots of the shape CEH seems to be ambiguous. Some of these roots such as *gwă- "go" or *dhë- "put" exhibit rather agentive semantics (see about *gwă- 3.2.1.4) and they can be analysed, corresponding to the CEC roots, as roots with bi-consonantal radicals. Other verbs such as *bhă- "shine"; "speak" (Pokorny 1959:104,105)97 can be considered as inactive98 and the laryngeal as corresponding by its status to the semivowel or sonant which can form the second component of a diphthong in the wider sense (CEI,CEH) or can appear in vocalized form (CI, CH) in the so-called reduced grade of a root.

There are also verbs with inactive semantics (and morphology) which seemingly have the same CEC root shape as active verb. Again, the derivation from the original inactive root is clearly reconstructible in many cases. So for the root/stem *ues- "kleiden" (Pokorny 1959:1172f.) we can reconstruct the original root *eu- "anziehen" (exuo etc., Pokorny (1959: 346). In other cases morphological processes applied to the stem show that the CEC-like structure is not original. The root indicated as *tek- "reichen" in Pokorny (1959:1057) is represented in Lithuanian by the nasal present tenkù, inf. tékti. The appli­cation of the nasal 'infix' suggests that *te-k- is a stem with -k suffix. The homonymic root *tek- "laufen" (Pokorny 1959:1059) is active, whereas another seemingly homonymic root "weben" (Pokorny 1959:1058) exhibits an -eh1/-d1 alternation, hence it is a *teHk- root.

Therefore we can postulate a hypothesis that the CEC roots with two radical consonants were confined to active verbs, whereas the final consonant or laryngeal in inactive CEIC, CERH etc. roots were suffixai 'determina­tives'.99 The second radical elements in inactive roots are sonants (semi-

Page 146: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

ACTIVE AND INACTIVE VERB CLASSES 129

vowels, nasals, liquids) and laryngeals. Yet until the IE verb has been sys­tematically analysed from this point of view, we cannot use the difference in root/stem shape as a reliable classificatory criterion.100

3.2.2.2 Representative examples of the active and inactive verbs

We shall discuss some semantic and morphological criteria which seem to be relevant for the classification of verbs as active or inactive by analysing representative examples.

(15) active verbs *gwhen- "strike" 0I pres.3sg. hánti, Gr. aor.lsg. *uegh- "carry, drive" OI aor.lsg. avăkşam, Lat. vexi *deik- "show" Avest. aor.2sg.(inj.) dăis, Gr. aor.lsg. *bher- "carry" 0I pres. lsg. bharami, Gr. pres.lsg. , Lat. fero *sed- "sit down, sit" 0I aor.3sg. asadat

vs. *ăs- "sit, be sitting" (see under (4) *legh- "lie, put" Gr. aor.lsg.

vs. *kei- "lie, be lying" (see under (5)

For the active verb category, which in the integration process had a tendency to attract inactive verbs (the active present being the central category of the integrated system) the negative evidence, i.e. absence of non-marked inactive forms is important. Positive evidence is provided by the sigmatic and reduplicated type of aorist as it is indicated for the roots *gwhen-, *uegh-, and *deik-. The primary durative verb *bher- is 'praesens tantum' in Greek and Latin, with suppletive aorist in Greek and perfect in Latin. The classification of *sed-, *legh- as active lies in the primary character of their perfective meaning vs. the primary stative character of the contrastive inactive roots. The sigmatic aorist is attested for Hegh- in Homeric Gr. The OI aorist asadat is one of the aorist forms with normal (full) grade which represent a simple thematic form of active verb (3.2.3.3).

(16) inactive verbs *as-, *kei- — see under (14), (4) and (5) *ueid- "see" Lat. vidi; cf. under (3)

Page 147: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

130 FROM INDO-EUROPEAN TO LATIN

*uert- "turn" tr.intr. Lat. verti 0I med.aor.3sg. varti (inj.), avarti (ind.) 0I med.pres.3sg. vartate vs. act.perf.3sg. vavarta intr. Lith. virstù

Lat. verto tr. Lith. verčiù

*leikw- "let, leave" Lat. liqui Gr. aor.lsg. 0I med.aor.2sg. (inj.) rikthas, 0I med.aor.3sg. areci intr. "to remain"

Lith. liekù tr. "to let"

*peh2g- "be firm, firm" Lat. pegi < pagi, pepigi tr. Gr. perf.lsg.(Doric) Gr. aor.lsg. intr.

*uelh1- "will" Lat. volui OI med.aor.lsg. avri, 3sg. avrta

*ueld- , *uelp- Gr. med. tantum lsg. *genh2- "be born, bear" Lat. genui tr.

OI perf. lsg. jajana tr. Gr. perf. lsg. intr. to med. tantum (pres.),

(aor.) OI med. aor.lsg. ajani, 3sg. ajani,jăni (inj.) intr.

Semantically (the verbs are arranged according to the root shape), two main groups of inactive verbs are represented: the psychosensory states/ processes ascribed to the personal experiencer (*ueid-, *uelh1

-); the diathet-ically vague verbs allowing both transitive actor-oriented and intransitive undergoer-oriented interpretation (other verbs). Diathetic ambiguity or vagueness is a characteristic feature and hence a diagnostic criterion of inactive verbs. The ambiguity is attested for the same form in the same language in the case of Lat. veno or Lith. liekù. The perfects of the roots *peh2g- and *genh2- are intransitive in Greek, whereas they are transitive in 0I and Lat. This can be explained only on the basis of original ambiguity characteristic of the inactive verb.

For the inactive character of morphology the following observations are relevant:

The Latin perfects are unmarked, hence compatible with inactive verbs, or they are -vi/-ui perfects, which probably were formed only from inactive verbs (3.3.4).

In 0I the verbs *uert, *leikw-, *uelh1- and *genh2- are primary media

Page 148: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

ACTIVE AND INACTIVE VERB CLASSES 131

(or media tantum as for *uelhr) with archaic Vedic medial aorists as indicated above. The basic tendency of 0I morphology was, however, to form system­atic active - middle oppositions by using the two ablauting forms of the inactive verbs and providing the full grade form with active endings and the zero grade form with middle endings (cf. dveśmi lsg.pres.act. vs. dviśe lsg. pres.med., etc.) Thus, only a few verbs attest so clearly to their inactive origin (see 3.2.3.6 on another group of 0I media).

For *uelh1- we also indicated the Greek media tantum formed from other stems of the same root. The Greek aorists are of thematic character with zero grade , intransitive - aorists or medial aorists with full grade . The correlation act.perf. vs. med.pres., well attested in Greek (see under *genh2-), is also attested for the verb *uert- in OI (with correspondence in Lat. revertï).

3.2.3 Morphology of the active and inactive verb classes: Injunctive-based system developing into present-based system

3.2.3.1 Personal endings of the active verb

For the opposition of so-called secondary forms (without -i) and primary forms (with -i) of active verbs, as well as for the comparison of active and inactive endings, the singular persons and the 3pl. present good evidence:

(17) secondary primary lsg. -m -m-i 2sg. -s -s-i 3sg. -t -t-i 3pl. -nt -nt-i

For the lpl. and 2pl. the basic endings can be reconstructed as -me, -te. The opposition of primary and secondary endings and, as we shall see further on, the inactive forms cannot be reconstructed for these persons. We have also dual endings attested in several IE languages with -ve as lpl. marker. In the following discussion only singular persons and 3pl. will be sys­tematically taken into account.

Page 149: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

132 FROM INDO-EUROPEAN TO LATIN

3.2.3.2 'Primary' forms with the postinflectional element -i: The IE injunctive

The formal difference between so-called secondary forms, which are fundamental, and primary forms, which are derived from secondary ones by the addition of the suffix/particle -i, is directly attested by the corresponding evidence of Aryan, Greek and Hittite, where endings with and without -i occur. In Latin, the difference between secondary -t and primary -ti is attested only in the 3sg. and pl. by archaic inscriptions where -t is represented by -d, and -ti by -t:

(18) -t > -d FECED fecit, KAPIAD caplat, SIED siet, sit -ti > -t IOVESAT iurat

The inverse classification of these forms as 'secondary' for fundamental and 'primary' for derivative forms in the tradition of IE linguistics was given with respect to their semantics and functions. The forms without -i have preterital and modal meaning, hence marked functions from the point of view of the paradigmatized structure of attested IE languages, whereas the forms with -i are presents: cf. the OI imperfect a-dha-m vs. (reduplicated) present da-dhă-m-i "put". However, we must assume for IE that the basic verb form which can be labelled 'injunctive'101 expressed non-actual/ideal verbal action (process or state), whereas by means of the post-inflectional suffix/particle -i the action was related to the hic-et-nunc situation.102 The injunctive had preterital103 or modal function, and, as attested in the Rgveda, was used also as a generic, non-actual present. We can assume that the -i element was probably only optional in its original application, serving as an actualizing particle. Like the injunctive, the form with -i also connects modal and temporal interpretation in its functional content, representing the indicative present. Its modal and temporal values are, however, considered as non-marked and basic in the paradigmatized IE structure (see also above 3.1.4).

3.2.3.3 Simple athematic and thematic forms

Simple athematic forms with and without -i were grammaticalized as members of an inflectional opposition only to a limited extent in IE languages. From most verbs one or both of the members were represented by derivations

Page 150: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

ACTIVE AND INACTIVE VERB CLASSES 133

formed by means of suffixation or reduplication. These derivations had originally various specific meaning belonging to 'Aktionsart' (3.1.4) and were grammaticalized, on the basis of certain common 'semantic features, as expressions of aspecto-temporal oppositions.

Cf. the following Greek and 0I examples:

(19) Gr. imperfect lsg. vs. present lsg. "say"

(20) 0I root aorist lsg. a-dhă-m vs. reduplicated present lsg. da-dha-mi "put"

Among these derivations the so-called thematic forms, i.e. forms with the consonantal endings -m, -s, -t, -nt and preceding so-called thematic vowel have a prominent position:

(21) imperfect lsg. *e-bher-om (0I abharam, Gr. ) 3sg. *e-bher-et (0I abharat, Gr. )

We do not consider it necessary to search for some phonological or morphonological explanation of the origin of this formant, which can be treated in a similar way as other simple and primary suffixes. The thematic vowel can be considered as a vocalic suffix with certain vague and no longer reconstructible meaning. It also occurs in nominal morphology, where we have 'thematic' -e/o- formants under similar morphonological conditions: stability of accent within the paradigm, no alternations between root and suffix. Unlike the consonantal suffixes of the type VC/C which themselves occur in full grade (with preceding vowel) or reduced grade (without preceding vowel) and alternate with root morphemes, the -e/o- suffix is incapable of FG - RG alternation and allows the root vowel to occur in the full grade in forms such as the 3sg. imperfect *e-bher-e-t, 0I abharat, or in the reduced grade in forms such as 0I 3sg. present tud-a-ti, 3sg. imperfect atudat "strike" or Gr. lsg. aorist "leave". However, with respect to the regularities in the form of active and inactive roots observed in 3.2.2.1, roots of the type *leikw-/*likw- are typical of inactive verbs. Thus, the 0I 'aoristic' roots of the tudati type (presents of the 0I 6th class) and the Greek/Aryan thematic aorists with reduced grade such as Gr. "leave", 0I lsg. avrtam "turn" seem to go back to the inactive verbs.

We assume that the -e/o- suffix was originally used in the active verb only, whereas in the inactive verb the long vocalic suffixes -ă- and -ê-

Page 151: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

134 FROM INDO-EUROPEAN TO LATIN

occupied its place, occurring in parallel functions (see 3.5). If is true, then the and tudati type is a secondary, more recent formation subsequent to the

formation of the thematic present as a category of the integrated verbal system.104

The distribution of -e- and -o- grade, one of the unclear points of IE morphology, seems in verbal endings to be determined by the following consonant:

(22) lsg. *(e)-bher-om lpl. *(e)-bher-ome 2sg. *(e)-bher-es 2pl. *(e)-bher-ete 3sg. *(e)-bher-et 3pl. *(e)-bher-ont

These thematic basic forms (injunctives) with 'secondary' endings -m, -s, -t, -nt were grammaticalized in 0I and Greek in preterital meaning (imperfects and aorists with augment). They occurred also with the -i marker of actual action (originally only optional), so that the following basic athematic and thematic forms can be reconstructed for IE:

(23) athematic basic form 3sg. (e)-dhě-t(i)105 "put" thematic basic forms 3sg. (e)-bher-e-t(i) "bear"

However, the thematic form with 'primary' endings of the type bher-e-ti was grammaticalized only as variants of the thematic present, which represents a common formation of both verb classes, i.e. of the integrated verbal system. Our explanation of the thematic present presupposes the reconstruction of inactive endings (see below 3.2.3.5). I will, however, deal with the thematic present endings in the following paragraph, since the thematic present is a formation which in the IE languages is allomorphic to the active present with athematic endings, if they are attested: cf. Gr. athematic presents or with nasal suffix and thematic presents etc.

3.2.3.4 The thematic present

This type of present represents a common formation of both classes, central to the verb system. Its characteristic feature is the long lsg. ending -ö (with fusion of the thematic vowel and person marker). Specific endings differing from the -es(i), -et(i) endings are attested for 2 and 3sg. in Greek and Lithuanian.106 The characteristic -ö ending or its reflex is attested in all

Page 152: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

ACTIVE AND INACTIVE VERB CLASSES 135

the representative IE languages with the exception of Hittite.107

The thematic present as a formation common to both classes does not occur in Hittite. Hittite preserves the endings of both classes in its present conjugations, but these conjugations are allomorphic inflectional classes, not semantically distinct verb classes. The distribution of -hi and -mi endings is no longer semantically relevant and probably does not reflect the original distribution.108

The thematic present endings can be explained by analogical contamination of the active and inactive verb endings:

(24) thematic present reconstruction Latin Greek

lsg. -oh2(i) linqu-o < - o h 2 < -oh2 2sg. -e-s(i)/-ei+s linqu-is < - e s i < -ei-s 3sg. -e-t(i)/-ei+t linqu-it < -eti < -ei-(t)

The starting point of this common formation was probably the identification of the 3sg. inactive -e with the thematic vowel -e-.109 The correlation -e/(-o) vs. -e-t(i) has led to the formation of the contaminated lsg. ending -o-h2(i) with thematic vowel -o- followed not by the active ending -m(i) but by the inactive ending -h2(i).110 As for the 2sg., its identity with the 3sg. should be presupposed, which can be reconstructed as an archaic feature of IE verb (see 3.2.3.5.4). In both 2 and 3 sg. the active consonantal endings were added to the underlying vocalic ending.

The historical connection of thematic present endings with inactive endings seems to be well supported by the linguistic data111, i.e. by the existence of a vocalic ending in lsg. and traces of vocalic endings in 2sg. and 3sg. The explanation of thematic present endings by contamination of two sets of attested personal endings (consonantal/active and vocalic/inactive) is in correspondence with its function as a basic present formation common to the active and inactive verb. Functionally the new present (with thematic present as basic form) included a part of the non-actual value of the original injunctive, i.e. the part which is —preterital and —modal (generic, non-deictic present).

Page 153: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

136 FROM INDO-EUROPEAN TO LATIN

lsg. -a < -h2e 2sg. -tha < -th2e 3sg. -el-o 3pl.indef. -r

3.2.3.5 Personal endings of the inactive verb

(25)

For the 1 and 2pl. no common IE forms can be reconstructed.

(26) singular persons — evidence Gr. perf. Lat. perf. - a - a - i vid-i - t h a - i s - t a - i vid-is-ti -e -ei-t vid-it

REDIEIT (see 3.3.3) Hitt. med.pres. - "sit" 0I med.pres -"say" -ha eš-ha-hari -ai bruv-e -ta eš-ta-ri -sai bru-se -a eš-a-(ri) -(t)ai bruv-e, bru-te

The singular perfect endings as documented by the Greek and Latin perfect are identical with the simplest m-p forms of Hittite. The difference between the 1st person with laryngeal and the 3rd person without laryngeal is directly attested in Hittite.

In 0I primary middle endings, only the 'mixed' ending (see 3.4.4) with consonantal element -s from the active class is attested for the 2nd person. In the 3rd person the vocalic ending is attested by some Vedic forms such as bruve, śaye (see above under 5), etc. and by the middle of perfect, type cakr-e "make". See on 0I secondary middle endings lsg. -i 3.2.3.5.2, 2sg. -thas 3.4.4.1.

The vocalism -o in the 3rd person is reconstructed on the basis of the Greek middle-passive 'mixed' -to(i) — see 3.4.4. The vocalism -o is typical of the middle, but we have good reason to assume that both variants, -e and -o, were originally used in the process category — see below 3.2.3.5.3.

Page 154: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

ACTIVE AND INACTIVE VERB CLASSES 137

(27) 3pl.indef. — evidence 0I perfect112 Lat. perfect vidur videre < *uidêri

The plural -r endings, attested also in Tocharian and Hittite preterites, contain probably the same -r element which occurs in the impersonal deagentivum (see 3.4.4) of Osco-Umbrian, Celtic and Venetic. On this basis the impersonal/indefinite 3rd person ending can be reconstructed, whose original domain was the inactive verb. In a part of IE area, the 'mixed' middle-passive endings were formed with the help of this variant of inactive ending (3.4.4.2).

The reconstruction of the inactive endings raises some problems which are briefly discussed in the following subsections.

3.2.3.5.1 The laryngeal interpretation

The usual reconstruction of the inactive personal endings suggests for the 1st person the sequence of laryngeal + vowel -e (-h2e), whereas for the 3rd person a pure vocalic ending -e is postulated. This reflects the situation in Hittite where only in the 1st person is the laryngeal represented by h, whereas in the 3rd person we have a vocalic ending without any overt traccc of the laryngeal. The interrelation of the 1st and 3rd person endings was then conceived of as H + e (and similarly tH + e for the 2nd person) vs. e. In this notation and interpretation the 3rd person was considered as the basic form and the 1st and 2nd persons as combinations of person marker (-H, -tH) + basic form (-e). This, moreover, led to the conclusion that the -o variant as reconstructed for the 3rd person should also be projected into the other persons, which then would have had the variants -h2o, -th2o, interpreted by Neu (1976) as variants used in middle function.

These assumed processes, however, do not correspond to the non-additive character of IE morphology and the underlying concept is not supported by other observable relationships between a and e. The fundamen­tals of laryngeal theory seem to be applied in an inconsistent manner in the usual interpretation of the inactive/'perfecto-mediaT endings.

3.2.3.5.2 The -d2 and -d1 variants of personal endings

In 0I the ending -i, evidently of inactive origin, is attested for both the 1st and 3rd persons: the 1st person forms represent the normal secondary middle ending, the 3rd person forms of the type aśravi, ajani "hear", "be

Page 155: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

138 FROM INDO-EUROPEAN TO LATIN

born/bear" represent the non-paradigmatized variant of the middle-passive aorist.

It can be assumed that these endings both represent reflexes of the vocalized laryngeals h2, and h1 (d2, d1).113 Because of the inconsistency of the laryngeal interpretation, only the 1st person ending was explained as a product of the vocalized laryngeal (see Kortlandt 1981, García-Ramón 1985).

3.2.3.5.3 The -e/-o variants of the 3rd person

The original distribution of the two full vowel endings (-e/-o) recon­structed for the 3rd person is not clear. In the languages with both perfect and middle as oppositional inflectional categories, the -e variant occurs in the perfect, and the -o variant in the middle. This is attested in Greek, but is also reconstructible for 0I. With respect to this distribution we can speculate that the (marked) -o variant was an agreement marker referring to the animate subject.

3.2.3.5.4 The primitive two member system of inactive endings

In the inactive system of personal endings, the 1st and 3rd persons are marked by vocalic/laryngeal suffixes only, whereas the 2nd person has a consonantal element in all formations which are supposed to go back to the inactive endings: th in the perfect, s in the middle and the thematic present. In the perfect system, the vocalism of the 2nd person follows the vocalism of the 1st person, whereas in the thematic present it follows that of the 3rd person. This situation seems to suggest that originally only two basic vocalic/laryngeal suffixes characterized the inactive forms.114

We can further speculate that these suffixes -a vs.-e (-h2e vs. -h1e) are ablauting variants of the same formants which are represented by the long vocalic suffixes -ă- and -ě- (-eh2-, -eh1-.) whose important function in the inactive system will be analysed especially in the section 3.5. The presupposed common semantic feature would be personal/dynamic for the ă suffix and non-personal/static for the e suffix.

If these long vowel suffixes are variants of short vowel ale suffixes, the reduced variants d2, d2 would be more understandable. Possibly, the laryngeal character of the -e as -h1e holds for inactive verbs, whereas the ablauting 'thematic' -e/o- vowel occurred in the active system. These considerations may suggest that the second variant of the 3rd person endings that we characterized as personal (animate) was originally not -o, but -a, i.e.

Page 156: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

ACTIVE AND INACTIVE VERB CLASSES 139

the full grade suffix 'dynamic/personal'. In this case the o vocalism must have developed as a middle marker in the active - middle oppositions of the integrated verbal system.

3.2.3.6 Other morphological processes: Stem alternation. Suffixation

3.2.3.6.1 Stem alternation: Internal inflection

The reconstruction of the original two-class system can be relevant for research on the IE stem alternations. For instance, the 0I verbs with full grade middle forms explained by Narten (1968) and her followers by acrostatic or proterodynamic accentuation correspond in their morphology (they are media tantum in part) and semantics to the inactive class115: as "sit", uh "assert", cakś "see", vas "be settled", śas "instruct", śi "lie", stu "praise".116 As the focus of our interest lies in those features of the IE verb which have direct relevance for Latin structure, we shall not try to reconstruct the original layer of IE accent - ablaut alternations.

For our purposes, only the general assumption is important that the ablauting grades of the inactive verbs were involved in the expression of intraclassal aspectual distinction (3.2.1.2) state/non-progressive vs. process/ progressive. The inflectional forms of the original inactive verb going back to the process class show a full grade vocalism in thematic active and middle presents (cf. Gr. indicative , subjunctive , middle , cf. the basic OI thematic 1st conjugation class: bhavati, patati, vartate (middle) "be", "fly", "turn") and also in Greek athematic middles such as (cf. 0I śaye, śete to root śi above), "hang", "fly", etc. It can probably be assumed — also with respect to the above 0I -e- grade middles — that the 0I zero grade middles developed in the innovative inflectional active - middle oppositions. The reduced grade occurs in the long vowel injunctives (cf. uid-ě- in Lat. video, Gr. -e- aorists such as to presents

"send", "destroy"), whose marked imperfective/progressive meaning is expressed by means of suffixation. We can, however, consider the possibility that the distinction state/non-progressive vs. process/progressive also had a (non grammaticalized, derivative-flectional) expression in the formal opposition on the type *loikw-é vs. *leøkw-o.

As follows from our treatment of the inactive roots/stems, the inactive stems whose final consonant is an original suffix (determinative) are capable of 'Schwebeablaut' (i.e. of the CEIC - CIEC alternations; see Antilla 1969) which, however, was not involved in the grammatical oppositions. They were

Page 157: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

140 FROM INDO-EUROPEAN TO LATIN

also subjected to nasal 'infixation', i.e. to the insertion of a nasal suffix between root and stem 'determinative'117, which was a rather frequently used process of forming the active (in the sense of non-marked voice) present of the inactive (in the sense of verb class) verb.

Another stem modification consisting in reduplication, was a facultative and emphatic means of expression used in both the active and inactive verb. Hence the Latin reduplicated perfects are partially aorists and partially perfects (3.3.4). The distribution of i vocalism for the present (Gr. present ) and e vocalism for the aorist and perfect (Gr. perfect ) is of iconic character. It is reminiscent of the i vs. e vocalism of the present vs. preterite particle (augment) and observable also in deictic elements.

3.2.3.6.2 Suffixation

The systematic description of active and inactive verb morphology should answer the question of what the original domain of application of the grammatically relevant verbal suffixes was and what role they played in the integration of the verb system. Only a few suggestions can be made here. 1) The question of what the original application of the thematic -e/o- suffix was before the constitution of the thematic present as a formation common to both classes can be answered with a high degree of probability. The distribu­tion of verbs with the simple -e/o- suffix and the situation in Hittite118

suggest that the -e/o- thematic vowel was applied to active verbs in the original two-class system. The thematic aorist of the type Gr. as opposed to the imperfect is a rather recent formation (see 3.2.3.3, note 104). 2) The long vowel suffixes -e-, -à- were applied originally to inactive verbs, as their distribution in cases where they do not function as generalized grammatical devices (i.e. in present stem formations) indicates. Furthermore, in some preterital formations the intransitive (Greek) and imperfective/ progressive value of these suffixes is detectable.

We assume that both the thematic -e/o- vowel and the long vowels -ě-, -ă- had very vague Aktionsart-semantics in the sense of 'more extended, more in progress'. Their use as preterital and modal/future markers may be explained on the basis of the IE 'injunctive' (see 3.1.4; 3.2.3.2, note 101). Similarly, as the non-characterized basic form, the injunctive proper, could express a non-actualized action or progress (with preterital, modal and generic-present functions as realizations of this non-actualized value) the same is to be assumed also for the forms characterized by suffixes. In this sense we

Page 158: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

ACTIVE AND INACTIVE VERB CLASSES 141

occasionally speak of the long vowel injunctives (3.5) or the injunctival functions of -s- suffixal formations, etc. (3.6). 3) The IE consonantal suffixes which expressed various lexico-derivative meanings — some of them having grammatically relevant Aktionsart-value — are in Latin and other IE languages involved in the expression of the inflectional present stem (linquo vs. liqui, cresco vs. crevi, etc.) or were grammaticalized to express the marked temporal and modal functions. It is the latter development which will interest us in the final sections of this study. The formations with aoristic -s- suffix were applied primarily to the active verb, whereas the -dh- suffix which combines with long vowel suffixes belongs to the inactive sphere. The sigmatic aorist is primarily applied to transitive active verbs. This is attested by the situation in the Rgveda as desc­ribed by Narten (see 3.3.4.1, note 128), by the distribution of the Latin sigmatic perfect and by the transitive/causative semantics of the Greek sigmatic aorist. The latter is apparent especially in oppositions of the type

"brought forth, begot" vs. "grew, sprang forth", etc. The modal -s- suffix and the possible inclusion of aoristic and modal -s- suffixes under the same suffixal formation with perfectivizing value will be discussed in 3.6.

Page 159: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)
Page 160: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

3.3 Inactive and Perfect: The Latin perfect

3.3.0 The Latin perfect and its relationship to the categories of the aorist and perfect of the Greek/Aryan verbal system is treated within the concept of the two-class system in a way which differs very consplcuously from traditional views. Languages such as Greek or 0I, where the aorist and perfect form two semantically opposed inflectional categories, and languages such as Latin, where this does not hold, are now to be considered as a result of alternative developments from the original two-class system, whereas in the traditional treatment the Latin system was regarded as a result of a secondary modifica­tion of the original system which was reconstructed after the model of Greek and OI.119

3.3.1 Greek and Aryan opposition of perfect to aorist vs. their identification in Latin and other IE languages

The Latin perfect is a formation which with its personal endings goes back to the inactive class, forming, however, the basic preterite of the verbal system of Latin. The perfect stems go back partially to the active and partially to the inactive verb (see 3.3.4). Here we faccc the first fundamental difference between IE languages in the realization of the inflectional categories as suggested in (6) of section 3.2. Only in Greek and Aryan have the aorist and perfect been realized as distinct categories standing in semantic opposition of aspecto-temporal character to each other. In other IE languages the original active preterite, labelled 'aorist'120 in our schemes (in 3.2 — (6), (7), etc.) and the 'perfect', going back to the original inactive state class, form a single category of preterite only, common to the whole verbal system.

Hittite also belongs to the fundamental group of languages not distin-

Page 161: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

144 FROM INDO-EUROPEAN TO LATIN

guishing between aorist and perfect. The active and inactive preterites did not develop into a semantic opposition in Hittite. In the preterites of -mi and -hi verbs we have only allomorphic formations without any semantic distinction (3.2.3.4).

It is typical of the Latin perfect that the double origin of this formation is very conspicuous, the sigmatic perfect corresponding clearly to the sigmatic aorist of Greek and Aryan. This was considered in traditional IE linguistics as sufficient support for assuming that the Latin system arose as a simplification of the richer system of Greek and Aryan. And if the inflectional oppositional categories were regarded as the starting point, as was the case in the traditional approach, then the innovation appeared to be unequivocally on the side of Latin. If, however, we consider the lexical, non-oppositional classes as original, then Greek and Aryan are innovative, too. In the original structure, the aorist was formed from the active verb alone and the perfect from the inactive alone. In Greek and Aryan the perfect was secondarily added to the active verb too and the aorist to the inactive one, as is exemplified by the following Greek verbs (active "show" and inactive perfecto-present

"know") and compared with Latin:

(28) active inactive aorist + perfect perfect + aorist

+ + dixi vidi

The aorist and perfect developed in Greek and Aryan into two distinct and mutually opposed inflectional categories. This opposition of perfect vs. aorist and present as a third aspecto-temporal stem (stative vs. perfective vs. imperfective, or perhaps more appropriately: ±perfective/—progressive vs. +perfective vs. —perfective/+progressive)121 represents an innovation restricted only to Greek/Aryan. In Latin any similar extension of perfect and aorist to the active and inactive class respectively did not take place, but the original distribution was in principle preserved.122 The aorist of the active verb, however, was structurally identified with the perfect of the inactive verb and they formed the components of a single inflectional category of preterite.

Page 162: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INACTIVE AND PERFECT 145

3.3.2 Semantic and morphological basis of the identification of aorist with the perfect in Latin: The central position of the present

The semantic basis of this identification of aorist with the perfect was the identification of the state achieved and of an action or process completed, the achieved state having been regarded as a result of the completed action or process (see K. H. Schmidt 1964). This identification, however, does not pre­suppose the existence of the aorist and perfect forms of the same verb. The opposition of the state achieved and of the process having its culminative point (its terminus) in this state could be realized within the sphere of the inactive verb, where the mutual relation of stative and process forms is of an aspectual character (non-progressive vs.progressive) as stated in 3.2.1.2.

As explained in 3.2.3 we assume that some intraclassal morphology preceded the formation of an integral system from the presupposed two-class system. The morphological basis of the opposition 'state achieved vs. process in progress' was provided by the stative and process forms belonging to the same inactive root. The Greek 'perfect vs. thematic present active' and 'per­fect vs. middle' oppositions suggest that the formal opposition *loikw-é vs. *leíkw-e/o could serve as (non-grammaticalized) expression of the 'state/non-progressive vs. process/progressive' distinction. Furthermore, the long vocalic injunctive is to be reconstructed as a marked imperfective/durative form of the inactive verb, cf. *uidë- in Latin present video, vides, etc. These process forms provided the bases for the present formations, whereby the pure themat­ic and long vocalic present as well as nasal present of inactive verbs are well attested. Cf. the following Latin examples:

(29) state process perfect present

pure thematic nasal long vocalic liqui linquo verti verto vidi video

The formation of the thematic present as a common category of both classes was decisive for the identification of perfect and aorist. In connection with this, the present became a fundamental category, central for the whole system of the verb. The relationship of the perfect vidi to the present video

Page 163: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

146 FROM INDO-EUROPEAN TO LATIN

was structurally identified the with relationship of the aorist dixi to the present dico:

(30) imperfective perfective process state (present) aorist (medium) perfect

thematic present dixi dico

video vidi

The common present of both classes and of the whole integrated verbal system (the athematic present and medium are not considered here) is here labelled as 'thematic present' with respect to the formation which represents the basic form of the present in languages such as Greek and Latin, i.e. thematic -e/o- + personal endings of mixed type (see 3.2.3.4), cf. e.g. verto, dico, Gr.

The presents of inactive verbs, however, also had other forms besides the pure thematic one. They were, as indicated above, formed from the long vocalic injunctives (Latin primary/non-denominative verbs of the 1st and 2nd conjugations) and also the nasal present (thematic in Latin, but athematic in Greek and Aryan) appears to be a common formation of both classes. All these presents could express the process which had its terminal point in the state expressed by the corresponding perfect and this was decisive for the identification of the perfect with the aorist of the active verb, which had a similar relationship to its respective present. Here the completed action expressed by the aorist represents the terminal point of the action in progress expressed by the present. Both forms became functionally equivalent, forming components of a single category of the preterite. This holds not only for Latin, but also for other IE languages not participating in the Greek/Aryan innovation, i.e. in the formation of the aorist from the inactive roots and the perfects from the active ones and in the development of aorist and perfect into two distinct inflectional categories. The different system of Greek and Latin is exemplified by the active verb Gr. "show", Lat. dico "say" and by the inactive verb Gr. Lat. linquo "leave" in the following diagram (the medium is not considered here). The innovative forms of Greek are marked by preceding +.

Page 164: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INACTIVE AND PERFECT 147

(31) active inactive imperfective perfective process state present aorist (medium) perfect

thematic present +

dico

linquo iqui

It is the thematic aorist that represents here the innovative formation of inactives in Greek and similarly in Sanskrit, cf.Gr. (lsg.) and 0I (3sg.) aricat, avidat. The innovative character of the thematic aorist with a reduced grade is also confirmed in new research from other points of view (see note 104).

We therefore need no longer assume that the same process we observe in Greek (see Chantraine 1927, K. H. Schmidt 1964) was realized in the prehistoric development of Latin, i.e. that first the perfect with the so-called resultative function was formed from the active transitive verb (cf. from and analogically the aorist from the inactive verb (cf. from

from etc.) and then the functional identification of both categories took place (with subsequent elimination of one of the synonymous forms). Latin founded its preterite on the original distribution of the perfect and aorist in the reconstructed two-class system. Both forms standing in the structural relationship to the present as a common category of both classes were functionally identified on the above defined semantic basis. We shall now specify this general picture by providing an analysis of Latin perfect forms with respect to the active and inactive morphology as outlined in 3.2.3.

3.3.3 The Latin perfect endings

The personal endings of the Latin perfect123 in the relevant forms of sg. and 3pl. go back to the inactive endings; see 3.2.3.5 under (25).

(32) Latin perfect endings lsg. -i < -a-i < -h2e-i vidi 2sg. -isti < -ista-i < -isth2e-i vidisti 3sg. -ft < -e-i-t vidit

-it < -eti vidit

Page 165: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

148 FROM INDO-EUROPEAN TO LATIN

lpl. -imus < -omos vidimus 2pl. -istis < -istes vidistis 3pl. -ere < -ěr-i videre

-erunt < -ër-ont viderunt -ěrunt < -ër-ont viderunt

< -is-ont?

The sg. endings represent the inactive endings followed by the particle -i (3.2.3.2) which is probably present also in the original -ere variant of 3pl., not accommodated to the generalized 3pl. ending -ont. In 3sg. the original perfect ending is -eit > -it, attested by inscriptions (REDIEIT) and by prosody. The ending -it is a generalized form of the 3sg. act. ending; lpl. is a generalized pres. act. ending; 2pl. is based on 2sg. with final -s introduced according to the model of the 2pl. pres. act. ending -itis < -etes.

It is the 2sg. and 3pl. which need a more essential commentary. The 2sg. is supposed to contain the element -is- which is usually characterized as an 'aoristic element' occurring also in the 3pl. variant -ěrunt < -is-ont and in the derived forms of the Latin perfect system such as the future videro < *uoid-is-o, the subjunctive viderim < *uoid-is-im, etc. The identity of -isl-er contained in the derived perfect stem with -is-/-er- element in the personal endings of 2sg.pl. and 3pl. (variant -ěrunt) is, however, based only on uncertain speculation.124 It is rather implausible that the marker charac­terizing a derived stem would be transferred to the basic perfect stem, and at that to some personal forms only. As for the 2nd person, we have -st(h)a ending as a regular preterital ending of Tocharian and Hittite and we find it also in the individual archaic forms in 0I and Greek.125 Therefore, the -s-element is to be explained as part of the personal ending and not as going back to the aorist suffix. 126 As for the 3pL, note that the -is- form with s preserved is not attested, since the so-called rhotacism (VsV > VrV change) was accomplished in the prehistory of Latin. The element -is is only reconstructed here on the above mentioned uncertain basis, which was supported by the tendency to search for the aoristic elements in the Latin perfect endings. Note further that the variant -ěrunt occurs only sporadically and in verses. The normal classic ending -ěrunt goes back to the original -ere accommodated to the pres. act. endings; -ere is attested in the archaic and archaising Latin.127 As stated in 3.2.3.5, this ending represents the original inactive 3 person ending with -r- element used as an impersonal/indefinite deagentivum.

