The Tocharian Gender System. A Diachronic Study in Nominal Morphology 9789004532892, 9789004532885

Citation preview

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

The Tocharian Gender System

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Brill’s Studies in Indo-European Languages & Linguistics Series Editors Olav Hackstein, Ludwig-Maximilians-University of Munich, Munich, Germany Michael Weiss, Cornell University, Ithaca, NY, USA Founding Editor Craig Melchert, UCLA, Los Angeles, California, USA Editorial Board Paola Cotticelli, University of Verona, Verona, Italy José Luis García Ramón, University of Cologne, Cologne, Germany David Goldstein, UCLA, Los Angeles, California, USA Stephanie Jamison, UCLA, Los Angeles, California, USA Ronald Kim, Adam Mickiewicz University in Poznań, Poznań, Poland Alexander Lubotsky, Leiden University, Leiden, the Netherlands Melanie Malzahn, University of Vienna, Vienna, Austria Alan J. Nussbaum, Cornell University, Ithaca, New York, USA Georges-Jean Pinault, EPHE - Écoles Pratiques des Hautes Études, Paris, France Jeremy Rau, Harvard University, Cambridge, Massachusetts, USA Elisabeth Rieken, Philipps University of Marburg, Marburg, Germany Stefan Schumacher, University of Vienna, Vienna, Austria

Volume 25

The titles published in this series are listed at brill.com/bsiel

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

The Tocharian Gender System A Diachronic Study in Nominal Morphology By

Alessandro Del Tomba

LEIDEN | BOSTON

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Cover illustration: The recto of THT 496, of which lines 3–4 are cited as B496a3–4 on p. 89. The fragment is part of the “Depositum der Berlin-Branderburgischen Akademie der Wissenschaften in der Staatsbibliotek zu Berlin — Preußischer Kulturbesitz — Orientabteilung”. The image is reproduced by courtesy of the Berlin-Brandenburgische Akademie der Wissenschaften. The Library of Congress Cataloging-in-Publication Data is available online at https://catalog.loc.gov LC record available at https://lccn.loc.gov/2023001994

Typeface for the Latin, Greek, and Cyrillic scripts: “Brill”. See and download: brill.com/brill-typeface. issn 1875-6328 isbn 978-90-04-53288-5 (hardback) isbn 978-90-04-53289-2 (e-book) Copyright 2023 by Alessandro Del Tomba. Published by Koninklijke Brill NV, Leiden, The Netherlands. Koninklijke Brill NV incorporates the imprints Brill, Brill Nijhoff, Brill Hotei, Brill Schöningh, Brill Fink, Brill mentis, Vandenhoeck & Ruprecht, Böhlau, V&R unipress and Wageningen Academic. Koninklijke Brill NV reserves the right to protect this publication against unauthorized use. Requests for re-use and/or translations must be addressed to Koninklijke Brill NV via brill.com or copyright.com. This book is printed on acid-free paper and produced in a sustainable manner.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

In lasting memory of a kind, strong man. Lost but never forgotten

∵

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Contents Preface and Acknowledgments xi Abbreviations and Symbols xiii List of Tables xviii 1 Introduction 1 1.1 Tocharian 1 1.2 Tocharian Grammatical Gender 4 1.3 Aim 6 1.4 Structure and Outline of the Book 8 2 The Gender System of Tocharian A Synchronic and Typological Overview 9 2.1 Grammatical Gender: Methodological and Theoretical Background 9 2.2 The Gender System of Tocharian 19 2.2.1 The Status of the Tocharian Genus Alternans 22 2.2.2 On the Terminology of the Third Gender and the Status of the Tocharian “Neuter” 31 2.3 The Gender Assignment System of Tocharian 36 2.3.1 Inflectional Patterns 37 2.3.2 Derivational Patterns 45 2.3.3 Semantic Patterns 50 2.3.4 Summary of the Gender Assignment Patterns 56 2.4 The Gender System of Proto-Indo-European: State of the Art and Open Questions 58 2.5 Summary and Conclusions 64 3 Gender in the Inflection of the Noun 66 3.1 Tocharian Nominal Categories 66 3.2 Tocharian Noun Declension 68 3.3 Aim and Structure of the Chapter 73 3.4 Nouns Denoting Female Entities 76 3.4.1 The śana-type 77 3.4.2 The aśiya-type 91 3.4.3 Conclusion 98

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

viii

Contents

3.5 Grammatical Gender and the Development of Inflectional Classes from class vi (nom.pl. -ñ) 100 3.5.1 The kantwo-type 101 3.5.2 The okso-type and the arṣāklo-type 145 3.5.3 The wertsiya-type 172 3.6 The Evolution of the PIE Neuter in the Noun Inflection of Tocharian 185 3.6.1 The Evolution of the Thematic Neuter Singular 185 3.6.2 The Evolution of the Thematic Neuter Plural 196 3.6.3 Alternating Gender and the Origin of the Plural Endings TB -na and TA -äṃ 209 3.6.4 On Some Tocharian B Pluralia Tantum 238 3.7 Summary and Conclusions 243 3.7.1 Evolution of the Feminine in the Noun Inflection 244 3.7.2 Evolution of the Neuter in the Noun Inflection 245 4 Gender in the Inflection of the Pronoun and Adjective 246 4.1 General Aim, Methodology and Structure of the Chapter 246 4.2 Overview of the Tocharian Pronominal System 249 4.3 Evolution of the Tocharian Demonstrative Pronouns 251 4.3.1 Introduction to the Tocharian Demonstratives 252 4.3.2 Paradigm of the Singular 253 4.3.3 Paradigm of the Plural 258 4.3.4 Origin of the Feminine Plural 264 4.3.5 Evolution of the Tocharian Demonstratives 266 4.4 Evolution of the Pronominal Adjective TB allek, TA ālak ‘other’ 270 4.5 Conclusion 276 4.6 Overview of the Tocharian Adjectival System 276 4.6.1 Metrical Constrains and Adjectival Agreement 279 4.7 Reconstruction of the Proto-Tocharian Adjectival Paradigms 283 4.7.1 The Thematic Type (class i) 283 4.7.2 The Athematic Type (classes ii, iii, iv) 307 4.7.3 The Proto-Tocharian Adjectival System: A Summary 314 4.8 Evolution of the Gender System: From Proto-Indo-European to Tocharian 316 4.8.1 Evolution of the Masculine Paradigm and the Neuter Singular 316 4.8.2 Evolution of the Feminine Paradigm and the Neuter Plural 322

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

ix

Contents

4.8.3

Summary of the Evolution of the Gender System in the Adjectives 345

5 Retrospective and Conclusion 347 5.1 Synchronic Analysis 347 5.2 Diachrony of the Masculine 348 5.3 Diachrony of the Feminine 350 5.4 Diachrony of the Genus Alternans 352 5.5 Outlook 354 Bibliography 357 Index Verborum 408

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Preface and Acknowledgments This book represents a thoroughly revised and partially restructured version of my doctoral dissertation, which I defended in March 2020 at Leiden University. The thesis was written under the supervision of Michaël Peyrot, Alexander Lubotsky, and Giorgio Banti, as the outcome of a three-year jointly supervised doctorate at Leiden University and Sapienza University of Rome. The work was accepted for publication by Brill two years later. I thank Olav Hackstein and Craig Melchert for allowing me to publish in their series as well as Elisa Perotti for her collaboration and constant support throughout the publication process. This book could not have been written without the support of many professors, teachers, and friends, whom I was lucky enough to meet during my studies. The words below are intended as a sincere token of gratitude to all of them. First of all, I am grateful to Michaël Peyrot, who has invested a lot of time and energy in my personal growth and training. I am indebted to him for his teaching and his constant support, and not a day goes by that I do not miss our discussions. Sasha Lubotsky has always been there to help with any questions about Tocharian historical linguistics and Indo-European reconstruction, duly providing food for thought and illuminating suggestions. I hope their teachings have borne the fruit they deserve. I thank Claudia A. Ciancaglini, who introduced me to linguistics, and Mauro Maggi, who taught me how to think about texts and languages. Under their guidance, I grew up first as a student and then as a scholar, and they have always been there for help and advice over the years. From them and other teachers from Sapienza, among whom Paolo Di Giovine and Marco Mancini need to be mentioned, I have learned a lot about historical linguistics and philology. In 2020 I won a postdoctoral position in Sapienza tutored by Marco Mancini. I thank him once again for enabling me to work on turning my thesis into a book. I owe a special debt of gratitude to Melanie Malzahn, who was a member of my Promotiecommissie and gave important feedback and constructive criticism to the work. Also, I am indebted to Ronald I. Kim, who has always provided detailed, objective, and constructive advice, without ever imposing his own opinions. His suggestions have allowed me to improve many sections of this book. Over the past years, I have greatly benefited from thriving discussions with other scholars, who commented on this work since it was a thesis and discussed with me several points of typological linguistics, Indo-European reconstruction,

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

xii

Preface and Acknowledgments

and Tocharian philology and linguistics. I thank Douglas Q. Adams, Giorgio Banti, Vincenzo Faraoni, Olav Hackstein, Artemij Keidan, Alwin Kloekhorst, Leonild Kulikov, Anna Pompei, Tijmen Pronk, and the anonymous reviewer of the manuscript for their comments. Special thanks go to Luca Alfieri, who read a first draft of this book and whose keen comments have produced several improvements. I would also like to thank my colleagues and friends, who were always willing to share their expertise, successes, and failures, in particular Chams Bernard, Corinne D’Antonio, Federico Dragoni, Marta Frenguelli, and Tao Pan. Working on this book would not have been the same had it not been for my family, who have always kept me driving on. I thank in particular my mother Bruna and my grandfather Sergio, who both invested a lot of energy in my education. My final heartfelt thanks go to Matteo for his undying support and care along this journey together. Alessandro Del Tomba Pannone (Trento), November 2022

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Abbreviations and Symbols

Grammatical Abbreviations

abl. ablative acc. accusative act. active adj. adjective all. allative alt., a. alternating adv. adverb arch. archaic, cf. Peyrot (2008) class. classical, cf. Peyrot (2008) coll. colloquial, cf. Peyrot (2008) com. comitative Cx controller agreement dat. dative dem. demonstrative det. determiner du. dual fem., f. feminine gen. genitive ger. gerundive inf. infinitive instr. instrumental ipf. imperfect loc. locative masc., m. masculine mid. middle neg. negation nom. nominative NP nominal phrase nt. neuter obl. oblique opt. optative part. particle perl. perlative pl. plural

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

xiv

Abbreviations and Symbols

prs. present prt. preterite ptc. participle sbj. subjunctive sg. singular Tx target agreement voc. vocative

Languages Alb. Albanian Arm. Armenian Att. Attic (Greek) Av. Avestan Bctr. Bactrian BMAC Bactria-Margiana Archaeological Complex Celt. Celtic Chin. Chinese Croat. Croatian (Standard) Du. Dutch E English Fr. French Germ. Germanic Gk. Greek Goth. Gothic Hom. Homeric IE Indo-European IIr. Indo-Iranian Ir. Iranian It. Italian Khot. Khotanese Khwar. Khwarezmian Lat. Latin Latv. Latvian Lesb. Lesbian Lith. Lithuanian Luw. Luwian

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

xv

Abbreviations and Symbols M Middle Mo Modern MP Middle Persian NP New Persian O Old OAv. Old Avestan OCS Old Church Slavonic OD Old Dutch OE Old English OHG Old High German OLG Old Low German ON Old Norse OP Old Persian Osc. Oscan OSIr. Old Steppe Iranian, cf. Bernard (forth.) Oss. Ossetic P Proto(-) Phr. Phrigian PIE Proto-Indo-European Pkt. Prākrit Port. Portuguese pre-PT pre-Proto-Tocharian pre-TA pre-Tocharian A pre-TB pre-Tocharian B Pruss. Prussian PT Proto-Tocharian Rom. Romanian Russ. Russian Skt. Sanskrit Sl. Slavic Sogd. Sogdian Sp. Spanish TA Tocharian A TB Tocharian B Tum. Tumshuqese Uy. Uyghur Ved. Vedic Sanskrit W Welsh YAv. Young Avestan