By eliminating the 'aoristic' -is- from the system of Latin perfect

Page 166: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INACTIVE AND PERFECT 149

endings, we reached a more consistent system, based on the inactive endings only, which were accommodated to the basic, non-marked active endings in the integrated verbal system. No underlying aoristic element participates in the formation of Latin perfect in our conception.

3.3.4 The Latin perfect stems

In accordance with the complex character of flectional expression typlcal of verb morphology (3.1.1), the opposition of 'perfect vs. present' is marked both by endings and by stem. Both devices are complementary in the sphere where one of the formal characteristics is absent, i.e. either the perfect and present endings (as in 3sg. and 1 and 2 pl.) or the perfect and present stems are identical. The innovations leading to the identity of personal endings were rendered possible by the predominant formal marking of the perfect vs. present stem opposition. The originally larger domain of the non-marked perfect stems was restricted in the development of Latin, especially by means of the morphological (analogical) lengthening of the stem vowel or by the reduplication.

In accordance with another typlcal trait of flectional morphology, several allomorphic morphological processes participate in the expression of perfect stem. Thus we have suffixal -si and -vi/-ui perfects, reduplicated perfects, perfects with lengthened stems (type veni, with subclass going back to diphthongs, type vidi), and non-marked perfects identical or non-identical with presents. We will begin with both the suffixal formations which are symptomatic with respect to the double source of the Latin perfect: active -si perfect/aorist (3.3.4.1) and inactive -vi/-ui perfect (3.3.4.2).

The positive marking of the perfect stem is not a unique morphological characteristic of the perfect vs. present stem distinction. There is also a positive marking of the present stem by suffixation in the presents like inf. dom-a-re vs. perf. dom-ui, inf ven-i-re vs. perf. ven-i, by infixation in the nasal presents like frango vs. perf. fregi, or by reduplication in the presents such as gigno vs. perf. gen-ui. This bilateral marking reflects the fact that the respective formants did not originally serve to the marking of present vs. perfect, but were d-f formants with a specific 'Aktionsart' value. From the perfect formants only the -vil-ui perfects are regularly derived from the stems with present characteristics: cf. inf.pres. vac-a-re, del-e-re vs. perf. vocavi, delevi.

Page 167: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

150 FROM INDO-EUROPEAN TO LATIN

3.3.4.1 Sigmatic perfects

The Latin sigmatic perfect goes back to the sigmatic aorist/preterite of the active verbs. Characteristic of the Latin sigmatic perfect is the fact that this formation is restricted to the original domain of the application of the sigmatic aorist, as attested also by the situation in Vedic Sanskrit.128 The Latin sigmatic perfect is formed from primary verbs with stems terminating in a consonant. In Latin, we have no extension of the sigmatic formation to so-called set bases, i.e. stems with final -H (d), or to the derived stems with a long vowel, which would be similar to Greek and 0I development (cf. Gr. aor. vs. Lat. perf. plevi, cf. Gr. aorists such as , , etc.). This is in keeping with the fact that unlike Greek no functional opposition of aorist to perfect was constituted in Latin, which would motivate the innovative extension of aorist to inactive verbs.

The sigmatic perfects are formed especially from the typical CEC roots such as vexi, texi, rexi (with lengthened stem which is found also in 0I, cf. lsg. aor. avăkşam corresponding to Lat. vexi). In part, the stems CEIC/CIC exhibit the sigmatic formation: cf. duxi to duco, dixi to dico, iussi to iubeo, etc. The semantics of these verbs is active/agentive (see 3.2.2.1).

3.3.4.2 Perfects in -vi/-ui

Though this formation has no satisfactory explanation and no direct correspondence in other IE languages, it is beyond any doubt that it belongs to the original perfect, i.e. inactive state category. In the suggested explanations the formant -u is deduced from some formation within the perfect proper (-u in 0I perfect of the type dadau, H with labial element (ET) as presupposed for the lsg. perfect ending, -ues participle, etc.— see Leumann 1977:596ff.). It is this formation which represents the perfect of the Latin productive long vowel classes amavi, delevi, audivi. In Greek the -KOL perfect has a corresponding position.

Note again the difference between Latin plevi and Greek vs. , Latin amavi and Greek vs. .In Latin the -vi/-ui

formation is the basic preterite. In Greek the -KOL formation represents the perfect proper, which is opposed to the aorist.

Page 168: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INACTIVE AND PERFECT 151

3.3.4.3 Reduplicated perfects

Reduplication was not a formal characteristic of the original perfect/ state category. This device was grammaticalized as a characterization of the marked 0I and Gr. perfects, standing in opposition to aorists. The redupli­cation was not restricted to the perfect proper; we have also reduplicated aorists, which within the system of the Greek aorist are of a clear-cut tran­sitive/causative type.129 Respecting this we cannot designate all reduplicated forms of the Latin perfect as original perfects. E.g. tetigi, a perfect from CEC root, is aoristic, as indicated also by the Hom.Gr. aorist (part.) Otherwise, reduplication was used in the Latin perfect system as a device compensating for the absence of other markers in the stems of the type CERC, CENC (where the lengthening was not applicable), cf. tutuli, tetendi, peperci.

3.3.4.4 Perfects with lengthened stems

The long vowel perfects with -<e-, -â-, -ö- should be divided into two groups: 1) The perfects with alternating long -ē-, -a- vs. -à- are the original inactive verbs: feci vs. facio, cepi vs. capio, fregi vs. frango, pegi vs. pango. The diathetic vagueness (3.2.1.5) is attested by the difference between Lat. pegi (< *pagi) and Gr. "I stiffen something" vs. "I am stiff". 2) The perfects with -ë- vs. -ě- (-ö- vs.-ö-) alternations probably go back to active verbs: lēgi vs. lego, sēdi vs. sedeo, fodi vs. fodio. In the case of vēni we can consider the possibility that two root/stem forms are contaminated (*gwă- and *gwem/n-, with analogical accommodation of the original -a- to -ě-of present or to -ë- as most extended perfect marker, cf. also pagi > pegi). Even if we can find some further motivation of the long vowel in the individual cases, on the whole the -ë- vs. -ë- alternation is to be explained as a morphological device employed within the active system. Unlike the long vowel of the first group, which represents the original full grade -eH vs. reduced grade -H (d), the long vowel in the second group is a secondary alternant.

Both types of the Latin long vowel perfects (1 and 2) have cor­respondences in the Germanic preterites: cf. Goth. sëtum, qëmun, brekum (lpl. preterits of the verbs "sit", "come", "break"; see Schmid 1985 on this Latin - Germanic agreement).

Page 169: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

152 FROM INDO-EUROPEAN TO LATIN

The Latin perfects with a long -f- or -ü- going back to the diphthongs -oi-/-ei- and -ou-/-eu- are original inactive state/perfects with typlcal CEIC, CEUC roots: vfdi vs. video, liqui vs. linquo, rūpl vs. rumpo, fūdi vs. fundo, etc. Semantically the characteristic diathetic vagueness or ambiguity is reconstructible in some verbs, cf. OI intr. rupyati "ache, feel spams" with Lat. rumpo - rupl "break".130 The nasal presents as the characteristic pre­sents of inactive tr. verbs are attested in this group of perfects.

The long vowel perfects together with the reduplicated perfects, both going to inactive as well as to active verbs, could play a certain role in the process of fusion of the perfect and aorist into one category.

3.3.4.5 Unmarked perfects

These are represented by several verbs of the type CERC where the conditions for reduplication were absent: verti, defendi. Probably they are a remainder of a larger group which was reduced by the extension of the formal markers of perfect or present stems.

3.3.5 The Latin perfect system vs. present system

The perfect indicative which we have dealt with so far is the centre of the whole perfect system, in the same way that the present indicative is the centre of the present system. We shall analyse other morphological forms of the perfect system in 3.6, as it is given by the organization of our exposition on verb morphology. Here only a few general thoughts on the mutual rela­tionship and internal organization of the perfect and present systems will be expressed as long as they are relevant to our characterization of the Latin p-f structure.

The relationship between the perfect and present systems has been much discussed since Meillet (esp. 1977:28ff.) showed the striking parallelism in the internal organization of the two systems:

(33) parallelism of the present and perfect systems videbam video videbo videram vidi videro viderem videam vidissem viderim

Page 170: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INACTIVE AND PERFECT 153

These systems represent a particularly good specimen of a highly or­ganized Latin paradigmatic structure. As has been clearly set out in recent research (see esp. Serbat 1980, Chodorkovskaja 1981, Pinkster 1983), this structure has different semantic/functional basis than the Greek present vs. aorist opposition, based primarily on aspectual distinction. As I do not intend to follow the results of the semantic-syntactic analysis in detail, I will concen­trate on what is relevant to our considerations about morphological semantics and about the difference between alternative morphological verb systems.

What should be noticed in the first placcc is the different position which the Latin perfect and the Greek aorist occupy in their respective systems. While the Latin perfect stands in direct opposition to the present, the Greek aorist is in direct opposition to the imperfect. The aspectual distinction perfec­tive vs. imperfective combines in Greek with the distinction past vs. non-past as expressed by the personal endings and augment:

(34) the position of Greek aorist past non-past

imperfective imperfect present perfective aorist

In Latin two types of temporal distinctions are in play: the absolute tense distinction holding between the members of the same system and the relative tense distinction (noted as pro-) holding between the members of the opposite system which are in direct mutual opposition.

(35) Latin absolute and relative tense oppositions past present future imperfect present future videbam video videbo

pro-past pro-present/past pro-future pluperfect perfect futurum exactum videram vidi videro

As for the notion 'relative tense' it should be stated that the point is not only a purely temporal anteriority, but also various connotations connected with it which partly differ in the individual dimensions of absolute tenses. Note further that the representative of the absolute tense should not always be explicitly present in the context, but only implied. The anterior future (futurum exactum) often has an emphatic affirmative value as in the following

Page 171: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

154 FROM INDO-EUROPEAN TO LATIN

example from Plautus:

(36) Pinacium: Cape Mas scopas. "Here, take that broom." Gelasimus: Caplam. "I'll do it." Pinacium: Hoc egomet, tu hoc convorre.

"I'll sweep here and you sweep there." Gelasimus: Ego fecero. "I am just doing that."

Plaut.Stich.351

What does the notion of anteriority which accounts for the mutual relationships between the members of both systems mean with a view to the relationship between the basic forms, i.e. present and perfect indicatives? It means that the perfect expresses the past action seen in its relevance to the present situation. This was shown by a comparative analysis of perfect and imperfect and formulated particularly clearly by Chodorkovskaja (1981).

With this, however, we again come to the original perfect as an important component of the Latin perfect. By the definition of the perfect as a past tense with relevance to the present situation the continuum is suggested between the past perfect and the so-called present perfect of the verbs like memini, odi.

To sum up, the Latin perfect standing in direct opposition to the present differs from the Greek aorist which stands in direct opposition to the imperfect. It is the Greek perfect which has a direct relation to the present, corresponding to the Latin perfect in this respect.

(37) position of the Greek perfect past non-past

imperfective imperfect present future

perfective aorist non-progressive pluperfect perfect perfect future

In neither of its basic components, i.e. the active preterite/aorist and inactive state/perfect indicatives, does the Latin perfect go back to the perfective aspect. The active aorist was defined as a past tense in languages like Latin, not as aspect. The anteriority feature goes back to the non-progres-sive/stative value of the perfect proper.

We have, however, another situation pertaining to the other members of the Latin perfect system. There the anteriority is probably based on the original

Page 172: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INACTIVE AND PERFECT 155

perfectivity expressed by the suffixal -is- formation. Note, however, that this expressly marked perfectivity is different form the grammaticalized non-marked perfectivity of the Greek aorist. Nevertheless, the mutual relationship between the Latin subjunctives and infinitives of present vs. perfect is comparable to a similar relationship between the Greek subjunctives and infinitives of present vs. aorist. It is only under the pressure of the basic indicative system (with pluperfect and futurum exactum as anterior tenses and with the indicative perfect defined as pro-present) that we can also interpret the relationships between subjunctives and infinitives as temporal (i.e. anterior vs. non-anterior/absolute) in Latin whereas as aspectual (i.e. perfective vs. imperfective) in Greek.

(38) parallelism of the Greek and Latin subjunctives and infinitives of present vs. perfect

imperfective subj. absolute

Greek Latin pleam

imperfective inf. absolute

plere

perfective subj. anterior

pleverim

perfective inf. anterior

plevisse

The innovative Latin subjunctives of the imperfect and pluperfect were formed as a result of the anteriority relationships basic for the perfect vs. present systems, i.e. as part of the structure based on relative tenses.

With the double origin of its basic components (active preterite/aorist + inactive state/perfect) and with its derived perfect stem marked by the suffix -is-, the Latin perfect system presents a good example of the para-digmatized system which developed by the grammaticalization and integration of various formations with originally distinct semantics.

3.3.6 Inactive and perfect: Conclusions

In the basic scheme under (6) we have indicated the main derivational lines going from the reconstructed active and inactive classes and their subclasses to the inflectional categories which we find in IE languages. In this scheme the

Page 173: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

156 FROM INDO-EUROPEAN TO LATIN

perfect and the middle/medium (middle-passive) are indicated as the main and most direct derivations from the state and process subclasses of the inactive verb. Analysing then, in the section 3.3, the Latin perfect in comparison with the Greek perfect we have attempted to define their relation to the categories implied in the classes and subclasses of PIE verb. Both morphological forms appeared to be the result of an alternative development of the two-class system. In Latin the perfect as a derivation from the state class was identified with the 'aorist' as a derivation from the perfective action class, which in languages such as Latin was realized as an aspectually neutral, but non-durative past tense. Thus the category labelled as perfect in Latin is a continuation of both active (non-durative) preterite and inactive non-progressive category. In Greek the perfect developed into an inflectional category distinct from the aorist and stood in semantic/functional opposition to it. This was achieved by the innovative formation of perfects from active verbs and aorists form inactive verbs. The different inflectional systems of Latin and Greek are, therefore, to be considered as a result of the alternative, in both cases innovative, development of the original two-class system.

What I would like to emphasize here is the inadequacy of the widely accepted oplnion holding that the Greek perfect is closer to the original state category than the Latin perfect is. This statement only has a limited value. Forming its perfect vs. aorist oppositions, Greek is based on the original semantics of the state class, so that some features of the original inactive state class appear in the Greek perfect as opposed to the aorist (intransitive/stative character of the perfect). However, in the perfect vs. aorist oppositions of the verbs expressing actions, the original semantics of the perfect was changed (with the resultative notion prevailing) and it resulted in the expressive formal transformation of the original perfect: the reduplication and stem alternations were used as the systematic markers of the oppositional category and a specific perfect suffix -ka was formed. With particular respect to generalized reduplication we can say that the Greek perfect is formally more characterized and more transformed in comparison with the reconstructed state class/category than that part of the Latin perfect which derives from the state category (non-reduplication, less marked suffixation etc.). In the case of the perfect it is evident that the formation of the morphological opposition resulted in the transformation of the category which enters in the newly formed opposition. This is an important prerequisite for the understanding of the relationship of the middle passive to the inactive, which will be examined in the next section. The fact that the middle-passive developed as a category in opposition to the active, whereas the original inactive was a non-oppositional class, must be respected in all its consequences.

Page 174: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

3.4 Inactive and Middle-Passive: The Latin deponent/passive

3.4.0 In our survey of verb classes and inflectional categories in (6) above, we indicated that the middle is an historical follower of the inactive/process class. As already explained, this simplified plcture serves as a general orienta­tion only, ascribing to a given subclass the inflectional category most directly reflecting its semantics. The middle, if developed, represents a semantic continuation of inactive/process class, exhibiting the features 'non-intentional' and ' introvertal' (3.2.1.4).

The middle function, however is not attested as an oppositional function of morphological diathesis in all IE languages. There is no inflectional opposition 'active vs. middle' in Latin. The Latin -r form marks a non-oppositional lexical class of deponents, which reflects the original inactive/process class131, and further the oppositional inflectional category of verbal voice (diathesis), exhibiting passive and direct-reflexive functions only, and not exhibiting those functions typlcal of the Greek middle (3.4.3.3).

The marked morphological diathesis (verbal voice) which we tradition­ally call the 'middle-passive' (in splte of the restricted occurrence of the oppositional middle outside Greek and Aryan) is extremely varied in the IE languages, when seen from both a formal and a functional point of view. Our discussion will concentrate on two main theses: 1) the middle-passive is of inactive origin; 2) the Greek/Aryan middle-passive cannot be projected onto IE.

3.4.1 Inactive origin of the middle-passive: Preliminaries

Desplte their diversity, the middle-passives of the individual IE languages have a common origin in the IE inactive verb class. The distinction

Page 175: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

158 FROM INDO-EUROPEAN TO LATIN

between active vs. inactive classes was of a diathetic character, as shown especially in section 3.2.1. Diathesis as a semantic-syntactic category (3.1.3) was not expressed by the morphological category of diathesis (verbal voice) in IE, but by the morphological class of active and inactive verbs. Both the IE morphological class of active vs. inactive verbs and the morphological category of diathesis of the individual IE languages are involved in the fundamental diathetic distinction of agent vs. non-agent. However, the marked term of morphological diathesis, i.e. the middle-passive, is positively marked for 'non-agent', signalling that the subject of the verb has a semantic role differing form that of the agent, the latter role being non-marked and primary. On the other hand, in the IE two-class system the inactive class was not a positively marked term of the diathetic distinction. On the contrary, if the distinction between the positive and the negative (neutral, non-marked) terms can be applied to the active and inactive verb classes, we are even more justified in classifying the inactive class as non-marked, especially with respect to the non-orientedness of the inactive verb whose primary actant can either be the actor or the undergoer in a one-actant construction (see 3.2.1.5; also, the primary thetic character of statements with inactive verbs confirms non-markedness).

3.4.2 Discussion of alternative explanations

The diachronical relationship between middle-passive and the form characterized above as 'inactive' (with -h2e, -th2e -e/-o endings in singular — 3.2.1.1; 3.2.3.5) is a complicated one and was realized in several variants in the individual IE languages, as will be explained later. In part of modern IE research this relationship has not been accepted.132 The proposal was made to define the form with vocalic ending -o(i) in 3sg. as a stative category. Oettinger (1976) projects both the 'stative' and the middle-passive onto IE as two distinct inflectional categories which are independent of each other.133

Within the framework of a two-class system of active vs. inactive verbs, Perel'muter (1977) defines the inactive class as a stative class and forerunner of only the perfect. The middle, according to Perel'muter, is of independent origin and its primary function is the direct-reflexive function.134 However, the direct-reflexive is only one of the realizations of the oppositional middle-passive. Other oppositions cannot be easily explained on the basis of a primary reflexive function (passives of the type caeditur "he is beaten", oppositions like "terrify" vs. "fear", "make to go,

Page 176: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INACTIVE AND MIDDLE-PASSIVE 159

carry" vs. "go" (3.4.3.1), where the active/transitive member seems to be secondary). All these functional realizations are understandable as different results of the process of forming the oppositional inactive counterparts to active verbs and the active counterparts to inactive verbs.

Another diachronical explanation of the middle, also based on the two-class system, has been proposed by Gamkrelidze & Ivanov (1984:267ff.). They explain the middle-passive endings as having arisen from both active and inactive forms through addition of agglutinative elements (see below about middle-passive endings). As for the semantics, no connection exists, according to this conception, between the middle-passive, which is defined as centripetal (introvertal) diathesis, and the inactive, which is conceived as an agreement category reflecting the nominal 'inactive' (1.2.1.5).

However, a semantic connection between the middle-passive and the inactive cannot be denied. Whatever differences the marked middle-passive shows in its various realizations and however greatly it differs from the non-marked inactive, both the middle-passive and the inactive share a common basic diathetic function, indicating that the subject (primary actant) is not identical with the agent (subject = agent 3.2.1.3). This, together with the formal relationship between the middle-passive and the inactive, serves as a sufficient basis for explaining the middle-passive as an inflectional category developed from the inactive class.

3.4.3 Trom inactive to middle-passive': Semantic analysis

The development of a diathetic inflectional opposition (with middle-passive as a formally and semantically marked member) from the diathetic distinction 'agent vs. non-agent' expressed by the verb classes was a complicated process. Its realization at its very beginning only in part belongs to the common IE period. The marked diathesis, if attested, has a very different character in the individual IE languages. We will contribute to the description of this development from the inactive to the middle-passive by characterizing the clear-cut difference between Greek/Aryan and Latin middle-passive (deponent/passive).135 These characteristic features will be evaluated in the framework of broader IE relations and the traditional view, according to which the Greek/Aryan system was projected onto IE, will be challenged.

The whole development is complicated by the fact that the middle-passive was not the sole successor of the inactive/process subclass. The

Page 177: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

160 FROM INDO-EUROPEAN TO LATIN

following scheme will suggest the development of the process subclass, which only partially resulted in marked diathesis:

(39) inactive process

thematic present medium tantum middle-passive

The inactive (as a non-marked category) became simultaneously a part of the active present, being identified with the active in part of its use. The transition from the non-marked inactive/process subclass to the middle-passive was characterized by the development of semantic and formal markedness. The marking of the subject as a non-agent by the marked middle-passive was connected with various specific semantic features. The IE languages differ in the semantics and functions of their middle-passives and these differences are characteristically related to formal differences, as we shall show further on.

3.4.3.1 The Greek middle-passive vs. the Latin deponent/passive: Syntactic diathesis and semantic version

The semantic differences between both classical languages in their morphological diathesis are very great. The Latin -r form as opposed to the active form has passive and direct-reflexive functions and personal or impersonal character. We not only have personal passives in Latin such as damnatur, datur, formed from transitive verbs, but also impersonal passives such as itur, (bene) valetur. In all of its usages the oppositional -r form is accompanied by syntactic derivation. Syntactic passivization has two components: 1) demotion of the agent from the subject position; 2) promotion of the patient from the position of transitive object to that of subject. Agent-from-subject demotion takes placcc in both the personal and impersonal passive (aliquis it → itur, aliquis damnat Socratem → Socrates damnatur lab aliquo/), whereas patient-to-subject promotion in the personal passive only (aliquis damnat Socratem → Socrates damnatur). In the direct-reflexive function patient-to-subject promotion occurs without agent-from-subject demotion. The participant occupying the subject position has both agent and patient roles in the case of the direct-reflexive function.136

Latin sentences with marked diathesis are therefore derived from their corresponding active sentences by the syntactic derivational process of agent demotion ± patient promotion. We can define Latin morphological diathesis as syntactic diathesis.137 This does not hold for Greek morphological

Page 178: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INACTIVE AND MIDDLE-PASSIVE 161

diathesis, which should be defined as 'semantic version'138 with respect to its primary use. Syntactic diathesis is a secondary function of the Greek middle-passive and is structurally bound to the primary use of the middle as semantic version (introvertal diathesis). We shall not discuss here the question of how the primary character of the Greek middle-passive as semantic version manifests itself in its secondary use as syntactic diathesis, i.e. in passive and direct-reflexive functions (see Kurzová 1987 on the differences between Greek and Latin in the passive and direct-reflexive functions). Here I will discuss those uses of the Greek middle-passive where its character of semantic version is clearly observable (see especially Perel'muter 1977:137ff. and Rijksbaron 1984:126-157 for discussion of the types of active vs. middle oppositions). Consider the following verbs with active vs. middle oppositions:

(40) active vs. middle as semantic version "marry" (of a m a n ) " m a r r y " (of a woman) " t e r r i f y " " f e a r " "make to go, c a r r y " " g o "

"counsel, a d v i s e " " c o n s u l t , ask advice" "prepare s.t." "get ready, prepare

s.t.for oneself'

In all of these cases, the sentences with middle forms are not syntactic derivations by agent demotion and patient promotion of the corresponding active sentences. The middle is not syntactic diathesis but semantic version. The active and middle forms express two semantic variants (versions) of the same basic meaning of the verbal lexeme. The semantic difference between these variants has diathetic character: subject = agent in the active, subject = agent in the middle-passive.

Also in cases where a passive or direct-reflexive interpretation is possible, the character of the respective oppositions as a whole cannot be determined by way of these interpretations. Cf. the following examples of a passive or direct-reflexive interpretation:

(41) "I take to wife" "I am taken to wife" (passive interpretation)

"I give myself to wife" (direct-reflexive interpretation)

"I put to flight" "I am put to flight" (passive interpretation)

Page 179: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

162 FROM INDO-EUROPEAN TO LATIN

"I marry" (of a woman) is not merely the result of a syntactic derivation of agent demotion and patient promotion from

A passive interpretation is possible for certain uses of the verb only. Yet, "I fear" is not derived from

. The same basic lexical meaning exists in two variants (versions) with different actant roles: agent vs. experiencer in vs. , agen-tive vs. non-agentive actor for vs. •

3.4.3.2 The origin of the oppositional and marked middle-passive in the non-oppositional and non-marked inactive

The middle-passive has, in contradistinction to the original inactive, a marked character. The subject is not merely signalled as a non-agent by the middle-passive. A specific relationship exists between the verbal process and the subject and can be described with the help of various terms and definitions. The term introvertal diathesis adequately expresses the internal relationship between the verbal process and subject.139 Another way of describing the positive markedness connected with the middle is to define this relationship by way of nominal semantic roles. For most middle verbs it is possible to specify the participant role as experiencer (as with ) or benefactive (as with or ). In many cases, however, the subject of the middle form is non-agentive actor with no marked experiencer or benefactive role. The subjects of "I go" or "I lie"140 can be characterized as non-agentive actors, whereas the subjects of their active counterparts "I carry", "I cheat someone" are agents. Therefore, the non-marked character of the original inactive survives to a marginal extent in middle usage.

The diachronic connection between the middle and the inactive is confirmed by this semantic continuum of non-marked and marked non-agen-tivity. The primary character of the non-oppositional inactive can be demonstrated for many Greek active vs. middle pairs. The verbal stem , *bhegh- also had the primary inactive form "I flee", to which the derived verb with -o- grade and secondary middle-passive form

were added. The formation of a factitive/active counterpart to the primary inactive verb, partially resulting in a disambiguation of the transitive and intransitive meanings already coexisting in the inactive verb, was decisive for the development of the Greek and Aryan active vs. middle oppositions.

The theories which have regarded the (oppositional) reflexives and passives as primary for Indo-European are invalidated by the semantic

Page 180: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INACTIVE AND MIDDLE-PASSIVE 163

continuum indicated above and by the clear diachronic relationship between the oppositional middle and non-oppositional inactive verb forms. The passives and direct-reflexives are, of course, syntactically relevant and cross-linguis­tically largely attested functions. They developed in all different realizations of marked diathesis in IE languages. However, in the original IE d-f structure the semantic orientation of the morphological distinctions was primary.

3.4.3.3 The innovative character of the Greek/Aryan oppositional middle

We have shown in the preceding section that the semantic and formal relationship between the non-oppositional inactive and the oppositional middle of the Greek/Aryan type should be interpreted along the lines of diachronic continuity. This, however, does not mean that the oppositional middle of the Greek/Aryan type can be projected onto IE.

Diathetic oppositions based on semantic version were probably formed only in some of the IE dialects and are attested in Greek and Aryan only as elaborated system of oppositions. We shall continue our comparison of Greek and Latin in order to define two different types of marked diathesis.

We can divide the Greek active vs. middle oppositions into two major types, those with experiencer (and, in part, non-agentive actor) in the semantic role of subject ( ) and those with benefactive in the semantic role of subject ( ). Neither type occurs in Latin. The above mentioned verb pairs with secondary factitive/active form and primary middle/inactive form are not attested in Latin. In Latin, the respective verb pairs are formed by lexical and derivational oppositions: terreo - timeo,fugo -fugio, etc.

Also not attested in Latin is the middle of the so-called in-direct-reflexive type141 in a large sense, i.e. with the benefactive as subject role. This type is represented in Greek by verbs such as "I prepare for myself", 'T ask advice" vs. "I advice, counsel".142 The middle of the indirect-reflexive type appears to be a common innovation of Greek/Aryan. Cf. also the often-mentioned parallel: 0I yajate, Gr. in the meaning "I offer in my own interest". No certain evidence has been brought for this type of the middle-passive outside Greek/Aryan area. The Tocharian examples of indirect-reflexive function quoted by K. T. Schmidt (1974) represent insufficient evidence (see Thomas 1985:97ff.). On the one hand, the Sanscrit influence could have played a role, given the close connection of the diathetic meanings with religious thought (see Hartmann 1954) also observed by K. T. Schmidt. On the other

Page 181: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

164 FROM INDO-EUROPEAN TO LATIN

hand, it seems that in many cases the indirect-reflexive interpretation is not necessary, but that the transitive use of the middle corresponds instead to that of the non-marked inactive with no specific semantic features. The existence of the transitive use of the inactive in the sense examined in 3.2.1 (type / know it; I break it) should be taken into account in historical explanations of middle-passive functions. The Greek/Aryan innovation, which consists in the development of indirect-reflexive middle, should be understood as the semantic polarization of active vs. middle oppositions formed from transitive verbs.

Therefore, the Latin oppositional middle-passive is restricted to passive and direct-reflexive functions. Individual oppositional -r intransitives are formations on lexico-derivational level in cases such as gravor "to feel oppressed, refuse" with respect to gravo "to oppress or overpower". Intransiti­ves such as agor "to betake oneself, go" can be interpreted as having a di­rect-reflexive function; they are used beside ago me, similarly as the real direct-reflexives such as accingor = accingo me. Allomorphic (non-opposi-tional) active/deponent verbs such as amplecto(r), misereo(r) attest semantic continuity (graduality) between the non-marked inactive and the marked mid­dle-passive/deponent.

These outstanding differences between the marked diathesis in Latin and Greek make the revision of the traditional views about the relationship of the Latin and Greek/Aryan verbal systems necessary also in the sphere of diathesis.

3.4.3.4 Impersonal passive: Latin backgrounding passive vs. Greek foregrounding passive

As stated above, Latin has an impersonal passive formed from intransitive verbs such as itur, valetur. In addition, the Latin passive is regularly combined with a non-personal subject as in liber legitur, dabitur (alicui), etc. The passive with a non-personal subject has the same basic function and character as an impersonal passive, with the effect of demoting or backgrounding the personal actant from the position of subject. Both functions can be subsumed under the terms 'deagentivum' or 'backgrounding passive'.143 The Latin equivalents of the above-mentioned Greek verbs

are not used in the middle function corresponding to Greek, but occur regularly as backgrounding passives with non-personal subjects: aliquid praeparatur, consulitur (in addition to

Page 182: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INACTIVE AND MIDDLE-PASSIVE 165

impersonal consulitur). The situation in Greek differs with respect to both types of passive,

impersonal and non-personal. In particular, Greek, does not possess an impersonal passive formed from intransitives of the type itur, etc. Only verbs of saying such as may serve as examples of the impersonally used passives. They, in fact, are transitives with an internal object

The occurrence of the backgrounding passive with a non-personal subject is limited in Greek. The basic character of the Greek passive is foregrounding, with the personal patient in the position of subject. In Greek, the passive function with the patient role of subject is only part of the broader middle-passive function with experiencer or benefactive as primary roles of subject. The subject of the Greek middle-passive is predominantly a personal subject, as with the subject of the Greek sentence in general.

Though the impersonal and backgrounding passive is not restricted to -r middle-passive only, as especially 0I shows by its large impersonal/back­grounding use of the passive, the difference between Latin and Greek together with the impersonal character of the Celtic passive support the explanation of the -r middle-passive as impersonal/indefinite deagentivum (3.4.4.2).

3.4.4 Formal characteristics of the Greek and Latin middle-passive

The semantic/functional differences defined above are related to similarly outstanding formal differences between both categories, i.e. the Greek middle-passive with its (primary) endings -mai, -sai, -tai (in the Ionic/Attic variant) and the Latin middle-passive (deponent/passive) with its respective endings -or, -ris, -tur. As we have seen, the non-marked inactive was transformed into two semantically distinct types of marked diathesis. According to the principle of iconicity (or naturalness), formal markedness developed parallel to semantic markedness, again in two different formal variants. The innovative character of the Greek oppositional middle, as a semantic category, suggests that formal characteristics of the Greek middle-passive are also innovative and that they cannot be projected onto common IE. With respect to both types of middle-passive endings, the Greek type with 3sg. -to(i) and the Latin type with 3sg. -tor, the third person appears to be decisive for the development of marked diathesis.

Page 183: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

166 FROM INDO-EUROPEAN TO LATIN

3.4.4.1 The middle-passive endings of the -to(i) type

Middle-passive endings represented by Greek -mai, -so(i)/-sa(i), -to(i) developed from vocalic inactive endings through their contamination with active consonantal endings -m(i), -s(i), -t(i). The continuity of the vocalic ending and the consonantized/mixed ending144 is attested for the 3sg. by the 0I and Hitt. middle endings -e < -oi, -a(ri) < -o, which were preserved in addition to the consonantized middle-passive endings 0I primary -te < -toi, secondary -ta < -to, Hitt. -ta(ri). For the first person, the diachronic continuity of the endings -a(i) and -ma(i) can be proven by comparing 0I and Greek: the Greek ending corresponds to the 0I ending -e < -ai, both endings being so-called primary endings of the middle-passive. Secondary en­dings have zero-grade vocalism -h2(d2) in 0I (-i < -h2) — see 3.2.3.5, whereas vocalism -eH in Greek , here again with preceding -m- (as a representative marker of the 1st person). In the 2sg. the mixed (consonan­tized) ending is attested in 0I in the primary series only: bharase < *bhere-sai. The secondary ending is -thas in 0I, i.e. the original inactive ending. This restriction is important in order to recognize the secondary character of the mixed -mai, -sai/-so, etc. series.145 The vocalism is -so in secondary Greek endings, whereas in the primary endings we have a dialectal difference between -sai for 2sg., -tai for 3sg. of Ionic-Attic, etc. vs. -soi, -toi of Arca-do-Cyprian146 and Mycenaean. With respect to the ambiguous position of the 2sg. between the lsg. and 3sg. (see 3.2.3.5.4) we may assume that there was a variation between -sa(i) and -so(i) vocalism and that this variation was probably a precondition for the dialectal generalization of -a- vocalism in the entire primary series, including the 3sg. -tai.

The whole singular series of mixed/consonantized endings of the type -mai, -sa(i), -to(i) is reconstructed for Albanian (see M. E. Schmidt 1930, Mann 1977:167, Demiraj 1985:730ff., llllff.) and for Baltic147. In Germanic, only 2sg. -za and 3sg. -da are attested148, no -m- form in the lsg. In the languages with -r middle-passive the mixed ending with the consonantal -t- element is attested only for 3sg. (see below).

Note that if we accept the above-mentioned explanation of the thematic present (3.2.3.4), two types of mixed, contaminated endings should be recon­structed for IE 3sg.:

Page 184: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INACTIVE AND MIDDLE-PASSIVE 167

(42) IE endings of 3sg. basic mixed

active -t(i) active -e-t inactive -e/-o middle-passive -t-o

Mixed endings should not be conceived of as agglutinative combina­tions of active and inactive markers. Diachronically, analogical contamination took place in both cases. In the case of the middle-passive it resulted in a marked series corresponding to the marked semantics of the middle-passive as diathesis. The consonantal elements -m-, -s-, -t- function as representative person markers, whereby the final vocalism {-o in the 3sg.) signals diathesis.

We can conclude that the formation of mixed endings of the type -mai, -sa/-so(i), -to(i) is an innovation which in its final form was realized in the individual languages/dialects. For the PIE period only the beginning of this tendency may be assumed, i.e. the mixed ending of the 3sg. The main structural effect and motivation of the mixed endings is to form marked middle-passive endings corresponding to the marked middle-passive meaning.

3.4.4.2 The -r endings

A characteristic feature of the Latin passive is its final -r element, which we find in the middle-passive endings of Italic, Celtic, Hittite and Tocharian, and which is also attested in Phrygian and (without consonantal -t element) in Venetic.149 The original domain of the -r element is the 3rd person. The -r ending is attested as a plural ending of the inactive verb as represented by the Lat. perfect videre, -runt or OI perfect vidur (3.2.3.5). In Celtic and Osco-Umbrian, the forms of the type *bher-er, *bher-or are used as an impersonal passive/deagentivum (see above).150 We know that 3pl. with an indefinite, non-specified subject can function as a deagentivum. In Russian, the 3pl. is a common way of expressing the passive: "I am injured". The assumed diachronic relationship between the impersonal -r passive and the inactive plural -r forms thus has a well proven semantic basis.

The element -r, which originally signalled the impersonal/indefinite deagentivum, was combined with the consonantal ending -t to form the marked diathesis. We find the forms -tor, -ter, -tr in the middle-passive of Hittite and Tocharian and in the deponent/passive of Celtic and Italic; the form -tor is also attested in Phrygian. Functionally, this form of the middle-passive was probably characteristic for languages, in which the deagentivum/background-

Page 185: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

168 FROM INDO-EUROPEAN TO LATIN

ing function of the passive was basic. The Celtic passive is of the impersonal type.151 In Hittite the impersonal backgrounding use is well attested.152 The backgrounding character of the Italic passive is attested by Osco-Umbrian -er/-or deagentivum and by the predominant backgrounding use of the Latin passive, which is especially well represented in Plautus comedies.153

Formally, the -r element is not added in all attested forms to the middle-passive endings of the type -to. The ending -tr has been reconstructed for Toch. -tär and can also be assumed for Celtic and Italic.154

The ending -tr shows that the personal -o variant of the inactive 3rd person ending was not the sole basis for mixed middle-passive endings. Here, the consonantal -t ending is contaminated ('mixed') directly with the in­definite/impersonal variant of the 3rd person ending, without the inter­mediation of the -to ending. This corresponds to the prevailing semantic orientation of the -r middle-passive, i.e. its backgrounding character.

Therefore, the ending -t-r can be added to the mixed endings, as indicated above in (42). Here again the -t element functions as a person marker, the final -r element signals diathesis:

(43) IE endings of the 3rd person basic mixed

active -t(i) active -e-t inactive -e/-o middle-passive -to impersonal/indefinite -r -t-r, -to-r

The extension of the -r element to other persons of the middle-passive was realized in principle in Latin, Hittite, Tocharian and Celtic (the Irish deponent). Differences exist in the form and distribution of these endings. In general, however, the extension of the -r element to persons other than the 3rd is also part of the dialectal isoglosses represented by the -r middle-passive.

3.4.5 Foregrounding -to(i) middle-passive vs. backgrounding -r middle-passive: Dialectal difference in IE area

We find, therefore, in IE languages two types of the middle-passive, both of which originate from the inactive verb. Their formal and functional characteristics vary and go back to the different formal and functional features of the inactive verb. Functionally, the -toft) middle-passive of the Greek type has its origin in the foregrounding use of the inactive, with the personal

Page 186: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INACTIVE AND MIDDLE-PASSIVE 169

subject in the role of experiencer. On the other hand, the -r middle-passive of the Latin type originates in the backgrounding use of the inactive verbs with non-personal actants or with general, impersonal value. Formally, the Greek type is characterized by the innovative tendency to form a mixed (contami­nated) series of endings of the type -mai, -sa(i)/ -so(i), -to(i), the Latin type by the -r element, whose original application was in the general, impersonal/ indefinite function. The -r element belongs to archaic, PIE elements as its occurrence in OI attests (3.2.3.5, and 3.4.4.2).

Both types are innovative morphological forms of marked diathesis. The functions of oppositional middle found in Greek and OI are innovative, especially the indirect-reflexive middle, which is not well attested outside the Greek/Aryan area. The impersonal and backgrounding passive/deagentivum clearly predominates in Celtic and Italic; in Hittite the backgrounding middle-passive (impersonal use included) is well attested (see above 3.4.4.2, and 3.2.1.5). This function was part of the original use of the non-oppositional inactivum and developed, to varying degrees, as a function of the oppositional middle-passive. Its extreme development is found in Italo-Celtic, its extreme abandonment in Greek. Similarly, the foregrounding use of the original inac­tive developed differently in the individual realizations of the oppositional middle-passive. Its extreme development is found in the Greek/Aryan middle, with its experiencer/benefactive semantics.

With respect to both formal and functional features we can draw the dividing line between two main types of the middle-passive: 1) the fore­grounding -to(i) middle-passive, which developed in Greek/Aryan, and is also attested in Baltic, Albanian and Germanic, and 2) the backgrounding -r middle-passive, which developed in Italic, Celtic, Hittite and Tocharian, and is also attested in Armenian, Phrygian and Venetic.155 Only with reser­vations can this difference be identified as a dialectal isogloss, since the decision whether the innovations are based on areal agreement or on parallel development is, as in many other cases, difficult. Most likely, the areal relationship characterized by concrete form - function agreements should be recognized for Greek/Aryan (with respect to its indirect-reflexive function) and Italo-Celtic (with respect to its backgrounding/impersonal character and other special formal and distributional agreements).156 Essential for our purposes is the conclusion that both types of middle-passive, realized in clear-cut fashion in Greek and Latin, are the result of an alternative development from a common basis of the non-oppositional inactive. Again, the Latin situation should not be reconstructed on the basis of the Greek system and conceived as

Page 187: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

170 FROM INDO-EUROPEAN TO LATIN

its secondary modification.

3.4.6 The endings of the Latin deponent/passive: A revised analysis

The above mentioned differences in the development of marked dia­thesis also make a revision of the traditional explanation of Latin passive endings and their relationship to the endings of the -to(i) type necessary.

We have shown above that the Latin 3sg. -tur need not have its origin in the -to(i) ending of the Greek/Aryan type. The contamination of the deagen-tive -r form with the personal ending -t (originally restricted to active verbs, but soon becoming a basic personal ending of the integrated system) was probably realized without involving the personal -o variant of the inactive ending in this process. The underlying form can be reconstructed as -t-r: *dice-tr, *seque-tţ, where -r represents an impersonal/indefinite variant of the inactive ending. The 3pl. -ntur and both 1st person endings -or and -mur can best be explained as a combination of the active endings with the -r element, developed in analogy to the 3sg.

The 2pl. is a special case, having an ending without the -r element, whose origin is to be found in either the infinitive (cf. Greek infinitives in -men, -menai, OI -mane) or participlal form (cf. the verbal adjective in -menoi occurring as participle in Greek and OI and as lexical formations such as alumnus in Latin). Our main interest will thus concentrate on the 2sg. Instead of the traditional explanation, which sees the origin of this form in -so/-se, another possibility should be considered:

(44) Latin 2sg. deponent/passive proposed explanation traditional explanation dice-re < *dice-ri *dice-so? seque-re < *seque-ri *seque-so?

We know that the classical ending -ris (diceris, sequeris) is a modification of the original -re ending (dicere, sequere), which is attested in both archaic Latin and as an archaism and poetism throughout the classical period. This original -re ending can have its origin in the deagentivum with -r ending combined with the particle -i.157 In a similar manner (but with different vocalism of the verbal stem), -ri can also be reconstructed for the 3pl. of the perfect: videre < *videri, dixere < *dixeri (3.3.3).

The fact that the impersonal/deagentive form established itself in the

Page 188: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INACTIVE AND MIDDLE-PASSIVE 171

2sg. is understandable, considering the widely attested generalizing use of the 2sg. (ubi maxime gaudebis, ibi maxime metues, etc.) on the one hand, and the indefinite/impersonal expressions in polite adress on the other. In Armenian, whose -r forms should not, in principle, be separated from the -r forms of the other IE languages, the -r element occurs above all in the 2sg. and in the imperative.158

As explained above (3.4.4.1), the Greek ending -so which was (in the variant -se) suggested for Latin dicere, sequere, was an innovative form and only occurred in languages possessing a personal foregrounding middle-passive (whereby its occurrence in these languages was also limited).

3.4.7 Inactive and middle-passive: Conclusions

According to our conception, both the perfect and the middle-passive are inflectional categories derived from the inactive verb, the perfect being a continuation of the non-progressive/state class and the middle-passive of the progressive/process class. In the process of integrating the PIE two-class system into the verbal system of the IE languages, both categories developed into marked inflectional categories, whereas the original inactive was a non-marked class. The markedness developed and manifested itself not only in the formal but also in the semantic characteristics of the perfect and the middle-passive.

The main difference between both categories comes about for two reasons: 1) Only in the Greek/Aryan area did the perfect develop into an inflectional category distinct from the active preterite. The tendency to develop marked verbal voice, however, was common to the whole IE area, although its realization varied. 2) The innovative formal characteristics of the perfect were achieved by stem marking, whereas the formal characteristics of the middle-passive consisted in marked verbal endings. Therefore, the endings of the perfect are a more direct reflection of the original inactive endings. Because the change in the marking of personal endings was considered more fundamental than the change in stem marking, the diachronic relationship between the middle-passive and the inactive has been denied in certain explanations of the middle-passive. We, on the other hand, have attempted to show that both types of the middle-passive are based on the formal and semantic characteristics of the inactive verb class.

Page 189: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)
Page 190: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

3.5 Thematic -e/o- suffix and Long Vocalic Suffixes: Latin subjunctives and futures: Latin present conjugations

3.5.0 An outstanding difference between Latin and Greek lies in the different role which has the alternating -e/o- suffix and the long vocalic -à-, -ē- suffixes in verbal morphology. This difference is particularly evident in the imperfects and subjunctives/futures. The Greek imperfect of thematic verbs such as is formed with -e/o- suffix as a present stem marker and augment + second­ary endings as preterital markers; cf. lsg.imperf., 3sg.imperf. In Latin, the imperfect is formed with he help of the suffix -bă- < -dhă- (see 3.7). The vocalic element of this suffix is -ă- suffix which is in the pure form present in the imperfect of the athematic verb sum, esse: eram < *esăm. As we have already explained in 3.1.4, 3.2.3.6.2, etc. the suffix -ē- together with a parallel suffix -ě- form long vocalic injunctives whose functional value is similar to athematic or thematic injunctives. Now, we shall examine the functional parallelism of thematic and long vocalic formations in a more detailed way, showing and interpreting the differences between Latin and Greek in the use of these formations.

The Greek subjunctive of thematic verbs is formed by the alternating -e/o- suffix, whereas the Latin verb dico exhibits two long vocalic formations with subjunctive/future value:

(45) imperfects and subjunctives of thematic verbs in Greek and Latin Greek etc. Latin dicebam, dicebas, etc. < *dicë-dhā-m, *dicê-dhâ-s, etc. Greek etc. Latin dicam, dicas, etc. present subjunctive

dicam, dices, etc. future

Page 191: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

174 FROM INDO-EUROPEAN TO LATIN

The long vocalic suffixes asserted themselves in the present system of Latin as present stem markers (1st and 2nd conjugations) and as preterital (imperfect) and modal markers. This complex of functions goes back to the basic, underlying form of the verb, so-called injunctive.

The occurrence of athematic presents in Greek qualifies Greek as more archaic than Latin159 and similarly in many other respects Greek has preserved the original form variety of d-f structure. However, the difference between Greek and Latin in terms of the suffixal formations which were expanded and grammaticalized as imperfects and subjunctives/futures is again a result of an alternative way of grammaticalization. The Greek situation can­not be conceived as reflecting the original IE structure. Both languages differ in the preference given to the formations with widely parallel functions both of which go back to the types of injunctive: injunctive with alternating thematic -e/o- vowel vs. long vocalic injunctive.

3.5.1 Functional parallelism of -e/o- injunctive and long vocalic -ā-, -ë-injunctives

We find thematic -e/o- formation and long vocalic -ă-, -ë- formations in parallel functions in IE languages.160 Both thematic -e/o- formations and formations with -ă- and -ë- suffixes occur as markers of inflectional stems of presents or of the preterites with durative or intransitive value. The present formations with -e/o- suffix are attested in the whole dialectal area of IE.161

In Latin, the verbs of the 3rd conjugation like dico are thematic verbs with pure -e/o- suffix. The imperfect of the Greek/Aryan type is a preterital formation with the same inflectional stem as respective presents; cf. Greek present lsg. vs. imperfect lsg. "speak".

As for -ā-, -ë- suffixes the Latin 1st and 2nd conjugations, with parallels in other IE languages, contain the -ā- presents like secare, iuvare, etc. and the -ë- presents such as videre, delere respectively.162 The inflectional stem in -ă- is represented by Slavic preterital/infinitival stems of the type bra- (to -e/o- present bero): brach OCS lsg.aor., brati OCS inf. Similarly the inflectional stem in -ë- is represented by Slavic preterital/ infinitival stem of verbs of the type OCS boleti inf. to pres. boliQ OCS lsg.

The suffixal formations in -ă- and -ë- asserted themselves as grammati­calized morphological markers of the preterites or subjunctives/futures. With respect to the complex and allomorphic character of inflectional morphology,

Page 192: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

THEMATIC AND LONG VOCALIC SUFFIXES 175

the difference between markers of inflectional stems and subsidiary or more central markers of morphological category is gradual only. In the imperfects of the Greek/Aryan type it is the -e/o- stem which signals the specific aspectual — non-perfective — type of preterite, the augment and secondary ending being markers of preterital and non-actual value in general. The subjunctive with -e/o- suffix formed from athematic verbs is well attested in Greek, though in residual forms only. The thematic subjunctive with alter­nating long -ë/ö- is as early as Homeric Greek, and is also used in the case of athematic verbs. In Latin, only one form of this type is attested: the future ero < eso. We can assume that the -e/o- subjunctives (futures) of athematic verbs were formed (and probably not systematically, but on a derivative level only) in connection with the formation process of thematic present. In opposition to the athematic present (Latin sum, Greek lsg. "go") the new productive form asserted itself in the modal/future interpretation (Latin ero, Greek lpl. "let us go").

The -ă- and -ë- formations are well attested as morphological markers of preterites and subjunctives in IE languages. Both formations were grammaticalized as basic preterites in Baltic languages, with no oppositional preterite category with respect to which they could assert their durative or intransitive meaning, which otherwise is characteristic for long vocalic prete­rites. In Latin, the pure -a- suffix occurs in eram < *esăm only, otherwise the -bă- < -dhă- suffix serves as a marker of the imperfect. The -ë- suffix functions as a marker of Greek aorist of intransitive type lsg. "I enjoyed"). The suffix -ă- in preterital function is attested in Tocharian, where the -ă- suffix is a characteristic preterital marker used with most verbs. The above-mentioned preterital/infinitival stems of Slavic belong more to the inflectional level. The comparison of these Slavic formations with evidently morphologized -ă- and -ë- preterites in Baltic and with the intermediary situation in Tocharian shows the graduality of the function of -ă- and -ë-suffixes as morphological markers and as markers of inflectional stem. Note the difference between Baltic, where the long vocalic preterites are a unique type of preterite and where no distinction of present and preterital endings exists, and Tocharian, where other preterital formations, especially the -s-preterite, are used and where the present endings are distinct from preterital ones. Given the apparent derivational origin of -ă- and -ë- suffixes in the function of inflectional stem markers, there is no reason for rejecting the opinion that the -ā- and -ë- as derivational suffix (secare, delere) and -ă- and -ë- preterite or subjunctive markers have a common origin, i.e. injunctives

Page 193: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

176 FROM INDO-EUROPEAN TO LATIN

with the derivational suffix -ă- or -ë-. As a subjunctive (future) marker the -ă- suffix was grammaticalized in

Latin, Celtic and Tocharian, while the -ë- suffix was grammaticalized in Latin only, as we shall see in the section 3.5.4.

3.5.2 The preterital and modal value of IE injunctive

The whole system of functions which we find in long vocalic forma­tions (marking of inflectional present and preterital stem, preterite and subjunctive marking) goes back to the IE injunctive. As we have explained in 3.1 the preterital and modal interpretation relates to the non-actual value of the basic form called injunctive. The non-actual meaning was interpreted in opposition to the actual present (with 'primary' endings of the type -mi, -si, -ti) either as subjunctive, if the opposition to the indicative was relevant, or as preterite, if the opposition to present mattered, the 'primary' form being a complex expression of the indicative present. In the process of the develop­ment of formalized paradigmatic oppositions the basic form became restricted to one of these functions, while the other function was expressed with the help of new derivations. This holds also for Latin long vocalic injunctives, which were grammaticalized in subjunctive functions, whereas in the preterital func­tion the marked suffix -dhā- > -bă- became a marker of imperfect (cf. subj. dele-as vs. imperf. dele-bas < *dele-dhăs). In Baltic on the contrary, the long vocalic injunctives were grammaticalized in the preterital function.

3.5.3 The semantics of the vocalic suffixes

The meaning of derivations with the -e/o- suffix and with long vocalic suf­fixes was probably very vague. With respect to the iconicity/naturalness principle (1.3.2.2) we can assume for the derived form the meaning 'more extent, more in­tensive'. Both types of suffixes probably belonged to the active and inactive verb respectively. This is a basis for the conception of these formations as alternative expressions of the previously noted system of verbal functions (stem markers, modal and preterital markers). In Greek the active -e/o- suffix, in Latin the inac­tive -à-, -ë- suffixes were preferred as expressions of subjunctives and imperfects.

The semantics of the formations with suffixes -ā- and -ë- must be treated from two points of view, i.e. with respect to the semantics of the suffixes on the one hand and with respect to the semantics of the verb classes on the other. These

Page 194: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

THEMATIC AND LONG VOCALIC SUFFIXES 177

two aspects of meaning combine in the semantics of long vocalic formations and account for the differences between the individual IE languages. Both suffixes probably had a durative/progressive and intensive value in their intraclassal relationships to the non-marked form. Perhaps we can assume that the -ă- suffix was more dynamic, whereas the -ë- suffix more static (3.2.3.5.4). The inactive and predominantly intransitive value, however, was given by primary appurte­nance of these formations to the inactive class. The differences between IE languages as to the character of the -ë- suffixal formations with respect to intran-sitivity vs. transitivity depend on whether the intransitive as a typlcal class value of inactive verb asserted itself as a relevant component of the morphological semantics. This can account for the differences between the intransitive Greek -ë-aorists and transitive Baltic -ë- preterites. In any case, the Latin and Baltic paral­lelism in the distribution of the -ë- variant of the preterite and future respectively, both of which occur in the present class of thematic (widely transitive) -e/o-verbs, is striking. In Latin, we have the -ë- suffix not only in the future dices, etc., but also in the stem vowel ('enlargement') of the imperfect dice-bam. In the grammaticalized preterites and subjunctives/futures, the diathetic/inactive value of -ë- and -ā- suffixes is not tracccable, it is, however, tracccable, together with aspectual/durative value, in the present classes, i.e. in the verbs with -ë- or -ā- as stem markers. In the derivative oppositions to the verbs with -e/o- or -ie/io-suffixes in particular, the -ë- and -ă- verbs exhibit their intransitive/stative (iaccco to iacio, pendeo to pendo) or subjective medial value (paro to pario). Cf. also the deponents of the 1st conjugation in opposition to active verbs of the 3rd con­jugation: consplcor to consplcio, susplcor to susplcio, aspernor to asperno, where the deponents with subjective/medial value are also formed with the -ā suffix (see Perel'muter 1977:93,104, Vendryes 1910:300 with Italo-Celtic correspondences).

3.5.4 Latin subjunctives and futures with -ā- and -ē- suffixes

Considered on the above basis, the Latin subjunctives/futures in -ă- and -ë- appear to have developed by the grammaticalization of long vocalic injunctives with modal interpretation. The -ă- and -ë- injunctives are archaic components of IE d-f verb; however, their grammaticalization as expressions of flectional mor­phological categories had taken placcc in the later development of dialectally differentiated IE. We have agreements between IE languages in the grammati­calized long vocalic preterites or subjunctives, which can be conceived of as dialectal isoglosses or as the results of parallel developments.

Page 195: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

178 FROM INDO-EUROPEAN TO LATIN

The Latin -ă- injunctive of 2nd - 4th conjugations (deleas, dicas, audias) has equivalents in Tocharian and Celtic -a- injunctives. The common source of these forms in the -ă- injunctive is also confirmed by the fact that all these lan­guages use largely an -ă- suffix as a marker of the inflectional present and preteri-tal stems. In Tocharian, both -a- presents and -ă- preterites occur, the -ü- suffix being the characteristic inflectional stem of the preterite for most of the verbs (see van Windekens 1982:117ff., Krause & Thomas 1960:196ff.,221,237ff.). In Celtic, -ă- presents represent one of the two basic Old Irish classes of 'weak' verbs (see Thurneysen 1909:313,330). This speaks very clearly in favour of the explanation of -ă- subjunctives as original -ā- injunctives, it makes other explana­tory efforts unnecessary, and it reduces the validity of objections which — with no compelling evidence — were expressed against this theory in recent research.163

The Latin -ë- future is more problematic. We consider the future of the 3rd and 4th conjugations (dices, audies) as going back to suffixal -ë- formation, while supposing that the -ë- subjunctive of the 1st conjugation (ames) has its origin in the optative *amâ-iê-s as is commonly accepted. The Latin -ë- future has no direct correspondence in other IE languages. The grammaticalization of the -ë-formation under similar distributional conditions is, however, a common feature of Latin and Baltic. The large functional parallelism of -ă- and -ë- injunctives makes it probable that the -ë- injunctive also could have been grammaticalized not only in preterital, but in modal function as well. A common feature of both Latin and Lithuanian -ë- formations is their assertion in the domain of active transitive verbs, which is secondary for -ă- and -ë- injunctives. In Latin, the complementary paradigm of the -ë- future (with -ā- formation in the lsg. dicam) reveals the basic and primary character of -ă- injunctive with modal function. The fact that in the first person no differentiation between subjunctive and future expression was needed can be explained semantically with respect to the fact that in the case of the 1st person the speaker as subject of the modality is identical with the subject of the sentence.

We have also modal formants with -ë- vocalism and a preceding conso­nantal element. The optative formant -ieH/iH- which is probably a common IE modal formant, exhibits -ë- vocalism. In Latin it represents an allomorph of the subjunctive in the athematic forms sim, velim, edim, and probably in the 1st con­jugation subjunctive amem < *amā-ie-m. Again, the opposition of optative vs. subjunctive was not grammaticalized in CIE164, it is a characteristic feature of the maximized Greek/Aryan verbal system. Furthermore, the formation with -ë-vocalism and modal function is also attested in the Latin subjunctive imperfect

Page 196: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

THEMATIC AND LONG VOCALIC SUFFIXES 179

*dice-sě-t > diceret. However problematic the position of the -ë- variant is, there are no

compelling arguments for considering this formation a secondary modification of the alternating -ë/ö- subjunctive of the Greek thematic type. We shall examine this question in the next section.

3.5.5 Greek thematic imperfects and subjunctives

There is no support for the assumption that Latin possessed imperfects and subjunctives of the Greek thematic type. The imperfects with pure thematic vowel as or with consonantal thematic suffixes as are Greek/Aryan innovations connected with the development of the aspectual opposition aorist vs. present. Imperfects are formed systematically from present stems, and are opposed to the presents by the type of ending (secondary vs. primary) and to the aorists by the inflectional stem and in part (with respect to the sigmatic aorist) by the type of endings. These grammaticalized oppositions are not original for IE; they developed in the Greek/Aryan system, which is characterized by the maximization of semantic oppositions and by semantic and formal/distributional features going back to d-f morphology.

Similarly, the Greek alternating long vocalic -ë/ö- subjunctive of the thematic type , etc. (in Aryan the alternation is not discernible) is an innovative formation which either accompanies or follows the formation of the thematic present. Most probably, it was formed together with the thematic present indicative from which it differs by the non-presence of the actualizing particle -i.165 The grammaticalization of this formation in the subjunctive function seems to be a later dialectal development and we cannot exclude the possibility that the -ë injunctives in modal functions, which can be presupposed as derivational (non-grammaticalized) formations for IE verb, supported the assertion of a long vocalic variant of the thematic -e/o- formation in the function of subjunctive. In any case, there is no support for the explanation of the Latin -ë- future as having developed by secondary modification of the alternating -ë/ö- subjunctive of the Greek type. This Latin formation is, as explained above, a component of the non-alternating long vocalic suffixal formations (-ā- and -ë- suffixes) which asserted themselves in Latin in the functions where Greek preferred the alternating -ë/ö-, -ë/ö-formations.

Page 197: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

180 FROM INDO-EUROPEAN TO LATIN

3.5.6 Conclusions: Latin present conjugations. Basic tendencies of Latin paradigmatization

We have defined the difference between Greek and Latin in the structural positions which have non-alternating long vocalic -ā-, -ë- formations in both languages. We find them in Latin in the functions where Greek prefers the alter­nating -e/o- suffix. The subjunctives of the present and the future of the 3rd and 4th conjugations are expressed by long vocalic suffixes in Latin. The Latin imper­fect goes back to the long vocalic formation in its indicative and in its subjunctive as well (3.6.3). The most frequent and productive 1st and 2nd conjugations have -ă- and -ë- suffixes in their inflectional stems. This systematic difference between both languages is to be explained as a difference in use and grammaticalization of two types of derivational formations of d-f IE structure, hence as a result of alternative developments. The alternating -e/o- suffix probably goes back to the active class in its original form, (3.2.3.3) becoming then a part of the thematic present (3.2.3.4). The -ā-, -ë- suffixes belonged to the inactive class.

The Latin present conjugations are a good example of paradigmatized flectional morphology.166 In examining the morphemic structure of finite verbal endings in the 1st Part of this study we have seen the main manifestations of the cumulative/fusional morphology: the fusion of inflectional with modal markers, to which the fusion with person marker joins in the case of 1st person in -ö; the additional elements such as -ë- in dicebam; the involvement of a root vowel in inflections (dat). The expression of categorial distinctions is complex and integral, with the determination going from stem to endings having an integrating effect.

The paradigmatization is advanced in the Latin present system in both morphological and inflectional paradigmaticity. The inflectional classes are reduced with respect to the original form variety and are well defined by the in­flectional marker of present indicative stem and by the type of subjunctive/future. We have analysed the Latin paradigmatized structure in comparison with Greek in 1.3.3. In this respect a comparison with Tocharian would be very instructive as it could reveal another important aspect of advanced paradigmatization in Latin: the non-existence of homonymal forms which the polyfunctionality of for­mants can produce. In Tocharian, the same formant -ă- has the same amount of functions (-ă- presents, -ă- preterites, -ă- subjunctives), but the entire structure is less paradigmatized and, unlike Latin, there are cases of indicative-subjunctive homonymy.

Page 198: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

3.6 Sigmatic Forms of the Latin Present and Perfect Systems: The Indo-European -s- Suffix

3.6.1 The aoristic and modal -s- suffix

Among the consonantal formations the -s- suffix or suffixes played a prominent role in the verb morphology of IE languages. The sigmatic aorist/ preterite is represented in most of the IE languages in an aspectual or temporal formation. In Latin, it represents an allomorphic expression of the perfect (3.3.4.1). The sigmatic aorists and preterites of IE languages are, however, a result of alternative grammaticalization realized in very different ways and to different degrees in individual languages. The Greek and OI sigmatic aorists as well, representing an aspectual formation in both cases, exhibit rather expressive differences in form, distribution and function.167

The systematic examination of -s- suffixal formations with modal and future meaning, as analyzed especially by Schmid (1963), gives results which are in keeplng with our concept of d-f structure. The sigmatic aorist as the grammaticalized expression of aspecto-temporal category cannot be considered a basis from which all other sigmatic forms with modal or future meaning are derived (see already Meillet 1908:103ff., further Schmid 1963, K. H. Schmidt 1966:23f., Kortlandt 1984:180f.). They represent another type of grammati­calization of the derivative formations, independent of the -s- aorist indicative. The modal sigmatic forms are attested in the athematic form168 and with various long vocalic suffixes (-ē-, -ă-, -i-); the sigmatic futures have thematic -se-/-so- or -sie-/-sio- endings.169

As already stated in 3.3.4.1 the sigmatic aorist/preterite was formed primarily from active verbs. One should examine whether the modal sigmatic formations also followed this restriction in the d-f structure. As shown by Schmid (1963: 44ff), the OI and Greek sigmatic futures are formed from in-

Page 199: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

182 FROM INDO-EUROPEAN TO LATIN

active ('subjective', 'medial') verbs and have inactive formal characteristics: OI -sio- futures have the inactive/stative -io- suffix, Greek futures of the type

(to present "hear") have medial form. These futures are, however, grammaticalized forms of the integrated verbal system with no other allomorphic expression of the future. The subjective/medial characterization must be considered with respect to the concept of futurity. The OIr -s- sub­junctive is restricted in its occurrence to about 50 verbs (see Thurneysen 1946:380, Watkins 1962:128ff.), whereas the complementary allomorphic -ă-subjunctive is a basic form. Therefore, we must leave the problem of the ori­ginal distribution of modal -s- formation open at present, but at least it can be stated that no such difference in distribution with respect to the active or inactive sphere has been observed which would contradict the derivation of the aoristic and the modal -s- from the same suffixai formation. In our conception, the formation with perfectivizing meaning can also account for the modal -s-which would represent the modal realization of its 'injunctival' value in the case of -s- subjunctives. The perfectivizing formation is also a possible source or one of the sources (besides the desiderative; see Schulze 1933, K. H. Schmidt 1966:21) of -s- futures.170 In the process of grammaticalization both the aoristic/preterital and the modal/future -s- formations developed indepen­dently and cannot be derived from each other; rather, each of them can be derived from the underlying derivational formation.171

Summing up we can say that 1) the sigmatic forms originated in derivational suffixal formation with perfectivizing value; 2) the sigmatic aorist represents only one type of grammaticalized sigmatic formation which devel­oped into the aspectual stem with own inflectional system in Greek and OI (subjunctive, infinitive, etc.) but even in these languages it did not attract other independently grammaticalized or semigrammaticalized realizations of the underlying derivational sigmatic formation. With this background we shall now briefly examine the Latin sigmatic forms. However, we must first recall the interpretative possibities offered by the morphology of IE consonantal formations for the explanation of morphological systems and inflectional patterns of individual IE languages.

3.6.2 The morphology of formations with consonantal suffixes: Latin sigmatic forms

The derived verb stems with consonantal suffix, be it the widely gram-

Page 200: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

SIGMATIC FORMS 183

maticalized -s- suffix or other suffixes (-dh-, -d-, -k-, etc.), could accept the same sets of vocalic 'endings' which applied to the verbal roots and primary stems: the endings with thematic vowel -e/o-, with the long vowel -ā-, -ë- or with the specific modal 'optative' suffix -ieH/iH-.172 In Latin, the following sets of endings are applied (besides the specific perfect indicative endings restricted to the inflectional perfect stem): thematic present endings going back to the primary IE series -ö, -esi, -eti, long vocalic endings -ām, -ās, -āt; -ēm, -ēs, -ēt, and modal endings -im, -ís, -it. As stated in 3.2 and 3.5, the secondary thematic endings -om, -es, -et which form the Gr. and OI imper­fects from the present stems including the derived stems of the type ,

, etc. ( ) were not established as inflectional forms of Latin morphology.

These derivational possibilities were a source of enrichment and restoration of the morphological systems in later IE, and in using the possibilities given by the grammaticalization of suffixal derivations, the individual languages differ, but also show characteristic agreement. In the present and the following section 3.7 we will examine this part of Latin verb morphology in comparison with similar formations represented in other IE languages.

3.6.2.1 The derivational basis of Latin sigmatic forms

With respect to the derivational basis of the -s- suffixes, the cyclical character of IE suffixation must be recalled (1.3. 1.1). The same suffix can be applied to the primary lexical roots/stems or to the inflectional stems, such as the Latin present and perfect stems.173 Among the Latin sigmatic forma­tions,174 there is a subsystemic formation including a small group of verbs such as faxo, dixo, etc. (3.6.4). The imperfect subjunctive is a sigmatic formation based on the present stem (3.6.3). The whole system of sigmatic formations is based on the perfect stem (3.6.5).

3.6.2.2 The vocalic endings of the Latin sigmatic forms

In the individual IE languages the -s- suffix combines with various vocalic suffixes/endings (3.6.1; 3.7.3). In Latin the thematic -e/o- suffix, -ë-suffix and -iH/ieH- suffix are found in modal and future forms, the -ă- suffix in the preterital form of pluperfect (3.6.5).

Page 201: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

184 FROM INDO-EUROPEAN TO LATIN

3.6.3 Imperfect subjunctive: The sē- formation

The imperfect subjunctive is a unique sigmatic form integrated into the present system. It is an -ê- 'injunctive' of sigmatic formation,175 with the present stem as the derivational basis. Attempting to characterize the contribution of the individual (cyclically applied) formants to the final semantic and functional effect of this morphological form, we can say that the present stem accounts for its non-anteriority, the aspectual (perfectivizing) -s-formant for its preterital value and the -ē~ injunctive has its modal realization here, which in combination with ' + preterital' gives '+potential, —real' vs. '+ideal, ±real' of present subjunctive (see Trost 1939).

Not only different combinations of formants: -i- + -ê- in Greek and -s-+ -ê- in Latin, but also different systems of basic inflectional oppositions account for the similarities and differences in the meaning and use of the Greek present optative and the Latin imperfect subjunctive (see Funk 1985). Here the general parallelism of these two formations is mentioned in order to indicate the formational capacities connected with IE suffixation. We can also observe additional formal modification characteristic of IE morphology in both the Latin subjunctive and the Greek optative. The derivational basis of the Latin imperfect subjunctive is the root in individual athematic formations: *es-sě-m, *uel-sě-m > vellem, *bhu-sě-m > forem176 and the present stem with characteristic class-marking vowel in individual conjugations, including the 3rd conjugation where the imperfect subjunctive is formed from the stem with -e- vowel: *dic-e-sě-m > dicerem. Also the Greek optative , where the stem vowel -e- has no motivation in the conjugation class, shows the same adaptational capacities of IE morphology. In Latin, the formation processes were regularized and paradigmatized. These paradigms, however, are results of a rather complicated derivational history in which cyclical suffixation was connected with additional modifications.

The imperfect subjunctive is a formation which shows well the insufficiency of the explanations deriving suffixal formations of IE languages from inflectional categories reconstructed for IE on the basis of the Greek/ Aryan system. The vague semantics of suffixal formants and the complex rules of their applications only allow an approximate analysis of the historical and functional content of morphological forms such as the Latin imperfect subjunc­tive. Yet the clear non-acceptability of the assumption that the sigmatic aorist

Page 202: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

SIGMATIC FORMS 185

of the Greek type would provide a basis for the Latin imperfect subjunctive demonstrates the essential advantage of the concept explaining the inflectional categories of individual IE languages as a result of alternative development, i.e. alternative grammaticalization based on a common inventory and common derivational capacities of the IE d-f structure. From this point of view we shall now examine other sigmatic forms of Latin.

3.6.4 Futures and subjunctives of the type faxo, faxim (-se/so- and -si-formations)

The subsystemic formation of the type faxo, faxim is a sigmatic forma­tion which is not integrated into the basic verbal system of Latin constituted by the present vs. perfect stem opposition. Only a few verbs display this formation in archaic Latin: faxo, faxim, dixo, dixim, ausim, etc.