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

xvi

Abbreviations and Symbols

Symbols /x/ phonological form [x] phonetic form; uncertain reading in Tocharian texts; addition in translation (x) restoration in Tocharian texts |x| morphological form (cf. Peyrot 2013a) ⟨x⟩ orthographic form {x} emendation in Tocharian texts « x » editorial supplement in Tocharian texts *x reconstructed form x* inferred form **x projected, but wrong form x | x nom. form | obl. form > developed phonologically into < developed phonologically from >> developed analogically into *-ā(-) are addressed, since it still represents a debated issue of Tocharian historical phonology. As to how the neuter gender evolved in Tocharian nouns, I investigate the outcome of selected classes of nouns, continuing both athematic and thematic paradigms. The ultimate goal is to understand how the PIE neuter lost its functions as a category of target gender and how it evolved into the Tocharian genus alternans. In order to achieve these aims, I also investigate the morphophonological mergers between case forms and gender markers that must have affected (pre-)Proto-Tocharian in an unattested stage. These mergers were caused by a general consonant apocope that led to confusions between the inherited case endings. In addition, I consider typological comparisons in the evolution of languages which synchronically show a gender system similar to that of Tocharian. In particular, in the evolution from Latin to the Romance languages, the neuter gender underwent a gradual depletion. In a transitional stage, however, the neuter displayed an alternating agreement pattern, which is fully parallel to the Tocharian genus alternans. The same system was attested in some Old Italian dialects and survives today in e.g. Romanian. The application of this comparison in diachronic typology may be of benefit to our analysis of the Tocharian gender system. Indeed, the developments that have taken place between Latin and its Romance descendants may provide insight into the developments between the gender system in (pre-)Proto-Tocharian as compared to the system attested by Tocharian A and B. In this study, I try to keep to the generally accepted principles of comparative linguistic reconstruction. Furthermore, in my understanding of the historical

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

8

Chapter 1

development of Tocharian, I always bear in mind that Tocharian underwent a long period of both regular and irregular linguistic developments before it reached the stage in which its gender system is attested; in this unattested period, Tocharian underwent a great deal of phonological decay that eroded the morphological markers. As gender is organised in form of systems, its evolution cannot be captured in terms of individual diachronic developments; rather, when gender systems evolve there is always co-evolution of connected subsystems (Wälchli & Di Garbo 2019: 201–202). Overall, this study does not aim to reconstruct the specific lexical and typological trajectory of inherent gender in Tocharian nouns. It rather aims to investigate the derivational and inflectional morphology of Tocharian gender markers in both the controllers and the targets from the point of view of Indo-European comparative reconstruction. 1.4

Structure and Outline of the Book

The book is organised as follows. Chapter 2 discusses the typology of grammatical gender and gives an overview of the gender system of Tocharian. In addition, it deals with the analysis of the Tocharian genus alternans from the perspective of synchronic typology and offers an introduction to the gender assignment system of Tocharian. Chapter 3 addresses issues related to the inflection of nouns, clarifying the origin and evolution for each of the selected inflectional classes. Chapter 4 provides an overview of the pronominal and adjectival system of Tocharian and treats the development of the three inherited genders across the inflection of both pronouns and adjectives. Chapter 5 recapitulates the findings and provides concluding remarks. The main body of the book consists of the second, the third, and the fourth chapters. As each of these chapters needs independent clarifications on the aim, the structure, and the methodology, such matters are discussed in the introduction of the relevant chapter. Since the category of gender requires the investigation of a wide range of inflectional and derivational patterns as well as its intersection with the other nominal categories of case and number, this book may be seen as a contribution to the study of Tocharian nominal morphology as a whole.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Chapter 2

The Gender System of Tocharian A Synchronic and Typological Overview

To pave the way for the investigation of the category of gender in Tocharian, I present in this chapter a general overview of the terms and concepts that are crucial to the analysis of grammatical gender, from both a synchronic and a diachronic perspective. In general linguistics, the literature on this topic is quite inconsistent, especially with regard to the terminology used. Therefore, an introduction to central notions like noun class, agreement, agreement controller, and agreement target is required, as well as some theoretical and methodological remarks relevant to the investigation of gender from the perspective of Canonical Gender Typology (§ 2.1). Particular attention is paid to the problematic status of the third Tocharian gender, the so-called genus alternans (§ 2.2). I propose typological arguments and cross-linguistic comparisons to demonstrate that the genus alternans is a separate gender in Tocharian (§ 2.2.1, § 2.2.2). This is followed by a synopsis of the principles of gender assignment in Tocharian, from both a formal and a semantic point of view (§  2.3). The chapter ends with a discussion of the reconstruction of the gender system in Proto-Indo-European, including the question whether the feminine gender was a recent innovation (§ 2.4). 2.1

Grammatical Gender: Methodological and Theoretical Background

In linguistics, gender (from Lat. genus ‘origin, kind, species’, via Old French gendre) represents a grammatical category that has attracted a great deal of study — both synchronic and diachronic — in languages across the world. In western linguistic scholarship, gender became a matter of special interest in the fifth century BCE, when the Greek philosopher Protagoras (c.480– c.410 BCE) recognised three genders in Ancient Greek (τὰ γένη τῶν ὀνοµάτων), classifying and dividing the nouns in ἄρρενα ‘masculine’, θήλεα ‘feminine’, and σκεύη ‘inanimate, pertaining to things’. The names Protagoras gave to these category values clearly suggest that he associated the notion of gender with a supposed inherently gendered quality of the referent (cf. Plato’s ὀρθότης τῶν

© Alessandro Del Tomba, 2023 | doi:10.1163/9789004532892_003 Alessandro

Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

10

Chapter 2

ὀνοµάτων “correctness of words”).1 The analysis of Protagoras is reported in the Rhetorics iii 5, 1407b7–8 of Aristotle (c.384–322 BCE), who, in view of the lack of a sex correlation for the σκεύη gender, claimed that it should be defined as τὸ μεταξύ ‘that which stays in the middle’ (Poetics 1458a).2 The term οὐδέτερον ‘not either, neuter’ appears in later grammatical traditions (Stoycs and Dionysius Thrax).3 Therefore, early Greek scholars already recognised that there is often no straightforward correspondence between natural and grammatical gender. That is to say, the sex of the referent must be distinguished from linguistic gender. Strictly speaking, gender refers to a grammatical category, i.e. grammatical gender, which fulfils two basic functions: classifying nominals and referring to constituents through agreement patterns. According to a famous definition by Hockett (1958: 231), gender is reserved for (1) “classes of nouns reflected in the behaviour of associated words”. An important analytical tool in understanding “the behaviour of associated words” is the distinction between gender and agreement class, where the latter is defined as (2) a set of nouns whose members take the same agreement form under the same conditions.4

1 Cf. the ambiguous statement by Protagoras about the gender of such feminine nouns as μῆνις ‘wrath’ and πήληξ ‘helmet’, which he attempts to interpret as masculine (reported by Aristotle Soph El. 173b17–174a4). It is clear that Protagoras is referring to the language of Homer. Similarly, Aristophanes makes a parody in Neph. 6586–93 in which he alters the ending of nouns so that they conform to the natural gender of their referents. See Wackernagel (2009: 399–408). 2 As Belardi (1985: 82–83) clarified, Aristotle believed that, in Greek, the stem of masculine nouns had to end with an ἡμίφωνον (i.e. Ν, Ρ, Σ [and Ψ, Ξ]), the stem of feminine nouns with a φωνῆεν μακρόν or a δίχρονον (“two-timed”, i.e. long vowels, except for Ι and Υ), while the stem of the third class of nouns, i.e. the neuter, could end either with an ἡμίφωνον or a δίχρονον (thus, τὸ μεταξύ ‘intermediate’). As a consequence, what Aristotle did was to transpose Protagoras’ distinction between ἄρρενα, θήλεα, and σκεύη “dal piano delle caratteristiche del denotato al piano delle caratteristiche del segno linguistico […]” (p. 83). 3 For a synthetic account of Greek and Latin linguistic terminology on grammatical gender, see Kilarski (2013: 59–82) and references therein. 4 Cf. Zaliznjak (1967: 30) and Corbett (1991: 105). Corbett (1991: 147–148, 2007: 243) refines the definition of agreement class as “a set of nouns such that any two members of that set have the property that whenever (i) they stand in the same morphosyntactic form and (ii) they occur in the same agreement domain and (iii) they have the same lexical item as agreement target then their targets have the same morphological realization”.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

The Gender System of Tocharian

11

The most significant pattern according to which gender is identified is consequently agreement, which commonly refers to (3) “some systematic covariance between a semantic or formal property of one element and a formal property of another”. (Steele 1978: 610) This relation is very often made by means of specific markers on one or all the elements that are linked together morphosyntactically. It follows that agreement provides the most reliable basis for defining gender and establishing the number of genders that a given language has.5 Nouns belong to the same agreement class if they take the same agreement form under the same conditions; if a given language has nouns that belong to different agreement classes, this language usually has more than one gender. In the literature, we sometimes find noun class as a blanket term for gender (Aikhenvald 2000: 18–20). Properly, a noun class is a specific group of substantives that have some characteristics in common, either semantic (e.g. the meaning and the features of the referent) or formal (e.g. phonological and/or morphological). However, this nomenclature is mostly found in studies on non-Indo-European languages: languages with noun classes have more than three “genders”, sometimes without a distinction between masculine and feminine.6 As a consequence, the difference between gender and noun class largely correlates with grammatical tradition rather than linguistic data. As mentioned above, languages use grammatical categories to group together words or morphological forms that share semantic and/or formal features. Morphosyntactically, agreement allows for the overt marking of the relationship between an adjective and the noun to which it refers. It provides 5 See Corbett (1991: 105, 2000: 348), Aikhenvald (2000: 28–50), Audring (2016), Corbett & Fedden (2016: 505–506, 2018: 12–13). In the literature, the terms “concord” and “agreement” are often use in contradictory ways (Corbett 2006: 5–7): some linguists use agreement as the superordinate term, while some others favour the use of agreement for the verbal domain and concord for the nominal domain (Baker 2008: 6–7). Since there is no evident advantage in using such a distinction for Indo-European studies, I will consistently use the term “agreement” to refer to both nominal and verbal domains. 6 See Corbett (1991: 146, 2007: 241), who argues that there is no real difference between “gender” and “noun class”: the former is preferred in Indo-European and Dravidian studies; the latter in Caucasian, African, and Australian studies (Aikhenvald 2000: 19–20). Cf. also Kilarski (2013: 8): “[T]he term ‘gender’ is usually reserved for the relatively small, sex-based system of the Indo-European type”. An in-depth and clear discussion on the differences between noun class, agreement class, and inflectional class has been offered by Babou & Loporcaro (2016) in a paper dealing with the noun classes of Wolof, a Niger-Congo language.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

12

Chapter 2

a morphosyntactic concord of various elements (the agreement target) with their head, usually a noun (the agreement controller). Typical agreeing elements are articles, adjectives, verbs, and pronouns (Audring 2016). Thus, a language has a gender system if noun phrases have an agreement target that shows gender marking (Corbett 2006: 4–5). Take for instance the following example from Latin (Virgil, Aen. 2.360): nox ātra cavā circumvolat night(f).nom.sg dark.nom.sg.f hollow.abl.sg.f fly.around.3sg.prs.act umbrā shadow(f).abl.sg “The dark night envelopes by (its) hollow shadow”. domains type controller target features values

nox ātra nominal nox ātra number, gender, case sg., fem., nom.

cavā umbrā nominal umbrā cavā number, gender, case sg., fem., abl.

nox circumvolat verbal nox circumvolat person, number third, sg.

In this example there are three domains of two types: the first type is represented by the noun nox ‘night’ and its modifier āter ‘dark, black’, and by the noun umbra ‘shadow’ and its modifier cavus ‘hollow, deep’ (nominal domain); the second type is between the subject of the sentence, i.e. nox, and its predicate circumvolō (verbal domain). The noun nox is the agreement controller in the first nominal domain and in the verbal domain, while the adjective ātra and the verb circumvolat are its agreement targets. The features expressed are case (nominative, ablative), gender (feminine), number (singular), and person (third). This implies that members belonging to the same agreement class must have the same morphosyntactic specifications where they occur in the same agreement domain with the same lexical item as their agreement target (Corbett 2012: 80). The canonical conditions that govern the agreement method may be expounded as follows (after Audring 2014: 8): (4) 1. Controller and (its) targets distinguish the same gender values. 2. All targets distinguish the same gender values. 3. Controllers are consistent in the gender value they trigger on a particular target. 4. Given the same controller, all targets show the same gender in all circumstances. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

The Gender System of Tocharian

13

As noted by Audring (2009: 15–16), the controller and the target stand in asymmetric relation to each other. This asymmetry has both a formal and a semantic component. On the semantic side, the indexation pattern in the agreement marking is only pertinent to the controller. On the formal side, the target depends directly on the controller, implying that changing the controller may result in modifications on the target, but not vice versa. In the Indo-European domain, the category of gender generally has a different inflectional status with respect to certain other nominal categories, such as number. Indeed, if a given language expresses both the category of gender and the category of number, a noun usually has a set of inflected forms that depends on the number values that this language has. It follows that number pertains to inherent inflection and is an inherent property of each word form (Thornton 2005: 105).7 On the other hand, nouns typically cannot have different inflected forms according to gender, given that it is inherently stored in each substantive (Fedden & Corbett 2017: 3). Gender is an inherent property of the nominal lexeme. Taking the nominative of Lat. umbra as an example, we have two different inflected forms for the category of number, the singular umbra and the plural umbrae, but we cannot produce an inflected masculine or neuter form of the same noun, because the feminine gender value of the noun is an inherent property of the lexeme. Therefore, Latin has overt coding of gender on nouns, which generally have a constant gender value (Audring 2016, 2019: 21–22).8 Conversely, both gender and number are marked on the targets by (contextual) inflection (Corbett 2013: 90–91; Booij 1994, 2005: 108–109). In other words, in Latin the gender value of the target is inflectional and has a syntactic function, while the gender of the controller is lexically specified and stored. It follows that the inflectional ending of a controller noun depends on the inflectional class of the nominal lexeme and its syntactic role in the sentence, while the ending of a target adjective depends on the inflectional class of the adjectival lexeme and on the grammatical values, including gender values, of the targeted noun.