The thematic formation -e/o- and subjunctive/optative in -f- (-iH/ieH-) are also connected in the present paradigms of irregular verbs with athematic forms: volo (pres.ind.) vs. velim (pres.conj.); ero < *eso (future) vs. sim, and further in the -isol-isim perfect formation (3.6.5). With respect to the cen­tral position of thematic -e/o- form in the Latin system the -f- subjunctives are classified as subjunctives of the respective -e/o-, -se/so-, -ise/iso- formations. The Latin subjunctives in -sim, -sis are considered as part of the inflectional system where the -se/so-, -ise/iso- form occuples a central position. Origi­nally, however, the modal athematic suffix -s-iH/-s-ieH- was independent of thematic formation. In OI, Baltic and Celtic we find -sie/sio- futures which can be interpreted as a result of the thematization of -s-iH/-s-ieH- forma­tions.177

3.6.5 The perfect forms with sigmatic suffix

Here I follow Jasanoff (1987) who rejected the hypothesis of the aoristic -is- element in Latin sigmatic perfect forms and explained the forms videro < *uidiso, viderim < *uidisim as formations parallel to faxo, faxim, but containing the connecting vowel -/'-, hence -i-so, -i-sim, not -is-o, -is-im. The connecting element may originally have been applied to the perfect stems terminating in consonants, type *uoid-i-so. The assumption that the bisyllabic roots/stems of the type gend- 178 motivated or supported the assertion of the vocalic element is possible,179 though not necessary.

Page 203: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

186 FROM INDO-EUROPEAN TO LATIN

Problems arise with respect to the remaining forms of the perfect system, i.e. pluperfect *uidisam > videram and subjunctive pluperfect vidissem. If the -sö/-sim forms were formed by the -s- suffix with prospective or desiderative meaning, then the application of such a suffix to give a preterital form would be impossible. To explain the pluperfect form we should presuppose a purely formal analogy with eram, or properly with visus eram. I prefer to consider the perfectivizing -s- suffix as a common formational basis for the both aoristic and modal/future sigmatic formations. On this basis the application of the -am, -as injunctive with preterital interpretation to the sigmatic formation is possible and has its parallel in the application of this injunctive to the -dh- suffixal formation (3.7). Here, in interpreting the complex suffixal formations, we must bear in mind that IE morphological processes were not realized on an agglutinative basis, i.e. by the additive combination of suffixes. IE suffixation consisted of cyclical application of the formants based on the 'decomposition via paradigmatization' in paradigmatic oppositions (1.1.3; 1.3.1.1). Considering this, we must not assume that the more external -ă- suffix always applied after the -s- or -dh- suffixes, but, on the contrary, the consonantal -s- and -dh- suffixes could be inserted into the already existing vocalic formation. The long vocalic -ă- injunctive with simple -ă- suffix is paradigmatized, with the exception of the form eram, in modal/ subjunctival realization. The preterital interpretation was realized by means of consonantal suffixes with corresponding aspectual (Aktionsart) value: perfectivizing or terminative (3.7). The system was completed by the formation of the pluperfect subjunctive vidissem which, too, is not only a result of the analogy with eram (visus eram), but resulted from the systematic parallelism of the present and perfect systems and was formed after the model of the imperfect subjunctive.

Page 204: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

3.7 Latin Imperfect in -bam and Future in -bō: The Indo-European -dh- suffix

3.7.1 The analytic explanation of the Latin indicative imperfect

The Latin indicative imperfect and the future in -bō of the 1st and 2nd conjugations have been explained as periphrastic form, composed of a nominal component and the auxiliary verb *bhü- "be, become". In splte of the difficul­ties implied, there are still attempts to advocate the analytic explanation. Baldi (1976) defines the nominal component as present participle, whose construc­tion with the verb "to be" is well attested in Latin syntax. The phonological difficulties, however, are, great, and Baldi is aware of them. Jasanoff (1978:120ff.) explains the nominal component as instrumental 'infinitive', with support in OI adverb guhā "hidden, in hiding". But the phonological and semantic difficulties connected with the second component remain {*bhuăm as autonomous word should give fuam which is attested as subjunctive only). The assumed development of the auxiliary verb into the suffix lacks any parallel and does not correspond to morphological processes characteristic of the evolutional period in question.

3.7.2 The flectional explanation of the indicative imperfect

The attempts to give a flectional explanation of the indicative imperfect as a suffixal formation were connected especially with its parallelism with the Slavic -ě-achb imperfect. This parallelism is also the starting point of recent explanation given by Erhart (1961 — see also 1982: 178,195, and 1983) and Pohl (1977). Both the Latin and Slavic imperfects are -ă- preterites, containing an -ă- suffix with preterital value. Common to both formations is the long vocalic stem in -ê- used also in primary verbs (3rd conjugation in Latin). Cf.

Page 205: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

188 FROM INDO-EUROPEAN TO LATIN

the parallelism:

(46) Latin OCS mone-bam to moneo m ně-a-ch to m ně-ti (inf.) dice-bam to dic-o nesě-a-ch to nes-ti (inf.)

Not only verbs with present stem in -ě- (Lat. vide-o, OCS (inf.) vidě-ti, but also primary verbs like Lat. dic-o, OCS (inf.) nes-ti contain the -ē- (Slavic -ě-) suffix in their imperfects.

Erhart does not treat the problem of the inserted consonant in the Latin -bam imperfect, Pohl characterizes it as a derivational formant without meaning used for inflectional purposes, enabling the roots with final vowel to take a formant with initial vowel.

We may assume that this consonant goes back to the IE -dh- suffix180

whose semantics and grammatical application, including the Greek -thē- aorists, were characterized by Benveniste (1935: 188ff.) and by W.P. Lehmann (1942, 1943) who explained the Germanic weak preterite as a formation with -dh- suffix. By adding the Latin -bam < -dham preterite to these Greek and Germanic prete­rites with -dh- suffix, we obtain a characteristic agreement of three IE languages in the innovative preterital formations.

3.7.3 The -dh- suffix in Greek, Germanic and Latin preterite: The Latin -bo future

Two important characteristics of the -dh- suffix follow from the analysis given by Benveniste and Lehmann: 1) its application to inactive roots; 2) its terminative aspectual (Aktionsart) value (achievement of a state).

As a lexico-derivational formant this suffix is well attested in Greek. It appears in a rather large group of verbs with clear inactive meaning and form, including media tantum such as "perceive", "be aggrieved", etc., the verb (aor. ) "learn" belonging to the inactive root *men- (presento-perfect ), the verb , med. "rejoice", etc. Among verbs in which the -dh- form is only one variant of the verbal lexeme (e.g. beside "spin", etc.), the intransitive value is apparent in the form "be full" vs. "fill". With respect to the transitive/active vs. intransitive/inactive value of the forms with -s- suffix vs. forms with -dh- suffix we adduce the -s- future of this verb as opposed to the -dh- form:

Page 206: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

LATIN IMPERFECT AND FUTURE 189

(47) I shall fill I am full

Similarly, as we observed in the case of the -s- suffix, the -dh- suffix, too, can combine with various types of vocalic endings. The IE languages differ in their use of the common derivational possibilities and capacities of the d-f structure, so that they also differ in the types of endings used in the grammaticalized formations with the -s- and -dh- suffixes. As for the sigmatic formations cf. the -se/so- futures of Latin and Greek vs. the -sie/sio- futures of OI, Baltic and Celtic, modal -se in Latin, -să in Celtic and -st in Latin, Baltic and Celtic (see note 168 on -si in Gaulish), preterital -să in Latin, etc. (see Schmid 1963b). Similar diversity of endings is found in the languages which used the consonantal suffix -dh- to express the intransitive (Greek), the imperfective/durative (Latin)181 or the basic (Germanic) preterite of weak verbs. It is the long vocalic -ê- suffix in Greek, the -ă- suffix in Latin, and the alternating -ë/ö- suffix in Germanic.

The Latin -bo/dho- future goes back to the -dh- suffix with thematic -e/o- endings which was grammaticalized to express the future of derived verbs already containing the long vowels -ê- or -ă- in their stems (or having roots with corresponding final vowel: nebam, plebam, dabam182 etc.). This type of future probably presupposes the already existing -bam < -dhām imperfect, and was established on the basis of the systemic analogy with the imperfect.

Page 207: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)
Page 208: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

3.8 Conclusions

In Part 3, we have dealt with the verb and concretized the charac­terization of the derivative-flectional and paradigmatic-flectional stages of the IE development outlined especially in section 1.3. We have tried to reconstruct the injunctive-based two-class system of PIE with its simple inventory of personal endings (two basic series) and tracc its development into the present based verbal system with paradigmatized inflectional categories. Lack of grammaticalized inflectional oppositions, but rich morphological diversity of the derivative-flectional formations expressing various Aktionsart-meanings are the typlcal features of the PIE system (section 3.1 and 3.2). The inflectional morphological categories developed from the reconstructed basic system along several lines. This development has been examined with special attention to Latin morphology and the difference between the Latin and Greek/Aryan verb which was the basis of the classical IE reconstruction. This difference results from the alternative ways of development of the integral verbal system from the original two-class system and of the grammaticalization of derivative-flectional formations. In sections 3.3 and 3.4, where the development of the basic aspecto-temporal and diathetic categories is outlined, it is shown that the simpler system of Latin has its correspondences in the verbal systems of other IE languages, whereas the three aspecto-temporal stems (present/imperfect vs. aorist vs. perfect) and the extensive use of the oppositional middle, are features limited to the Greek/Aryan area. The morphological basis of these differences in the fundamental aspect/tense system and in the diathesis was provided by the different use and constitution of personal endings and of inflectional stem characteristics.

The basic components of the PIE verb were also the vocalic suffixes -e/o- (thematic suffix), and -ă-, -ë-. In section 3.5 the role of the long vocalic suffixes -ā-, ē- in Latin, where they express the subjunctives and futures and serve as an underlying form to express the imperfect, is examined, again in

Page 209: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

192 FROM INDO-EUROPEAN TO LATIN

contrast with the situation in Greek. In sections 3.6 and 3.7, the consonantal suffixal formations serving in Latin morphology as an expression of marked tenses and moods (the sigmatic forms of present system, the -bam imperfect and -bō future) are analyzed. The synthetic explanation of the Latin imperfect has a broader impact on the conception of the later IE structural development. The slogan 'today's morphology is yesterday's syntax',183 does not yet hold for this stage of development as it does not yet reflect the basic structural tendency. The marked and innovative forms of Latin are also explained within the paradigmatization of derivative-flectional morphology, i.e. within the structure, where 'today's inflection is yesterday's derivation'.

Page 210: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

Summary

In the first Part of the book two developmental stages of Indo-European morphosyntactic structure are identified and characterized typologically: the derivative-flectional stage of reconstructed Indo-European and the paradigmatic-flectional stage of individual Indo-European languages. The characterization is based on an enlarged notion of the flectional type, which includes not only the level of formal morphology, but also that of morphological semantics and the syntactic-semantic level. The semantic interpretation of the flectional type, it is maintained here, reveals the underlying principles which determine this type of language structure and its development.

Flectional expression of the grammatical categories is not realized on an additive, compositional basis, i.e., by means of a certain segment with a discrete grammatical meaning, but on an integral/synthetic and decompositional basis. It is the internal modification of the word which serves to express the grammatical meaning. Its individual components are identified by morphological oppositions to other forms of the same derivational set or of the same inflectional paradigm.

Internal binding of grammatical meaning to the word is a common feature at both stages in the development of the flectional type. However, the derivative-flectional structure, reconstructed for Indo-European on the basis of the analysis of certain morphosyntactic phenomena of ancient Indo-European languages, clearly shows the semantic motivation for and the underlying principle of the flectional type, namely, the principle of primary grammatical relevance of the lexico-derivational categories.

The main manifestations of this principle are to be found in the following features of Indo-European morphosyntax: 1) Lexical noun and verb classes, i.e., animate vs. inanimate nouns and active vs. inactive verbs, differ as to morphological (case and number markers, personal endings, aspecto-temporal and modal suffixes) and semantic-syntactic charac­teristics. Yet, as shown in section 1.2, the existence of these classes in Indo-European does not imply that IE morphosyntactic structure was of an active or

Page 211: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

194 FROM INDO-EUROPEAN TO LATIN

ergative character. 2) Morphological semantics was of the word-category-inherent, non-relational type. Morphological processes expressed the semantic distinctions which asserted themselves within the word-category in question. They were not directly oriented to express the syntagmatic and contextual relationships with other categories. Primacy of aspect before tense as a deictic category related to the hic-and-nunc situation, and the semantics of Indo-European cases as opposed to agglutinative affixes are the clearest examples of this feature. 3) Inflectional processes were closely related to derivational processes by their distribution (application to certain word classes only) and semantics (word-category-inherent distinctions). Broad variation in form and allomorphy, which are characteristic of flectional morphology, are in part semantically motivated (cf. the specific meanings of aspecto-temporal and modal formants).

Implicit semantic distinctions were made explicit in the process of paradigmatization. For instance, nominal lexemes which were originally vague as to number, were introduced into oppositions of number.

Contrary to the prevailing oplnion according to which Indo-European flectional morphology had an analytic origin and developed on a compositional basis, the concept of derivative-flectional structure suggests the synthetic origins of the flectional morphology. The development from derivative-flectional to paradigmatic-flectional structure is an evolutional process working on a decompositional basis and proceeding from a non-differentiated, synthetic word structure to a differentiated and more analytic structure.

In the second and third Parts of the book the development of the nominal and verbal systems is treated in more detail within the selected typological frame-work (see the summaries in 2.6 and 3.8). Latin is a good example of a language with highly developed paradigmatic inflection and therefore a good basis for contrasting the two stages of evolution we are presenting in this study. At the same time, Latin paradigmatization is also contrasted with alternatives in other Indo-European languages and groups of languages. The comparison of the Latin verbal system with the Greek/Aryan verbal system, which has been the basis for the traditional reconstruction of the Indo-European verb, is given special attention in the present work. Seen in a broader comparative perspective, agreements between Latin and other language groups become obvious. The main features of the alternative paradigmatization of the Indo-European verb are bipartite vs. tripartite aspecto-temporal systems, different types of middle-passives, different formations of subjunctives and imperfects, and differences in the application of sigmatic and other consonantal suffixes.

Page 212: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

Notes

1. I use the terms 'flection', and 'flectional' (type, structure, language)' in reference to types of language structure (as opposed to agglutination and isolation), whereas the terms 'inflection', and 'inflectional' refer to the grammatical part of morphology (as opposed to derivation, word formation). — The terms 'derivative, derivational' used in the combination 'derivative-flectional' or in characterizing IE flectional morpho­logy as 'derivational' can be misleading if they are taken in the strict sense. This might suggest that the original IE structure was a structure without 'grammar', i.e. without expression of semantic-syntactic categories and relations. What this term implies in our conception is that the morphological devices used to express grammati­cal functions were semantically, distributionally, and morphologically related to the morphological devices serving the purpose of derivation proper, i.e. functioning at the lexical level. — Observations concerning the derivative-flectional character of IE morphology have been already given in Brugmannian tradition. Yet, the derivational theory as an explanatory theory of flection has not been systematically elaborated. The so-called adaptational theory as opposed to agglutinative theory (see below, and note 2) by Ludwig (1873) and his followers (see Oertel & Morris 1905) differs from our conception by considering the elements subjected to morphologization as originally semantically empty.

2. This widely accepted theory dates back to August Schleicher, see esp.Schleicher (1950:14). On Schleicher's evolutional theory and its biological foundations, see Koerner (1981, 1989), Romportl (1989). On this conception of typological develop­ment, see also Koerner (1989:334), Drobin (1980).

3. This aim of classical typology is well defined by Coseriu, see especially Coseriu (1980:199-200, 1988,I,185ff.). Coseriu's contributions to typology are published in Albrecht et al. 1988, I,161-224 and discussed in III,1-208. Cf. also Sasse's (1988) concept of 'immanent typology'. An analysis of the various conceptions of typology has been given by Ramat (1987). See also the discussion about typology in Folia Lin-guistica 20, where Plank (1986) and Sgall (1986) represent important contributions to Flection/Agglutination/Isolation typology. — The terms 'typology' and 'typo­logical' are used also for general comparative research represented by Comrie (1983), Mallinson & Blake (1981), Shopen (1985), and other, whose aim is not to define the

Page 213: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

196 FROM INDO-EUROPEAN TO LATIN

types of languages. This is justified to the extent to which these contributions aim at typlzation of linguistic features. Typology investigating structural connections going across the individual functional domains of the language structure, as represented esp. by Croft (1990), is highly relevant for the semantic interpretation of the IE flectional type and its development (see note 5). Among projects on morphosyntactic typology cf. especially Seiler's UNI-TYP project as outlined in Seiler & Lehmann (1982) and the projects presented by W.P.Lehmann (1978, 1986, 1990). — After some stagnati­on, the questions of typology, so stimulatingly analyzed by Skalicka (e.g. 1979), are now being intensively discussed in Prague (see esp. Popela 1985, Sgall 1988). Of course, there are differences among the individual scholars which I cannot charac­terize here. My personal attitude has been expressed in these introductory remarks. — On the typology of Latin, see C. Lehmann (1979), Trost (1962).

4. See Schleicher (1850:14), Sgall (1986:15). See discussion by Popela(1985:45-46) in favour of flection-agglutination continuum as emphasized by Skalicka after the separation of 'introflection' as a special type. In our conception, the main typlzation line lies between flection (including introflection) and agglutination, both these types being based on different (decompositional vs. compositional) principles. — As for the general characterization of flectional morphology, I have consulted esp. Anderson (-1985), Matthews (1972), and Plank (1981, 1986).

5. Especially with respect to these differences between European languages, Skalicka (1979:335ff.) expressed the oplnion that languages are essentially polytype-structures. By using a flexible base of typlzation, not restricted to the formal morphological differences we get another plcture. The more synthetic and more analytic forms are not distributed at random, they correspond to the semantic graduated hierarchy of 'non-relational, internal, (word category inherent) vs. relational, external' — see Coseriu (1988), Kurzová (1974). This iconicity seems to have a general if not universal validity; — see Bybee (1985,esp.20-33), Croft (1990,esp.176-180); see already Saplr (1921:1OI), and Greenberg (1966:95) as for number vs. case rela­tionship (cf. below 2.3.2.1).

6. I use the term morpheme in the sense of minimal sign with a certain form and meaning.

7. Lexical subcategorization is also a morphosyntactic phenomenon inasmuch as it concerns the grammatically relevant semantics affecting the semantic-syntactic structure of the sentence.

8. I have used Szemerényi (1980) and Erhart (1982) as modern handbooks of IE linguistics. I cite these handbooks only in those instances when referring to information not generally accepted.

9. If not indicated otherwise, I will follow the usual practice of giving the lsg. forms of Latin and Greek verbs and the 3sg. forms of OI verbs. Translations are given for Greek and OI only in cases where the meaning is relevant (which is not the case here).

Page 214: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOTES 197

10. The terms 'Pre-Indo-European' or 'Pre-Proto-Indo-European' (so Rix 1988) are sometimes used if the reconstruction relates to the more remote stages of reconstructed Proto-Indo-European. I often use the term 'Indo-European', without further specification, also for the reconstructed IE structure in its relatively common or dialectically more-or-less differentiated stages of development. If the specification is preferable, I use the term 'Proto-Indo-European (PIE)' for the structure before the separation of Hittite, whereas the term 'Common Indo-European (CIE)' for the structural traits not shared by Hittite.

11. Further on, the term 'nominative structure' is used because it is the nominative which includes the agentive subject in its function, and is thus comparable to the ergative or active case as cases expressing an agent.

12. See the critical analysis of the ergative hypothesis by Villar (1983). I have formulated the main arguments against ergative/active explanation of IE in Kurzová (1986). — Among recent contributions in favour of the ergative hypothesis Schmalstieg (1980) presents original views especially on nominative - genitive relationship.

13. See Klimov (1973, 1978), K. H. Schmidt (esp. 1979, 1986 — a modified version of the active hypothesis, that also considers new contributions, i.e. esp. Villar's (1983) hypothesis about the priority of the accusative). On verbal activity see also Perel'mu-ter (1977), Gamkrelidze & Ivanov (1984:267ff.).

14. I use active vs. inactive as established terms, though I would prefer the clearer expressions agentive vs. non-agentive. The homonymy with the term active (as opposed to passive) in verbal voice is avoided by referring to the opposite term in the respective contexts.

15. Ivanov (1981) suggests the analytical origin (as a demonstrative pronoun) also for the IE -s formant, not explaining, however, why we have word inflection instead of the expected group inflection (the demonstrative elements are placcd only once for the noun phrase). This is a typlcal example of how the differences of language structures have been neglected by scholars propounding ergative theories.

16. We must consider the function of the IE accent in connection with the entire character of the IE word, which due to its complex and hierarchical structure, was sufficiently marked as a unit so that there was no need of a specific means for its demarcation. Therefore, the accent was free to become involved in the expression of paradigmatic oppositions.

17. The long vowel in Latin words is marked only if it is relevant for the discussion. Otherwise only long vowels in the formants (separately quoted endings and suffixes) are marked.

18. This research, following suggestions made by Kuiper (1942) and Pedersen (1926), is represented especially by Schindler (1967a, 1975ab, 1977), Rix (1976:122ff.), and Beekes (1985) for nominal morphology. For the verb, see especially Narten (1968), Insler (1972). Cf. also 2.1.3.1, note 32, 3.2.3.6.1, notes 115, 116 (below).

Page 215: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

198 FROM INDO-EUROPEAN TO LATIN

19. The absolutive and the oblique cases of heteroclitic neuters have different stems: abs. -r- stem vs. obl. -n- stem. Cf. abs. femur vs. obl.feminis, abs. iecur vs. obl. iecino-ris from iecinis. See Szemerényi (1980:158; 1956:191; 1975:332ff.= 1987a:264ff.).

20. On morphology, see esp. Dressler et al. (1987) with Panagl's examination of his­torical problems; further, cf. Dressler (1985, 1986), Mayerthaler (1981), Panhuis (1980, 1981), Wurzel (1984). On iconicity in general, see esp. Bouissac, Herzfeld & Posner (1986).

21. This is particularly evident with respect to the 1st persons, where it is clear that nos does not mean many egos. In the other two persons, the singular and plural terms dif­fer by illocutory features related to their different position in the discourse situation, the plural notion being thus semantically complex, and including such features as 'in­definite, polite', etc.

22. We find the pres.dep. - perf.act. correlation only in revertor - reverti in Latin (and in a few later artificial forms such as assentior - assensi).

23. The cross-linguistic variation on noun - verb distinction, its degrees and parameters, see especially Walter (1981), Broschart (1987), Sasse (forthcoming).

24. On the definition of nouns as referential expressions vs. verbs as predicational expressions, see Sasse (forthcoming). The terminology defining denotational correlates of the sentence and word categories is not unified. The correlate of the sen­tence is labelled 'situation' (as in our use) or 'state of affairs' (which in our use is a correlate of the verbal predicate). The denotational correlate of the prototyplcal noun is labelled 'individual' or 'object' from the logico-semantic point of view, and 'par­ticipant' if the semantic-syntactic aspect prevails; the correlate of the prototyplcal verb is called 'event'. See also 2.4 (below).

25. As the term 'property' is too unequivocally associated with adjective and attribute, the term 'aspectual' as used by Wilensky (1990) would be perhaps appropriate, which means a denotational correlate of nouns in general, and is thus capable of covering the continuum between adjectival and substantival nouns. 'Aspectual' in Wilensky's sense of the term is opposed to 'absolute', a term which refers to an individual whose reference is already established prior to the sentence forming act (cf. 2.4.3.2 below).

26. The fundamental position which the intellectualized language style in modern IE languages obtained (under a strong influence of Latin elaborated prose) was decisive for this development. I cannot pursue this subject in detail here.

27. On IE agreement, cf. also Boeder (1973) and Kurzová (1974).

28. Meillet characterizes the IE sentence structure as 'appositional' (1934:350, and similarly p.360): "l''apposition' est le procédé qui caractérise d'une manière essen­tielle la syntaxe indo-européenne". The IE subject is also conceived by Meillet as a kind of apposition to the subject person incorporated in the verb.

Page 216: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOTES 199

29. On the allegedly non-marked character of the SOV word order in IE, see esp. W. P. Lehmann (1974:39ff.), Gamkrelidze & Ivanov (1984:320ff).

30. More about this type of subordinate clause Greek and in archaic Latin, see Kurzová (1989:34).

31. In this schematic representation of alternations -é- represents a full grade; the distribution of the -o- realization is not examined. For the nominative the lengthened vowel is possible reconstruction, hence pěd(s).

32. As for the teminology used by individual scholars see especially Eichner (1972: 91,n.33), Rix (1976:123), Beekes (1985:1). As I am interested (see 1.1.3.4) in mor­phological alternations and not in their motivation by accent, I use a simplified ter­minology, not distinguishing the static inflection as a specific type, though the inflection with stable root accent seems to be attested for both nominal and verbal (see 3.2.6.3) morphology. - With respect to noun inflections I use the following terms: 'proterodynamic' for alternations between root and case suffix, 'hysterodynamic' for alternations between stem suffix and case suffix, and 'am-phidynamic' for alternations between root and stem suffix. Therefore, the term 'pro-terodynamic' includes all inflections in which the stem suffix is not engaged in alternations. Either a root, e.g. in the type nom. *nokwt-s vs. gen. *nekwt-s or nom. döm(s) vs. gen. dem-s (see Schindler 1967a), or a case suffix, e.g. in the type nom. *ou-i-s vs. gen. *ou-i-os, exhibit alternations. I leave the question open whether the vocalized gen. suffix in *ou-i-os is secondary, i.e. introduced analogically from root nouns (*ped-s vs. *ped-os) or from hysterodynamic type (*patêr(s) vs. *patr-os).

33. This was shown by Benveniste (1935:especially 52ff.). His characterization of non— ablauting -i- 'élargissement' and ablauting -i/ei- suffix as two different formants has, however, become superseded by the more recent research.

34. Here only nom. vs. gen. alternations are given. The loc. vs. dat. alternations are not attested by OI -i- and -u- stems, where the loc. ending goes probably back to n o n -marked locative (2.4.6.1, note 63).

35. The occurrence of both possibilities with the same noun is probably a secondary phenomenon, given the fact that the original distribution reflects the accentual types of words.

36. Root alternations are not directly attested in IE languages, but the well attested fact that different languages generalized different grade forms of the root nouns justifies the reconstruction of root alternations: cf. Lat. *ped-s > pes vs. *ped-es > pedis and Gr. *pod-s > vs. *pod-os > . The alternation -e- vs. -o- is supposed to have replaced the alternation -0- vs. -e-.

37. See Knobloch (1950-52:147), Watkins (1969:16,24,28), Blümel (1972:110ff.), and recently Rix (1988:107).

Page 217: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

200 FROM INDO-EUROPEAN TO LATIN

38. The IE type of inflection was designated as 'stem inflection' ('stammflektierend'in Finck 1923:126) because of the fact that the different stems have different inflectional forms. This character of IE inflections explained by the d-f origin of IE inflection and by the wordstructure of both d-f and p-f subtypes. The term 'stem inflection' referring to the type of word inflection is not used in this study.

39. However, only contact position, not fixed relative order, is obligatory in the group--inflected languages. In Bulgarian for instance both positions, i.e. adjective - noun and noun - adjective are possible with obligatory contact position and with postponed article standing always in the first member of the group:

. — As shown in section 1.2.1.2 above, the agreement presents counter-evidence against the analytical origin of cases from the demonstrative elements.

40. The arguments given for the second alternative are not compelling, as shown by Meid (1979:167ff.)-

41. On the opposition animate vs. inanimate as a characteristic feature of PIE structure, see esp. K. H. Schmidt (1979, 1980, 1986, and elsewhere.)

42. Cf. Meillet (1921:215ff.). The fact that the same languages have in both instances the same (animate or inanimate) class speaks in favour of semantic/notional relevance of this distinction.

43. See Hardarson (1987:81) for the evidence for this construction. As for the situation in Greek (regular only in Attic), see Humbert (1954:74-75).

44. This does not mean that the classification of masculine vs. feminine is in principle impossible without the preceding classification of animate vs. inanimate. We have masculines and feminines without neuters in Semitic and Hamitic languages. On areal extension and structural characteristics of the male vs. female (non-male) distinction, see Bechert (1982).

45. Yet the semantic motivation for gender assertion should be presupposed also for the collective/abstract in -ě < -eh2 with respect to the fact that verbal root abstracts are also feminine (vox, etc.). The use of-eh2 grade for feminine/abstract vs. -h2 grade for inanimate/concrete (n.pl.) can be considered as marking the animate nominative sub­ject by alternation.

46. Non-oppositional as not forming regular grammaticalized oppositions for the same lexical stem. Also in cases where the -o- masculines and -ă- feminines are formed from the same nominal root or stem (amicus vs. arnica, dominus vs. domina, servus vs. serva) this opposition belongs to the lexical word formation level and joins other derivational formations expressing motion.

47. I have consulted Welmers (1973) in particular.

Page 218: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOTES 201

48. For -i- and -u- stems we must assume two types of non-sg. forms with -e/-h1 suffix: the above mentioned -iH/~uH forms with zero grade of the stem vowel -i- and -u-(dual forms of OI) and -eie/oue- forms with full grade of stem vowel, forming a basis for the nom.pl. forms -eies, -oues; cf. OI dual matī vs. pl. matayah.

49. Though the -n variant is attested in more languages (Gr., Celt., Germ., Hitt., Venet., Messap.) than the -m variant (Aryan, Italic), -m seems to be an original form of this suffix. Note the possibility of explaining the variant -N just with respect to the relation between acc.sg. and acc.pl. In acc.pl. the nasal -m- was probably assimilated to -s (dental series) and the assimilated sound -n- was then transferred to the singular. If this is true, the value of -s as a representative marker of the plural is confirmed by this presupposed change: -m vs. -ns > -n vs. -ns.

50. The nom.pl. in -ös is not attested in the above mentioned group of languages. In fact, the only support for assuming that the -ōs form was original ending in this group is the Olr opposition of nom.pl. fir < *uiroi vs. voc.pl. firu < *uirös which can be explained by the fact that the innovative -oi form was applied only in nom., whereas in voc. the old form continued to be used; see K. H. Schmidt (1975). In Latin the homonymy of nom. and acc.pl. which we have in 3rd-5th declensions was prevented by the use of the -oi form.

51. The reconstruction of the -e- and -o- grades of the given suffix is based on the observable distribution in the attested forms: both -es, -os; only -om; only -ei. In general, the distribution of the -e- and -o- grade is not clear.

52. Kuryłowicz (1964:195) considers the opposition -C vs. -VC only for loc. and dat. which are both local cases in his conception. (In other theories dat. is a grammatical case; see e.g. Jacobsohn (1923) — cf. K. H. Schmidt (1975:277) on the theories of local and grammatical cases).

53. See also de Groot (1956a, 1956b) and, for Latin, Lavency (1969). On the conceptions of case oppositions, see Calboli (1972:82ff.), Hubka (1980).

54. Modern analyses of grammatical meaning are inspired especially by Fillmore (1968), Chafe (1970) and Lyons (1977). The concept of 'participation' is elaborated recently in Seller' s UNI-TYP project (see Drossard 1986, Broschart 1987, C. Lehmann 1988:33ff.).

55. The necessity of considering referential roles as distinct from relational roles like agent etc. is acknowledged in recent research. See on this Genuisienè (1987:44ff.) with further references. See also the discussion on IE subject by Sasse (1982) and in 2.4.3 (below).

56. The relationship between possessivity and subject/object functions is analyzed by Seiler (1983), especially with respect to the verb affixes expressing subject or object which also apply (often in derived forms) to the possessive structures.

Page 219: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

202 FROM INDO-EUROPEAN TO LATIN

57. The lengthened nominatives have been explained by compensatory lengthening following the elimination of -s (see Szemerényi 1980:109) or by other phonological or morphonological change. See the discussion by Villar (1983:49ff.).

58. In the main points, I follow Sasse (1982) where a critical analysis is made of the other conceptions of IE subject. On the role of agreement for the constitution of subject - verb relation, see Heger (1982).

59. The nominative marking was also understood as marking animate class in traditional IE research. The relationship between class marking and case marking has been analyzed by Baudiš (1925).

60. See Villar (1981). The dat. and loc. were explained as two ablauting variants of the same suffix already by Brugmann (1911:122) and by Kuryłowicz (see note 52 above).

61. The shortening before -r is regular. For the shortening of final -ă Leujeune (1949) and Bliimel (1972:39) presuppose a phonological change which took place in Latin and Osco-Umbrian. The common explanation is that it occurred by extension from iambic words. Moreover, structural motives can be considered: dehomonymization with respect to abl.sg. in -ă; the tendency of nominatives sg. to be less marked and voluminous, which is to be observed in all declensions with the exception of the type nom. vates vs. gen. vatis.

62. Explained by syncopation and elimination of -s in final cluster: *agros > *agrs > *agers > ager.

63. See Brugmann (1911:174-175), Benveniste (1935:87ff). According to Neu (1980), all instances of a Hittite non-marked locative (used also in directive function) are neuters except -t- stem adverbal forms siuat "am Tage", lukat "am (nächsten) Morgen". According to Starke (1977) the notional distinction animate vs. inanimate strongly determines the distribution of cases in Old Hittite texts. The situation with -i- and -u- stems is more complicated in consequence of the mutual influence of inanimate and animate declension types.

64. Cf. Schmid (1973) on the restricted adequacy of the traditional names of IE cases.

65. See, however, Haudry (1979:35, 1980:135ff.) who explains the OI (Aryan) instrumental in -ă as a postposition, by referring to the group inflection in Vedic formulas navyasä vacah (= navyasa vacasa).

66. So Coleman (1987). For other possible instances of instrumental sg., see Bartoněk (1987:123).

67. From both morphonological (-t as a phoneme characteristic of formatives) and etymological (adverbs with -t elements) points of view, -t seems to be a more plausible reconstruction. See on this esp. Schmid (1973): -d was reconstructed on the basis of Latin where, however, -d can go back to -t as well.

Page 220: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOTES 203

68. The postinflectional/postpositionalcase suffixes were already presupposed by Meillet (1922:50-51) who also included the locative in -i in this group. On dental suffixes, see esp. Schmid (1973). On Haudry's explanation of OI (Aryan) instrumental in -ă as postpositional, see note 65 above.

69. The Baltoslavic -o- stem genitives (Lith. -o, OCS -a) are supposed to reflect -öt/d, with dental non-attested directly but by the closed character of the syllable (see K. H. Schmidt 1977).

70. The semantic difference between instrumental without -d and ablative with -d is, according to Prat (1975:67), well preserved in -o- and -ă- stems, whereas -/- and -u-stems have forms with -d in all functions.

71. In the new inscription from Satricum (about 500 B.C.) published by Stibbe in Stibbe et al. (1980), see de Simone (1980:82-83) and Untermann (1986). Until now the form -osio was attested within Italic group only in Faliscan (see Giacomelli 1963:142-4).

72. See especially Schmid (1973) for the postinflectional character and the etymology of these suffixes. The historical relationship of the underlying form in -a, which is traced back to the long -ă suffix by Laroche (1970:46) — with correspondence in ad­verbs like Gr. aod, Lat. eo, etc. — is not quite clear.

73. In plural, the gen. in -an is attested in old Hittite texts. According to Starke (1977:19) the .variant -as is not attested for plural in old Hittite texts; see, however, Neu (1979:192). In the newer variants of Hittite the gen. in -as is generalized; see Neu (1979:192-193).

74. See 2.4.5 above. Both forms (-i and -ei) are also allomorphic in Mycenaean. See Ruijgh (1967), Coleman (1987).

75. According to common interpretation — see Morpurgo Davies (1966), Coleman (1987), Bartoněk (1987) — the -bhi form is paradigmatized as an instr.plural in Mycenaean (though local functions are also attested), whereas its use in Homeric Greek is greater. This difference is, however, not essential, the alleged paradig-matization being of transitory character.

76. See Porzig (1954:90-91) who considers two differences in case formation as relevant: gen.sg. and dat./instr.pl.

77. A very systematic analysis of the intraclassal and interclassai relationships within the Latin case system is given by Risch (1977).

78. The main point is an analysis of the Latin fifth declension which was constituted from different IE formations (diphthong stems as dies, -ê- (-eh1.) stems as fides, root nouns as spes, etc.). See Pedersen (1926), Risch (1977:230), Beekes (1985).

79. Whereas other variants with -i go back to the original -i- stem forms, this seems not to be the case of the acc. in -im, as the original -im would also give -em: cf. quern < quim. The ending -im goes back to -On of the -i- stem feminines like vim, neptim, cutim, etc. See Meillet (1906:30, 1925:131); Novotný (1946:127). For other opinions

Page 221: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

204 FROM INDO-EUROPEAN TO LATIN

on the matter, cf. Sommer (1914: 374); Blümel (1972:72).

80. According to Haudry (1983) the starting point of the mutual influence of consonantal and -i- declensions was dative which after the simplification of *mnt-ey-ey to *ment-ei terminated in both declensions in -ei > -ī.

81. See Risch (1977:244): "-ium (ist) günstiger als -um [...] wegen der volleren Form und der Morengleichkeit mit den übrigen Pluralformen (— oder - - ) " .

82. As is known, another structural tendency which played a role in the distribution of the gen.pl. in -ium vs. -urn and abl.sg. -i vs. -e is the characterization of adjectives. The historical source of this tendency was supplied by the -i- stem adjectives.

83. The systematic homonymy of the nom. and acc. of neuters (= original absolutive) is not included in this survey.

84. The observation that paradigmatically related cases (nom.acc. and dat.abl.) tend towards homonymity was made by Williams (1981:267ff.), who, however, did not distinguish between cross-number and same number homonymy. See the critical discussion by Joseph & Wallace (1984).

85. 'Non-oppositional' in the sense of not forming morphological oppositions within the same lexeme. The distinction '(non-oppositional) morphological class (i.e. grammatically relevant lexical class) vs. (oppositional) morphological category' was introduced already in the Part 1, section 1.1.5.1 and used also in Part 2.

86. The general characterization of research problems concerning IE verb is given inter alia by Stang (1942:7ff.), Lane (1959:157), Adrados (1963), Bader (1972), Meid (1979), Kortlandt (1984), Erhart (1988). The conception of original verb system corresponding to our d-f structure is briefly outlined by Kuiper (1937:37-38).

87. Cf. Rix (1986) who suggested a similar use of the term Aktionsart to cover the IE verbal formations.

88. This is denied by Perel'muter (1977:117ff., 1984).

89. As known, it was Kuryłowicz (1932) and Stang (1932) who deduced the endings of perfect, middle and Hittite -hi conjugation from original series of person markers.

90. Here, and in similar cases where basic structures and developments are analysed, I provide a detailed breakdown of IE morphological forms for those readers not specialized in IE studies. Otherwise, when discussing more specific questions a basic knowledge of IE research is assumed.

91. For the sake of simplicity I use the term 'aorist' also with respect to the basic active preterite of languages (such as Latin) where no aspectual opposition imperfective vs. perfective was grammaticalized. This is possible due to the solidarity or preference relationship between perfective and preterital meaning (see e.g. Salkie 1987:139).

Page 222: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOTES 205

92. I.e. with presupposed non-systematic active vs. inactive counterparts formed from the same verbal root via derivation.

93. On the distinction progressive vs. non-progressive, see Comrie (1976:32ff.), Dahl (1985:24ff.) and the discussion by Salkie (1987).

94. The distinction agentive/active vs. non-agentive/inactive is multidimensional and its semantic focus varies cross-linguistically with respect to one type of the formal device (verb classes; diathesis/voice as verbal category; agentive case) and intrastructurally with respect to different formal devices present in the given language structure (agent as subject of active verb in general, including verbs not forming passives; agent as participant subjected to passivization, etc.). For IE diathesis the opposition-external/extrovertal vs. internal/introvertal was decisive. See also discussion in 3.4

(below).

95. Cf. K. H. Schmidt (1986:93ff.) about the lack of evidence for this hypothesis.

96. They are examined by Specht (1935, esp.42ff.) and Porzig (1954:91-92); see below 3.2.2.2 under (15).

97. In individual forms the inactive character comes to light: Gr. 3sg.med.fut. "he will appear", as unique form of non-containing the -n- suffix; Gr. 3sg.med. , part. to "speak, say". — The verbs of saying are inactive if they go together with the verbs of thinking, seeing, and other verbs of psychosensory processes, cf. also Lat. aio going back to eH- root which appears with velar or dental 'enlargements' in IE.

98. We can consider the possibility that these roots received consonantal endings instead of non-distinct vocalic/laryngeal endings in the process of the integration of both clas­ses (as I suppose also for the long vocalic 'injunctives' in -ă- and -ê-). Probably, the -s endings of Hitt. preterite of -hi conjugation (2sg. and 3sg.) and possibly also Tocharian -s endings can be explained by the indistinct character of inactive endings (especially in the case of roots terminating in a vowel). This element was probably applied postinflectionally (perhaps functioning as agreement device), its identity with derivational -s- suffix (presupposed by Watkins 1962) is not probable.

99. With respect to the verbs like dico, iungo (see above) it should be assumed that certain suffixal determinatives could change the inactive verb into the active one.

100. It is interesting to note that in his proposal to distinguish two different verb roots instead of a unique one, Sihler (1977) ascribes the active meaning to the CEC, whereas the inactive meaning to the CEHC root (without elaborating a systematic difference between active and inactive verb class): *( -"direct, rule" (Lat. rego, Gr. ) vs. *rehjg- "be efficacious, have mana" (Lat. rex, Gr. ). This is not accepted by Strunk (1987) who assumes the unique acrostatic verb with a generalized long grade in Aryan, whereas a generalized full grade in Latin, Greek and Germanic. See, however, 3.2.3.6 (below) about the possible inactive origin of acrostatic inflection.

Page 223: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

206 FROM INDO-EUROPEAN TO LATIN

101. With this wider sense the term 'injunctive' was applied to cover the use of Vedic non-augmented verbal forms with secondary endings and preterital, modal (subjunctive), but also generic present functions. See especially Hoffmann (1967). As stated already in 3.1 (above), I use the term 'injunctive' not only with respect to the simple athematic and thematic forms, but also with respect to suffixal formations with -ă-, -ě-, -s- suffix, etc. which, too, attest both preterital and modal realizations of the same underlying semantic/functional unit defined as 'non-actual'.

102. The interpretation of the form with -i' as present is confined to the active verb, in the inactive verb both present (Gr. and 0I middle-passive presents) and preterital (Lat. perfect, Hitt. and Toch. middle-passive preterites) value of the forms with -i are attested. This suggests the wider actualizing use of this particle, in statements where a hic-and-nunc situation is relevant and respected.

103. The preterital value was signalled (probably also optionally as especially the Avestan and Homeric use attests) by another particle, preposed to the verb form, the so-called augment. This e- particle is attested only by Greek-Aryan and by individual Armenian forms.

104. On the recent character of type see Oettinger (1979:314) with bibliographical references, Bammesberger (1985) with reference to Cardona (1960) which I have not consulted.

105. Root aorist OI lsg. adhām, 3sg. adhāt is attested, whereas the present forms are of the reduplicated type in OI as well in Greek: lsg. dadhămi, 3sg. dadhăti,

< -ti. See above under (20).

106. Cf. for the 2sg. in -ei Lith. nest < *(d)nekei and Olr conjunct form -bir < *bherei. Greek has in 2sg. -ei-s with consonantal ending -s added. See Meillet (1924:191-192), and discussion by Szemerényi (1980:218ff.).

107. OI bharămi also continues IE thematic ending -ā < -oh2 with -mi added from the athematic forms. The thematic -ă ending without added -mi is attested in Gatha-Avestan; see Beekes (1988:154).

108. See the discussion about Hitt. -hi conjugation in Neu, ed. (1979) with contributions by Cowgill, Jasanoff, Kuryłowicz, and Meid. See also Oettinger (1979, 1985).

109. Possibly, both formants originally differed when we assume that inactive -e ending contains laryngeal -h1e, whereas the thematic vowel of active class was purely vocalic -e (see 3.2.3.5.1 below). By the realization of -h1e as -e both formants became homonymic.

110. The preference for the inactive ending in the lsg. is to be understood with respect to the internal relationship between person and verbal process in the case of the inactive verb vs. external relationship between agentive person and verbal action in the case of the active verb.

Page 224: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOTES 207

111. The thematic present endings on the basis of inactive endings can be explained in more than one way (see explanations by Watkins 1969:103ff., Jasanoff 1978:50ff.). The vocalic/laryngeal base of both series of endings speaks clearly in favour of their historical relationship. Nevertheless, the derivation of thematic present from inactive endings is not commonly accepted; see Cowgill (1985).

112. The -ur ending occurs as a secondary ending of 3pl. not only in the perfect.

113. The stem vocalism long vowel before d2 and short vowel before d1 follows the same rule which holds for the perfect paradigm. For the form ajani with short vowel for both lsg. med.aor. and 3sg.med-pass.aor. see Narten (1964:117). The explanation of -i < -d1 as a variant of inactive ending is in keeping with Migron's (1975) characterization of the Vedic middle-passive form in -i as impersonal (see section 3.2.1.5, on the non-oriented character of the 3rd person of inactive verb, and cf. above 3.2.3.5. on the impersonal/indefinite -r ending).

114. The identical endings for the 2nd and 3rd persons are also attested in the Hittite preterite of the -hi conjugation and middle-passive preterite of both conjugations.

115. See Bader (1980a) on the origin of the 'proterodynamic' presents in the basic category of verb which corresponds to our inactive.

116. Quoted after Insler (1972). In pres.act. the verbs śas and stu, and further the verbs takš "hew" and daś "honour" also exhibit a full grade on the place of the normal null grade in the plural; in the singular they have a long grade. This seems to be a secon­dary phenomenon with -ă- vs. -å- gradation having replaced -e- vs. -0- gradation. Certainly, the long grade cannot be explained directly by the accentuational type; cf. Strunk (1987:385ff.). The correspondences to the above verbs are also in Avestan; see Kellens (1984:91ff.), Beekes (1988:164ff.). For Hittite, see Oettinger (1979:99ff.) and Bader (1980a).

117. See Kuiper (1937). On the -ne/n- and -neufnu- presents, see Strunk (1967).

118. See on the Hitt. -mi verbs with simple thematic vowel Oettinger (1979:257ff.; 1985).

119. Doubts about the IE origin of the tripartite aspecto-temporal system present vs. aorist vs. perfect have been expressed in many contributions made to the revised reconstruction of the IE verb. Cf. e.g. Bader (1972:19).

120. See note 91 above. If the aorist is opposed to the imperfect, then the imperfect is the marked member of the opposition, in all its realizations (i.e. imperfect as an aspectual category of Greek/Aryan as well as imperfect as an aspecto-temporal category of durative preterite).

121. The perfect in its original stative or later resultative value can be viewed as +perfec-tive(confective) with respect to the present/imperfect, but as —perfective with respect to the aorist, the constitutive feature of the perfect being non-progressivity. See above 3.2.1.2 and note 93 on the distinction progressive vs. non-progressive. As to the development of IE perfect, especially its resultative value, see K. H. Schmidt (1964).

Page 225: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

208 FROM INDO-EUROPEAN TO LATIN

122. The individual doublets such as par si beside peper ci, etc. are to be considered as parts of IE 'form-variety' or diversity, i.e. within the d-f structure with its various derivational possibilities of particular verb classes and verbs. They are not relics of the grammaticalized inflectional opposition of aorist vs. perfect.

123. The explanation of the Latin perfect endings after the reconstruction of perfecto-medium/inactivum was mainly given by Untermann (1968); on 3pl. endings, see Bader (1968), also de Simone (1980:75-76).

124. See Leumann (1977:608): "das gleiche Element ist nach unsicherer Vermutung auch im Indik. perf. in 3. Plur. -ěrunt und 2.Sing, und Plur. -istl, -istis erhalten".

125. See Petersen (1933:28), Bonfante (1941:205-206), Cowgill (1965:172-173), van Windekens (1982:281-282). The Greek -stha forms are attested in several Homeric forms such as imperf. , etc. The -stha ending is explained either as an original form of perfect ending or by the contamination of the -tha ending with the secondary -s ending.

126. For Lat. -istï this was stated by Schmid (1960:300), Watkins (1962:1 Iff.), Bader (1968).

127. The long vocalism of -ere has a correspondence in Tocharian.

128. The sigmatic aorist is restricted in the Rgveda to the roots of the CV and CEC type, not occurring in the CEIC, CEUC, CERC roots typical of inactive verbs. See Narten (1964:18).

129. For the diathetically based distribution of Greek aorists, see Kølln (1969; on the reduplicated aorist p. 14-15).

130. We can say that the intransitivity or transitivity is given by the type of present: the intransitive -io present on the one hand (OI) vs. the transitive nasal present on the other (Lat.). However, the fact that basic present formation is intransitive in one case (OI), whereas transitive in another (Lat.) attests to the original diathetic vagueness of an inactive root.

131. In many instances, we can reconstruct the original inactive meaning of the deponents which express external and intentional actions. For hortor (honor) see Walde & Hoffmann (1938:658): root *gherH- "begehren, gem haben" (cf. Gr. » "rejoice, be glad", OG geron, gem, etc. "begehren, gierig"). For orior, see Bader (1980b): root *d1er- "être debout". However, not all deponents seem to go back to the inactive roots. Also the originally active verbs could develop meanings leading to the -r marking, e.g. on the reflexive basis, so perhaps also sequor "join, follow". — We must probably assume that the -r marking of the deponents was a later renewal of the inactive class marking. When the thematic present (and similarly the subjunctive) became a generalized category common to both active and inactive verbs, the formal expression of inactivity/non-agentivity was lacking in the present system. The renewed marking of the lexical class of inactive/non-agentive verbs (which themselves cannot be subjected to syntactic passivization) by the same formant as is used for the

Page 226: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOTES 209

marking of passive was modelled on the perfect system where the synthetic perfect became an active category only, whereas the passive and inactive/deponent perfect was expressed by analytical formation (participle in -tus + auxiliary verb sum).

132. Among the classics of IE studies there is a difference between Delbrück (1897:412-437), who based his discussion of the middle on media tantum, thus recognizing their original character, and Wackernagel (1920 I,130-134) who, on the contrary, considered the oppositional middle-passive of the Greek type to be original. See the discussion by Hatcher (1942:13ff.), supporting the primacy of media tantum and showing that Latin had no middle of the Greek type.

133. Stative' is also a category distinct from the perfect: stative -o(i) vs. perf. -e(i) in 3sg. See Oettinger (1979: 520ff.), Rix (1977:134ff.; 1988:104).

134. See especially Perel'muter (1984); see the objections by Trost (1985). Recently, also Rix (1988) explains middle-passive as original reflexive, considering the m-p endings as incorporated object pronouns.

135. As already mentioned in 3.4.0 (above), we use the term middle-passive for both types of marked voice, Greek and Latin, though the middle is represented only by deponents in Latin. The direct-reflexive function of the Latin -r form shares with the passive the syntactic derivation consisting in the promotion of the patient from the direct object to the subject, hence it can be considered as a part of the passive. The term deponent/passive serves to emphasize the non-occurrence of the oppositional middle in Latin. It defines adequately the Latin -r form. However, not all languages exhibiting the -r type of the middle-passive share with Latin its extremely 'syntactic' diathesis (see note 137) with no grammaticalized active middle oppositions. For Hittite, the term middle-passive is fully adequate also with respect to the functions of the oppositional diathesis.

136. The identity subject = agent does not hold for the direct-reflexive function either, since the subject of the direct-reflexive has both agent and patient roles, i.e. subject = agent + patient.

137. With the term 'syntactic diathesis' we refer here not to the diathesis as a semantic-syntactic category (see 3.1.3), but to the morphological diathesis connected with the syntactic effect, i.e. accompanied by syntactic derivation.

138. We find this term (which has been only occasionally used in the discussions about verbal voice) a good device for clarifying the difference between the syntactically defined and semantically defined diathesis of Latin and Greek, respectively.

139. The terms used are introvertal vs. extrovertal or centrifugal vs. centripetal. The internal character of the subject verb relation stands in contrast to the external cha­racter of the agent - verb relation in the case of active. See also the definition of the middle by Strunk (1980:322): "Eine generelle Funktion des Mediums [...] besteht darin, ein durch das Verbum bezeichnetes Verhalten wesentlich in Beziehung zum Verhältnistráger zu setzen."

Page 227: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

210 FROM INDO-EUROPEAN TO LATIN

140. I have drawn from the analysis of middles and deponents given in an unpublished paper by Untermann (1984).

141. For the indirect-reflexive function of the Greek middle see Kurzová (1987). The instances where another benefactive is present in particular show that the reflexive component is part of the middle as an introvertal diathesis; its definition by means of the benefactive role is approximate only, reflecting certain relationships between the diathetic verb semantics and nominal roles. See e.g.

"and they prevailed on Dionysius not to send some warships which he had then prepared for the Lacedaemonians" Lys. 19,20 with the dative representing a benefac­tive, whereas the reflexive component relates to the subject Dionysios.

142. There are also oppositions such as "I order" vs. "I accept order" which can be classified as a special type of indirect-object-to-subject promotion (see Feldman 1978). This 'indirect passive' together with the indirect reflexive is a component of the middle-passive with a benefactive subject role.

143. The terms deagentivity, deagentive function were introduced by Grepl (1973) and applied to Latin by Tesařová (1988). For the term 'backgrounding', see Foley & Van Valin(1984:149ff.).

144. Besides the research devoted to the perfecto-medium problem (see 3.2.0 above) cf. also Birwé (1956) on the Greek/Aryan evidence of original vocalic (non-consonantized) endings.

145. Both secondary and primary -so(i) ending is attested in Avestan (see Reichelt 1909:131).

146. Before the decipherment of Mycenaean Ruiperez (1952) had expressed the opinion that the Arcado/Cyprian -soi, -toi is original and this opinion was generally accepted after the Mycenaean attestation of -o- vocalism.

147. Cf. OPr asmai and Old Lith. reflexive endings -mies(i), etc. (see Stang 1966:406ff.). There is no category of m-p in the Baltic languages.

148. Cf. the Gothic forms nimada (lsg. and 3sg.), nimaza (2sg.), nimanda (all pl. persons). See Streitberg (1906:130).

149. On -r endings, cf. especially K. H. Schmidt (1963a: 167), Neu (1968: 8ff,161ff.), Jankuhn (1969:30ff.), G. Schmidt (1977). As regards Armenian, see below 3.4.6, and note 158).

150. The forms of the type toler, tolar are also attested in Venetic (see K. H. Schmidt 1963a).

151. The Celtic passive is impersonal with inserted object pronouns. An OIr innovation is a specific form for 3pl. (Ernout 1908-1909:274ff., K. H. Schmidt 1963b).

Page 228: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOTES 211

152. For impersonal uses see the examples in Neu (1968b); e.g. for the verb ak(k)-"sterben", see Neu (1968b:2), Ivanov (1963:163).

153. On the impersonal character of the Italic and Latin passive, see Ernout (19081909). On the backgrounding use of the Latin passive, see Kurzová (1987). Not only does the impersonal passive of intransitive verbs have a backgrounding character, but also the 'personal' passive with non-personal subjects (type liber legitur) and both forms are used with the same semantic-syntactic effect by Plautus. Cf. the example from Plaut.Merc. 775-777: Cocus: scio iam quid velis. Nemp' me hinc abire velis. ("Now I know what you want: why, of course, you want me to get out"). Lysimachus: Volo inquam! ("Indeed, I do!") Cocus: Abibitur. Drachmam dato. ("It will be fulfilled. Tip me, a shilling"). Lysimachus: Dabitur. ("You will be given"). Both impersonal passive abibitur and 'personal' passive dabitur with the unexpressed non-personal subject {drachma) serve to the same effect of depersonalization (intensional avoidance of the expression by 1st person: abibo, dabo). See Scherer (1975:59).

154. See G. Schmidt (1977). For Toch. -tar cf. Thomas (1985:86). On Latin see below 3.4.6.

155. In Phrygian, the -tor form is attested, in Venetic the -er/-or deagentivum and -to middle-passive cooccur. See K. H. Schmidt (1963a). On Armenian see below 3.4.6, and note 158.

156. Cf. especially the -er/-or deagentive form in Celtic and Osco-Umbrian and the non­occurrence of -r form in the 2pl. of both Irish deponent and Latin deponent/passive paradigms.

157. It may be assumed that the -r deagentivum was (besides the nominal -s- formation) one of the sources of the Latin infinitive in -re.

158. See a survey of Armenian -r forms in Solta (1984[1985]: 65-66). The -r imperative is attested also in Hittite and Tocharian. For Tocharian, see Krause & Thomas (1960:256).

159. Also the present paradigm of the unique Latin athematic verb esse was remodelled after the pattern of thematic verbs. The original paradigm *esm(i), *ess(i), *est(i), *smos, *ste(s), *sent(i) has been transformed into the paradigm where the distribution of es- and so- stems corresponds to the distribution of -e- and -o- stems in thematic flection (see Hamp 1985:224-225).

160. Due to the fact that in the traditional reconstruction of the CIE verbal system the inflectional categories and oppositions of Greek/Aryan are projected into IE, the para­llelism of d-f formations with -e/o-, -ā-, and -ê- suffixes was acknowledged and examined only marginally, especially by Hirt (1939:97ff.; 1901:212ff.). In contemporary IE research the parallelism of -ă- and -ê- formations has been systematically examined by Perel'muter (1977:93ff.), where I also find important suggestions with respect to the diathetic inactive semantics of these formations. Otherwise, this parallelism was especially acknowledged by Lane in many of his studies (1949, 1959, 1970:73ff.), based primarily on the analysis of Tocharian, where

Page 229: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

212 FROM INDO-EUROPEAN TO LATIN

this parallelism is particularly evident. See also van Windekens (1982:119). With our definition of d-f structure we have obtained a necessary interpretational basis for understanding this parallelism.

161. For the Hittite situation, see 3.2.3.4 (above). The thematic present (with -ō in the lsg.) as the formation common to both classes is not attested in Hittite, the thematic inflectional stems of -mi verbs are, however, well attested; see Oettinger (1979:257; 1985:296ff.).

162. Present stems with primary -ă-, -ê- suffix form the nuclear component of the 1st and 2nd conjugations. Other components are denominatives on the one hand and roots in -ă- or -ě- on the other, with no clear boundary between stems with -ā-, -è- suffixes and roots in -ă- or -ê-; see about root/stem continuum 1.3.3.3 (above). A systematic account of historical sources of Latin conjugations is given by DeWandel (1982).

163. See Oettinger (1984), cf. the critical remarks by Thomas (1986:366). For Olr, see Kortlandt (1984). Kortlandt's objection to the derivation of lsg.subj. -ber from *bherăm is based on the incorrect assumption that the acc. of -ă- stems túaith is a re­sult of a phonetic development (he follows Cowgill 1975:49). Yet, it is an analogical form (see Thurneysen 1909: 181), as is now confirmed by the accusatives of -ā-stems in -im found in the 'plomb du Larzac' (see K. H. Schmidt 1990). See also the criticism of Cowgill's explanation by de Bernardo-Stempel (1987:34ff. and 62, notes 115, 119). Very important for the -ă- injunctive reconstruction are concrete cor­respondences such as that examined by Schindler (1967:240) for the verb *kwrei-"buy": preterite *kwriiâ- Toch. *karyă- (cf. Toch.B. lpl. kăryam, etc.), subjunctive Olr ni-cria < *kwriiăt.

164. On the problems in reconstructing the IE moods, see especially Strunk (1984, 1988). Hittite has no subjunctive.

165. A common explanation, also followed by Strunk (1988) for other than lsg., is that --e/o- suffix as marker of subjunctive (cf. the subjunctive of athematic verbs like

"was doubled and finally contracted in thematic verb stems like e.g. *bheret(i) vs. subjunctive *bher-e-e-t(i) > *bher-ě-t(i)" — Strunk (1988:292). Originally, the thematic subjunctive of the Greek type was a voluntative mood, according to Strunk, and restricted only to the lsg. in -ö.

166. See the synchronic and diachronic analysis of the Latin present conjugations by DeWandel (1982).

167. See especially Meillet (1908), Watkins (1962). There are only a few verbs which have a sigmatic aorist in both 0I and Greek. Unlike the Greek sigmatic aorist, the OI aorist has a lengthened root vowel, which we find also in the Latin sigmatic perfects of the type vexi (see 3.3.4.1 above).

168. Athematic forms are attested in Baltic, Italic and Celtic. Celtic evidence appears to be particularly important for the reconstruction and interpretation of IE -s-formations. On the athematic character of OIr -s- subjunctive see Watkins (1962: 126), Schmid (1963b:227), Rasmussen (1985), Kortlandt (1984), and Hamp (1987)

Page 230: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

NOTES 213

who also explains the lsg. as athematic. A systematic analysis of the OIr subjunctives and futures is given by K. H. Schmidt (1966). The -sio- future is also attested in Gaulish (see K. H. Schmidt 1990:21).

169. See the survey of vocalic endings used in sigmatic subjunctive/future in various languages by Schmid (1963b).

170. The present form of perfective verbs has, as is well known, future meaning in the Slavic languages. Therefore, the sigmatic futures can go back to the indicative of a perfectivizing formation. (Another frequent source of the future forms is the volitive mood, i.e. subjunctive, as we presupposed for long vocalic futures in 3.5 above.)

171. Recall the modal/subjunctive and preterital interpretation of the 'non-actual' injunctive (3.1.4; 3.2.3.1 above), whereby the 'ideal' value of the subjunctive can be considered as more elementary that the 'realized' ( + real, but —actual) value of the preterite.

172. This suffix was established as the modal (optative) marker in IE languages. We have, however, probably the same suffixal formation in the present stems containing -i-, -ī-and exhibiting in Latin, Baltic, and Slavic a semi-thematized conjugation (with a -io ending in lsg., but an athematic 2sg, and 3sg.).

173. It is a latter possibility which is regular in the paradigmatized structure, and we can observe the shifts in the application of the same formants from the lexical stem to the inflectional stem. Cf. the archaic subjunctives of the type adven-at, tag-am, based on lexical stem vs. the classic subjunctives (adveni-at, tang-am) based on inflectional present stem.

174. On the lexical level the -se/so- suffix appears in quaeso, and viso.

175. An explanation of the Latin subjunctive imperfect on the basis of the modal/future sigmatic formation (not on the basis of the sigmatic aorist) is given by Erhart (1983:27). The formation is desiderative according to Erhart, whereas I prefer to consider it as perfectivizing.

176. In Plautus Pseud. 499 the form faxem is attested (see Safarewicz 1969:236), which belongs to the faxo, faxim subsystem (3.6.4 below), also with respect to its function corresponding to fecissem.

177. Another possible explanation is the application of the -ie/io- suffix to the athematic formation with pure -s- suffix. The original status of the -si- subjunctive as a forma­tion which is parallel to the thematic -sie/sio- or -se/so- formations and not dependent on them is valid in both instances.

178. For these verbs, the secondary attachment to the inflectional perfect stem must be presupposed, by which the original *gen-i-sö was transformed to *genu-isö > genuerö. See Narten (1973) for *monisö > *monuisö > monuerö.

179. Burrow's (1979) analysis of the OI treatment of laryngeals (see also Nyman 1985), according to which the reduced grade of long vowels in IE roots was -a- in OI as in other IE languages, also suggests a revision of the treatment of vocalized laryngeal

Page 231: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

FROM INDO-EUROPEAN TO LATIN

in set-bases.

This derivation is possible phonologically as both -d- and -b- are substitutes of IE word internal -dh-, without clear conditions of their distribution. See the discussion by Muller (1921), and Leumann (1977:168). After a long vowel see plē-b-ês < *plě-dh-.

See section 3.5.3 (above) on the solidarity between intransitive and imperfective meaning, with respect to -ă- and -è- suffixes and their use in Latin and Greek.

In da-barn a reduced root variant dă- occurs, which was generalized in cases where the long vowel is not required by inflection(as in 2sg. dās).

On this rule governing further morphological innovations of the individual European languages in their most recent prehistorical and historical development, see Givón (1971), and Anderson (1980).

Page 232: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

References

Abraham, Werner. 1987. "Zur Typologie des Mediums in der Westgermania". Linguistik in Deutschland: Akten des 21. Linguistischen Kolloquiums, Groningen 1986 ed. by Werner Abraham & Ritva Århammar, 3-23. Tübingen: Max Niemeyer.

Adrados, Francisco Rodriguez. 1963. Evolucion y estructura del verbo indoeuropeo. Madrid: Instituto "Antonio de Nebrija".

Albrecht, Jörn, Jens Lüdtke & Harald Thun, eds. 1988. Energeia und Ergon: Sprachliche Variation — Sprachgeschichte — Sprachtypologie. Studia in honorem Eugenio Coseriu. 3 vols. Tübingen: Gunter Narr.

Anderson, Stephen R. 1980. "On the Development of Morphology from Syntax". Historical Morphology ed. by Jacek Fisiak, 51-69. The Hague: Mouton.

—. 1985."Inflectional Morphology". Language Typology and Syntactic Description ed. by Timothy Shopen. Vol.3. Grammatical Categories and the Lexicon, 150-201. Cambridge: Cambridge Univ. Press.

Antilla, Raimo. 1969. Proto-Indo-European Schwebeablaut. Berkeley & Los Angeles: Univ. of California Press.

Austefjord, Anders. 1984. "Zur Gestaltung des germanischen Tempussystems". Indogermanische Forschungen 89.160-168.

Bader, Françoise. 1968. "Le système des désinences de troisième personne du pluriel du perfectum Latin". Bulletin de la Société de Linguistique de Paris 62.87-105.

—. 1972. "Parfait et moyen en grec". Mélanges de linguistique et de philologie grecques offerts à Pierre Chantraine, 1-21. Paris: Klincksieck.

—. 1980a. "Présents moyens hittites à vocalisme -e- et formations de présents indo­européens". Recherches de Linguistique: Hommages à Maurice Leroy ed. by Jean Bingen, André Coupez & Francine Mawet, 21-40. Bruxelles: Éditions de l'Univ. de Bruxelles.

—. 1980b. "De lat. arduus à lat. orior". Revue de Philologie 54.37-61, 263-275. Baldi, Philip. 1976. "The Latin Imperfect in *bā-". Language 52.839-850. Bammesberger, Alfred. 1985. "Thematische und athematische Wurzelaoriste im

Indogermanischen". Schlerath & Rittner 1985. 71-78.

Page 233: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

216 FROM INDO-EUROPEAN TO LATIN

—. 1986. Der Aufbau des germanischen Verbalsystems. Heidelberg: Carl Winter. Bartoněk, Antonín. 1987. "Pádový systém v mykénštiné [Case System in Mycenaean]".

Sborník prací filozofické fakulty brněnské univerzity E 32.121-126. Baudiš, Josef. 1925. "Poznámky k vývoji indoevropské struktury gramatické [Notes on

the Development of the Indo-European Grammatical Structure]". Sborník filosofické fakulty University Komenského v Bratislavě 3.3-17.

Bechert, Johannes. 1982. "Grammatical Gender in Europa: An areal study of a linguistic category". Papiere zur Linguistik 26.23-34.

Beekes, Robert S. P. 1977. "Indo-European Neuters in -f. Cardona & Zide 1977. 45-56. —. 1985. The Origins of the Indo-European Nominal Inflection. Innsbruck: Institut fur

Sprachwissenschaft, Univ. Innsbruck. —. 1988. A grammar of Gatha-Avestan. Leiden: E.J. Brill. Benveniste, Émile. 1935. Origines de la formation des noms en indo-européen. Paris:

Adrien Maisonneuve. Bernardo Stempel, Patrizia de. 1987. Die Verbreitung der indo-germanischen Liquiden

und nasalen Sonanten im Keltischen. Innsbruck: Institut fur Sprachwissenschaft, Univ. Innsbruck.

Birwé, Robert. 1956. Griechisch-arische Sprachbeziehungen im Verbalsystem. Walldorf, Hessen: Vordran.

Blümel, Wolfgang. 1972. Untersuchungen zu Lautsystem und Morphologie des vor-klassischen Lateins. München: Kitzinger.

Boeder, Winfred. 1973. "Probleme der verbalen Kongruenz in den indogermanischen Sprachen". Indogermanische und allgemeine Sprachwissenschaft: Akten der IV. Fachtagung der Indogermanischen Gesellschaft, Bern, 28. Juli -l. August 1969 ed by Georges Redard, 1-10. Wiesbaden: Ludwig Reichert.

Bonfante, Giuliano. 1941. "The Latin and Romance Weak Perfect". Language 17.201-211. Bossong, Georg. 1980. "Syntax und Semantik der Fundamentalrelation: Das Guarani als

Sprache des aktiven Typus". Lingua 50.359-379. Bouissac, Paul, Michael Herzfeld & Roland Posner, eds. 1986. konicity: Essays on the

Nature of Culture. Festschrift for Thomas A.Sebeok on his 65th birthday. Tübingen: Stauffenburg.

Brettschneider, Gunter & Christian Lehmann, eds. 1980. Wege zur Universalienforschung. Sprachwissenschaftliche Beitrăge zum 60. Geburtstag von Hansjakob Seller. Tübingen: Gunter Narr.

Broschart, Jürgen. 1987. Noun, Verb and Participation. Köln: Institut für Sprach­wissenschaft, Univ. Köln.

Brugmann, Karl. 1911. Vergleichende Laut-, Stammbildungs- und Flexionslehre nebst Lehre vom Gebrauch der Wortformen der indogermanischen Sprachen. 2nd ed., Vol.II.l. Strassburg: Karl Trübner.

—. 1916. Vergleichende Laut-, Stammbildungs- und Flexionslehre nebst Lehre vom Gebrauch der Wortformen der indogermanischen Sprachen. 2nd ed., Vol.II.3. Ibid.

Page 234: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

REFERENCES 217

Bühler, Karl. 1934. Sprachtheorie. Die Darstellungsfunktion der Sprache. Jena: Gustav Fischer. (English translation, Amsterdam & Philadelphia: John Benjamins, 1990.)

Burrow, Thomas. 1979. The Problem of shwa in Sanskrit. Oxford: Clarendon Press. Bybee, Joan L. 1985. Morphology: A study of the relation between meaning and form.

Amsterdam & Philadelphia: John Benjamins. Calboli, Gualtiero. 1972. La linguistica moderna e il latino: I casi. Bologna: Riccardo

Pàtron. Cardona, George. 1960. The Indo-European Thematic Aorist. Yale Ph.D. Dissertation. — & Norman H. Zide, eds. 1987. Festschrift for Henry Hoenigswald: On the occasion of

his seventieth birthday. Tübingen: Gunter Nam Chafe, Wallace L. 1970. Meaning and the Structure of Language. Chicago & London:

Univ. of Chicago Press. Chantraine, Pierre. 1927. Histoire du parfait grec. Paris: Honoré Champion. Chodorkovskaja, B. B. 1981. "K probleme vida v latinskom glagole. (Sémantika

imperfekta i perfekta v rannej latyni)". Voprosy jazykoznanija 1981/4.93-104. Coleman, Robert. 1987. "Early Greek Syncretism and the Case of the Disappearing -phi".

Studies in Mycenaean and Classical Greek Presented to John Chadwick ed. by J.T. Killen, José L. Melena& Jean-Pierre Olivier, 113-125. Salamanca: Univ. Pais Vasco.

Comrie, Bernard. 1976. Aspect: An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge Univ. Press.

—. 1983. Language Universals and Linguistic Typology: Syntax and morphology. Oxford: Basil Blackwell.

Coseriu, Eugenio. 1980. "Partikeln und Sprachtypus: Zur strukturell-funktionellen Fragestellung in der Sprachtypologie". Brettschneider & Lehmann 1980. 199-206. (Repr. in Albrecht, Lüdtke & Thun 1988 I, 185-194.)