7 The distinction between inherent and contextual inflection has been theorised by Booij (1996: 2, 2005: 103–112). Although both number and gender display inherent inflection in the controller, the main difference between them is that gender is an inherent property of the lexeme, while number is an inherent property of the word form (Loporcaro 2018: 9). Kibort (2010: 78) has further refined this distinction suggesting that, as far as nouns are concerned, gender is an inherent and (lexically) fixed feature, while number is an inherent and selected feature. 8 For an informative typological survey of the interaction of gender and number in some African languages, see Di Garbo (2014), especially pp. 105–109 about the cumulation of gender and number on nouns. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

14

Chapter 2

Although many languages show oppositions of gender, the way in which these oppositions are expressed is not always the same. First of all, gender is not a universal category: many languages lack it entirely (e.g. Turkic languages), while others display gender distinctions only in particular lexical classes, as in the pronominal gender system of Modern English.9 The mechanisms by which nouns are allotted to genders — the assignment system  — involve two basic principles: the meaning and the form (Corbett & Fraser 2000; Thornton 2009; Van Epps & Carling 2019: 2–3). The first principle is found in those languages where the gender system is assigned by semantic patterns. A characteristic of a canonical gender system is when “the meaning of a noun is sufficient to allow for gender assignment” (Corbett & Fedden 2018: 31). There are strict semantic assignment systems and predominantly semantic assignment systems: in the former, genders are assigned to nouns solely on the basis of semantics; the latter systems are still clearly based on semantics, but there are groups of exceptions (Corbett 1991: 8–30, 2013: 110–113). In many other languages, however, semantic rules are not enough to assign gender to nouns and other rules are required. Usually, if there is an opposition between masculine and feminine, one of the semantic factors that encodes gender is sex: nouns denoting males are masculine; nouns denoting females are feminine. However, this is often only a tendency (cf. the common example Germ. Mädchen ‘girl’, which requires neuter agreement in a noun phrase).10 Indeed, the vast majority of nouns in these languages are classified according to formal mechanisms, i.e. the signifier. As a consequence, they have what is generally called a formal assignment system, as they assign gender on the basis of both semantic and formal criteria. Formal criteria may include, for example, phonological information such as initial phonemes, final phonemes, or the prosodic structure of a given word. Another type of formal gender assignment is morphological, where the gender of a noun can be detected from morphological information such as inflection, derivation, or compounding. A morphological assignment system usually requires knowledge of the inflectional classes. However, gender crucially differs from the notion of inflectional class, the latter being defined as

9 For a thorough analysis of pronominal gender systems, see Audring (2008, 2009). 10 The lexical gender of Germ. Mädchen is more likely to be respected in local agreement than in distant agreement, where the referent-based gender of the noun prevails (Wälchli & Di Garbo 2019: 225–226, 280).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

15

The Gender System of Tocharian

(5) “a set of lexemes whose members each select the same set of inflectional realizations”. (Aronoff 1994: 182) That is to say, an inflectional class includes nouns with the same inflectional characteristics, but it may consist of nouns with different genders. It follows that in formal assignment systems gender may be particularly difficult to predict, because the gender of many nouns turns out to require knowledge of the inflectional classes. Most of the Indo-European languages, including Tocharian, show this typical formal assignment system (see  §  2.3 above), i.e. a core of semantic assignment rules (e.g. the sex of the noun referent or animacy) is supplemented by phonological or morphological principles (Matasović 2004: 79–164). To pave the way for my synchronic investigation into the category of gender in Tocharian, some basics on Canonical Typology must be introduced. This approach was first developed by Corbett for agreement (Corbett 2006) and morphosyntactic features (Corbett 2012), and was then specifically refined to establish the distinctiveness of grammatical gender (Corbett 2013; Corbett & Fedden 2016, 2018).11 In fact, since gender is a morphosyntactic phenomenon involving nominal agreement, the clusters of principles and criteria developed by Corbett & Fedden (2016, 2018) are specifically based on those principles and criteria previously theorised by Corbett for canonical agreement and canonical morphosyntactic features. As far as canonical agreement is concerned, Corbett (2006: 8–27) proposes three fundamental principles: (6) canonical agreement principle i Canonical agreement is redundant rather than informative. principle ii Canonical agreement is syntactically simple. principle iii The closer the expression of agreement is to canonical inflectional morphology, the more canonical it is as agreement. These principles can be established applying many criteria that relate to the canonical aspects of controllers, targets, domain, features, and conditions. To simplify the thorough account given by Corbett (2006: 9), canonical gender agreement avoids any syntactic complications; the controller is present, has overt expression of features, and is consistent in agreement, while the target 11

For an introduction to Canonical Typology, see Bond (2019).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

16

Chapter 2

has obligatory and regularly marking, which doubles that of the noun (i.e. the controller). There are five Canonical morphosyntactic principles that specifically apply to gender (Corbett & Fedden 2016: 499–517, 2018: 14–27). For our analysis of the Tocharian gender system, we will mostly concentrate on principle i with its four associated Criteria (7). (7) canonical morphosyntactic feature principle i Canonical features and their values are clearly distinguished by formal means. Criterion 1 Canonical features and their values have a dedicated form. Criterion 2 Canonical features and their values are uniquely distinguished across other logically compatible features and their values. Criterion 3 Canonical features and their values are distinguished consistently across relevant parts of speech (word classes). Criterion 4 Canonical features and their values are distinguished consistently across lexemes within relevant parts of speech. The interplay between these four criteria allows us to deal with another conceptual distinction that will prove relevant from the perspective of Tocharian. This distinction is that between controller gender and target gender.12 According to Corbett (1991: 151), (8) the controller gender is the gender into which nouns are divided, while the target gender is the gender which is marked on the modifier. This means that there is no specific set of forms in the modifiers that specifically mark a given gender. In other words, the controller gender is lexically marked on a given noun, while the target gender provides, on the morphosyntactic level, the creation of sets of agreement patterns that are related to the gender of the noun. On several occasion, Corbett (e.g. 1991: 150–154, 2012: 82–83 and 168–169, 2013: 93–95) exemplifies the distinction between target and controller gender 12 This terminology is more widely used in the literature. The pair target vs. controller gender mirrors Hockett’s selective vs. inflectional gender (1958: 230) and late Corbett’s non-autonomous vs. autonomous gender (2011: 459, 2012: 167–188). Following Croft (2013), Di Garbo (2014: 7–13, 2016) refers to agreement patterns as indexation patterns and accordingly defines the entities whose inflectional morphology signal gender as gender indexes and the entities that trigger an indexation pattern as indexation triggers. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

17

The Gender System of Tocharian

using Romanian, the only modern Romance standard language for which three genders are assumed by the vast majority of scholars, although the sets of distinctive agreeing forms available to mark gender values on modifiers are just two:13 Table 1

Target and controller gender in Romanian

singular

plural

(1) masculine

băiat bun ‘(a) good boy’

băieți bun-i ‘good boys’

(2) neuter

scaun bun ‘(a) good chair’

scaune bun-e ‘good chairs’

(3) feminine

fată bun-ă ‘(a) good girl’

fete bun-e ‘good girls’

Although the adjective bun ‘good’ displays only two distinct sets of forms, one for the masculine and one for the feminine, we have three substantives (băiat m. ‘boy’, scaun nt. ‘chair’, fată f. ‘girl’) in agreement with the same modifier, which shows three different agreement environments according to the gender of the substantive with which it agrees. According to Canonical Typology (Corbett 2013: 91–110; Corbett & Fedden 2016; Audring 2019: 19–22), the Romanian masculine and feminine are quite canonical: despite the fact that some cumulative exponence occurs, gender is expressed by means of suffixes on targets and is marked consistently and obligatorily. Both the controller and the target express these gender values overtly. Conversely, the Romanian neuter has no unique and dedicate form to justify it: it is non-canonical because neuter nouns do not have any agreement forms which are used uniquely for them (Corbett & Fedden 2016: 515). Thus,

13 Actually, the analysis of the Romanian gender system has become a disputed argument among specialists of Romance languages. A three-gender analysis is today maintained by several scholars, e.g. Matasović (2004: 51–52), Igartua (2006: 60–63), Acquaviva (2008: 135–140), Loporcaro & Paciaroni (2011), Loporcaro (2016, 2018: 92–110). In contrast, a two-gender analysis is argued by e.g. Maiden (2016), Bateman & Polinsky (2010). For a discussion on the term “neuter” for the third Romanian controller gender, see Maiden (2016: 40–41). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

18

Chapter 2

the Romanian neuter is non-canonical in violating criterion 1 and making the relation between gender and number more complex (criterion 2).14 A final concept to be introduced is that of inquorate gender. The term was coined by Corbett (1991: 170–172, 2012: 83–84), referring (9) to those agreement classes which comprise a small number of nouns that should be marked as exceptions. This peculiar type of agreement class is “inquorate” because it has insufficient members to deserve being labelled a gender value. Corbett & Fedden (2016: 508) exemplify the notion of inquorate gender using Serbo-Croat, which has three gender values (masculine, feminine, and neuter) plus a fourth agreement class. This fourth agreement class is for those nouns which take neuter agreement in the singular but feminine agreement in the plural. Theoretically we could set up an additional gender value for these nouns. However, there are only two members of this agreement class: oko ‘eye’ and uho ‘ear’. Therefore, Corbett & Fedden (2016: 508, 2018: 20–21) prefer to treat this agreement class as inquorate and thus label the two nouns as lexical exceptions (Corbett 2011, 2012: 188–189). According to Canonical Typology, inquorate values share with controller genders the violation of criterion 1. In both cases agreement provides for a combination of forms in the targets available for agreement with nouns in the normal gender values: that is, the targets have no dedicated forms for the candidate gender. Inquorate and controller genders, however, differ in so far as inquorate values also violate criterion 4, according to which “canonical features and their values are distinguished consistently across lexemes within relevant parts of speech”. Both controller and inquorate genders are non-canonical. However, while controller genders comprise a substantial number of nouns and form productive category values, inquorate genders represent limited groups of nouns which are synchronically unproductive. That is, the number of nouns belonging to the inquorate category is not sizeable: it represents a closed category (Igartua 2006). To conclude, genders are paradigmatic classes of nouns, established on syntagmatic evidence, since they can be identified on the basis of agreement with nominal modifiers. They can be assigned according to semantic and/or formal principles and form a relatively small, closed system. In the following sections, the theoretical background outlined above will be used to analyse the gender system of Tocharian.

14

See Corbett (2013: 93–95) and Corbett & Fedden (2016: 514–515). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

19

The Gender System of Tocharian

2.2

The Gender System of Tocharian

Although the factors according to which languages express gender contrasts are not always typologically clear, for the great majority of the Indo-European languages there is generally no dispute as to the number of genders they have. For a few others, however, the matter is more complex. Tocharian is one of those languages. Like in most languages with gender, in Tocharian the element triggering gender agreement is usually a noun, which is the agreement controller. Gender agreement occurs in adjectives, numerals from ‘one’ to ‘four’, demonstratives, relative and some interrogative pronouns used attributively, some participles and gerundives: these are the agreement targets. According to a classical theory, Tocharian displays only two grammatical genders in both the controller and the target. These are the masculine and the feminine. Remnants of the Indo-European neuter are indeed limited to some frozen forms, like the singular of the demonstrative pronoun TB te, ta ta- < PIE *tod (Skt. tát, Gk. τό, etc.; see § 2.2.2, § 4.3.2).15 In the following, some typical examples of masculine agreement are presented:16 THT1113 a2–3 (Karmavācanā) sg. Tṣe Cṣamāne postaññe te[a3]ṅkäñ-c one.m.nom.sg monk(m).nom.sg even stop.3sg.sbj-2sg.suff “If only one monk stops you”. (cf. Peyrot 2013a: 311) B337 a1 (Vinayavibhaṅga) pl. Cṣamāni Tmakci naumīyenta pareṃ monk(m).nom.pl self.m.nom.pl jewel(a).obl.pl bring.3pl.prs “The monks carry off the jewels by themselves”. (cf. Ogihara 2009: 327) A394 b1 (literary)17 sg. Tṣom Ckoṃ Twäc Ckoṃ one.m.obl.sg day(m).obl.sg.m second.m.obl.sg day(m).obl.sg pe ¦ śwāsi mā tāp tricäṃ part food(a).obl.sg neg eat.3sg.prt third.m.obl.sg koṃ ᛫ day(m).obl.sg “On the first and on the second day, he did not eat food; [also] on the third day …”. (cf. Carling dta 162) 15 16

For the term “frozen form”, see Trask (2000: 127). In the examples below, t x and c x indicate the agreement target and the agreement controller respectively. 17 Metre: 4×5¦7. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

20

Chapter 2

A151 a1 (Garbhasūtra) pl. Tcesäm ṣpät Ckoṃsaṃ kälytär this.m.obl.pl seven day(m).loc.pl stand.3sg.prs “For these seven days he stands”. Examples of feminine agreement are given below: IT248 b5–6 (Vinayavibhaṅga) sg. omte krui Caśiya Tṣär(ps)e[b6]maneñña there if nun(f).nom.sg pointing.out.f.nom.sg Tstmausa tākoy stand.prt.ptc.f.nom.sg be.3sg.opt “If a nun were standing there, giving instructions”. (cf. Peyrot 2013a: 348) AS18B a2 (Vinayavibhaṅga) pl. Ttoy Caśiyana po Tlalāṃṣuwa this.f.nom.pl nun(f).nom.pl all carry.out.prt.ptc.f.nom.pl stare be.3pl.prs “These nuns have arranged all”. (cf. Meunier 2013: 155) A187 a1 (literary) sg. Tlyāki Tkälkālyi Ctkaṃ naṣ flat.f.nom.sg accessible.f.nom.sg earth(f).nom.sg be.3sg.prs “The earth is flat and walkable”. (cf. Knoll 1996: 16) A59 a1 (Unmādayantījātaka) pl. sarkk oki tākar Tñäkcyāñ Ctkañi sequence like be.3pl.prt divine.f.nom.pl earth(f).nom.pl “The heavenly earths were like a series of steps”. (cf. Sieg 1952: 42) In these examples the substantives TB ṣamāne ‘monk’, TA koṃ ‘day, sun’ and TB aśiya ‘nun’, TA tkaṃ ‘earth’ are the controllers, while the various modifiers — adjectives, demonstratives, participles — are the targets. As demonstrated by the agreeing modifiers, the first pair of nouns is masculine, while the second is feminine. In addition, Tocharian has a large and productive class of nouns that constitutes a third, separate category: the genus alternans. As pointed out by Igartua (2006: 58), the term genus alternans “was coined to cover the specific nature of the third gender in Tocharian, which combines agreement traits of

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

21

The Gender System of Tocharian

the other two, the masculine and the feminine”. The genus alternans is exemplified below: B11 a5 (Udānālaṅkāra)18 sg. päst kl(au)tkoträ Clakle Tse ¦ away turn.3sg.prs.mid suffering(a).nom.sg this.m.nom.sg “This suffering turns away”. (cf. Schmidt 1974: 273) B88 b5 (Araṇemiijātaka) pl. sū Ttoṃ Cläklenta lkāṣṣäṃ he(m).nom.sg this.f.obl.pl suffering(a).obl.pl see.3sg.prs “He sees these sufferings”. (cf. Schmidt 2001: 318) A341 b3 (Koṭikarṇāvadāna) sg. Tcaṃ Coko wärpnātär that.m.obl.sg fruit(a).obl.sg enjoy.3sg.prs “[She] enjoys that fruit”. (cf. Sieg 1952: 40) A57 b2–3 (Rūpyāvatāvadāna) pl. Tsukaṣinās Cokontu eṣäntās pñintu happy.f.obl.pl fruit(a).obl.pl giving.prs.ptc.pl merit(a).obl.pl ese[b3](ñc) give.3pl.prs “[They] give merits giving fruits of happiness”. Although the agreement targets display only two distinct sets of forms, one for the masculine (nom.sg. TB se ‘this’; obl.sg. TA caṃ ‘that’) and one for the feminine (nom.pl. TB toṃ ‘these’; obl.pl. TA sukaṣinās), they stand in agreement with the same noun (TB lakle ‘suffering’ and TA oko ‘fruit’), revealing a third agreement environment that combines traits of both the masculine and the feminine. This agreement is designated alternans because it “alternates” masculine agreement in the singular and feminine agreement in the plural. From a historical point of view, the genus alternans in part mirrors the Proto-Indo-European neuter, because a number of alternating nouns historically reflect Indo-European neuters (e.g. TB yasar, ta ysār alt. ‘blood’ < PIE *h1ésh2-r̥ or the collective *h1ésh2-ōr nt.; see § 3.6). It could therefore be argued that Tocharian has an opposition between two target genders  — the masculine and the feminine  — and three controller 18 Metre: 4×7¦7¦4.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

22

Chapter 2

genders — the masculine, the feminine, and the genus alternans — which are regularly defined on the basis of the agreement between a noun and its modifier(s) (see (8) above) (Kim 2009a: 73–74; Hartmann 2013: 26–28; Fellner 2014a: 16). Table 2

Correlation between gender and number in Tocharian

gender

masculine genus alternans feminine

number singular

plural

masculine agreement masculine agreement feminine agreement

masculine agreement feminine agreement feminine agreement

In other words, the intraparadigmatic opposition between the three Tocharian agreement classes is based on the fact that the feminine is opposed to the genus alternans in the singular, while the masculine is opposed to the genus alternans in the plural. The masculine and the feminine are opposed to each other both in the singular and in the plural. Indeed, the Tocharian gender system does not meet all conditions pointed out in (4) for the agreement method, as it violates the first condition (controllers and targets distinguish the same gender values) by having three controller genders and two target genders: the agreement elements only have morphological markers for two genders, the masculine and the feminine. From the point of view of Canonical Typology, the agreement classes of the Tocharian masculine and feminine come fairly close to being canonical values. Indeed, these values are clearly distinguished by formal means and the agreement is syntactically simple. Despite the fact that the target markers cumulate the encoding of gender, number, and case distinctions, the controller nouns have inherent gender, and the targets have dedicated forms for features and values that must be conveyed grammatically, including the masculine and the feminine. Conversely, the status of the Tocharian third agreement class, the so-called genus alternans, is more complex. 2.2.1 The Status of the Tocharian Genus Alternans The Tocharian system of gender is relatively rare within the Indo-European domain, and both synchronic and diachronic analyses of it have raised controversy. From a synchronic point of view, the main matter of debate has been the linguistic analysis of the genus alternans. This problem is linked to a central Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

The Gender System of Tocharian

23

working question: how many genders did Tocharian have?19 The answer is not obvious, as one might imagine. In what follows, I will deal with the synchronic status of the third Tocharian gender, applying the principles and criteria of Canonical Typology and putting forward typological arguments and cross-linguistic comparisons with the Romance languages in general and with Romanian and Standard Italian in particular. I argue that the Tocharian genus alternans is to be regarded as a fully-fledged gender value, formally and semantically opposed to both the masculine and the feminine. Certain grammars and handbooks on Tocharian begin the discussion on the category of gender by reporting that both Tocharian languages display only two genders, the masculine and the feminine (but cf. Winter 1998: 159).20 Such a claim is present, for example, in the Elementarbuch (teb §§ 65–66), in the introduction to Tocharian by Krause (1971), more cautiously in the two excellent handbooks on Tocharian by Pinault (1989a: 67, 2008: 461), and also in Schmidt (2018: 215–216; cf. also Kim 2006: 726). In other works on Tocharian nominal morphology, the genus alternans has usually been treated as a “group of nouns”, or, more specifically, as an “agreement class” (Kim 2009a: 73–74; Fellner 2014a: 16). It is consequently not referred to as a gender in its own right, but as a gender-like category, paradigmatically different from the masculine and the feminine. The books and articles referenced above are mainly historically orientated and a discussion on the gender system from a synchronic and typological point of view is not expected. A new publication in this field is Hartmann (2013), whose aim is to provide a detailed account of the synchronic aspects of the category of gender in Tocharian (pp. 26–28). Hartmann claims that both Tocharian languages display two target genders (masculine and feminine) and three controller genders (masculine, feminine, and the alternating gender). He also argues that, in the literature on gender (he refers to Busmann 2008), the term Genus is sometimes employed as a synonym of Nominalklasse. However, Hartmann affirms that it is more correct to use Genus instead of Nominalklasse for Tocharian, because male entities are sorted into the masculine gender and female entities into the feminine gender (p. 26).21 No matter whether we

19 This question implies the definition of grammatical gender as given in § 2.1. 20 Some handbooks (e.g. Krause & Thomas 1960) report that Tocharian has three genders including the neuter. On the status of the Tocharian neuter gender, i.e. frozen forms that go back to the PIE neuter, see below (§ 2.2.2). 21 Hartmann (2013: 26) further argues that the term Nominalklasse, i.e. noun class, should be used as a synonym of Deklinationsklasse, i.e. inflectional class. I cannot agree with this terminological choice for the reasons given above (§ 2.1). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

24

Chapter 2

accept this argument or not, it does not say anything new on the analysis of the genus alternans. Given the special role Tocharian has acquired within the study of gender systems, I believe it is important to shed new light on the typological status of the genus alternans.22 The central problem here is not establishing whether nouns of this class are both masculine and feminine, or neither masculine nor feminine. Rather, we must establish whether we have to consider a controller gender like the Tocharian genus alternans as a real gender or not. Thus, the issue is not purely definitional. We have already seen that the agreement class (cf. the definition in (3) above) is the tool by which we deduce gender. That is, the gender of a noun is inferred from the gender-marking on associated elements. As a consequence, suggesting that Tocharian has two genders and three agreement classes does not say anything on the status of the genus alternans: once we have recognised that a language has two or more agreement classes, we must proceed further to establish whether those agreement classes can be analysed as real gender values. Usually a given language has as many genders as it has agreement classes. That is, canonical gender values match agreement classes (Corbett & Fedden 2016: 505, 2018: 8). On the other hand, there are some cases that make the relation between gender and agreement class less than straightforward. For our discussion it is fruitful to make a comparison with Standard Italian. Like most other Romance languages, Italian has reduced the three-gender system of Latin to a two-gender system, losing the neuter as a category value. As a consequence, we would expect only two agreement classes, one for the masculine and one for the feminine. However, Standard Italian shows a limited class of nouns that behaves exactly as the Tocharian genus alternans. Some examples are given below:23

22

23

Suffice it to mention that the term genus alternans, which is today used to label gender values that combine agreement traits of other agreement classes (Igartua 2006), was first coined by Krause & Thomas (teb § 66) for Tocharian: “Das Genus eines Substantivs ist daher im allgemeinen erst durch das Genus des attributiv oder prädikativ hinzugefügten Adjektivs oder Demonstrativpronomens bestimmbar. Dabei ergibt sich die Tatsache, daß gewisse Substantiva in den Singularformen Adjektiva oder Pronomina mit mask. Endung, in den Pluralformen solche mit fem. Endung zu sich nehmen. Wir verwenden dafür den Ausdruck Genus alternans (a.)”. I do not mention the gender system of Central-Southern Italo-Romance dialects, where the alternating gender is to be analysed differently. See Loporcaro & Paciaroni (2011: 410–424) and the relevant sections in Loporcaro (2018).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

25

The Gender System of Tocharian Table 3

Italian “alternating” nouns

singular masculine il braccio il dito il lenzuolo il paio l’uovo

plural ‘the.m arm’ feminine le braccia ‘the.f arms’ ‘the.m finger’ le dita ‘the.f fingers’ ‘the.m bed sheet’ le lenzuola ‘the.f bed sheets’ ‘the.m pair’ le paia ‘the.f pairs’ ‘the.m egg’ le uova ‘the.f eggs’

This group of nouns shows masculine agreement in the singular and feminine agreement in the plural, as is clearly demonstrated by the article, which is inflected as masculine in the singular (It. il, lo) and feminine in the plural (It. le).24 Applying the rules given above, it should be concluded that the Italian type braccio : braccia constitutes a third gender value. However, there is broad consensus among scholars that this class does not constitute a separate gender in Standard Italian. The main argument adduced to support the latter analysis is that this kind of agreement is limited to only one inflectional class with more or less thirty members.25 This class is very unproductive and closed, and it has been progressively eroded over the last centuries, developing a more recent masculine plural variant that is clearly based on the (masculine) singular form, e.g. il braccio : i bracci, il lenzuolo : i lenzuoli.26 Given that this group cannot constitute a new gender, nor even a controller gender (as in the case of Romanian and Tocharian), one could say that the Italian type braccio : braccia is an inquorate gender (cf. the definition in (9) above), as nouns belonging to this type have no dedicated target forms and the class is unproductive and close: it is a dying category (Thornton 2011: 460; Loporcaro 2018: 82–86, 245–246).