—. 1988. "Der romanische Sprachtypus: Versuch einer neuen Typologisierung der romanischen Sprachen". Albrecht, Lüdtke & Thun 1988 I, 207-224.

Cowgill, Warren. 1965. "Evidence in Greek". Evidence for Laryngeals ed. by Werner Winter, 142-180. The Hague: Mouton.

—. 1975. "The Origins of the Insular Celtic Conjunct and Absolute Verbal Endings". Rix 1975. 40-70.

—. 1979. "Anatolian i- Conjugation and Indo-European Perfect: Instalment II". Neu & Meid 1979. 25-39.

—. 1985. "The Personal Endings of Thematic Verbs in Indo-European". Schlerath & Rittner 1985. 99-108.

Croft, William. 1990. Typology and Universals. Cambridge: Cambridge Univ. Press. Dahl, Osten. 1985. Tense and Aspect Systems. Oxford: Basil Blackwell. Delbrück, Berthold. 1897. Vergleichende Syntax der indogermanischen Sprachen. Vol.11.

Strassburg: Karl Trübner. Demiraj, Shaban. 1985. Gramatikë historike e gjuhës shqipe. Tiranë: "8. Nëntorit". Devine, Andrew Mackay. 1970. The Latin Thematic Genitive Singular. Stanford: Univ.

Page 235: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

218 FROM INDO-EUROPEAN TO LATIN

DeWandel, Nancy Carol. 1982. The Origins and Development of the Latin Present System. Ph.D. Diss., Ohio State Univ., Columbus, Ohio.

Dixon, Robert M.W. 1977. "Where Have All the Adjectives Gone?". Studies in Language 1.19-80.

Dressier, Wolfgang. 1985. "Zur Bedeutung der Sprachtypologie in der natürlichen Morphologie". Albrecht, Lüdtke & Thun 1985 III, 199-208.

—. 1986. "Explanation in Natural Morphology, Illustrated with Comparative and Agent-Noun Formation". Linguistics 24.519-548.

—, Walter Mayerthaler, Oswald Panagl & Wolfgang U. Wurzel, eds. 1987. Leitmotifs in Natural Morphology. Amsterdam & Philadelphia: John Benjamins.

Drobin, Ulf. 1980. Indogermanische Religion und Kultur? Eine Analyse des Begriffes indogermanisch. Stockholm: Inst. for Comparative Religion, Univ. of Stockholm.

Drossard, Werner, 1986. Kasusmarkierung und die Zentralität von Partizipanten. Köln: Institut fur Sprachwissenschaft, Univ. Köln.

Eichner, Heiner. 1972. "Die Etymologie von heth. mehur". Münchener Studien zur Sprachwissenschaft 31.53-107.

Erhart, Adolf. 1961. "Indoevropská préterita s dlouhými vokály ă, ē [Indo-European Preterits with Long Vowels ă, ë] ". Sborník prací filozofické fakulty Brněnské univerzity A 9.17-33.

—.1982. Indoevropské jazyky: Srovnávací fonologie a morfologie [Indo-European Languages: Comparative phonology and morphology]. Praha: Academia.

—. 1983. "Das lateinische System der verbalen Modi und seine indoeuropäischen Grundlagen". Sborník prací filozofické fakulty brněnské univerzity A 31.25-30.

Ernout, Alfred. 1908/1909. "Recherches sur l'emploi du passif latin à l'époque républicaine". Mémoires de la Société de Linguistique de Paris 15.276-333.

Faust, Manfred, ed. 1983. Allgemeine Sprachwissenschaft, Sprachtypologie und Textlinguistik: Festschrift Jur Peter Hartmann. Tübingen: Gunter Narr.

Feldman, Henry. 1978. "Passivizing on Datives in Greek". Linguistic Inquiry 9.499-502. Fillmore, Charles. 1968. "The Case for Case". Universals in Linguistic Theory ed. by

Emmon Bach & Robert T.Harms, 1-88. New York: Holt, Rinehart & Winston. Finck, Franz Nikolaus. 1923. Die Haupttypen des Sprachbaus. 2nd ed. Leipzig & Berlin:

B. G. Teubner. Foley, William A. & Robert D. Van Valin, Jr. 1984. Functional Syntax and Universal

Grammar. Cambridge: Cambridge Univ. Press. Funk, Wolf-Peter. 1985. "On a Semantic Typology of Conditional Sentences". Folia

Linguistica 19.365-413. Gamkrelidze, Thomas V. & Vjač. Vs. Ivanov 1984. Indojevropejskij jazyk i indojevropejcy:

Rekonstrukcija i istoriko-tipologičeskij analiz prajazyka i protokuVtury. Vol.I. Tbilisi: Izdatel'stvo tbilisskogo univerziteta.

García-Ramón, José L. 1985. "Die Sekundärendung der l.Sg. Medii im Indogermani-schen". Schlerath & Rittner 1985. 202-217.

Page 236: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

REFERENCES 219

Genuisienè, Emma. 1987. The Typology of Reflexives. Berlin, etc.: Mouton & de Gruyter. Giacomelli, Gabriella. 1963. La lingua falisca. Firenze: Istituto di Studi Etruschi ed

Italici. Givón, Talmy. 1971. "Historical Syntax and Synchronic Morphology: An archeologisťs

fieldtrip". Papers from the Regional Meeting of the Chicago Liguistic Society 7.394-415. Chicago.

Gonda, Jan. 1979. The Medium in the Rgveda. Leiden: E.J. Brill. Greenberg, Joseph H. 1966. "Some Universals of Grammar with Particular Reference to

the Order of Meaningsful Elements". Universals of Grammar, ed. by Joseph H. Greenberg. 2nd ed., 73-113. Cambridge, Mass.: MIT Press.

Grepl, Miloš. 1973. "Deagentnost a pasívum v slovanských jazycích [Deagentivity and Passive in Slavic Languages] ". Československé přednášky pro 7. mezinárodní sjezd slavistu ve Varšavě Vol.I, 141-151. Praha: ČSAV.

Groot, Albertus Wilhelm de. 1939. "Les oppositions dans les systèmes de la syntaxe des cas". Mélanges de linguistique offerts à Ch. Bally, 107-127. Genève: Georg.

—. 1956a. "Classification of the Uses of a Case Illustrated on the Genitive in Latin". Lingua 6.8-65.

—. 1956b. "Classification of Cases and Uses of Cases". For Roman Jakobson. Essays on the occasion of his sixtieth birthday ed. by Morris Halle et al., 187-194. The Hague: Mouton.

Hamp, Eric P. 1985. "Two Problems of Latin Alternation". Symbolae Ludovico Mitxelena septuagenario oblatae ed. by José L. Melena, 223-225. Vitoria: Univ. del país vasco.

—. 1987. "The Athematic s- Subjunctive". Ériu 38.201. Hardarson, Jón Axel. 1987. "Zum urindogermanischen Kollektivum". Münchener Studien

zur Sprachwissenschaft 48.71-113. Hartmann, Hans. 1954. Das Passiv: Eine Studie zur Geistesgeschichte der Kelten, Italiker

und Arier. Heidelberg: Carl Winter. Hatcher, Anna Granville. 1942. Reflexive Verbs: Latin, Old French, Modern French.

London: Oxford Univ. Press. Haudry, Jean. 1979. L' indo-européen. 2e éd. Paris: Presses Universitaires de France. —. 1980. "La 'syntaxe des désinences' en indo-européen". Bulletin de la Société de

Linguistique de Paris 75.131-166. —. 1983. "Notes de morphologie Latin". Hommages à Robert Schilling ed. by Hubert

Zehnacker & Gustave Hentz, 477-485. Paris: Les Belles Lettres. Havránek, Bohuslav. 1968. "Quelques problèmes de l'étude diachronique de la structure

syntactique surtout en slave". Travaux linguistiques de Prague 3.9-16. —. 1971. "Příspěvek k autonomní komplexnosti starého slovanského slovesa (na materiálu

staroslověnského jazyka) [Contribution to the Complex Autonomy of Old Slavic Verb (based on the Old Church Slavonic data)]". Studia Palaeoslovenica ed. by Bohuslav Havránek, 127-131. Praha: Academia.

—. 1973. "Větná struktura v nejstarších fázích slovanských jazyků". Československé

Page 237: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

220 FROM INDO-EUROPEAN TO LATIN

přednášky pro 7.mezinárodní sjezd slavistă ve Varšavě Vol I, 5-13. [German Summary, "Die Satzstruktur in den alten slawischen Sprachen" p. 13] Praha: ČSAV.

Heger, Klaus. 1982. "Nominativ - Subjekt - Thema". Festschrift für Helmut Stimm zum 65. Geburtstag: Fakten und Theorien ed. by Sieglinde Heinz & Ulrich Wandruszka, 87-93. Tübingen: Gunter Narr.

Hirt, Helmuth. 1901. "Kleine grammatische Beiträge". Indogermanis che Forschungen 12.195-241.

—.1939. Die Hauptprobleme der indogermanischen Sprachwissenschaft. Tübingen: Max Niemeyer.

Hjelmslev, Louis. 1935. "La catégorie des cas. Première partie". Acta Jutlandica 7.1-184. —. 1937. "La catégorie des cas. Étude de grammaire générale. Deuxième partie". Acta

Jutlandica 9.1-78. —. 1971 [1956]. "Sur F indépendence de l'epithète". Essais linguistiques, 199-210. Paris:

Minuit. Hoffmann, Karl. 1967. Der Injunktiv im Rgveda: Fine synchronische Funktions-

untersuchung. Heidelberg: Carl Winter. Hubka, Karel. 1980. "Strukturální a systémový popis syntakticko-sémantických jevů: Nad

Lavencyho studií o latinských pádech [Structural and Systemic Description of Syntactic-Semantical Facts: To Lavency's study on Latin cases]". Moderní lingvistika a klasické jazyky ed. by Helena Kurzová, 9-31. Praha: Kabinet pro studia řecká, římská a latinská ČSAV.

Humbert, Jean. 1954. Syntaxe grecque. Paris: Klincksieck. Insler, Stanley. 1972. "On Proterodynamic Root Present Inflection". Münchener Studien

zur Sprachwissenschaft 30.55-64. Ivanov, Vjač. Vs. 1963. Chettskij jazyk. Moskva: Izd. vost. lit. —. 1981. "Die historisch-vergleichende Indoeuropäische Sprachwissenschaft und die

Typologie". Zeitschrift jur Phonetik, Sprachwissenschaft und Kommunikations-forschung 34.415-420.

Jacobi, Hermann. 1897. Compositwn und Nebensatz: Studien über die indogermanische Sprachentwicklung. Bonn: Friedrich

Jacobsohn, Herrmann. 1923. "Kasusflexion und Gliederung der indogermanischen Sprachen". Antidoron: Festschrift Jacob Wackernagel, 204-216. Göttingen: Vandenhoeck & Ruprecht.

Jakobson, Roman. 1936. "Beitrag zur allgemeinen Kasuslehre: Gesamtbedeutungen der russischen Kasus". Travaux du Cercle linguistique de Prague 6.240-288.

Jankuhn, Harald. 1969. Die passive Bedeutung medialer Formen, untersucht an der Sprache Homers. Göttingen: Vandenhoeck & Ruprecht.

Jasanoff, Jay H. 1978. Stative and Middle in Indo-European. Innsbruck: Institut fur Sprachwissenschaft, Univ. Innsbruck.

—. 1979. "The Position of the -hi-Conjugation". Neu & Meid 1979.79-90. --.1987. "The Tenses of the Latin Perfect System". Cardona & Zide 1987.177-183.

Page 238: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

REFERENCES 221

Jong, Jan R. de. 1983. "Word Order within Latin Noun Phrases". Latin Linguistics and Linguistic Theory ed. by Harm Pinkster, 131-141. Amsterdam & Philadelphia: John Benjamins.

Joseph, Brian & Rex Wallace. 1984. "Latin Morphology: Another Look". Linguistic Inquiry 15.319-328

Kellens, Jean. 1984. Le verbe avestique. Wiesbaden: Ludwig Reichert. Kiparsky, Paul. "Tense and Mood in Indo-European Syntax". Foundations of Language

4.30-57. Klimov, Georgij A. 1973. "Tipologijajazykov aktivnogo stroja i rekonstrukcija protoindo-

jevropejskogo". Izvestija Akademii nauk SSSR, Serija literatury i jazyka 32.442-447. —. 1977. Tipologija jazykov aktivnogo stroja. Moskva: Nauka. —. 1978. "Obščeindojevropejskij i kartvel'skij: K tipologii padežnych sistem/". Voprosy

jazykoznanija 1978/4.18-22. Knobloch, Johann. 1950-52. "Zur Vorgeschichte des indogermanischen Genitivs der

ø-Stamme auf -sjo. Vorbemerkung". Die Sprache 2.131-149. Koerner, Konrad. 1981. "Schleichers Einfluß auf Haeckel: Schlaglichter auf die

wechselseitige Abhängigkeit zwichen linguistischen und biologischen Theorien im 19. Jahrhundert". Zeitschrift für vergleichende Sprachforschung 95.1-21. (Repr. in Koerner 1989:212-231, with Postscript 1988 on p.232.)

—.1989. Practicing Linguistic Historiography: Selected Essays. Amsterdam & Philadelphia: John Benjamins.

Kølln, Herman. 1969. Oppositions of Voice in Greek, Slavic, and Baltic. København: Einar Munksgaard.

Kortlandt, Frederik. 1981.1st sg. middle *-H1. Indogermanische Forschungen 86.123-136. —. 1984. "Old Irish Subjunctives and Futures and their Proto-Indo-European Origins".

Ériu 35.179-187. Krause, Wolfgang & Werner Thomas. 1960. Tochańsches Elementarbuch. Vol.1.

Grammatik. Heidelberg: Carl Winter. Kuiper, F.B.J. 1937. Die indogermanischen Nasalprâsentia: Ein Versuch zu einer

morphologischen Analyse. Amsterdam: Noord-Hollandische Uitgevers Maatschappij. —. 1942. Notes on Vedic Noun-Inflection. Ibid. Kuryłowicz, Jerzy. 1932. "Les désinences moyennes de l'indo-européen et du hittite".

Bulletin de la Société de Linguistique de Paris 33.1-4. —. 1949. "Le problème du classement des cas". Biuletin Polskiego Towarzystwa

Jezykoznawczego 9.20-43. —. 1964. The Inflectional Categories of Indo-European. Heidelberg: Carl Winter. —. 1979. "Die hemitische hi-Konjugation". Neu & Meid 1979.143-146. Kurzová, Helena. 1970. "Der lateinische Infinitiv im Vergleich mit dem

Griechischen". Živa antika 20.83-93. —. 1973. "Die Wort- und Satzstruktur des Indogermanischen". Listy filologické

96.1-15.

Page 239: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

222 FROM INDO-EUROPEAN TO LATIN

ezikoznanie/Linguistique balkanique 17.41-50. —. 1981. Der Relativsatz in den indoeuropüischen Sprachen. Praha: Academia; Hamburg:

Helmut Buske. —. 1983. "Zur linguistischen Charakteristik des Griechischen im Vergleich mit dem

Lateinischen". Concilium Eirene 16. Proceedings of the 16th International Eirene Conference, Prague 31.8.- 4.9.1982 ed. by Pavel Oliva & Alena Frolíková. Vol.II, 257-262. Praha: Kabinet pro studia řecká, římská a latinská ČSAV.

—. 1985. "Zum Typus des Altgriechischen". Listy filologické 108.3-13. —. 1986. "Typologie und die Rekonstruktion der indogermanischen Morphosyntax".

Folia Linguistica 20.49-80 —. 1987a. "Strukturell-typologische Sprachbetrachtung und die Morphosyntax des

Griechischen und Lateinischen". Listy filologické 110.202-215 —. 1987b. "Morphological Semantics and Syntax". Explizite Beschreibung der Sprache

und automatische Textbearbeitung 14. Probleme und Perspektiven der Satz- und Textforschung, 81-89. Praha: Matematicko-fyzikální fakulta UK.

—. 1988. "Morphological Semantics and Syntax in the Nonformalized Sentence Structure of Greek". In the Footsteps of Raphael Ktihner ed. by Albert Rijksbaron, Hotze Mulder & G. C Wakker, 147-149. Amsterdam: Gieben.

—. 1989. "Subordination in Latin". Eirene 26.23-32. Lane, George S. 1949. "On the Present State of the Indo-European Linguistics".

Language 25.331-342. —. 1959. "The Formation of the Tocharian Subjunctive". Language 35.157-179. —. 1970. "Tocharian: Indo-European and Non-Indo-European Relationships". Indo-

European and Indo-Europeans: Papers presented at the Third Indo-European Conference at the University of Pennsylvania ed. by George Cardona, Henry M. Hoenigswald & Alfred Senn, 73-88. Philadelphia: Univ. of Pennsylvania Press.

Laroche, Emmanuel. 1970. "Études de linguistique anatolienne III.9: Le directif'. Revue Hittite et Asianique 28.22-49.

Lavency, Marius. 1969. "A propos de la syntaxe des cas en latin classique". Les Études Classiques 37.325-337.

Lehmann, Christian. 1979. "Zur Typologie des Lateinischen". Giotta 57.237-253. —. 1988. Studies in General Comparative Linguistics. Köln: Institut fur Sprach-

wissenschaft, Univ. Köln. Lehmann, Winfred P. 1942. "The Indo-European dh-Determinative in Germanic".

Language 18.125-132. —. 1943. "The Indo-European dh-Determinative as Germanic Preterite Formant".

Language 19.19-26. —. 1969. "Proto-Indo-European Compounds in Relation to Other Proto-Indo-European

Syntactic Patterns". Acta Linguistica Hafniensia 12.1-20. —. 1974. Proto-Indo-European Syntax. Austin & London: Univ. of Texas Press. ---. 1978. "The Great Underlying Ground-Plans". Syntactic Typology: Studies in the

Page 240: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

REFERENCES 223

phenomenology of language ed. by Winfred P. Lehmann, 3-55. Austin: Univ. of Texas Press; Hassocks, Sussex: Harvester Press.

—, ed. 1986. Language Typology 1985: Papers from the Linguistic Typology Symposium, Moscow, 9-13 December 1985. Amsterdam & Philadelphia: John Benjamins.

—, ed. 1990. Language Typology 1988: Systematic Balance in Language. Papers from the Linguistic Typology Symposium, Berkeley, 1-3 Dec 1987. Amsterdam & Philadelphia: John Benjamins.

Lejeune, Michel. 1949. "Sur le traitement osque de *-ă final". Bulletin de la Société de Linguistique de Paris 45.104-110.

Leumann, Manu. 1977. Lateinische Laut- und Formenlehre. (= Lateinische Grammatik von Leumann-Hoffman-Szantyr 1.). München: H. C. Beck.

Lewy, Ernst. 1942. Der Bau der europăischen Sprachen. Dublin: Hodges, Figgis & Co —. 1961. Kleine Schriften. Berlin: Akadem ie-Verlag. Ludwig, Alfred. 1873. Agglutination oder Adaptation? Eine sprachwissenschaftliche

Streitfrage. Prag: J.G. Calve'sche K.K. Univ. Buchhandlung Lyons, John. 1977. Semantics 2 vols. Cambridge: Cambridge Univ. Press. Mallinson, Graham & Barry J. Blake. 1981. Language Typology. Cross-linguistic Studies

in Syntax. Amsterdam & New York: North-Holland. Mann, Stuart E. 1977. An Albanian Historical Grammar. Hamburg: Helmut Buske. Marouzeau, Jean. 1922. L'ordre des mots dans la phrase latine. Vol.I: Les groupes

nominaux. Paris: Honoré Champion. Martinet, André. 1979. "Shunting on to Ergative or Accusative". Plank 1979. 39-43. Matthews, Peter H. 1972. Inflectional Morphology: A theoretical study based on aspects

of Latin verb conjugation. Cambridge: Cambridge Univ. Press. Mayerthaler, Willi. 1981. Morphologische Natürlichkeit. Wiesbaden: Atheneum. McLendon, Sally. 1978. "Ergativity, Case and Transitivity in Eastern Porno".

International Journal of American Linguistics 44.1-9. Meid, Wolfgang. 1971. Das germanische Präteritum. Indogermanische Grundlagen und

Ausbreitung im Germanischen. Innsbruck: Institut für Sprachwissenschaft, Univ. Innsbruck.

—. 1975. "Probleme der räumlichen und zeitlichen Gliederung des Indogermanischen", Rix 1977.204-221.

—. 1979. "Der Archaismus des Hethitischen". Neu & Meid 1979. 159-176. Meillet, Antoine. 1906. De quelques innovations de la déclinaison latine. Paris:

Klincksieck. —. 1908. "Sur l'aoriste sigmatique". Mélanges de Linguistique offerts à F. de Saussure,

81-106. Paris: Honoré Champion —. 1921. Linguistique historique et linguistique générale. Paris: Honoré Champion. —. 1922. "Du nominatif et de l'accusatif'. Mémoires de la Société de Linguistique de

Paris 22.49-55. —. 1924. "Remarques sur les désinences verbales de l'indo-européen". Bulletin de la

Page 241: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

224 FROM INDO-EUROPEAN TO LATIN

Société de Linguistique de Paris 70.64-75. —. 1934. Introduction à l' étude comparative des langues indo-européennes. 7th ed.

Paris: Hachette. —. 1977. Esquisse d'une histoire de la langue latine, Avec une bibliographie mise à jour

et complété par Jean Perrot. Paris: Klincksieck. Meinhof, Carl. 1936. Die Entstehung der flektierenden Sprachen. Berlin: Georg Reimer. Migron, Saul. 1975. "The Rgvedic Passive Aorist in -i: A functional study". Folia

Linguistica 8.271-310. Moravcsik, Edith A. 1978. "On the Distribution of Ergative and Accusative Patterns".

Lingua 45.233-279. Morpurgo Davies, Anna. 1966. "An Instrumental-Ablative in Mycenaean?". Proceedings

of the Cambridge Colloquium on Mycenaean Studies ed. by Leonard R. Palmer & John Chadwick, 191-202. Cambridge: Cambridge Univ. Press.

Muller, Frederik. 1921. "Lat. barba. Zur Hauchdissimilation". Indo germanis che Forschungen 39. 172-189.

Narten, Johanna. 1964. Die sigmatischen Aoriste im Veda. Wiesbaden: Otto Harrassowitz. —. 1968. "Zum "proterodynamischen" Wurzelprăsens". Pratidānam. Indian, Iranian, and

Indo-European studies presented to Franciscus Bernardus Jacobus Kuiper on his sixtieth birthday ed. by K. C. Heesterman, G. H. Schokker & V. I. Subramoniam, 9-19. The Hague: Mouton.

—. 1973. "Zur Flexion des lateinischen Perfekts". Münchener Studien zur Sprach-wissenschaft 31.133-150.

Neu, Erich. 1968a. Das hethitische Mediopassiv und seine indogermanischen Grundlagen. Wiesbaden: Otto Harrassowitz.

—. 1968b. Interpretation der hethitischen mediopassiven Verbalformen. Ibid. ---. 1976. "Zur Rekonstruktion des indogermanischen Verbalsystems". Studies in Greek,

Italic, and Indo-European Linguistics offered to L. R. Palmer ed. by Anna Morpurgo Davies & Wolfgang Meid, 239-254. Innsbruck: Institut fur Sprachwissenschaft, Univ. Innsbruck.

—. 1979. "Einige Ueberlegungen zu den hethitischen Kasusendungen". Neu & Meid 1979. 177-196.

—. 1980. Studien zum endungslosen "Lokativ" des Hethitischen. Innsbruck: Institut fur Sprachwissenschaft, Univ. Innsbruck.

—. 1985. "Das frühindogermanische Diathesensystem: Funktion und Geschichte". Schlerath & Rittner 1985. 275-295.

— & Wolfgang Meid, eds. 1979. Hethitisch und Indogermanisch. Vergleichende Studien zur historischen Grammatik und zur dialekt-geographischen Stellung der indoger­manischen Sprachgruppe Altkleinasiens. Innsbruck: Institut für Sprachwissenschaft, Univ. Innsbruck.

Novotný, František. 1946. Latinská mluvnice pro střední školy [Latin Grammar] Vol.III. 1. Praha: Jednota českých filologů.

Page 242: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

REFERENCES 225

Nyman, Martti. 1985. "ë/ö/a as an Ablaut Pattern in Indo-European". Indogermanische Forschungen 90.55-61.

Oertel, Hanns & Edward P. Morris. 1905. "An Examination of the Theories Regarding the Nature and Origin of Indo-European Inflections". Harward Studies in Classical Philology 16.63-122

Oettinger, Norbert. 1976. "Der indogermanische Stativ". Münchener Studien zur Sprachwissenschaft 34.109-149.

-~. 1979. Die Stammbildung des hethitischen Verbums. Nürnberg: Hans Carl. —. 1984. "Zur Diskussion um den lateinischen ā-Konjunktiv". Giotta 62.187-201. —. 1985. "Thematische Verbklassen des Hethitischen: Umlaut und Ablaut beim

Themavokal". Schlerath & Rittner 1985. 296-312. Panhuis, Dirk. 1980. "The Personal Endings of the Greek Verb: An exploration in the

subjectivity and non-arbitrariness of a paradigm". Studies in Language 4.105-117. —. 1981. "The Arbitrariness of the Lingual Sign as a Symptom of Linguistic Alienation".

Studies in Language 5.343-360. Pedersen, Holger. 1926. La cinquième déclinaison latine. København: Det Kgl. Danske

Videnskabernes Selskab. Perel'muter, Ilja Aronovič. 1977. Obščeindojevropejskij i grečeskij glagol: Vido-

vremennyje i zalogovyje kategorii. Leningrad: Nauka. —. 1984. "Indojevropejskij medij i refleksiv: Opyt funkcional'nogo analiza". Voprosy

jazykoznanija 1984/1.3-13. Petersen, Walter. 1933. "Hittite and Tocharian". Language 9.12-34. Pinkster, Harm. 1983. "Tempus, Aspect and Aktionsart in Latin: Recent trends 1961-

1981". Aufstieg und Niedergang der römischen Welt ed. by Hildegard Temporini & Wolfgang Haase, Vol.II.29/1, 270-319. Berlin & New York: Walter de Gruyter.

—, ed. 1983. Latin Linguistics and Linguistic Theory: Proceedings of the 1st International Colloquium on Latin Linguistics, Amsterdam, April 1981. Amsterdam & Philadelphia: John Benjamins.

Plank, Frans. 1979. "Ergativity, Syntactic Typology and Universal Grammar: Some past and present viewpoints". Plank 1979. 3-36.

—. 1981. Morphologische (Ir)regularitäten. Tübingen: Gunter Narr. —. 1986. "Paradigm Size, Morphological Typology, and Universal Economy". Folia

Linguistica 20.29-48. —, ed. 1979. Ergativity. Towards a theory of grammatical relations. London: Academic

Press. Pohl, Heinz Dieter. 1977. Slavisch und Lateinisch. Klagenfurt: Klagenfurter sprach-

wissenschaftliche Gesellschaft (= Klagenfurter Beitrge zur Sprachwissenschaft. Beiheft 3).

Pokorny, Julius. 1959. Indogermanischen Etymologisches Wörterbuch. Vol.1. Bern & München: A. Francke

Popela, Jaroslav. 1985. "K otázce kombinace typologických vlastností v jazycích".

Page 243: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

226 FROM INDO-EUROPEAN TO LATIN

Bulletin ruského jazyka a literatury 26.37-67. Porzig, Walter. 1954. Die Gliederung des indogermanischen Sprachgebiets. Heidelberg:

Carl Winter. Prat, Louis C. 1975. Morphosyntaxe de l'ablatif en latin archaïque. Paris: Les Belles

Lettres. Ramat, Paolo. 1987. Linguistic Typology. Berlin & New York: Mouton de Gruyter. Rasmussen, Jens Elmegård. 1985. "Der Prospektiv - eine verkannte indogermanische

Verbalkategorie?". Schlerath & Rittner 1985. 384-399. Reichelt, Hans. 1909. Awestisches Elementarbuch. Heidelberg: Carl Winter. Rijksbaron, Albert. 1984. The Syntax and Semantics of the Verb in Classical Greek: An

introduction. Amsterdam: Gieben. Risch, Ernst. 1977. "Das System der lateinischen Deklinationen". Cahiers Ferdinand de

Saussure 31.229-245. (Repr. in Kleine Schriften ed. by Annemarie Etter & Marcel Looser, 599-615. Berlin & New York: Walter de Gruyter, 1981.

Rix, Helmut, ed. 1975. Flexion und Wortbildung: Akten der 5. Fachtagung der Indogermanischen Gesellschaft, Regensburg 9.- 14. September 1973. Wiesbaden: Ludwig Reichert.

—. 1976. Historische Grammatik des Griechischen: Laut- und Formenlehre. Darmstadt: Wissenschaftliche Buchgesellschaft.

—. 1977. "Das keltische Verbalsystem auf dem Hintergrund des indoiranisch-griechischen Rekonstruktionsmodells". Schmidt & Ködderitzsch 1977.132-158.

—. 1986. Zur Entstehung des indogermanischen Modussystems. Innsbruck: Institut rur Sprachwissenschaft, Univ. Innsbruck.

—. 1988. "The Proto-Indo-European Middle: Content, forms and origin". Münchener Studien zur Sprachwissenschaft 49.101-119.

Romportl, Simeon. 1989. "Kořeny a smysl biologického myšlení v jazykovědných pracích Augusta Schleichera (1821-1868)". Listy filologické 112.8-23. [English Summary, "The Roots and Sense of the Biological Thinking in Schleicher's Linguistic Works" p.23].

Ruijg, C. H. 1967. Études sur la grammaire et le vocabulaire du grec mycénien. Amsterdam: Hakkert.

Ruipérez, Martin Sánchez. 1952. "Desinencias medias primarias indoeuropeas sg.l.a *-(m)ai, 2.a -soi, 3.a -(t)oi pl. 3.a -ntoi". Emerita 20.8-31.

Safarewicz, Jan. 1969. Historische lateinische Grammatik. Halle (Saale): Max Niemeyer. Salkie, R. 1987. Review of Carl Bache, Verbal Aspect. A general theory and its

application to present-day English. (Odense: Odense Univ. Press, 1985). Lingua 73.130-140.

Sapir, Edward. 1921. Language. An introduction to the study of speech. New York: Harcourt, Brace & Co

Sasse, Hans-Jürgen. 1982. "Subjektprominenz". Fakten und Theorien: Beitrăge zur romanischen und allgemeinen Sprachwissenschaft: Festschrift für Helmut Stimm zum

Page 244: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

REFERENCES 227

65. Geburtstag ed. by Sieglinde Heinz & Ulrich Wandruszka, 267-286. Tübingen: Gunter Narr.

—. 1987. "The Thetic/Categorical Distinction Revisited". Linguistics 25.511-580. —. 1988. "Der irokesische Sprachtyp". Zeitschrift für Sprachwissenschaft 7.173-213. —. Forthcoming. "Syntactic Categories and Subcategories". Handbuch zur Sprach und

Kommunikationswissenschaft. Syntax ed. by Theo Venneman & Joachim Jacobs. Berlin & New York: Walter de Gruyter

Scherer, Anton. 1975. Handbuch der lateinischen Syntax. Heidelberg: Carl Winter. Schindler, Jochen. 1967a. "Das indogermanische Wort rur 'Erde' und die dentalen

Spiranten". Die Sprache 13.171-205. —. 1967b. "Tocharische Miszellen". Indogermanische Forschungen 72.239-249. —. 1975a. "Zum Ablaut der neutralen s-Stămme des Indogermanischen". Rix 1975.259-

267. —. 1975b. "L'apophonie des thèmes indo-européens en r/n" .Bulletin de la Société de

Linguistique de Paris 70.1-10. —. 1977. "A Thorny Problem". Die Sprache 23.25-35. Schleicher, August. 1850. Die Sprachen Europas in systematischer Uebersicht. Bonn: H.

B. König. (New. ed., eith an introd. by E. F. K. Koerner, Amsterdam & Philadelphia: John Benjamins, 1983.)

Schlerath, Bernfried & Veronica Rittner, eds. 1985. Grammatische Kategorien. Funktion und Geschichte. Akten der VIL Fachtagung der indogermanischen Gesellschaft, Berlin, 20.- 25. Februar 1983. Wiesbaden: Ludwig Reichert.

Schmalstieg, William R. 1980. Indo-European Linguistics. A new synthesis. University Park & London: the Pennsylvania State Univ. Press.

Schmid, Wolfgang P. 1963a. Studien zum baltischen und indo-germanischen Verbum. Wiesbaden: Otto Harrassowitz.

—. 1963b. Review of Watkins (1962). Indogermanische Forschungen 68.224-228. —. Review of Albert Thumb, Handbuch des Sanskrit mit Texten und Glossar. Eine

Einflihrung in das sprachwissenschaftliche Studium des Altindischen. Vol. I: Grammatik, Teil 2: Formenlehre. Dritte stark umgearbeitete Auflage von Richard Hauschild (Heidelberg: Carl Winter, 1959). Indogermanische Forschungen 65.297-302.

—. 1973. "Sprachwissenschaftliche Bernerkungen zum hethitischen 'Direktiv'". Festschrift Heinrich Otten ed. by Erich Neu & Christel Rüster, 291-301. Wiesbaden: Otto Harrassowitz.

—. 1985. "Das Lateinische und die Alteuropa-Theorie". Indogermanische Forschungen 90.129-146.

Schmidt, Gernot. 1977. "Das Medium im vorhistorischen Keltisch". Schmidt & Ködderitzsch 1977.89-107.

Schmidt, Karl Horst. 1963a. "Venetische Medialformen". Indogermanische Forschungen 68.160-169.

Page 245: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

228 FROM INDO-EUROPEAN TO LATIN

—. 1963b. "Zum altirischen Passiv". Indogermanis che Forschungen 68.257-275. —. 1964. "Das Perfektum in indogermanischen Sprachen. Wandel einer Verbalkategorie".

Giotta 42.1-18. —. 1966. "Konjunktiv und Futurum im Altirischen". Studia Celtica 1.19-26. —. 1975. "Das indogermanische Kasusmorphem und seine Substituenten". Rix 1975.268-

286. —. 1977. "Zum Problem des Genitivs der o-Stămme im Baltischen und Slavischen".

Commentationes linguisticae et philologicae Ernesto Dickenmann lustrum claudenti quintum decimum ed. by Friedrich Scholz, Wilma Woesler & Peter Gerlinghoff, 335-344. Heidelberg: Carl Winter.

—. 1979. "Reconstructing Active and Ergative Stages of Pre-Indo-European". Plank 1979.333-345.

—. 1980. "Zur Typologie des Vorindogermanischen". Linguistic Reconstruction and Indo-European Syntax. Proceedings of the Colloquium of the "Indogermanische Gesellschaft", University of Pavia, 6-7 September 1979 ed. by Paolo Ramat, 91-112. Amsterdam: John Benjamins.

—. 1982. "Caucasian Typology as an Aid to the Reconstruction of Proto-Indo-European". Folia Linguistica 5.364-383.

—. 1983. "Indogermanisch als Diasystem". Faust 1983.341-348. —. 1984. "Rekonstruktion und Ausgliederung der indogermanischen Grundsprache".

Incontri Linguistici 9.127-133. —. 1986. "Zur Vorgeschichte des prädikativen Syntagmas im Indogermanischen". O-o-pe-

ro-si. Festschrift fur Ernst Risch ed. by Annemarie Etter, 90-103. Berlin: Walter de Gruyter.

—. 1990. " Zum plomb du Larzac". Celtic Language, Celtic Culture. A festschrift for Eric P. Hamp ed. by A.T.E. Matonis & Daniel F. Melia, 16-25. Van Nuys, California: Ford & Bailie.