24

25

26

A similar agreement environment can also be found in Modern French, though limited to just three substantives (amour ‘love’, délice ‘delight’, and orgue ‘organ’), which, like the Italian type braccio : braccia, show masculine agreement in the singular and feminine in the plural. In fact, other inflectional classes that show the same agreement environment as the type braccio : braccia can be found, for example, in il carcere ‘the.m prison’ : le carceri ‘the.f prisons’ and il gregge, ‘the.m flock’ : le greggi ‘the.f flocks’. However, this inflectional class is extremely marginal and more closed than the type braccio : braccia. See Loporcaro (2016: 950 fn. 16). See Dressler & Thornton (1996: 16), Thornton (2011), Loporcaro, Faraoni & Gardani (2014: 5–6), Acquaviva (2008: 155), and Loporcaro (2018: 81–87, 245–246).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

26

Chapter 2

Conversely, in the case of Tocharian, there is evidence for claiming that it has a three-gender system, including the genus alternans. Firstly, the third Tocharian agreement class fulfils the claim by Hockett (1958: 231) and Corbett (1991: 105) that genders are classes of nouns reflected in the behaviour of associated words (cf. definition (1) above), as gender values match agreement classes (Corbett & Fedden 2018: 18). Secondly, from the point of view of the noun inventory, we find a wide range of substantives in the genus alternans, which is also productive, since the most recent loanwords not referring to human entities are usually placed in this category. Thirdly, as Hartmann (2013) further demonstrated, the genus alternans is quite a coherent class also from a semantic point of view, since substantives pertaining to this gender never refer to animate entities. The situation in Tocharian therefore parallels that of Romanian, where, in a manner similar to Tocharian, so-called neuter nouns select agreement targets formally identical to the masculine in the singular and to the feminine in the plural. The entire paradigm of a Romanian neuter noun and its gender agreement show a combination of agreement forms that differ from those used for the masculine and the feminine. Applying the Criteria of principle i of Canonical Typology (cf. (7) above), we are forced to analyse the Tocharian controller and target differently with respect to criterion 1 and criterion 2. Indeed, just as in the masculine and the feminine, so too do the alternating nouns tend to belong to individual inflectional classes, with the peculiarity of having no formal distinction between nominative and oblique in either the singular or the plural (with some minor exceptions, cf. § 2.3.1 and § 2.3.3).27 Conversely, it is the target that has no dedicated and unambiguous forms for the genus alternans: alternating nouns do not have any agreement forms which are used uniquely for them, 27

From the perspective of the agreement controller, the genus alternans is somewhat more canonical than the Romanian neuter. In Romanian, both neuter and masculine nouns end in consonants or -u in the singular; in the plural, neuter nouns only end in -e or -uri (Maiden 2016). Of these, final -e is shared with most feminine plural nouns and only -uri is almost unique to neuter nouns. It follows that it is impossible to tell from the form of the singular alone whether a noun is neuter or masculine; in the same way, it is impossible to predict whether final -e in the plural implies a neuter noun. Conversely, in Tocharian the plural inflection of a noun is usually a good indicator for distinguishing alternating from masculine and feminine nouns (see § 2.3.1 and § 2.3.3), as alternating nouns generally have one of the following nominative-accusative plural endings (depending on the inflectional class to which they belong): TB -a, ta -ā, TB -wa, ta -u, TB -nma, TB -nta, ta -nt, -ntu. The only plural ending which is particular to both feminine and alternating nouns is TB -na. However, in this case the referential gender serves to disambiguate the grammatical gender, as all feminine nouns with nom.obl.pl. -na refer to female entities (see § 2.3.3 and further § 2.3.1 for a non-canonical inflectional class of masculine nouns with pl. -nta).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

27

The Gender System of Tocharian

and this leads to a mismatch between controller and target genders.28 The mismatch is induced by number (Corbett & Fedden 2018: 27, 33). Still, there is another piece of evidence that, in my view, demonstrates that the Tocharian alternating gender must be regarded as a gender value in its own right. Again, this evidence comes from a cross-linguistic comparison between Tocharian and Standard Italian in non-canonical agreement environments where two alternating nouns are syntactically coordinated (Paciaroni, Nolè & Loporcaro: 112–121; Loporcaro 2018: 85–86). See the following example: Il braccio e il dito sono def.m.sg arm(x).sg and def.m.sg finger(x).sg be.3pl.prs rotti. Questi sembrano davvero brutti. broken.ptc.m.pl this.m.pl look.3pl.prs really awful.m.pl “The arm and the finger are broken. They look really awful”. In the sentence above, the coordination of two alternating nouns, i.e. braccio ‘arm’ and dito ‘finger’, inflected as singular crucially results in a masculine plural agreement in the modifiers (rott-i ‘broken.m’, quest-i ‘these.m’, brutt-i ‘horrible.m’). This agreement seems to be ungrammatical, because we would expect the targets inflected as feminine plural. In turn, the expected agreement is found only when the agreement controllers are inflected in the plural, as in the example below: Le braccia e le dita sono def.f.pl arm(x).pl and def.f.pl finger(x).pl be.3pl.prs rotte. Queste sembrano davvero brutte. broken.ptc.f.pl this.f.pl look.3pl.prs really awful.f.pl “The arms and the fingers are broken. They look really awful”.

28

I wonder whether the non-canonical behaviour of the Tocharian third gender could also be described by applying principle v of Canonical Typology, according to which: “In a canonical system of morphosyntactic features the contextual values match the inherent values” (Corbett & Fedden 2016: 511). This principle implies that non-canonicity arises when an inherent value of a controller (like the singular number of a wordform or the masculine gender of a lexeme) is not directly matched in the contextual feature of a target (like an adjective). If this is correct, the Tocharian alternating gender is less canonical in this respect, because there is no dedicated form in the target specifically used to match the inherent value of the controller. Again, the target selects the masculine form in the singular and the feminine form in the plural when in agreement with a controller whose inherent gender value is alternating.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

28

Chapter 2

Although this is not a decisive argument for gender resolution in itself, since even same-gender conjuncts may require the application of specific gender resolution rules (Corbett 2006: 238–239), it shows that the Italian inquorate gender braccio : braccia has been losing consistency in its syntactic manifestations, particularly if compared with the situation of Old Italian and Central-Southern Italo-Romance dialects (Paciaroni, Nolè & Loporcaro 2013: 112–121).29 I have therefore tried to find examples of similar nominal agreement in Tocharian. The examples proposed below are in my opinion probative to draw up a strong distinction with respect to Standard Italian.30 A17 b5–6 (Puṇyavantajātaka) taṃne wkänyo knānmune pñintwäṣ pkä(nt) thus manner.instr.sg wisdom(a).nom.sg virtue(a).abl.pl separately mā pälkäṣ ṣyakk a(ts) (pa)[b6]t nu ṣokyo pälketsāñ31 neg shine.3sg.prs together part or but very shining.f.nom.pl mäskaṃträ tämyo tom pkänt pkänt be.3pl.prs therefore this.f.nom.pl separately separately sambhārntu wewñunt sambhāra(a).nom.pl call.prt.ptc.f.nom.pl “In this manner wisdom without virtue(s) does not shine, but rather together they are especially brilliant. Therefore, these are called sambhāras”. (cf. Sieg 1944: 21)

29 Corbett (2006: 243–244) has shown that in Slovene two gender resolution rules operate in coordinated agreement: (1) if all conjuncts are feminine, then agreement is feminine; (2) otherwise agreement is masculine (I thank Tijmen Pronk for bringing this point to my attention). This seems to suggest that agreement in these cases is not a good indicator of gender. However, Corbett (2006: 261) argues that the difference in languages like Slovene is that they also have semantic resolution rules: (1) if all conjuncts refer to female humans, agreement is feminine; (2) if all conjuncts refer to humans, whether all male or of mixed sexes, agreement is masculine; (3) in all other cases, agreement is masculine. In this respect the situation of Romance languages and Tocharian is very different. For this reason, I maintain that the coordinated agreement test can be used for our purposes. 30 See also Hartmann (2013: 106). 31 The nom.pl.m. of TA pälkets ‘shining’ is not attested, but it may be reconstructed as pälketse* (cf. TA wākmtse ‘distinguished, superior’ from wākmats). The nom.pl.f. pälketsāñ is also attested in A148 a2–3, where it regularly agrees with an alternating noun inflected as a plural, i.e. lyiyā-āpsā ‘limbs, limbs and joins’ (DTA 37; see also tg § 174).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

The Gender System of Tocharian

29

B5 a6 (Udānālaṅkāra)32 tary= akṣā-ne pudñäkte ¦ three.f announce.3sg.prt-3sg.suff Buddha(m).nom.sg teki ktsaitsñe srukalñe ¦ disease(a).obl.sg old.age(a).obl.sg death(a).obl.sg toṃ mā tākoṃ śaiṣṣene ¦ this.f.nom.pl neg be.3pl.opt world.loc.sg mā ṅke tsaṅko(y) pudñäkte ᛬ neg part rise.3sg.opt Buddha(m).nom.sg “The Buddha announced to him the three: disease (Skt. vyādhi-), old age (Skt. jarā-), death (Skt. maraṇa-). If they were not there in the world, then the Buddha would not arise”. (cf. Sieg & Siegling 1949: 10) Both Tocharian languages have a wide range of demonstrative pronouns, which always agree in gender and number with their antecedent (in both attributive and pronominal uses). In the fragments above, we have two feminine inflected forms, i.e. TA tom and TB toṃ, which are in anaphoric reference with coordinated alternating nouns (namely knānmune ‘wisdom’ and pñintu (pl.) ‘virtue’ in A17 b5–6; teki ‘disease’, ktsaitsñe ‘old age’, and srukalñe ‘death’ in B5 a6). This means that the feminine plural forms of the pronouns actually represent the plural of the coordinated alternating singular in the nouns. Also, TB tarya ‘three’ in B5 a6 is regularly inflected as feminine when in agreement with the coordinated alternating singular nouns teki, ktsaitsñe, and srukalñe. Another important example in this regard is the following from Tocharian A: A73 b5 (Mūghapakkhajātaka) c tuṅk āly(a)knaṃ mā t yāmlaṃ other.loc.pl neg do.ger.f.nom.pl love(a).nom.sg c ynāñmune ᛬ reverence(a).nom.sg “Love and reverence could not have been made to anyone else”. (cf. Thomas 1952: 34) There is no demonstrative pronoun attested here, but, as in the case of the previous examples, coordinated alternating nouns inflected as singular (TA tuṅk and ynāñmune) agree with the subjunctive gerundive yāmlaṃ, which is in turn inflected as a feminine plural. 32 Metre: 4×7¦7.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

30

Chapter 2

As far as we can tell from the evidence at our disposal, the resolution rules that obtain in Tocharian are the following:33 Table 4

Gender resolution in Tocharian

a. if all conjuncts are masculine b. if all conjuncts are feminine c. if all conjuncts are alternating d. if one conjunct denotes a male animate

→ → → →

masculine agreement feminine agreement feminine agreement masculine agreement

As for the alternating agreement, the situation of Tocharian is, again, parallel to that of Romanian (Corbett 1991: 289; Paciaroni, Nolè & Loporcaro 2013: 119–120):34 Frigider-ul şi televizor-ul sunt stricate. fridge(nt).sg and television(nt).sg are broken.f.pl Acestea trebuie să fie reparate det.f.pl must.be repaired.f.pl “The fridge and the television are broken. These must be repaired”. To sum up, on the basis of this very clear evidence, the following agreement rules with alternating nouns can be posited by comparing Tocharian and Standard Italian: 33

34

For an account of the gender agreement in coordinating environments, see Hartmann (2013: 104–109). I was not able to find unambiguous examples of conjuncts agreeing with a masculine noun not denoting male entities and another noun of a different gender. Selected examples are: (1) B375 b1 ista(k pañä)kt(e) käṣṣi cau wäntare śarsa Taṣanikeṃ (pl.m.?) Cśāriputreṃ (sg.m.) Cmaudgalyāyaneṃ (sg.m.) “Th(e Bud)dha, the teacher, immediately understood this fact [and] the venerable Śāriputra [and] Maudgalyāyana …” (cf. Thomas 1957: 120); (2) B107 a9 Cnānda (sg.f.) Cnandābala (sg.f.) weñāre se cisa śpālmeṃ tākaṃ cwi aiskem […] a10 T-Ctoy (pl.f.) Tkakkāccuwa (pl.f.) bramñikteṃś maitare “Nandā [and] Nandabalā said: « Who is better than you, to him we give it ». […] Having rejoiced, they set out to God Brahman” (cf. Pinault 2008: 158); (3) A395 a4 täm kaklyuṣuräṣ Tcem (pl.m.) priyadattes Cpācar (sg.m.) Cmācar (sg.f.) cam klopyo ime crakär “Having heard that, Priyadattas father [and] mother, through this suffering, lost consciousness” (cf. Krause 1971: 40; Zimmer 1976: 49–50); (4) B4 a2 Cteki (sg.alt.) Cktsaitsñe (sg.alt.) Ckes (sg.alt. or m.) Cyoko (sg.f.) ¦ T-C toṃ (pl.f.) ñya(tsenta) /// “Disease (Skt. vyādhi-), old age (Skt. jarā-), hunger (Skt. kṣudh-), thirst (Skt. tṛṣṇā-): these (are) the plagues” (cf. Sieg & Siegling 1949: 8). Note, however, that example (4) might not be relevant, as nom.pl.f toṃ can also agree with nom.pl.alt. ñyatsenta ‘plagues’, as the metre seems to indicate. For a discussion of gender resolution in Romanian, see Loporcaro (2018: 104–109).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

31

The Gender System of Tocharian Table 5

Alternating agreement in coordinated singular noun phrases

controller target anaphoric

controller target anaphoric

italian

tocharian

{“alt.”sg}noun + {“alt.”sg}noun {masc.pl}adjective {masc.pl}demonstrative

{alt.sg}noun + {alt.sg}noun {fem.pl}adjective {fem.pl}demonstrative

italian

tocharian

{“alt.”pl}noun + {“alt.”pl}noun {fem.pl}adjective {fem.pl}demonstrative

{alt.pl}noun + {alt.pl}noun {fem.pl}adjective {fem.pl}demonstrative

This comparison of Standard Italian and Tocharian has highlighted that in the former the group of substantives that show alternating agreement is a closed category, with masculine agreement in coordinated environments and in anaphoric reference; in the latter, alternating nouns represent a cohesive group, with different plural markers and, as far as we can tell from the fragmentary corpus, coherent agreement in all possible environments. The only remaining peculiarity is that the patterns of syncretism are such in the targets that there are no agreement markers unique to the alternating gender in Tocharian. The genus alternans is therefore a non-autonomous gender value, as alternating nouns do not have any agreement forms which are unique to them, and this leads to a mismatch between controller and target genders. All these crucial elements allow us to conclude that the genus alternans should be considered a real (non-autonomous) gender in Tocharian. Although it is a grammatical strategy that pertains to the domain of the controller gender, i.e. of a non-canonical gender value, it is fully embedded in the grammar of the language. In this way it displays all the features that a gender value must have, since it also represents a systematic property that belongs to the core of the category of gender. On the Terminology of the Third Gender and the Status of the Tocharian “Neuter” In the current literature, the name of the third Tocharian gender is not consistent. I have so far avoided this problem, labelling it the “alternating gender”. However, even in recent works on Tocharian, the terms “alternating” and 2.2.2

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

32

Chapter 2

“neuter” are frequently used interchangeably.35 This terminological mismatch can also be found in the two modern dictionaries of Tocharian, the Dictionary of Tocharian B by Douglas Q. Adams (2013) and the Dictionary and Thesaurus of Tocharian A (vol. 1) by Gerd Carling (2009). The first uses the term neuter, but the second uses “alternans”. Other specialists of Tocharian also diverge with regard to this nomenclature: Krause & Thomas (1960, teb), Pinault (e.g. 2008), and Hartmann (2013) use “alternans”, while Sieg, Siegling, & Schulze (1931, tg), Malzahn (e.g. 2011), Kim (2009a, 2014), and Fellner (e.g. 2014a) use “neuter”. The reason scholars use “neuter” when referring to the third gender is historically founded, since several nouns reconstructed as neuter for the protolanguage synchronically belong to this category. However, the third Tocharian gender is different from the PIE neuter. Indeed, it is the result of morphophonological mergers that led, on the one hand, to the functional loss of the neuter and, on the other hand, to the rise of an agreement class that in turn can be analysed as a new gender, namely the genus alternans (see § 3.6 and § 4.8). Employing the label “alternating” has also some terminological advantages with respect to a residual class of frozen forms for which the term “neuter” is more appropriately used. This relic class constitutes remnants of the historical neuter gender and it is limited to: (1) demonstrative pronouns, like TB te, ta ta- < PIE *tód (cf. Skt. tát, Gk. τό, etc.); (2) ordinal numerals, which derived from the corresponding cardinals with the addition of the suffix TB -te, ta -t < PIE *-to- (cf. Gk. τρίτος ‘third’, Lat. quārtus ‘fourth’, Av. puxδa- ‘fifth’, etc.). As demonstrated by Stumpf (1971: 5–7 and 47–62), the neuter gender of the demonstratives must be regarded as an archaism. This is not surprising, given the fact that pronouns have a special typological role in the rise, the further development, and the possible decline of gender values. Indeed, if demonstratives play a key role in the origin of gender markers, then they are also the category in which traces of a decayed gender might still be found. Corbett (1991: 310–312) and Luraghi (2011: 451) claim that the rise of gender systems is a grammaticalisation process that is expected to undergo the following development: 35

The third Tocharian gender cannot be labelled as “ambigeneric” (from Lat. ambo ‘both’). Etymologically, this term would imply that the genus alternans belonged in part to the class of masculine and in part to the class of feminine, and consequently that the alternating nouns must be considered masculine in the singular and feminine in the plural. This hypothesis is improbable in light of the analysis given in § 2.2.1.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

33

The Gender System of Tocharian

(10) generic nouns → classifiers → pronominal demonstratives → attributive demonstratives → determiners → agreement markers. This kind of development follows Corbett’s Agreement Hierarchy (cf. Corbett 2010; Di Garbo & Miestamo 2019: 23–24, 37). Conversely, when a gender value is lost, the opposite evolution is expected (see § 4.3.2). In Tocharian, neuter demonstrative forms are distinguished from the masculine and the feminine ones by three main differences: (1) they have only singular inflection; (2) they have a non-palatalised stem TAB t-; (3) they have only pronominal (i.e. anaphoric) function (see below). From a typological and cross-linguistic perspective, this situation is fully understandable. Indeed, in languages where a gender is lost but it continues to be formally differentiated in the pronouns, it can only be used with pronominal value, and never attributively, i.e. with adjectival value. Pronouns generally retain gender distinction when attributive modifiers have lost gender agreement (Corbett 1991: 137–139). Strictly speaking, it means that in a noun phrase the neuter demonstrative cannot be used as a nominal modifier, as it cannot occur with a noun phrase headed by a noun as a controller. Following Corbett (1991: 159), it would be more convenient to label these forms of the demonstratives as “neutral”, given the fact that they are only used with neutral agreement. See the following examples, which clarify the function of the neuter demonstratives (TB te, ta täm): B85 b4–5 (Araṇemijātaka) || te keklyau« ṣo »rmeṃ araṇemiñ lā[b5]nte dem.obl.sg hear.abs Araṇemi(m).gen.sg king(m).gen.sg pit maiwāte-ne k(eṃ)tsa klāya ᛫ bile.nom.sg tremble.3sg.prt-3sg.suff earth.perl.sg fall.3sg.prt.act “Having heard this, the bile of king Araṇemi quivered (= king Araṇemi fainted) [and] fell to the ground”. (cf. Schmidt 2011: 314–315) A346 a1 (literary) täm pälkoräṣ weyeṃ nāṃtsu nande dem.obl.sg see.abs surprised be.prt.ptc.nom.sg Nanda.nom.sg träṅkäṣ speak.3sg.prs “Having seen this, Nanda, being surprised, speaks”.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

34

Chapter 2

Frozen forms of the neuter demonstratives also occur with two functions: (1) as temporal or modal adverbs; (2) as conjunctions. In Tocharian A, neuter demonstratives with adverbial value usually show the particle TA -ne added directly to the basic form: from the pronoun of anaphoric deixis TA säm, sām, täm ‘he, she, it’, we have tämne ‘so’ or tämnek with further addition of the emphatic particle -k (cf. e.g. TB ykāk ‘still’, TB ṣek ‘always’, TA okāk ‘until’). Formations with secondary cases are also attested, especially when the pronouns are used as conjunctions, as in the old instrumental TA tämyo ~ tämyok ‘therefore (← *‘because of that’)’. As far as Tocharian B is concerned, the ablative TB tumeṃ and the perlative TB tusa, both from the pronoun of anaphoric deixis TB su, sāu, tu ‘he, she, it’, respectively mean ‘then, thereupon (← *‘from this’)’ and ‘therefore (← *‘through this’)’ (Stumpf 1971: 58–59). Outside of the demonstratives, other old neuter forms can also be found in the inflection of the ordinals for ‘second’ and ‘third’ (Winter 1992: 134–136; Pinault 2008: 558–559): TB wate, ta wät < PT *wəte < PIE *du̯ itom ‘second’; TB trite, ta trit < PT *trəyte PT *pewə > TB pew (Ringe 1996: 28; cf. Adams dtb 429, who reconstructs the suffix as *-u̯ en-). The outcome of the PIE possessive suffix *-u̯ ent- is well attested in Tocharian, as in TB pernew ‘glorious’, tallāw ‘miserable’, etc. (Pinault 2008: 524–526). The fact that TB -pew did not follow the same inflection of the regular outcome of the adjectives in *-u̯ ent- is plausibly explained by Winter (1962b) by analogy with the inflectional type of the bahuvrīhi-compounds. Since TB °pew is an inherited word (to be formally connected with Skt. padvát- ‘having feet, running’), the pair TB wi-pew* : śtwer-pew might match Skt. dvipád- : cátuṣpad- and Umb. dupursus : petupursus (Iguvine Table VIb 10–11).68 Furthermore, both Tocharian languages show two terms referring to ‘water’, namely TAB āp ‘body of water; river’ and TB war, ta wär ‘water’. The fact that the former is feminine while the latter is alternating matches the Vedic pair áp- (f.): udán- (nt.).69 In the R̥ gveda, the feminine noun clearly refers to water 68 On the meaning of Gk. ἀνδράποδα ‘slave’ as opposed to Gk. δοῦλος and τετράποδα, see Lazzeroni (1998b: 26–31). 69 Several PIE words for ‘water’ can be reconstructed, but the heteroclitic paradigm PIE *u̯ ód-r/n- is the most broadly attested. It is well-known that Vedic had many words referring to water. Two of them are Ved. udán- and vā́r, both of neuter gender. Lubotsky (2013) has argued that these nouns may belong to one and the same supplementary paradigm, so that the nominative and the accusative case of udán- are supplied in the singular by vā́r. He argues that these two nouns come from the same paradigm on the diachronic level, and that PIE *u̯ eh1r (or *u̯ oHr), from which Ved. vā́r derives, developed from an original *u̯ odr̥ . I agree with Lubotsky that vā́r and udán- are both synchronically and diachronically connected. To be more precise, since Ved. vā́r occurs only in the nominative and accusative singular and the udán-forms occurs in all other case-forms, they would

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

54

Chapter 2

as an active and living being, a personified natural force, while the other refers to a material and inactive entity, i.e. as the thing itself (Meillet 1921: 216–217). In Tocharian A, the first term is attested only twice in the locative singular āpaṃ ‘in the water’ (A226 a3 tāmäk āpaṃ ālyek nuṃ wrasañ tāloṣ klopasuṣ “in this very water again other unhappy and miserable living beings …” and A396.a a1 āpaṃ “in the water”). In Tocharian B, the term is not frequent either. I have identified the following certain attestations (twice as a nominative singular, once as an oblique singular, and once as an oblique plural):70 IT179 a4 (literary) ot śoliṣṣa71 āp wräṃtsaimeṃ mäske(tär) then of.hearth.f.nom.sg water(f).nom.sg from.opposite be.3sg.prs “then the water of hearth (?) is from the opposite [direction]”. (Adams dtb 47) IT179 b4 (literary) sāu āp det.f.nom.sg water(f).nom.sg “the water”. perfectly mirror the outcome of a heteroclitic *r/n-stem. However, I am not completely convinced by the evolution *-dr > *-h1r. One might say that the cluster *-dr# has been simplified in -r, and that the loss of the dental stop caused compensatory lengthening of the root vowel in some Indo-European languages. A parallel situation is perhaps attested in two different Latin words. According to de Vaan (2008: 641 and 644), Lat. unda ‘wave’ (< *udna; cf. also Lat. fundus ‘bottom’ < *bhudhno- < *bhudhmno-) is etymologically connected to the n-stem of *u̯ odr̥ /n-, which in turn became a first-declension noun (the a-stem is probably to be interpreted as an old neuter plural); on the other hand, Lat. ūrīna is built on ūr- plus the suffix -īno-. This ūr°, attested only indirectly, can be the outcome of a zero grade *udr-, with loss of *-d- and compensatory lengthening of the initial vowel. If Tocharian inherited this paradigm, TB war, ta wär may be the result of a merger between the r- and the n-stem. See recently Kim (2019b). The continuant of PIE *h2ép- is only attested in the Indo-Iranian branch (e.g. Av. ā̆p-, OP ā̆p- ‘water’), in Tocharian, in Oscan (acc.sg. aapam), and perhaps in Baltic (e.g. OPrussian ape, Lith. ùpė ‘river, brook’, see NIL 311–317). In most Indo-European languages this term has developed the specific meaning of ‘river’. 70 According to Adams (dtb 47) another possible attestation is in IT74 b1, a bilingual Sanskrit-Tocharian B fragment, where he reads (śt)w(āra) a(päṃ). However, the reading is doubtful, given the fact that the document is very fragmentary. Peyrot (2007 n° 74) reads /// (śt)w(āra) k⋅t⋅ ᛫ – ///. 71 Adams (dtb 47) seems to have followed Broomhead (1962: 1.240 and 2.238) in interpreting śoliṣṣa as a later form of TB śaulaṣṣa ‘pertaining to life’. It cannot be excluded, however, that śoliṣṣe* is to be derived from śolyiye* ‘hearth’ (only attested in the loc.sg. śolyine ~ śoline).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