— & Rolf Ködderitzsch, eds. 1977. Indogermanisch und Keltisch. Kolloquium der Indogermanischen Gesellschaft am 16. und 17. Februar 1976 in Bonn, unter Mitwirkung von R.Ködderitzsch herausgegeben von K. H. Schmidt. Wiesbaden: Lud­wig Reichert.

Schmidt, Klaus T. 1974. Die Gebrauchsweisen des Mediums im Tocharischen. Diss., Georg-August-Univ., Göttingen.

Schmidt, Manfred Erwin. 1930. "Untersuchungen zur albanischen Sprachgeschichte". Zeitschrift für vergleichende Sprachforschung 57.1-42.

Schulze, Wilhelm. 1933. Kleine Schriften. Göttingen: Vandenhoeck & Ruprecht. Seiler, Hansjakob. 1983. "Possessivity, Subject and Object". Studies in Language 7.89-

117. — & Christian Lehmann, eds. 1982. Apprehension. Das sprachliche Erfassen von

Gegenständen. Teil 1. Tübingen: Gunter Narr. Serbat, Guy, ed. 1980. Le sens du parfait de l'indicatif actif en Latin: Colloque de

Page 246: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

REFERENCES 229

Morigny, 2 décembre 1978. Paris: Univ. de Paris-Sorbonne. Sgall, Petr. 1979. "Die Sprachtypologie V. Skaličkas". Skalička 1979. 1-20. —. 1986. "Classical Typology and Modern Linguistics". Folia Linguistica 20.15-28. —. 1988. "Natürlichkeit, Syntax und Typologie". Zeitschrift Jür Phonetik, Sprach-

wissenschaft und Kommunikationsforschung 41.463-470. Sihler, Andrew L. 1977. "The Etymology of PIE *reg- king, etc.". Journal of Indo-

European Studies 5.221-246. Silverstein, Michael. 1976. "Hierarchy of Features and Ergativity". Grammatical

Categories in Australian Languages ed. by Robert M. W. Dixon. N.J.: Humanities Press.

Simone, Carlo de. 1980. "L'aspetto linguistico". Stibbe et al. 1980. 71-94. Skalička, Vladimír. 1979. Typologische Studien. Mit einem Beitrag von Petr Sgall. Ed. by

Peter Hartmann. Braunschweig & Wiesbaden: Friedrich Vieweg. Solta, Georg Renatus. 1984[1985]. Review of Gert Klingenschmitt, Das altarmenische

Verbum. (Wiesbaden: Ludwig Reichert, 1982). Kratylos 29.59-74. Sommer, Ferdinand. 1914. Randbuch der lateinischen Laut- und Formenlehre.

Heidelberg: Carl Winter. Specht, Franz. 1935. "Zur Geschichte der Verbalklasse auf -ē": Ein Deutungsversuch der

Verwandschaftsverhăltnisse des Indogermanischen". Zeitschrift für vergleichende Sprachforschung 62.29-115.

—. 1944. Der Ursprung der indogermanischen Deklination. Göttingen: Vandenhoeck & Ruprecht.

Stang, Christian S. 1932. "Perfektum und Medium". Norsk tidsskrift for sprogvidenskap 6.29-39.

—. 1942. Das slavische und baltis che Verbum. Oslo: Dybwad. —. 1966. Vergleichende Grammatik der baltischen Sprachen. Oslo: Universitatsforlaget. Starke, Frank. 1977. Die Funktionen der dimensionalen Kasus und Adverbien im

Althethitischen. Wiesbaden: Otto Harrassowitz. Stibbe, C.M., et al. 1980. Lapis Satricanus: Archaeological, epigraphical, linguistic and

historical aspects of the new inscription from Satricum. (= Archeologische Studien van het Nederlands Instituut te Rome: Scripta Minora 5.) 's Gravenhage: Staatsuitgeverij.

Streitberg, Wilhelm. 1906. Gotisches Elementarbuch. Heidelberg: Carl Winter. Strunk, Klaus. 1967. Nasalprăsentien und Aoriste. Ein Beitrag zur Morphologie des

Verbums im Indo-Iranischen und Griechischen. Heidelberg: Carl Winter. —. 1980. "Zum idg. Medium und konkurrierenden Kategorien". Brettschneider &

Lehmann 1980. 321-337. —. 1984. "Probleme der Sprachrekonstruktion und des Fehlen zweier Modi im

Hethitischen". Incontri linguistici 9.135-152. —. 1987. "Further Evidence for the Diachronic Selection: Ved. răsti, Lat. regit, etc.".

Cardona 1987. 385-392.

Page 247: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

230 FROM INDO-EUROPEAN TO LATIN

Szemerényi, Oswald. 1956. "Latin rēs and the Indo-European Long-diphtong Stem Nouns". Zeitschrift für vergleichende Sprachforschung 73.167-202. (= Scripta minora II, 781-816.)

—. 1987[1969]. "Unorthodox View of Tense and Aspect". Scripta minora I, 161-171. —. 1987[1975]. "Rekonstruktion in der indogermanischen Flexion: Prinzipien und

Probleme". Scripta minora I, 257-277. —. 1980. Einfimrung in die vergleichende Sprachwissenschaft. 2. Aufl. Darmstadt:

Wissenschaftliche Buchgesellschaft. —. 1985. "Strukturelle Probleme der indogermanischen Flexion. Prinzipien und

Modellfälle". Schlerath & Rittner 1985. 515-533. —. 1987a. Scripta minora: Selected Essays in Indo-European, Greek and Latin ed. by P.

Congidine & J.T. Hooker, Vol.1. Indo-European, Vol.11. Latin. Innsbruck: Institut rur Sprachwissenschaft, Univ. Innsbruck.

—. 1987b."The Origin of Aspect in the Indo-European Languages". Giotta 65.1-18. Tesařová, Drahomíra. 1988. "Die Kategorie des Genus verbi und der Deagentităt im

Lateinischen". Listy filologické 111.145-154. Thomas, Werner. 1985. Die Erforschung des Tocharischen (1960-1984). Stuttgart: Franz

Steiner. —. 1986. Review of Windekens (1982). Indogermanische Forschungen 91.361-368. Thurneysen, Rudolf. 1909. Handbuch des Altirischen Vol.1. Grammatik. Heidelberg: Carl

Winter. —. 1946. A Grammar of Old Irish. Dublin: Institute for Advanced Studies. Tischler, Johann. 1982. "Zur Entstehung der -#/-Konjugation: Überlegungen an Hand des

Flexionsklassenwechsels". Investigations philologicae et comparativae: Gedenkschńft fur Heinz Kronasser ed. by Erich Neu, 235-249. Wiesbaden: Otto Harrassowitz.

Trost, Pavel. 1939. "Zum lateinischen Konditionalsatz". Giotta 27.206-211. —. 1962. "Subjekt a predikát". Slavica Pragensia 4.267-269. —. 1964. "K lingvistické charakteristice klasické latiny". Zprávy Jednoty klasických

filologů 6.110-111. [Summary of lecture]. —. 1985. "Medium und Reflexiv". Studia Linguistica Diachronica et Synchronica:

Werner Carl Winter Sexagenario anno MCMLXXXIII gratis animis ab eius collegis, amids discipulisque oblata ed. by Ursula Pieper & Gerhard Stickel, 825-827. Berlin: Mouton de Gruyter.

—. 1987. "Subject and Predicate". Explizite Beschreibung der Sprache und automatische Textbearbeitung, vol.XIV: Probleme und Perspektiven der Satz und Textforschung, 145-150. Praha: Matematicko-fyzikální fakulta, Univ. Karlova. [= Engl, translation of Trost (1962) with Postscript on p. 150.]

Untermann, Jürgen. 1968. "Zwei Bemerkungen zur lateinischen Perfektťlexion". Studien zur Sprachwissenschaft und Kulturkunde: Gedenkschrift fiXr Wilhelm Brandenstein (1898-1967) ed. by Manfred Mayrhofer et al., 165-171. Innsbruck: Institut rur Sprachwissenschaft, Univ. Innsbruck.

Page 248: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

REFERENCES 231

—. 1984. "Reflexiv und Medium". Unpublished lecture presented in the plenary session of UNITYP project, Institut für Sprachwissenschaft, Univ. Köln.

—. 1986. "Lapis Satricanus". Göttingische Gelehrte Anzeigen 238.24-30. [Review of Stibbe 1980.]

Vendryes, Joseph. 1910. "Sur quelques présents en -a- du verbe italo-celtique". Mémoires de la Société de Linguistique de Paris 16.300-310.

Villar, Francisco. 1981. Dativo y locativo en el singular de la flexion nominal indoeuropea. Salamanca: Ediciones Universida.

—. 1983. Ergatividad acusatividad y genero en la familia linguistica indoeuropea. Ibid. Wackernagel, Jacob. 1920. Vorlesungen uber Syntax mit besonderer Berücksichtigung von

Griechisch, Lateinisch und Deutsch. 1. Reihe. Basel: Birkhăuser. —. 1928. Vorlesungen Uber Syntax mit besonderer Berücksichtigung von Griechisch,

Lateinisch und Deutsch. 2. Reihe. Basel: Birkhăuser. Walde Alois & Johann Baptist Hofmann. 1938. Lateinisches etymologisches Wörterbuch.

3., neu bearb. Aufl. von J. B. Hofmann. Heidelberg: Carl Winter. Walter, Heribert. 1981. Studien zur Nomen-Verb Distinktion aus typologischer Sicht.

München: Wilhelm Fink. Watkins, Calvert. 1962. Indo-European Origins of the Celtic Verb 1: The sigmatic aorist.

Dublin: Institute for Advanced Studies. —. 1963. "Preliminaries to a Historical and Comparative Analysis of the Syntax of the

Old Irish Verb". Celtica 6.1-49. —. 1967. "Remarks on the Genitive". To Honor Roman Jakobson: Essays on the occasion

of his seventieth birthday Vol.Ill, 2191-2198. The Hague: Mouton. —. 1969. Indogermanische Grammatik. Vol.Ill: Formenlehre. Part 1: Geschichte der

indogermanischen Verbalflexion. Heidelberg: Carl Winter. Welmers, William Everett. 1973. African Language Structures. Berkeley: Univ. of

California Press. Wierzbicka, Anna. 1980. Lingua Mentalis. The Semantics of Natural Language. Sydney,

etc.: Academic Press. Wijk, Nikolaus van. 1902. Der nominale Genitiv Singular im Indogermanischen in seinem

Verhăltnis zum Nominativ. Zwolle: Tiil. Wilensky, Robert. 1990. "Meaning and Knowledge Representation". Proceedings of the

Fourteenth International Congress of Linguists, Berlin, August 10 - August 15, 1987, vol.1, 77-104. Berlin: Akademie-Verlag.

Williams, Edwin. 1981. "On the Notions 'Lexically Related' and 'Head of a Word'. Linguistic Inquiry 12.245-274.

Windekens, A.J. van. 1982. Le tokharien confronté avec les autres langues indo­européennes. Vol.II.2: La morphologie verbale. Louvain: Centre International de Dialectologie Génerale.

Wurzel, Wolfgang U. 1984. Flexionsmorphologie und Natürlichkeit. Berlin: Akademie-Verlag.

Page 249: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)
Page 250: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

Index

A ablative 57, 78, 82, 90-92, 94, 97, 101

Hittite 95 Latin 52, 78, 82, 89, 91, 97

ablative function double range: with genitive, with

instrumental 97 ablaut 11, 25, 26, 29, 54, 55, 77, 110

in inactive verbs 139 see also alternation

ablauting 56, 131, 138, 139 absolute

and relative tenses in Latin 153 absolutive 11, 15, 16, 18, 26, 32, 51,

62, 83, 86, 90, 91, 97, 123-125 accent 25, 26, 29, 110 accentual word types 29 accentuation

acrostatic or proterodynamic 139 accentuation classes

of IE words 55 accusative 43, 50, 78, 80-82, 86, 87, 89,

124, 125 accusative of respect 40 accusative plural 72, 73, 92 Greek and Latin 40

acrostatic aerostatic or proterodynamic

accentuation 139 act of predication 85 actant - event correlations 40

actant roles 162 actants 16

actor and undergoer 122 vs. participants 122

active and inactive verb classes 115-141

active and inactive systems integration of 91

active hypothesis 16 active languages 17 active or ergative structure

of IE 15-21, 124-126 see also active/ergative hypothesis

active preterite/aorist + inactive state/perfect

as components of Latin perfect 155 active roots 128 active structure

of IE 15-21 see also active hypothesis

active structure as a type of fundamental syntactic relations 21

active verbs imperfective, perfective 118

active vs. ergative, nominative-accusative 87, 193

structure, language 13, 15 active vs. inactive 15, 26, 27, 35, 38,

39, 84, 112, 129, 144, 158, 182, 193 -e/-o suffix vs. -ă, -ë suffixes 176 active -si perfect/aorist vs. inactive -ui

Page 251: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

234 FROM INDO-EUROPEAN TO LATIN

perfect 149 active perfect vs. middle-passive

present 131 active preterite - aorist 143 and animate vs. inanimate 124 and diathesis 115 and thematic present 135 case 16, 17, 125 IE verbs — reconstruction 127 morphology 131 morphology — Latin perfect 147 perfect 143 person affixes 16 prefixes in Dakota 19 roots 127 vague semantic classification 127 verb classes of IE 20, 59, 109, 113,

115-141 verbs 19

active vs. middle oppositions Greek 161

active vs. middle-passive 35, 36, 117, 131, 138, 139, 156-158, 168

active perfect vs. middle-passive present 36

Greek active vs. middle oppositions 161

Latin deponent/passive 160 active vs. passive 112 active/ergative hypothesis 125 active/ergative morphosyntax 126 active/transitive

secondary 159 actor 84, 122, 123

non-agentive actor 162 actor - undergoer opposition 20 actual 134 actual knowledge 85 actualizing particle 113, 132 adaptational

capacities of IE morphology 184 addition 6, 7, 9, 38, 56, 72, 73, 76,

101, 159 additional

adaptation (modification, elements) 25, 33, 180, 184

additive 8

agglutinative morphology 50 adjectival verbs (predicates)

in active languages 20 adjective 19, 20, 41, 57

and agreement 42, 66 and gender 26, 59, 63, 64 and word order 66 Bantu 65 predicative 44 referential and nominal character in IE

41 adnominal genitive 40 Adygean 19 affected 81, 87 affinity

paradigmatic of cases 101 affixation 24 African languages 64 agent 19, 20, 79, 80, 84, 122, 125, 158,

159 vs. non-agent 158

agent - goal perspective 123 of transitive active verbs 121

agent vs. experiencer 162 agent vs. non-agent 120, 125, 159 agent-from-subject demotion 112, 160 agentive vs. non-agentive actor 162 agentive/active vs. non-agentive/inactive

128 agglutination-like

accusative plural 72 addition 56 structure 6, 72

agglutinative 6-9, 49, 50, 57, 72, 73, 159, 167

agreement 6, 10, 41, 42, 58, 63, 68, 80, 82, 85, 86, 101, 102, 126

and word autonomy 66 between subject nominative and subject

person 84 combination of nominative marking and

agreement 86 IE in case, number and gender 58 in Balkan languages 67 in Bantu 65 in inactive verbs 84, 126, 138

aktionsart 26, 60, 110-113, 119, 133,

Page 252: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INDEX 235

140, 149, 186, 188 Albanian 166, 169 alienable possesion 89 allomorphic 11, 26, 38, 53, 57, 67, 97,

109, 111, 134, 135, 149, 174, 181, 182

allomorphization 59, 86 allomorphized 26, 83, 98 allomorphy 27, 28, 35, 49, 50, 54, 56,

100, 103, 194 and paradigmatization 53 in d-f and p-f structure 53 in Latin noun inflection 51-54

alternating 55, 151, 173-175, 179, 180, 189

alternating thematic -e/-o suffix 173 alternation 10, 11, 24, 26, 29, 30, 37,

54 ,77 , 83, 88, 110, 128, 133, 139, 151, 156

types of 55 alternative 3-5, 7, 11, 96, 143, 153,

156, 169, 176, 180 grammaticalization 12, 92, 110, 174,

181, 185, 191 paradigmatization 3, 4, 12, 37, 109,

110, 194 ambiguity of ablative function

eliminated in Latin 97 American languages 117 amphidynamic 55 analogical 71, 74, 99

contamination 25, 135, 167 analogy 52, 71, 93, 189

non-proportional 71 analysability 76 analysable 7-9, 29, 49, 51, 60, 74, 108 analytic 3, 6-8, 19, 23, 30, 45, 46, 76,

77, 187 explanation of IE 8

animacy hierarchy 17 animate

and nominative marking 87 animate and Inanimate 61 animate experiencer 84 animate nominative and accusative vs.

inanimate absolutive allomorphic in Latin 97

animate nominative vs. inanimate absolutive 83

animate nouns nominative of 16

animate vs. inanimate 17, 18, 21, 26, 27, 38, 39, 51, 55, 59, 66-68, 96, 121, 124

and case system 77 and casus indefinitus 90 and nominative marking 83 and number 70 as disambiguating device 101

animate/masculine, feminine 59 animate/personal class 18 animateness 62

and nominative marking 83 animates and inanimates

morphological differences between 61 semantic differences in number and

case 59 animation 62, 64 anterior vs. non-anterior/absolute 155 anteriority 153 aorist 116, 118-120, 129, 143, 144, 154

and perfect 119, 120, 143-147, 150, 156

aspectually neutral past tense in Latin 156

complex expression of 37 Greek 11, 36, 37 sigmatic 141 sigmatic and reduplicated 129 sigmatic, reduplicated 36, 129 thematic 133, 140, 147

aorist of active verb identified with perfect of inactive verb

in Latin 144 aoristic

-s- suffix 141 and modal -s- suffixes 141 and modal -s-formation 182 element -is- 148, 185 perfect 120, 151 roots 133

aoristic and modal -s suffix 181 aoristo-present 115, 117 aorists

Page 253: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

236 FROM INDO-EUROPEAN TO LATIN

Greek -the- aorists 188 appositional 44

relative clauses 45 sentence structure 42, 43

Arcado-Cyprian 166 archaic Latin 57, 93, 132, 170, 185

-osio genitive 94 archaisms and innovations 37 Armenian 169, 171 Aryan 70, 91, 93 ,94 , 132 aspect 33, 40, 119, 193

and temporality 112 vs. tense 26

aspecto-temporal 12 bipartite vs. tripartite systems 194 distinctions 39, 46, 108, 111 opposition 133, 143 stem 24, 35, 144, 191 tripartite system 110

aspectual 26, 36, 60, 119, 155, 175, 186 aspectual/durative 177 distinctions 112, 115, 139, 153 formation 181 meaning 12 opposition 12, 145, 179 stem 182 terminative value 188

aspectual distinction perfective vs. imperfective

and past vs. non-past 153 aspectual subclasses 118, 119 athematic 134

forms 132, 134 modal sigmatic forms 181 subjunctive/optative 185

attributive syntagm 42 attributive syntagm of genitive +

nominative word order 88

attributive-predicative 44 augment 24, 37, 113, 153 Australian languages 17 autonomous 86, 87 autonomous/individual 62, 63, 81, 83

and nominative marking 81 autonomy

of nominal and verbal constituents 18

autonomy of adjective 64 auxiliary verb 187 Avestan 61, 63, 92

B backgrounded

actants 123, 126 backgrounding 164

device 124 middle-passive 168, 169 passive 164, 165, 167-169 use of inactive verbs 169

backgrounding passive, Latin vs. Greek foregrounding passive 164

Balkan languages 67, 68 Baltic 70, 93, 166, 169, 175, 185, 189

preterites 175 Baltoslavic 75, 91, 93, 94 Bantu 58, 65 basic case suffixes 54, 77 basic case system 77, 95

basic 95 Basque 16, 62 benefactive 79, 122, 162, 163, 165 bi-consonantal radicals 128 bicentral 86 binding device

article as 67 binding vowels 25 bipartite statements 7 bipartite structure of statement 85 bipartite temporal system

present vs. preterite 110 vs. tripartite 110

bipersonal verb 19 Bulgarian 67

C case 39, 45, 77

agreement 58 analytic vs. derivational origin 19 and animate vs. inanimate 51 , 59, 61 and flexive ending 31 and number 35, 49-51, 53, 69 case and number 32 case-number morpheme 4, 27, 32, 51 case-number-gender morpheme 60

Page 254: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INDEX 237

flectional expression of 27, 49, 50 Greek vs. Latin 40 in active languages 16, 17 in ergative languages 16 in ergative or active structure 16 indifferent, absolutive 11 paradigmatization 4 semantics 58 strong and weak cases 54

case category derivational origin 79

case ending 51 in Latin 98

case markers cooperation with person markers 121,

125 person markers and case markers 121

case paradigm differences in IE languages 95 in Greek 96 structural description 79

case suffix fusion with inflectional class marker

108 case suffixes

and alternation 25 and derivational suffixes 59 basic 54-56, 77 explained as analytic elements 77 fusion with declension marker 51 marked 95

case suffixes of -o- stems innovative 93

case system (paradigm) 58, 77 basic 77 development of 77 paradigmatization 81

cases fundamental 61 original form and meaning 77 polyfunctionality 79

casus commemorativus 90 casus indefinitus

as quotation form 90 casus indefinitus or primitivus 90 categorial/immanent semantics

in d-f type 79

categorical statement 84-86, 126 category-immanent semantics 112

primacy of 38 category-immanent vs. relational

semantics 37-39 category-inherent 87 category-inherent (immanent)

semantics of cases 103 category-inherent semantics 40, 58, 63,

87 causative affixes 123 Celtic 75, 137, 165, 167-169, 176, 178,

185, 189 and Latin 53

central vs. marginal 79 centralizing 18, 43, 44, 46 centripetal (introvertal) diathesis 159 clusters

of properties 6 collective plural 5 1 , 6 2 , 6 9 collectivity 11 Common Indo-European (CIE) 4, 178, 196 comparative characterization of Latin and

Greek 7 complement frame of the verb 84 complementary

devices 149 forms 178, 182 functions 125 units 13

complex 9, 25, 27, 28, 33, 35, 37, 38, 40, 42, 43, 46, 49, 50, 60, 73, 103, 108, 109, 149, 174, 180

morpheme 32, 33, 49, 57, 58, 69, 74, 76

complex morpheme 75 complexity 26, 28, 32, 36-38, 76 complexity-cumulativity 51 composites 46 compositional basis

of typological evolution 7 connecting vowel 185 connective particle 57 consonant

inserted consonant 188 consonantal auslaut

of plural forms 75

Page 255: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

238 FROM INDO-EUROPEAN TO LATIN

consonantal endings active 133, 166 active, aoristo-present 115, 116 aoristo-present 133, 166

consonantal formations 181 consonantal markers

of tense 34 consonantal stems 55 consonantal suffixal formations 192 consonantal suffixes 182

case suffixes 51 grammaticalization 109 marked modal and temporal 140

consonantal third declension Latin — representing basic case forms

78 consonantized/mixed endings of

middle-passive 166 constitution of the subject 62 contact position 88 contamination 76

formal 74 functional 74 of active and inactive endings in

thematic present 135 of inactive and active endings in

middle-passive 166 continuity (graduality)

between inactive and middle-passive 164

continuum 25, 34 coordination

of active and inactive verbs 19 copula 20, 88 countable

vs. non-countable, collective 70 cumulative 27, 49, 72, 73

case-number morpheme 4 case-number-gender morpheme 60 expression of categorial distinctions 27 expression of number and case 69

cumulative expression of categorial distinctions 27

cumulative semantics of nominative marker 87

cumulative/fusional 27, 28, 32, 46, 50 morphemes 76

morphology 65, 103, 180 cyclical

application of suffixes 24, 25 suffixation 26, 183

D d-f = derivative-flectional d-f and p-f morphology

main differences 30 d-f structure

original character of 31 dative 78

+ directed, + dynamic 89 dative see also locative/dative 82 dative and instrumental

in Greek 96 dative vs. locative 89 dative-ablative plural

Latin 52, 92, 93 dative/locative 93

in Greek 96 deagentivum 137, 148

or backgrounding passive 164 declension marker (stem vowel) 51 declension of inanimates, neuters 90 declensional class

as marker of gender 67 decomposable 29 decomposition 8, 9, 27

via paradigmatization 8, 186 decomposition in paradigmatic

relationships 25 decompositional 73 decompositional basis

of typological evolution, of evolutionary process 7

decompositional capacity of flectional structure 9

definiteness 67 dehomonymization

structural 102 deictic category

of tense 112 deictic elements 70 demotion

agent-from-subject demotion 160 dental ablative suffix 57, 91, 92

Page 256: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INDEX 239

dental cases in Hittite 95

dependency-frame of verb 44

dependent 81, 87 deponent/passive 157

endings 170 vs. Greek/Aryan middle-passive 159

deponents 11, 36, 177 and inactive/process class 157

derivation 12, 77 from derivation to inflection 10

derivational 21, 26, 38, 80 capacities 27, 110 capacities of IE morphology 185,189 category of Aktionsart 26 character of case suffixes 54, 77 character of flexive formants 59 form and semantics of cases 19 origin of case category 79 origin of flexive formants 175 semantics of case suffixes 80

derivational and inflectional 9, 10, 12, 24, 26, 28, 30, 39, 52, 58-60, 108, 110

grammaticalization 108, 179, 180 derivational basis 183, 184 derivational history 28, 184 derivational origin

of flexive formants 46 derivative-flectional 35

formations 108 morphology 49, 53, 57, 97, 107 stage, (sub)type 3, 4, 7, 10, 13, 21,

23, 50, 191 structure 57, 61, 66, 77

derivative-flectional type constitutive features 80

desiderative 182, 186 determinative 128, 139 dialectal isogloss

Greek/Aryan, Italo-Celtic 169 dialectal isoglosses 168, 177

vs. parallel developments 177 diathesis 11, 35, 40, 80, 107, 115, 167

active vs. inactive 115 expressed by verb classes 158

semantic vs. syntactic 40 diathesis as a semantic-syntactic category

158 vs. verbal voice 112

diathetic 36, 111, 163 inflectional opposition 159 verb classes 118

diathetic distinction active vs. inactive 112, 118, 120, 157 agent vs. non-agent 158, 159, 161

diathetic function of inactive and middle-passive 159

diathetic opposition agentive vs. non-agentive 120

diathetic oppositions and semantic version 163

diathetic vagueness 130, 151, 152 in Hittite 122

diathetic verb classes 109 diathetic/inactive 177 diathetically vague

inactive verbs 121-123 differentiated 7 direct object 40

and accusative 80 direct-reflexive 160

and middle 158 directed 89 directive/terminative pl.

in Hittite 75 directive/terminative sg.

in Hittite 95 disambiguating devices 101,122 disambiguation 10, 101, 112, 162

by animate vs. inanimate categorization 96

by syntactic devices 101 constructional 101 of Greek dative/instrumental 96 structural 100

disappearance of neuter 68 of nom. vs. accus. opposition 68

discourse categories 86 distribution

derivational of grammatical formants 10, 11, 12, 31

Page 257: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

240 FROM INDO-EUROPEAN TO LATIN

distributional properties of grammatical formants 11,

12, 31 diversity of formations 12 dominating principle 5, 6, 23, 61 double article 67 dual 58, 69, 70, 73-75 dual/non-singular 69, 70, 73, 87, 90, 92 duality

as simple non-singularity vs. plurality 74

durative/progressive and intensive value 177

dynamic 89 vs. static 177

dynamic/personal 138

E embedded conjunctional clause 45 endings

-r endings: 3pl.indef. 136 -r endings: middle-passive, 3pl.indef.

167 of Latin deponent/passive 170 of Latin perfect 147 plural -r endings 136 sets of Latin endings 183

endings of active and inactive verbs 26 English 122 enlargement 177 equipollent opposition 87 ergative case 16, 17, 62, 87, 88, 126

Adygean 19 ergative hypothesis 16, 88 ergative languages 13, 15, 16, 18, 62 ergative marking

Georgian 18 ergative or active structure

of IE, see also ergative hypothesis 15 of IE, see also ergative/active

hypothesis 13, 15, 18 ergative structure 15-17

of IE, see also ergative hypothesis 4, 15

ergative/active case 17 ergative/active hypothesis 13, 17, 18,

87, 125

ethno-cultural 62 European languages 6, 43, 85, 122 events 39, 41 evolutional

perspective 7 process 7

evolutionary perspective from d-f to p-f structure 103

experiencer 79, 84, 121-123, 130, 162, 163, 165, 169

explicit 7-9, 25 explicitness

IE tendency to 27 of p-f structure 28 restricted 35

extrapolation 15, 118

F factitive/active

to primary inactive 162 to primary middle-passive 163

Faliscan 202 feminine 26, 61-64, 66, 67 finite verb

and word autonomy 43 complex character of 40

fixed word order 66, 88 flection 7 flectional

expression 8, 23, 25, 30, 33 expression of number and case 49 morpheme 8, 46 morphology 8, 27, 28, 31, 34, 46, 65 opposition 27 structure 4, 5, 10, 25, 27

flectional ending 31, 32, 34, 35, 37, 58 flectional morphology of IE

complex 27 cumulative/fusional 27 form variation 27

flectional principle 45, 73 flectional structure

semantic interpretation 9 flectional type 4, 7, 13, 23, 24, 29, 30

enlarged conception of 6, 9 origin of 3

flectional vs. agglutinative 49

Page 258: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INDEX 241

focus 86 foregrounded

actant in subject position 124 foregrounding 165, 168, 169, 171 form variation 27

semantically motivated 28 form variety 35, 37, 110, 174 formalization

of sentence structure 6, 42, 67 founded 38, 78 founding 38, 78 full grade 78 fundamental relations

character of 15 fusion 27, 28, 30, 34, 49, 51-53, 56, 58,

60, 98 fusional 9, 27, 56 fusional alternation 55 fusional/cumulative 108 future 35, 36

-se/-so futures of Latin and Greek 189

-s- suffixal 181 -sie/-sio futures of OI, Baltic and Celtic

189 Greek medial futures 182 in -se/-so or -sie/-sio 181 Latin -bo/-dho future 189 Latin -ě future 177 Latin -ě- future 177-179 OI -sio futures 182

future in -bo Latin indicative imperfect and the

future in -bo 187 futures

-sie/-sio futures 185 -s futures 182

futures and subjunctives -se/-so and -si- 185

futurity 182 futurum exactum

with emphatic affirmative value 153

G Gaulish 189 gelenk-partikel

connective particle 57

gender 4, 32, 39, 51, 57-59, 61-68, 87 and agreement 63, 83 and natural class 64 morphological-paradigmatic function

66 non-oppositional in substantives 64 oppositional in adjectives 64

gender agreement and word autonomy 66

gender declension in Bulgarian 67

generalization 12 generic, non-actual present 132, 135,

140 genitive 15, 78, 82

ablative function 82 adnominal, adverbal 82 as derived (founded) form 80 gen. + nom. syntagm 88 gen.sg. in -ï 53, 94 gen.sg. of -o- stems 52, 56 gen.sg. in -osio/-oso 92, 94, 96 innovative, semantic restriction 92 Latin — adnominal 82 partial participation 82 possessive, partitive 57, 82 predicative 88 relation to nominative and accusative

80 singular vs. plural 82

genitive marking 81 ,82 genitive plural 92

in -ōrum, -ārum 94 Latin 52

genitive singular explanations 88 Latin 52

genitive/ablative in Greek 96

Georgian 16, 18, 62 Germanic 25, 30, 151, 166, 169, 188,

189 Gothic 61, 91, 94, 116, 117, 151 governed

not governed 84 government 6, 10, 42, 43, 80, 82, 101

non-formalized character of 43

Page 259: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

242 FROM INDO-EUROPEAN TO LATIN

grades of vocalic alternations (ablaut) 24

graduality between derivational and inflectional

morphology 59 grammatical categories

expression in flectional type 7 expression in language types 5

grammatical component and lexico-derivational c. 30, 31, 37,

38 grammatical relevance

of lexical categories 10, 11 of lexical classes 11 of lexical morphemes 10 of lexical, word inherent semantics 26 of lexico-derivational categories 23 of noun and verb classes in IE 21

grammatical semantics vs. lexical-derivative semantics 8

grammatical words 8 words 7

grammatical/abstract cases vs. local/concrete 78

grammaticalization 12, 53, 79, 81, 83, 108, 110, 155

alternative 174, 181, 185 and paradigmatization 80, 110 and privative oppositions 87 of aspectual opposition 26 of consonantal suffixes 109 of derivational formations 181 of long vocalic injunctives 177, 178

grammaticalized 66 grammaticalized meaning 12, 28 grammaticalized subject 86

development of 86 Greek 35, 91, 93, 94, 112, 132, 134,

143, 148, 153 Greek -kα perfect

and Latin -vi/-ui perfect 150 Greek and Aryan

innovation in perfect and aorist 144 Greek and Latin

difference in the orientation of morphological semantics 40

Greek aorist of intransitive type 175

Greek case paradigm 96 Greek innovations 36 Greek middle 157 Greek perfect

more marked than Latin 156 Greek present optative

and Latin imperfect subjunctive 184 Greek present vs. aorist opposition

vs. Latin 153 Greek/Aryan 119, 133, 191

innovative middle 169 opposition aorist vs. perfect 143 vs. Latin verbal system 109

Greek/Aryan innovation in aorist and perfect formations 146

Greek/Aryan middle-passive vs. Latin 159 .

group inflection 19, 46, 91 vs. word inflection 77

Guarani 20

H heteroclisis

of neuters 27, 63 hiatus 29 hiatus filling consonant 25 hic-and-nunc situation 112 hierarchically ordered structure

of flexive ending 34 hierarchy

of morphosyntactic devices 10 of persons in Guarani 20

Hittite 52, 56, 63, 75, 81, 91, 94, 96, 117, 122, 132, 135, 136, 140, 143, 148, 166-169

-hi and -mi conjugations 135 and traditional interpretation of verb

system 110 case paradigm (system) 95 gender 61 middle 122 not distinguishing between aorist and

perfect 143 Homeric Greek 12, 36, 129, 151, 175 homonymic 90 homonymy 34, 180

and affinity of cases 101

Page 260: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INDEX 243

and increased allomorphy 100 higher in plural paradigm 97 in IE and Latin 100 of Latin case endings 100-102 of nom.sg. and gen.sg. 102

Hungarian 50, 57, 72 compared with Latin 57 noun phrase 57

hypercharacterization and redundance 33

hysterodynamic 55

I iconic 140

natural isomorphy 32 principle 109

iconicity between sg. and pl. paradigm 75, 99 between sg. and pl. paradigms 99 in number and case marking 73 markedness iconicity 91, 99 or naturalness 29, 165, 176

iconicity principle 73 ideal 184 imperfect

-bam < -dhām imperfect 189 Greek thematic 179 Latin and Greek 173,174 Latin and Slavic 187 Latin indicative imperfect 187 of Greek/Aryan type 174, 175, 179 perfect and imperfect in Latin 154

imperfect subjunctive 183 vs. present subjunctive 184

imperfect subjunctive, Latin and Greek present optative 184

imperfective vs. perfective 118, 153 imperfective/durative 119, 189

in inactive verb 145 imperfective/presentic 12 imperfective/progressive 139 impersonal deagentivum 136 impersonal passive 169

impersonal/backgrounding passive 165 Latin 160, 164 or deagentivum 167

impersonal/deagentive form in 2sg. 170

impersonal/indefinite 3rd person 137 impersonal/indefinite deagentivum 148,

165 implicit

gender in noun stem 63 implicit and stem internal expression 28 implicit semantic distinctions 26 inactive, see also active vs. inactive

-dh suffix 188 absolutive or inactive case 16

agreement category 159 and middle-passive 157, 159, 162 and perfect 143-145, 147-149, 155 and stative 158 Hittite 122 non-marked 158, 159 non-oppositional, vs. oppositional

middle 163 prefixes in Dakota 19

inactive -ui perfect active -si perfect/aorist and inactive -ui

perfect 149 inactive ending -i

OI, secondary middle 137 inactive endings

primitive two member system 138 inactive morphology 147 inactive personal endings

laryngeal interpretation 137 inactive plural -r forms 167 inactive prefix

in Guarani 20 inactive roots 127, 128, 130, 139

inactive 129 inactive suffix 20 inactive verb, see also active vs. inactive

and agreement 84 diathetically ambiguous 121 non-personal, non-oriented third person

84 personal endings 135 two-actant 126 unmarked 122 without intentional goal-orientation 121

inactive verb class transitive, intransitive 121

inactive verbs 36, 188

Page 261: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

244 FROM INDO-EUROPEAN TO LATIN

and Latin perfects 130 and semantic roles of subject 122 as adjectival predicates 20 neutral with respect to diathesis 123 non-oriented, diathetically vague 121 states and processes 118 transitive, non-oriented 20

inanimate 51, 55, 61 and instrumental 91

inanimate (neuter) 61 inanimate participant

involved in the state of affairs 125 inanimate source 123 inanimate undergoer 84, 123, 126

with inactive verbs 124 inanimate undergoer (source) 123 inanimate/neuter 59 incorporation-like 6, 45 incorporative

pseudo-incorporative 23 indefinite, non-specified subject 167 independent 81 independent vs. dependent 79 indicative

vs. subjunctive 34 indicative imperfect, Latin

analytic explanation 187 flectional explanation 187

indicative present 132, 176 indifferent 11, 62 indirect-reflexive middle 164

innovation of Greek/Aryan 163 innovative 169

individual 39, 62 individual 81, see autonomous/individual

individual/autonomous 71, 72 individuals 39 individuals/objects 41 infinitive 170

Greek, of present vs. aorist 155 Latin, of present vs. perfect 155

infix nasal 128

infixation 139 in nasal presents 149

inflection 12, 57

as grammatically relevant derivation 77

from derivation to inflection 10 inflection with stable root accent 63 inflectional 12

and derivational 9, 59, 108, 110 and lexico-derivational 23

inflectional categories and lexico-derivational class 115 and lexico-derivative classes 118 as grammaticalized derivational

categories 12, 108, 111 developed from derivative formations

191 inflectional class 24, 28, 51 inflectional class marker 60, 108 inflectional ending 24, 38 inflectional marker 33 inflectional morphology

and derivational morphology 59 inflectional paradigm 31

and morphological p. 97 inflectional stem 24, 25, 31, 33, 39, 174 inherent 13

characteristics of roots/stems 26 class semantics 26, 27

injunctival 182 preterito-modal function 110

injunctive 110, 113, 132, 140, 184 and thematic present 135 long vocalic 173, 174 non-actual value of 176 non-actual/ideal 132 preterital, modal 132, 176 thematic 173, 174

injunctive-based system 112, 115 changed into present-based s. 113, 131 vs. present-based s. 107

injunctive-based two-class system 191 injunctives in -ă- and -ë-

as archaic IE components 177 innovation

type-changing, type-preserving 37 innovative aspectual opposition aorist vs.

present 179 innovative Greek oppositional middle 165 innovative Greek/Aryan middle 163

Page 262: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INDEX 245

innovative indirect-reflexive middle 169 innovative preterital formations 188 innovative thematic aorist 134, 147 inserted consonant 188 instrumental 78, 90

as case form of inanimates 91 its merging or substitution by dative in

Greek 96 Latin 89 marked plural forms 92 non-marked form -e/-H 95 with vocalic case suffix 90

instrumental in -oi semi-adverbal in Greek 96

instrumental form underlying to ablative 92

instrumental meaning 91 instrumental plural in -oi-s 93 integral modification of the word

in flectional type 5, 8, 10 integrated present category 119 integrating tendency

of flectional word structure 9 integration

of active and inactive verbal systems 91

of animate and inanimate case systems 91

of case forms into paradigms 89, 91, 95

intensive durative/progressive and intensive value

177 more extent, more intensive 176

intention referential 85

intentional 121 intentional goal-orientation 124 intercategorial 39

and relational 39 interclassai 98 internal 30 internal and external inflection 24, 33

gradient difference 108 internal inflection 110, 139 internal modification of the word 7, 23 internal reconstruction 5

intonation patterns 85 intracategorial 39 intraclassal 98 intraclassal morphology

of active and inactive verbs 145 intransitive 122, 189

Greek aorist 131, 175, 177 undergoer-oriented 130

intransitive active verbs 16, 121 intransitive verbs 15,36 intransitive/inactive

vs. transitive/active 188 intransitive/passive 36 intransitive/stative

Greek perfect 156 intransitives

in Latin 164 intransitivity vs. transitivity 121, 123,

177 introvertal 157 introvertal diathesis 157, 159, 161, 162 invariable of the flectional type 13 invariant 50 inversion

subject inversion 85 'involved 125 Ionic/Attic 165 Irish 168, see Old Irish isoglosses in case formation 96 isolating 3, 7, 8 isolating type 7, 10 isolation 3, 7 isolation → agglutination → flection

hypothesis 30 Italic 75, 167-169 Italo-Celtic 169, 177

L language types 5-7 laryngeal 127, 128, 136 laryngeal interpretation

of inactive endings 137 Latin and Baltic 177, 178 Latin and Germanic 151 Latin and Greek/Aryan system

alternative developments 143 Latin idicative imperfect

Page 263: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

246 FROM INDO-EUROPEAN TO LATIN

analytic explanation 187 Latin imperfect subjunctive

and Greek present optative 184 Latin indicative imperfect

analytic explanation 187 Latin p-f structure 152 Latin paradigmatic structure

highly organized 153 Latin passive

personal, impersonal 160 Latin perfect 143

as basic preterite 143 inactive personal endings 143

Latin perfect and Greek aorist different position in system 153

Latin perfects of inactive verbs 130

Latin present conjugations 180 Latin sigmatic forms 182 Latin subjunctives and futures 177 lengthened nominatives 83 lengthened stems

of Latin perfects 151 lengthening

as nominative marker 83 lexeme

sentence with more lexemes 7 lexical and derivational categorization

10,21 lexical and grammatical components 37 lexical class

grammaticalized 4 non-oppositional 115

lexical stem determining the morphological

processes 23 in d-f structure 26

lexical subcategorization as expression of grammatical semantics

10 lexical, non-oppositional categorization

115 lexical, non-oppositional classes 59 lexico-derivational

and grammatical 30 and inflectional 23

lexico-derivational classes

and inflectional categories 118 lexico-derivational component

determining position in d-f type 80 lexico-derivational formant 188 lexico-derivational semantics

vs. grammatical 8 lexico-derivational stem 24 lexico-derivational subcategorization

grammaticaly relevant 23 Lithuanian 74, 91, 93, 128, 130, 134

and Latin 178 local suffix 75 local/concrete cases

vs. grammatical/abstract 78 location

pure, actual: locative 89 virtual: dative 89

locative 78, 90, 91 Greek, semi-adverbal 96

locative plural suffix -si/-su 92 locative vs. dative 56, 89, 95

restricted grammaticalization 89 locative/dative 82

indirect participation 82 long vocalic injunctive

-ă- injunctive 186 imperfective/durative 145

long vocalic injunctives 113, 173 long vocalic present

of inactive verbs 145 long vocalic preterites

with durative or intransitive meaning 175

long vocalic stem in -ë 187 long vocalic suffixes 173

as present stem markers 174 as preterital (imperfect) and modal

markers 174 inactive 133

long vocalic suffixes -ă- and -ë- 133, 138, 140

and thematic suffix 173 long vowel roots 128

M marginal case functions

and form variety 97

Page 264: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INDEX 247

marginal marked form of ablative 78

marked form of nominative 15, 18, 19, 87

marked middle-passive vs. non-marked inactive 159

marked mood 33 marked suffixes

plural 76, 91 marked tense 33 markedness 29

semantic and formal 99, 160, 171 singular vs. plural 99

markedness correlations 109 markedness iconicity 99 markedness theory

and nominative - accusative opposition 87

marker of animate class 87 marker of nominative

and marker of animate class 87 marker of plural 3 2 , 7 1 , 7 3 marker of plural vs. dual 73 marker of present indicative 116 marker of singular 87 masculine vs. feminine 26, 61, 63 maximization of semantic oppositions

in Greek/Aryan 179 meaning-constitutive activity 8 media tantum 116, 117, 131, 188 medial aorists

archaic Vedic 131 Greek of inactive verbs 131

medium see middle 36

Megrelian 18 Messapic 200 middle 36

and direct reflexive 158 and perfect 108, 115, 116, 156 as introvertal diathesis 161 as semantic version 161, 163 developed from inactive 157 Greek 157 Hittite 122 oppositional middle 191 oppositional, vs. non-oppositional

inactive 163 restricted outside Greek and Aryan

157 middle (medium) 115 middle future

to active present 36 middle of indirect-reflexive type

not attested in Latin 163 middle present and active perfect 36 middle-passive 35, 36, 117, 157

and inactive 157, 159 Greek/Aryan 110, 157 introvertal diathesis 159 its semantic and formal connection with

inactive 159 marked 158, 159 marked vs. non-marked inactive 159 polyfunctional 112 two type of 168

middle-passive endings 166 middle-passive of Greek/Aryan

vs. deponent/passive of Latin 159 middle-passive, marked

its origin in non-marked inactive 162 mixed

consonantized/mixed endings 166 mixed (consonantized) endings 166 mixed middle-passive endings 117,137,

166 mixed, contaminated endings

two types of 166 modal 39

-să in Celtic 189 -s suffixes and sigmatic aorist 141 -se in Latin 189 -sï in Latin, Baltic and Celtic 189 and aoristic -s suffixes 141, 181-183

modal and future forms with vocalic suffixes 109

modal and future sigmatic formations 181

modal formants with -e vocalism 178 modal marker 34, 60, 176 modal sigmatic forms

athematic, with long vocalic suffixes 181

modal vocalic exponents 9

Page 265: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

248 FROM INDO-EUROPEAN TO LATIN

modal/future -s-formations 182 modality as semantic-syntactic category

vs. mood 111 modifications

additional 184 mood 107 mood as morphological category

vs. modality 111 morpheme 8, 9, 31, 49

complex morpheme 75 morpheme addition 7

vs. word modification 8 morphemic structure of noun and verb

60 morphemic structure of verb 107, 108 morphemic structure of word 9, 30, 31 morphological analogy 52 morphological and inflectional paradigms

97 morphological and semantic-syntactic

categories 111 morphological classes 11

active vs. inactive 112, 116, 158 animate vs. inanimate 51

morphological diathesis differences between Greek and Latin

160 morphological diathesis (verbal voice)

middle-passive 157 morphological expression

non-additive, internal 30 morphological formants

polyfunctional 10 vague 10

morphological modification 52 morphological naturalness 25, 109 morphological opposition 30 morphological paradigm 31,36 morphological processes 23, 109, 139 morphological semantics 12, 37, 39

orientation of 112 orientation of, Greek and Latin 40 type of 37

morphological systems 12 morphologized 66 morphology

active and inactive verb 131

morphonological change 51 morphonological difference between

active and inactive roots 27, 127 morphonological motivation

of morphological processes 29 morphonological processes 29 morphonological rules of neuters 63 morphosyntactic differences

and allomorphic morphological differences 67

morphosyntactic form 10 morphosyntactic relevance

of animate vs. inanimate distinction 61 morphosyntactic relevance of animate vs.

inanimate distinction 63 Mycenaean 91, 166

N nasal infix 128 nasal infixation 24, 139 nasal present

of inactive tr. verbs 152 of inactive verbs 145 thematic in Latin, athematic in Greek

and Aryan 146 nasal suffix 139 natural correlations

between form and meaning 34 natural morphology 98

and IE flection 29 naturalness

morphological naturalness 109 typologically relevant aspects of 29

naturalness or iconicity principle 29 negative evidence

of active morphology 129 neuters 11, 51, 83, 97

casus indefinitus 90 disappearance of 68 heteroclisis 63 inflection with stable root accent 63 plural 62

neutral 70, 74, 76, 87, 123, 158 nominal and verbal constituents

autonomy and mutual independence in IE sentence 18

nominal and verbal systems

Page 266: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INDEX 249

in d-f and p-f structure 59 nominal case and number morphemes of

adjectives 20 nominal inflection 49 nominal predicates 19, 20 nominal semantic roles 162 nominal sentences 88 nominal syntagm (noun phrase)

in Latin and Hungarian 57 nominal system 58 nominative 18, 43, 78, 86, 124

+autonomous, (-dependent) 87 and subject 80 and subject constitution 83 formed by -s suffixation 52 formed by lengthening 52 marked by stem alternation

(lengthening) 83 marked IE 15, 18 nominative/accusative of neuters 62 non-singular and plural 73 pli. vs. nominative dual 75 position in Latin paradigm 90 restriction in -s nominative 83

nominative and accusative relation to genitive 80

nominative and accusative plural 71, 92 nominative marker

and marker of animate class 87 nominative marking 86, 87

of privileged participant 83 vs. accusative marking 81

nominative marking and agreement combination of 86

nominative of non-singular and plural 73 nominative plural 71

in -oi 94 Latin 52 without -s 75

nominative vs. accusative opposition 11 nominative vs. genitive opposition 30 nominative-accusative

dual 70 nominative-accusative language

Indo-European 125 nominative-accusative morphosyntax 124,

126

nominative-accusative structure of Indo-European 15

nominative-accusative vs. ergative/active structure 15

nominativization 18 non-actual present

generic 132 non-actual value of injunctive 113, 132,

135, 176 non-actual /ideal 132 non-actualized 140 non-additive 30

and integral character of flectional morphology 31

flectional expression of grammatical categories 49

non-agent 15, 19, 62, 120, 123-125 non-agentive actor 84, 162, 163 non-arbitrary 29 non-contact position

of adjective and noun 66 non-differentiated 7 non-durative 156 non-formalized sentence

in d-f type 80 non-formalized syntax 66 non-intentional 157 non-marked perfect stems 149 non-markedness

of inactive class 158 non-oppositional 115 non-oppositional inactive and oppositional

middle 162, 163 non-oriented 84

diathetically vague inactive verbs 121 inactive verbs 20

non-orientedness of the inactive verb 158 non-personal 84 non-personal passive

impersonal and non-personal passive 165

non-personal /static -ë suffix 138 vs. personal/dynamic 138

non-progressive 145 state 119

non-sigmatic nominative pl. 75, 94

Page 267: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

250 FROM INDO-EUROPEAN TO LATIN

non-singular 35, 70, 71 non-singular forms without -s suffix 94 non-singularity

plurality, duality 74 noun

and word autonomy 43 referentially closed 85

noun and verb classes 23 basic of IE 26

noun and verb distinction 41 noun classes 17

animate vs. inanimate 61 in African languages 64

noun classes of Bantu languages and IE genders 65

noun phrase non-hierarchically structured 64

noun reference and predication 86 number 38, 58, 69-72, 74, 76 number and case 39

flectional expression 49

O object 15, 43, 123, 124

and accusative 80 object genitive 40 Old Church Slavonic 61, 70, 93, 94,

116, 174, 188 Old High German 91 Old Indian 24, 61, 63, 74, 91-93, 117,

125, 133, 136, 139, 143, 148, 150, 152, 165, 166, 167, 169, 181, 185, 187, 189

Old Irish 70, 94, 178, 182 Old Prussian 209 one-actant verb 16 one-word or bipartite statements 7 opposition nominative - accusative

privative vs. equipollent 87 oppositional 115 oppositional middle

restricted outside Greek and Aryan 157

oppositional middle-passive and non-oppositional inactive 162

oppositions 34, 73 optative 178

optative suffix-ieH/iH- 178,183 optative vs. subjunctive

not grammaticalized opposition in CIE 178

optional 92, 113 orientation of morphological semantics

112 Osco-Umbrian 94, 137, 167

P p-f = paradigmatic-flectional paradigm 58

morphological, inflectional 31, 34 paradigmatic affinity of cases 101, 102 paradigmatic opposition

intracategorial 39 paradigmatic oppositions 9, 81 paradigmatic-flectional 13, 23

morphology 107 stage, (sub)type 3

paradigmatic-flectional structure of Latin of Latin 97

paradigmaticity 31 paradigmatization 4, 8, 12, 28, 30, 31,

37, 80, 98, 107, 109, 113 decomposition via paradigmatization

186 grammaticalization and

paradigmatization 110 in Latin noun inflection 97 in Latin present system 180 of basic cases 81 of nominative - accusative - genitive

oppositions 86 paradigmatization of derivative-flectional

morphology 192 paradigmatization process

cyclical 25 paradigmatized 155 paradigmatized oppositions 8 paradigmatized structure

of Latin present classes 34 paradigms 38 parallel developments

vs. dialectal isoglosses 177 parallelism

of Latin and Slavic imperfects 187

Page 268: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INDEX 251

of Latin present and perfect systems 152, 186

partial - total 39 participant

dependent, affected 81 in accusative 81 in nominative 81 independent and privileged role 81

participant in the situation 79 'participant role 162 participants

agent, patient, experiencer, benefactive etc. 122

vs. actants 122 participation 79

and referentiality 80 participial form 170 particle 24 particle -i 132

in Latin perfect 148 particles 91 partitivity

and possessivity 82 partitivity and plurality

vs. possessivity and singularity 82 passive 20, 21, 35, 112, 122, 126 passive and direct-reflexive functions

of Latin passive 157, 160 passive future

Greek 36 passive or direct-reflexive interpretation

of Greek middle 161 passive transformation 112 past vs. non-past

and perfective vs. imperfective 153 patient 15, 19, 62, 79, 80, 84, 122, 165

and accusative 80 patient-to-subject promotion 112, 160 perfect 11, 35, 36, 115, 117

and aorist 119, 143 and aorist in Latin 145 and inactive 143, 150, 155 and middle 116 archaic unreduplicated 117 Greek and Aryan: added to active verb

144 IE formed from inactive verb alone

144 in Greek and Latin 154 in relation to present 154 Latin -vi/-ui perfect and Greek -KCL

perfect 150 Latin and Greek/Aryan 143 Latin in -vi/-ui 150 Latin reduplicated 151 Latin sigmatic 150 Latin unmarked 152 Latin with lengthened stem 151 Latin, double origin 144 Latin: in direct opposition to present

153 non-progressive 119 reduplication 151 relevance to the present situation 154

perfect and imperfect in Latin 154

perfect and middle 108, 115, 156 perfect endings

and inactive endings 147 Latin 147

perfect in Latin active preterite and inactive

non-progressive category 156 perfect indicative 9

centre of perfect system 152 present and perfect indicative 154

perfect indicative endings 183 perfect of inactive verb

identified with aorist of active verb in Latin 144

perfect stem vs. present stem Latin 25

perfect stems Latin 149-152

perfect system Latin perfect system vs. present system

152-155, 186 perfect vs. middle

in Greek 145 perfect vs. present 149 perfect vs. thematic present active

in Greek 145 perfective 144 perfective vs. imperfective 38

Page 269: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

252 FROM INDO-EUROPEAN TO LATIN

and past vs. non-past 153 perfective/aoristic vs.

imperfective/presentic 12 perfective/non-durative 119 perfectivity 155 perfectivizing 141, 182 perfectivizing -s suffix 186 perfectivizing formation

sigmatic aorist, modal 141 perfectivizing or terminative 186 perfecto-medium 115, 116, 118 perfects

and media tantum 116 neutral with respect to diathesis 123

person 107 person affixes 16

two series of 121 person and number

expression of in verbal endings 32 person marker 60

vs. modal and tense/aspect marker 108 person markers

cooperation with case markers 121, 125

person markers and case markers 121 person-diathesis

complex morpheme 32 personal affixes

active and inactive 16, 17, 19 and case marking 21

personal endings 23, 191 active verb 131 inactive verb 135 primary, secondary 131

personal prefix of the first person in Guarani 20

personal pronouns Georgian ergative marking 18

personal subject of Greek middle-passive 165

personal/dynamic -ă suffix 138

phonological change 51 Phrygian 167, 169 pluperfect 35, 186 plural 62, 71, 73

and non-singular 73

marked suffixes 91 of animate and inanimate nouns 62 of neuters 11, 26

plural cases 90 marked suffixes 76

plural forms more voluminous 75

plural forms in -s in Hittite 95

plural marker of neuters 69

plural paradigm 92, 95 nominative and accusative 71

plurality as individualized non-singularity vs.

duality 74 marker of 71 representative marker of 32

polarization 164 polyfunctional

cases 79 morphological formants 10

polyorientation of inactive verb 123

Porno Eastern Porno 17

positional and hierarchical ordering in Greek verb ending 36

positive evidence of active morphology 129

possessive/partitive structures 81 possessivity and partitivity 82 possessivity and singularity

vs. partitivity and plurality 82 postinflectional 6, 91-93, 96 postinflectional particle -i 46, 132 postinflectional suffixes 90, 91, 95, 104 postpositions

and analytic explanation of case suffixes 77

potential 184 praesens tantum 129 predicability 44 predicate 84 predication 85

noun reference and predication 86 predication base 84-86

Page 270: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INDEX 253

predicational 60 vs. referential 84

predicational expressions 41 predicative function

of genitive 88 of nominals 44

predicative syntagm 42 prefixation 24 preposed formants

reduplication, augment 30 prepositions 45 present 9

generic, non-actual 132 Greek present formations 12

present and perfect indicatives and anteriority 154

present and perfect systems Latin, parallelism of 186

present classes Latin 34

present conjugations of Latin as example of paradigmatized flection

180 present indicative

centre of present system 152 present optative, Greek

and Latin imperfect subjunctive 184 present participle

Latin 187 present perfect 154 present system

Latin perfect system vs. present system 152, 186

present vs. aorist 12 present vs. aorist opposition

aspectual in Greek 153 present vs. aorist vs. perfect 110 present vs. perfect opposition

Latin 12 present vs. preterite 110 present-based system 115, 131 present/imperfect vs. aorist vs. perfect

191 presento-perfects 11, 117, 188 presents

athematic 12 reduplicated 12

simple thematic 12 suffixal 12

presents of inactive verbs pure thematic, long vocalic, nasal 146

preterital formations innovative 188

preterital markers 176 preterital/infinitival stems of Slavic 174,

175 preterite

-ă preterites 187 Germanic weak preterite 188 Tocharian 175

preterites allomorphic Hittite of -mi and -hi verbs

144 Baltic 175 Baltic, and Greek aorists 177 long vocalic preterites 175

preterites with durative or intransitive value 174

primacy of word structuring to sentence structuring 13

primary endings 131, 132, 134, 136, 165, 166, 176

primary form indicative present 176

primary object 40 primitive language 7 privative opposition 87 privileged (prominent) participant

marked by nominative suffix 83 privileged participant

in subject position 62 privileged role of animate participant in

the sentence 87 process category

and marked diathesis 119 and thematic present 119

process vs. state 119 process/progressive

vs. state/non-progressive 139 processes 16, 115 progressive 39, 119, 139, 144 promotion

patient-to-subject 162 patient-to-subject promotion 160

Page 271: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

254 FROM INDO-EUROPEAN TO LATIN

pronominal declension (subsystem) 94 pronominal subsystem 53, 93 pronouns

and analytic explanation of case suffixes 77

property 85 proportional 73 prospective or desiderative meaning 186 proterodynamic 55

aerostatic or proterodynamic accentuation 139

Proto-Indo-European (PIE) 3, 4, 13, 77, 81, 95, 97, 110, 112, 113, 156, 167, 169, 171, 191, 196

prototypical domain of basic cases 78

prototypical subject 86 pseudo-incorporating elements

incorporation-like 46 psychosensory states/processes

expressed in inactive verbs 130

Q quasi-derivational distribution of IE

formants 110 quotation form

casus indefini tus 90

R real 184 reconstruction 3-5

internal vs. external 5 of older stage of flection 13 traditional 3

reconstruction of derivative-flectional type 10

reconstruction of IE active and inactive verbs 127

reconstruction of the IE active and inactive verbs 127

reduced grade 78 with vocalized sonants or laryngeals

128 reduplicated aorist 36, 129 reduplicated aorists

of transitive/causative type 151 reduplication 24, 37, 140, 156

grammaticalized in OI and Gr. perfect 151

in Latin perfect 149 in presents 133, 149

referential 60, 79 vs. predicational 84

referential expressions 41 referential intention 85 referential role of subject 41, 125 referentially closed 85 reflexive particle as a marker of

intransitivity 123 relational frame of verb 84 relational semantics 112

vs. category-immanent 37 relative adverbs 45 relative and absolute tenses in Latin 153 religious thought 163 repetition of the same marker 65 representative marker of plurality 32, 73,

74, 76, 92 representative person markers 167 residual form of instrumental 78 resultative 117, 156

perfect 147 Rgveda 132 rheme 86 rhotacism 148 Romance 68 root-perfects 117 Rumanian 67 Russian 167

S Sandhi 70, 75 Sanscrit 163 schwebeablaut 139 secondary endings 36, 131, 132, 134,

136, 137 secondary thematic endings 183 semantic and formal markedness

of middle-passive 160 semantic categories of agent, patient

and cases 79 semantic category

changed to inflectional category 68 semantic continuum of non-marked and

Page 272: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INDEX 255

marked non-agentivity 162 semantic interpretation of the flectional

structure 9 semantic orientation of the morphological

distinctions 163 semantic roles

agent vs. non-agent 158 agent, experiencer, patient 41 nominal 162 of subject 122 of subject of Greek middle 163 relational vs. referential 79

semantic version 160, 161, 163 semantic-syntactic and morphological

categories 111 semantic-syntactic categories and relations

11 semantic-syntactic relevance

of morphological processes and categories 4

of word categories 61 semantic/category-inherent orientation of

Greek morphology 42 40

semantic/prototypical subject 86

semantic/referential function of agreement 64, 66

semantics and distribution of grammatical

formants 10, 11 of category inherent type 27 of flectional structure 3 of nominal categories 57 relational, referential 79 word-category-inherent of case 58

semi-adverbal locative, instrumental in Greek 96

semi-grammatical (derivational) formations

as source of tenses/aspects and moods 113

semi-inflectional 36 sentence

non-formalized IE 13 with more lexemes 7

sentence formalization 40

in European languages 42 sentence frame 44 sentence structure

appositional 42 formalization of 6 free, non-compact of IE 18 non-formalized IE 42

sentence structure vs. word structure 4 sentence structuring 7

word structuring 13 separability of morphemes 33

relative and gradual in IE 31 separative 76 set bases 150 sexual opposition 64 shape

of active and inactive roots 127 shape of root and stem

of noun and verb classes 27 shortening

of lengthened nominatives 90 sigmatic aorist 36, 182

and sigmatic forms with modal and future meaning 181

Greek, transitive/causative 141 of transitive active verbs 141

sigmatic aorist/ preterite as allomorphic Latin perfect 181 aspectual or temporal 181

sigmatic aorist/preterite formed from active verb 181

sigmatic formation Latin subsystemic 183

sigmatic formations with modal and future meaning 181

sigmatic forms 182 Latin 181-183, 185

sigmatic futures 181 sigmatic perfect

Latin 141, 150 Latin, corresponding to sigmatic aorist

144 sigmatic perfect in Latin

corresponding to sigmatic aorist 144 singular and plural

nominative 73 situation 39, 79

Page 273: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

256 FROM INDO-EUROPEAN TO LATIN

Slavic 70, 112, 187 preterital/infinitival stems 174, 175

sonant or laryngeal as second radical element

of inactive roots/stems 127 spatiotemporal differentiation of IE 4 split

aspecto-temporal 18 ergativity/activity split 18

split ergativity 18 state achieved

identified with action or process completed 145

state achieved vs. process in progress 145

state category perfect 119

state of affairs 41, 79, 85 state/non-progressive vs.

process/progressive 139, 145 statements

bipartite 7 containing more than one lexeme 7 one-word 7

states 16, 115 states and processes

inactive verbs 118 static

vs. dynamic 177 stative 111, 115, 117

and inactive 158 stative category

with vocalic ending -o(i) in 3sg. 158 stem 10

long vocalic in -e 187 stem alternation 37, 156

as nominative marker 83 in active and inactive verb 139

stem alternations 30 in verb 139 system of 27

stem markers, modal and preterital markers 176

stem modification 24 stem vowel 51 strong accent

accentuation 88

strong cases 78 structural descriptions

of case paradigm 79 subject

and categorical statement 83 and nominative 80 as privileged participant 62 constitution of 62 expressed by the nominative and

subject-person marker 43 grammaticalized 86 prototypical 86 semantic/prototypical and

syntactic/generalized 86 subject - predicate relation

constitution of 83 grammaticalized 85

subject constitution and agreement 83 and nominative 83

subject inversion 85 subject marking

by nominative and agreement 126 subject nominative

and subject person 84 subject person 40

and subject nominative 84 subject = agent 159 subject — predicate relation

grammaticalized 85 subject-person marker 43 subjects of nominal predicates 19 subjunctive 35

OIr -ă subjunctive 182 archaic Latin 25 Greek thematic 179 Greek, of present vs. aorist 155 Latin, of present vs. perfect 155 OIr -s subjunctive 182

subjunctive imperfect Latin 178

subjunctive marker -ă in Latin 28

subjunctive pluperfect 186 subjunctive/optative in-T- 185 subjunctives

-ă- subjunctives 178

Page 274: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INDEX 257

-s subjunctives 182 subjunctives and futures with -ă and -ë

suffixes Latin 177

subjunctives of the imperfect and pluperfect

innovative Latin 155 subjunctives/futures

Latin and Greek 173 subordinating conjunctions 45 substitution 10 subsystemic sigmatic formation

Latin 183, 185 suffix 11

-dh suffix 188 -ă suffix with preterital value 187 -ă- in Tocharian 180 -s suffix 183 aoristic and modal -s- suffix 181

suffixal derivations as source of inflectional categories 108

suffixation 10, 24, 140 IE cyclical 183 in active and inactive verb 139 in present stems 133 non-additive character of 24

suffixes -ă- and -ë- suffixes 177 -dh, -d, -k suffixes 183

suppletive aorist 129 Swahili 65, 66, 68 syncopated 88 syncretism

alleged in Greek case system 96 syntactic categories

and cases 79 syntactic categories of subject, object 79 syntactic devices

of expressing grammatical semantics 10 syntactic diathesis and semantic version

160 syntactic diathesis in Latin

vs. semantic version in Greek 160 syntactic function of agreement 64 syntactic passivization 160 syntactic/generalized

subject 86

syntactic/relational orientation of Latin morphology 40

T temporal

vs. aspectual 155 temporality 112 temporality as semantic-syntactic category

vs. morphological categories of aspect and tense 112

tempus primacy of in Latin 40

tense as a 'deictic' category 112 tense-mood

complex morpheme 32 tense/aspect 107

expressed by inflectional stem 33 tenses/aspects and moods

their origin in the derivational formations 113

terminative 188 perfectivizing or terminative 186

terminus 145 thematic -e/-o suffix

alternating 173 and long vocalic -ă, -ë suffixes 173 originally active 140

thematic -ë/ö- subjunctive 179 thematic and long vocalic formations

functional parallelism of 173 thematic aorist 36, 131, 133, 147 thematic forms 132, 133 thematic imperfects and subjunctives

in Greek 179 thematic present 113, 119, 120, 145, 166

central position in verbal system 134, 145

contamination of the active and inactive verb endings 135

of inactive verbs 145 reconstruction 134

thematic suffix 133 thematic vowel 133

and ablaut 56 e-grade in vocative 90 incapable of FG - RG alternation 133 vocalic suffix with vague meaning 133

Page 275: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

258 FROM INDO-EUROPEAN TO LATIN

thematization 185 theme - rheme 86 thetic statement 84, 85, 126

event central 123 nominal, entity central 85 verbal, event central 85 with inactive verbs 158

thetic statements 126 Tocharian 35, 70, 136, 148, 163, 167,

168, 169, 175, 178 -ă presents, -ă preterites, -ă

subjunctives 180 compared with Latin 180 preterite 175

topic - focus 86 total vs. partial 40, 79 traditional

approach to IE verb 144 traditional reconstruction 3, 110

of verbal system 143 traditional views about verbal systems

164 transitive 122

actor-oriented 130 transitive active verbs 15

in ergative and active languages 16 transitive actor-oriented and intransitive

undergoer-oriented 130 transitive and intransitive

inactive verb 121 transitive Baltic -ë preterites

vs. intransitive Greek -ē aorists 177 transitive inactive 164 transitive/active vs. intransitive/inactive

188 -s suffix vs. -dh suffix 188

transitive/causative meaning of sigmatic aorist 36

transitivity vs. intransitivity 121 tripartite

vs. bipartite aspecto-temporal system 110

tripartite aspecto-temporal system present vs. aorist vs. perfect 110

two series of personal affixes 121 two-actant verb 16, 20 two-class system 107, 115, 122, 143

injunctive based two-class system 191 its transformation into integral verbal

system 118, 119 two-class system of active vs. inactive

verb classes 109 type of grammatical semantics 10, 13 type of morphological semantics 12, 37 typological approach 4 typological characterization 13 typological evolution

of IE from d-f to p-f stages 7 typological naturalness 29, 46 typological relevance of archaisms and

innovations 37

U undergoer 20, 84, 122-124, 126 underlying principle

of IE flectional type 26 unipartite and bipartite structure 85 unipartite structure of statement 85 unipersonal IE verb 20, 84 unipersonal verb

in Guarani 20 unmarked perfects

Latin 152

V vague 70, 133

Aktionsart-semantics 140 character of active vs. inactive

classification 127 meaning of case suffixes 81 meaning of flectional formants 10 meaning of vocalic suffixes 176

vague with respect to number 35, 74, 76, 93

vagueness 81 variety

in morphological and inflectional paradigms 36

of derivative-flectional formations 108 Vedic 131, 136 Vedic Sanskrit 150 Venetic 137, 167, 169 verb

centralizing function 18

Page 276: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)

INDEX 259

derivative-flectional morphology 107 IE, unipersonal 84 Latin and Greek/Aryan 191 paradigmatization 107 place of in the IE sentence 43

verb categories 107 /erb classes

active vs. inactive 20, 21, 59, 108, 115-141

and case marking 18 basic noun and verb classes of IE 26 expressing diathesis 113

verb morphology traditional vs. modern interpretation

110 verbal activity 19 verbal adjective 170 verbal ending

Greek 35 Latin 35

verbal nouns neutral with respect to diathesis 123

verbal system maximized in Greek/Aryan 178

verbal valency as disambiguating device 101

verbal voice < 0 verbal voice (verbal diathesis) as a

morphological category vs. diathesis as a semantic-syntactic

category, 112 verbo-nominal relations 39 version

semantic vs. syntactic diathesis 160 vocalic alternation 24

German 25 Latin 25

vocalic case suffix instrumental 90

vocalic ending -o 117 vocalic endings 189 vocalic marker

modal 34 vocalic suffixes

modal, future 109 thematic -e/-o, long vocalic 191

vocalic-laryngeal endings

inactive, perfecto-medial 116 perfecto-medium 115

vocalism final, signalling diathesis 167

vocalized form of sonants and laryngeals 128

vocative 78, 90

W weak cases 78 weak position in the paradigm 91 weak preterite

Germanic 188 word

as a complex and hierarchically structured unit 80

IE, as semantic complex 38 IE, morphemic structure 30 relatively autonomous and

hierarchically structured 13 word autonomy 6, 42, 57, 77

and agreement 66 in d-ftype 80

word categories 39, 41 word inflection 19, 42, 46, 91

vs. group inflection 77 word inherent expression of grammatical

semantics 43 word inherent semantics 26 word integration

IE tendency to 27 word modification

vs. morpheme addition 8 word modifications

primary significative value of 8 word order 42, 44

free, and word autonomy 44 in attributive genitive syntagm 88 non-grammaticalized 44

word ordering 7 word structure vs. sentence structure 4 word structuring

vs. sentence structuring 13 word-category-inherent (immanent) 39 word-category-inherent semantics 26 word-internal morphological modifications

80

Page 277: [Helena_Kurzová]_From_Indo-European_to_Latin_The(BookZZ.org)