55

The Gender System of Tocharian

B140 b4 (literary, Kāvya) āp saṃsā(rṣṣai)72 (no) sū kā water(f).obl.sg of.saṃsāra.f.obl.sg then det.m.nom.sg why swāsaṃ rain.3sg.sbj. “why does he rain then the stream of saṃsāra?”. (Adams dtb 46–47) IT23 a5 (literary) orotstsana āpäṃ// great.f.obl.pl water(f).obl.pl “great waters/rivers …”. As is clear, TAB āp does not mean simply ‘water’, but more specifically ‘river, rain’ or, more generally, ‘flowing water’. As far as the origin of this word is concerned, two different hypotheses can be formulated: either it goes back directly to PIE *h2ép- or it has been borrowed from Sanskrit or Middle Indian. Hartmann (2013: 445–448) claims that both explanations are possible. However, certain evidence may point to the reconstruction of TAB āp as an inherited word. In Tocharian B, the noun is surely feminine and has an oblique plural in -äṃ. It probably belongs to class v, which does not include loanwords of Indo-Aryan origin that do not refer to human beings. Indeed, if it were a loanword, a plural ending in TB -nma or TB -nta would be expected. As we have seen, the plural form of this word is attested only once in Tocharian B at the end of a fragmented line. As a consequence, one could say that āpäṃ// has to be restored as āpäṃ(ta) /ápənta/ or āpäṃ(ma) /ápənma/ (cf. for instance TB cakaṃma, plural of cāk ‘picul’). Of the two restorations, the latter would be preferable because monosyllabic loanwords usually take the plural -nma in Tocharian B. However, a peculiarity of nouns of class ii.2 is that they attract the accent in the plural (e.g. kālp /kálp/ ‘meaning, sense’ : kalpanma /kalpə́nma/), while in the hypothetical form **āpänma the accent would be fixed on the first syllable. For these reasons, and because the word is of feminine gender, TAB āp is more likely to be an inherited word (Van Windekens 1976: 166; dtb 47).73 72 Restorations by Sieg & Siegling (1953: 73). 73 Hartmann (2013: 446) lists a few loanwords that have retained the gender of the Indo-Aryan source. However, the majority of nouns retaining the feminine gender are from Sanskrit ā-stems or ī-stems (e.g. TAB anityāt from Skt. anityatā- ‘transiency’ [swtf 1.55]). Note that in Classical and Buddhist Sanskrit āp is only used in the plural (swtf 1.92; mw 47); the singular inflection is confined to Vedic Sanskrit. Thematic āpa (nt.) is found in Pāli.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

56

Chapter 2

Turning now to the second term, the situation is completely different with regard to both productivity and gender. Indeed, TB war and TA wär are very productive and they are of alternating gender, representing old neuter forms. From a semantic point of view, TB war and TA wär mean both material and flowing water. A polar semantic distribution with TAB āp is therefore opaque. As a matter of fact, the chronological distance between the culture attested in the Veda and the one attested in the Tocharian texts is huge. We can therefore hypothesise that the Tocharian words had the same semantic distribution of the Vedic pair, and that this distribution is still represented in the gender polarisation of the terms, even though it has become opaque in the historical attestation of the Tocharian languages. 2.3.4 Summary of the Gender Assignment Patterns Below I list the semantic and formal patterns of gender assignment discussed above (cf. also Hartmann 2013: 381–409). Semantic Principles 1. Animacy is a strong predictor of non-alternating gender. Inanimacy is a general condition for the members of the alternating gender, but inanimate nouns (both abstract and concrete) are also distributed across the feminine and the masculine. 2. Referential gender is a strong predictor of masculine or feminine gender. Words denoting human referents have a gender corresponding to the referential gender (i.e. biological sex). 3. Words referring to people without a specified sex are masculine. 4. Words denoting animals can be masculine or feminine, but the hypernym for ‘animal’ is alternating. Formal Principles (Inflectional) 5. Nouns that do not distinguish nominative and oblique in the plural are usually alternating (first macro-class). 6. Nouns distinguishing nominative and oblique in the plural are usually non-alternating (second macro-class). 7. Nouns that do not distinguish nominative and oblique in the plural but that refer to male or female entities are masculine or feminine (→ principle 2). 8. In Tocharian B, nouns with nom.pl. -i, obl.pl. -eṃ are usually masculine (→ principle 6).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

The Gender System of Tocharian

57

9.

In Tocharian B, nouns distinguishing nominative and oblique in the singular are typically non-alternating. 10. In Tocharian B, nouns with nom.sg. -ya and obl.sg. -yai are typically feminine (→ principle 9). 11. In Tocharian B, nouns with nom.sg. -a, obl.sg. -o, gen.sg. -oy are feminine (→ principles 7, 9). 12. In Tocharian A, nouns with obl.sg. -āṃ are feminine (→ principles 2, 6). 13. In Tocharian B, nouns with obl.sg. -ṃ are masculine (→ principles 2, 6). 14. Nouns with gen.sg. TAB -i, TA -y are non-alternating (→ principle 2). 15. Nouns with gen.sg. TB -ñ, TA -(y)āp are masculine (→ principle 2). Formal Principles (Derivational) 16. Nouns derived with the suffixes TB -eñca, TB -ntsa, TB -nta, ta -nt, TB -(a)uca, TB -uki are non-alternating. They are typically masculine, although they can more rarely refer to feminine entities (→ principles 2, 6; partial conflict with principle 10). 17. Agent nouns derived with the suffix TB -tse, ta -ts are masculine (→ principles 2, 8). 18. Agent nouns derived with the suffixes TB -tau, -(a)u are masculine (→ principles 2, 6). 19. Nouns derived with the suffixes TB -śke, -kke are masculine (→ principles 2, 8). 20. Nouns derived with the suffix TB -āñca, ta -āñc are feminine (→ principle 2). 21. Nouns derived with the suffix TB -ñña are feminine (→ principles 2, 7, 10). 22. Nouns derived with -a from masculine nouns with -e, -o are feminine (→ principles 2, 7, 10). 23. Nouns derived with the suffixes TB -śka, -kka are feminine (→ principles 2, 7, 10, 22). 24. Abstract nouns derived with the suffix TB -eñña, ta -eṃ, TB -auña are feminine (→ principles 6, 10). 25. Nouns derived with the suffixes TB -ññe, -auñe, -uññe, TA -una, -one are alternating (→ principles 5, 9). 26. Substantivised infinitives in TAB -tsi are alternating (→ principles 5, 9). 27. Nouns derived with the suffixes TB -or, -wer are alternating (→ principles 5, 9). 28. Deverbal nouns derived with the suffix TA -äm are alternating (→ principles 5, 9).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

58 2.4

Chapter 2

The Gender System of Proto-Indo-European: State of the Art and Open Questions

Since the late nineteenth century, the Proto-Indo-European category of gender has been one of the most enduring issues within Indo-European studies and it remains a topic of controversial analyses from a historical, typological, and areal perspective.74 In the Indo-European domain, certain semantic associations can be used to predict the gender of nouns. Indeed, the vast majority of the oldest Indo-European languages display a three-gender system, constituted by the masculine, the feminine, and the neuter, with a predominantly formal assignment system (Matasović 2004: 136–138). Natural sex certainly played an important role in the distinction between masculine and feminine. The neuter is typical of those nouns referring to referents which are neither male nor female, and to non-human entities in general. However, a given noun may of course be masculine or feminine even if it does not refer to an animate entity (e.g. Gk. πόλεμος ‘war’ m.; Lat. rosa ‘rose’ f. etc.). Furthermore, nouns denoting small children and young animals are often neuter in the Indo-European languages, suggesting that sex was irrelevant for humans and animals before the age of procreation (Luraghi 2011: 445). The three-gender system is attested in many ancient Indo-European languages, such as Sanskrit, Avestan, Ancient Greek, and Latin, and persists in several modern ones (e.g. in Modern Greek, German, Russian, etc.). Conversely, among others, the Romance languages (with notable exceptions), most of the modern Indo-Aryan languages (Hindi and the Rajasthani languages), certain modern Iranian languages (e.g. Pashto and Zazaki), and the Celtic languages have reduced the number of genders to just two: the masculine and the feminine.75 This type of binary system is not semantically based, because the 74

75

For recent bibliography and up-do-date discussions of several problems connected to the PIE gender system, see Ledo-Lemos (2003), Matasović (2004), Luraghi (2006, 2009a, 2009b, 2011, 2014), Melchert (2000, 2014), Pinault (2011b), the papers collected in Neri & Schuhmann (2014), and Carling & Cathcart (2021). See also the accurate overview by Lundquist & Yates (2018: 2094–2100). For the drift in the development of gender within Indo-European, see Priestly (1983) and Matasović (2004: 72–77). For a phylogenetic method applied to the evolutionary reconstruction of PIE gender, see Carling & Cathcart (2021: 572–573). Note that there are also Indo-European languages that have preserved the neuter rather than the feminine (e.g. Swedish and Danish), and even some that have increased the number of genders, as in the case of Central-Southern Italo-Romance. For the Scandinavian languages, see Van Epps & Carling (2019); for Central-Southern Italo-Romance, see Loporcaro (2018). Gender distinctions have been lost in Armenian and in several Iranian and Indo-Aryan languages.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

The Gender System of Tocharian

59

distribution of the substantives in a given gender is highly idiosyncratic. The idiosyncrasy mirrors the fact that the referents of most words have no sex. This has been a general property of Indo-European languages in so far as they preserve gender distinctions. In several ancient Indo-European languages, the correlation between gender and inflectional class is not a one-to-one relationship (Matasović 2004: 138–142). In Latin and Ancient Greek, for instance, ā-stems (the so-called “first declension”) are predominantly feminine and o-stems (the so-called “second declension”) are masculine or neuter, but still we find inconsistences in the distribution of the genders in these two declensions. For example, Lat. poēta ‘poet’, agricola ‘farmer’, frātricīda ‘fratricide’ and Gk. ὁ πολίτης ‘citizen’, ὁ νεανίας ‘young man’, ὁ στρατιώτης ‘soldier’ are all masculine but they belong to the first declension, while Lat. platanus ‘planetree’, domus ‘house’, alvus ‘womb, belly’ and Gk. ἡ νῆσος ‘island’, ἡ ἔπημος ‘desert, wilderness’, ἡ ὁδός ‘road’ are feminine second-declension nouns.76 In the athematic type (Greek and Latin third declension), which includes several inflectional classes, it is rather difficult to distinguish the masculine from the feminine a priori, although some derivational suffixes seem to be used either exclusively or primarily with one gender (cf. Hoffmann 2016: 823–824 for Latin). As discussed above (§ 2.1), the analytical tool that allows us to identify the genders of a language is agreement. The Indo-European tripartite system is accordingly identified through a morphosyntactic agreement system that involves nouns, adjectives, pronouns, participles, etc. (Luraghi 2014: 199). The same system of gender was also reconstructed for the proto-language, or at least it was up until the late 19th century. Indeed, even before Anatolian was discovered leading scholars had identified, through the application of internal reconstruction, that the feminine gender developed latest. At an early date, Delbrück (Grundriss 3.132–133) already proposed that the three-gender system of Proto-Indo-European appears to have originated in a binary noun class opposition between an animate and a neuter gender. The feminine would have arisen later. There is today a broad scholarly consensus around an original two-gender model of Proto-Indo-European, according to which the emergence of a grammatical feminine gender is a later phenomenon which resulted in a reconstitution of the entire gender system of the proto-language. In this respect, one of the most enduring questions within the Indo-European field has been the origin of the feminine gender, from the inception of modern historical linguistics through the work of the Neogrammarians (e.g. Brugmann 76 As far as Latin is concerned, it is worth noting that masculine nouns belonging to the ā-declension denote animate male referents (Luraghi 2006: 91; Hoffmann 2016: 822).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

60

Chapter 2

1891). However, a special impetus for this continuing debate was provided by the decipherment of Hittite. In the following I lay out the principal evidence for the reconstruction of the two-gender system for the oldest phases of Proto-Indo-European and then address the problem of Hittite and Anatolian.77 The French Indo-Europeanist Antoine Meillet (1921: 211–229, 1931) questioned the three-gender system for the older stage of Proto-Indo-European. He proposed that the distinction between masculine and feminine within the “animate” gender was a recent innovation: “[a]u gendre animé, marqué par le masculin, avec une différenciation éventuelle pour le cas particulier du féminin, s’oppose le genre inanimé, le « neutre »” (Meillet 1921: 213). Meillet further observed that in many ancient Indo-European languages we can find pairs of words with similar meanings, but one is either masculine or feminine, while the other is neuter. This peculiar gender contrast in limited sections of the lexicon represents the preservation of a former state of affairs where an opposition between active/agentive and inactive/inagentive entities can be reconstructed. Meillet (1921: 228) first pointed out that an important feature distinguishing animate from inanimate is the ability of the referent to move, perform an action, and cause an event to happen. Confirmation of such a subdivision comes from the lexicon. Indeed, by comparing old Indo-European languages, we can find substantives that describe the same referent as a dynamic entity, on the one hand, and as a static entity, on the other hand. The first kind of substantives pertains to the masculine or the feminine gender and the second to the neuter gender.78 This is an important feature that characterises animate nouns with respect to the neuter in the ancient Indo-European languages, where polarisations of features, including individuation, may have played a role in gender assignment (Lazzeroni 1998c; Luraghi 2011). In this reconstructed system, animate and inanimate gender were prototypical categories, where a given noun could be assigned to a given gender even if it did not share all the features of a certain category (Luraghi 2014). From an inflectional point of view, a similar conclusion is suggested by the fact that, in archaic inflectional classes, masculine and feminine gender do not attest formal differences in declension (Pinault 2011b: 131). On the other hand, feminine nouns are often characterised by suffixation, being therefore more marked (Luraghi 2006: 93; Pinault 2011b: 130–131; Neri 2017: 24–26; Lundquist & Yates 2018: 2097). Furthermore, where there was a necessity to explicitly distinguish between masculine and feminine, different nouns were 77 See Matasović (2004: 165–181), Luraghi (2009b, 2011: 452–453, 2014), Pinault (2011b: 130–133), Lundquist & Yates (2018: 2094–2100). 78 On these pairs, see also Lazzeroni (1998a, 1998b, 1998c).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

The Gender System of Tocharian

61

used. Examples are numerous from the kinship lexicon (Meillet 1937: 189–191): Skt. mātár- ‘mother’ f. (< PIE *méh2tēr) : pitár- ‘father’ m. (< PIE *ph2tḗr); Gk. υἱύς ‘son’ m. (< PIE suHi̯u-) : θυγάτηρ ‘daughter’ f. (< PIE *dhugh2tḗr); Lat. frāter ‘brother’ m. (< PIE *bhréh2tēr) : soror ‘sister’ f. (< PIE *su̯ esōr). Etymologically, the last word may contain PIE *sor-, which can be found in Anatolian as an independent feminine suffix, e.g. Hitt. išḫaššara- ‘lady, mistress’ from išḫa‘lord’, Hitt. ḫaššuššara- ‘queen’ from ḫaššu- ‘king’ (Ledo-Lemos 2003: 133–135). Some nouns that can be reconstructed for the proto-language are clearly epicene, like PIE *gwóu̯ - ‘bovine’, which had either gender, depending on the sex of the referent (Pinault 2011b: 131). The marking of sex with higher animate beings is sometimes encoded by means of different types of noun phrases, which involve (Hackstein 2010: 53–64, 2013: 96): (1) pure lexical marking (Gk. ὁ θῆλυς ὀρεύς ‘the female.m mule’); (2) archaic lexical marking with the generic word for ‘male’ and ‘female’ (Lat. equus fēmina ‘mare [i.e. “female horse”]’, lupus fēmina ‘female wolf’, porcus fēmina ‘sow [i.e. “female pig”]’, agnus mās […] fēminaque ‘male and female lamb’, mās et fēmina aquila ‘male and female eagle’; Gk. ἄρσενες ἵπποι ‘male horses’; Ved. áśvāso vr̥ ṣ́ aṇo ‘male horses’); (3) morphosyntactic marking through agreement (Gk. ἡ θήλεα ἵππος ‘mare [i.e. “the.f female.f horse”)’, ὁ θεός ‘the.m god’, ἡ θεός ‘the.f goddess’). The formation of feminine nouns through the suffix *-eh2 > *-ā is acknowledged to be a later morphological process, which has been independently standardised in the Indo-European languages (e.g. Skt. áśva- ‘horse’ vs. áśvā- ‘mare’). Furthermore, it is a well-known fact that in the Indo-European languages the so-called thematic *o-stems are mostly masculine or neuter. Still, it is possible to reconstruct certain feminine *o-stems for Proto-Indo-European. A famous example is PIE *snusó- ‘daughter-in-law’, whose original stem has been retained only in Greek νυός, Armenian now (gen.sg. nowoy), Tocharian B santse (Peyrot & Meng 2021), and partially in Latin nurus, while in all other Indo-European languages the noun has been remodelled to *snuséh2- (cf. Skt. snuṣā́-, OE snoru and CS snъxa).79 Some old Indo-European languages have adjectival classes that exhibit only a two-way set of forms, making no distinction between masculine and feminine, which are condensed into one form. This is different from that of the neuter (cf. Lat. trīstis m./f. vs. trīste nt. ‘sad’, Lat. immortālis m./f. vs. immortāle nt. ‘immortal’, Lat. levis m./f. vs. leve nt. ‘light’; Gk. ἄδικος m./f. vs. ἄδικον nt. ‘unjust, wrong’, Gk. φορός m./f. vs. φορόν nt. ‘bearing’). These differences are only limited to the masculine and the neuter (see Wackernagel 2009: 458–463). 79 See Pinault (2011b: 131). The ū-stem of Lat. nurus, -ūs is analogical after socrus, -ūs ‘mother-in-law’ (de Vaan 2008: 420).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

62

Chapter 2

A similar situation is also observed in some pronominal stems, as in the indefinite and interrogative pronoun *kwi-, cf. Gk. τίς m./f. vs. τί nt., Lat. quis m./f. vs. quid nt. The distinct feminine forms attested in e.g. Sanskrit are thought to be a later development (Beekes 2011: 230; Lundquist & Yates 2018: 2097). Further evidence for the later development of the feminine gender is also said to be found in the paradigm of the demonstratives, which are considered to have played a key role in the development of the feminine gender (Luraghi 2011: 452; Lundquist & Yates 2018: 2097). Indeed, the paradigm *so (m.), seh2 (f.), *tod (nt.) shows that the feminine *-eh2 developed by simply substituting the thematic vowel -o- of the masculine. Thus, the original situation implied a contrast between the *s- of the animate and the *t- of the inanimate (Luraghi 2011: 452). This contrast, however, would only be marked in the nominative singular, as in the rest of the declension the masculine, the feminine, and the neuter all inflect on a stem *t-. In this context, the position of the Anatolian languages is rather difficult. Indeed, thanks to the decipherment of Hittite texts, it was discovered that the Anatolian language completely lacks a grammatical feminine gender, since it only displays a distinction between common gender and neuter gender, as in Hitt. kāš antuḫšaš ‘this man’ (common gender) vs. kī ḫuitar ‘this animal’ (neuter gender).80 There are two different hypotheses concerning the lack of the feminine gender in Anatolian (Luraghi 1998: 190–191; Bianconi 2021: 17–19). The Schwundhypothese claims that the feminine was not attested in Anatolian simply because this Indo-European branch lost this category value that consequently should be reconstructed for the proto-language. On the other hand, the Herkunftshypothese (whose most radical variant is the Sturtevant’s IndoHittite hypothesis) claims exactly the opposite, stating that the proto-language started to grammatically encode feminine gender only after the split of Anatolian from the rest of the Indo-European family.81 As a consequence, there would never have been a separate feminine gender in Proto-Anatolian. The current communis opinio in the field tries to accommodate both hypotheses (Kim 2018b; Melchert forth.): it is commonly acknowledged that 80

For a recent discussion on the feminine gender in Hittite and the functions of the suffix *-eh2 in Anatolian and in Proto-Indo-European, see Melchert (2014) and Sasseville (2018). Both scholars bring evidence for an individualising function of the suffix *-eh2 in Anatolian. This property would be older than the one marking the feminine gender. See further Fellner & Grestenberger (2016). According to Goetze (1960) and Kloekhorst (2019: 223–230), traces of *-ih2 used as a motion suffix can be found in Kanišite Hittite. 81 The so-called Indo-Anatolian hypothesis has recently been gaining support, particularly from the Leiden School. See Kloekhorst (2008: 7–11, 2019: 224–227, 2022) and Kloekhorst & Pronk (2019: 2–7).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

The Gender System of Tocharian

63

the Anatolian situation is archaic, but less archaic than previously thought. As pointed out by Olander (2015: 22), “[t]he difficult exercise is to point out which specific traits in Anatolian and non-Anatolian Indo-European are archaisms and which ones are innovations; on this point opinions are strongly divergent”. A recent study by Carling & Cathcart (2021) contains controversial proposals with respect to the question of the two-gender system of Proto-IndoEuropean. Applying phylogenetic methods, the authors have shown that it is more likely for Anatolian to have collapsed a Proto-Indo-European three-gender system than to have preserved an ancient two-gender system. As far as the Proto-Indo-European reconstruction is concerned, there are two sets of problems here. The first is about the number of category values to be reconstructed for the proto-language and how it developed. The second, which is linked to the first, is whether some Indo-European languages, including Anatolian, inherited a two-fold or a three-fold gender system from the proto-language. I think that the first problem is settled and there is indeed a broad consensus among scholars that the three-gender system known from most Indo-European languages replaced an earlier animacy-based two-gender system. The second issue has long been a source of significant debate and it still remains to be completely settled, although it is commonly believed that, as far as the gender system is concerned, the Anatolian situation is archaic.82 Most recent scholarship has therefore adopted the position that the three-gender system emerged out of a two-gender system and, accordingly, more attention has been paid to the vexed question of the origin of the feminine. The main topics of debate have revolved around how the animacy system split up into a sexus distinction; what the original function of *-eh2 was; the connection between abstract, collective, and feminine; and the problem

82 See Rieken (2009), Luraghi (2011), Pinault (2011b), Jasanoff (2017: 223), Kim (2018: 164), Lundquist & Yates (2018: 2094–2100), Kloekhorst & Pronk (2019), Kloekhorst (2019: 223–230), and Melchert (2014, forth.). On the so-called “i-mutation” in Luwian and Lycian, see Starke (1990: 85–89) and Oettinger (1987). Rieken (2005) has argued that ́ the i-mutation has nothing to do with either the devī-́ or the vr̥kī-suffix. See Norbruis (2021: 9–50). Furthermore, in a lecture held at the 54th Annual Meeting of the Societas Linguistica Europaea, Carling et al. (2021) showed that preliminary tests applied to the lexical and typological trajectory of Indo-European gender evolution indicate that Indo-European gender was at first strengthening, with an increase of gender values, and then weakening, with a merging of genders. As far as the position of Anatolian is concerned, the problem still remains: did Anatolian depart/isolate when the Indo-European proto-language was in the process of strengthening the gender system or in the process of weakening it?

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

64

Chapter 2

of the rise of the agreement.83 From an inflectional point of view, the original twofold system consisted only of the masculine and the neuter (mirroring the Anatolian common and neuter genders), while the feminine gender was later formed through the addition of special suffixes. There is little agreement about the details of this development and, in particular, on how the suffixes *-ih2/-i̯eh2 (“athematic”) and *-(e)h2 (“thematic”) started to mark the feminine gender. In this field, the position of Tocharian is open to questions that still need to be definitively answered. Indeed, it has recently been claimed that Tocharian departed from Proto-Indo-European immediately after Anatolian and that the unexpected distribution of the grammatical feminine markers would be strong evidence for such an early split (see e.g. Kim 2009a, 2014; Hackstein 2012; Kortlandt 2017, 2020). We will deal with this diachronic issue in the following chapters. 2.5

Summary and Conclusions

The goal of this chapter has been twofold: on the one hand, it has introduced the category of gender from the point of view of general linguistics; on the other hand, it has dealt with specific problems relating to the gender system of Tocharian from a synchronic perspective. After discussing some terms and concepts revolving around the linguistic notion of gender, we moved on to the core of the chapter, discussing the synchronic problems of the Tocharian gender system. We considered the linguistic analysis of the so-called genus alternans. Nouns pertaining to this category show a peculiar agreement, since they combine agreement traits 83

See Harðarson (1987, 2014), Tichy (1993), Matasović (2004), Pinault (2011b), Luraghi (2011), Hackstein (2012), and Nussbaum (2014a). Stang (1945) was virtually alone in assuming that the gender system of Proto-Indo-European was fourfold. The fourth gender that he suggested to add to the commonly assumed three is the collective. He based this view on the peculiar agreement pattern of the collective nouns ending in *-ā (< *-eh2), like in the Ancient Greek type ἕτερα καὶ ἕτερα ὕδατα (nom.pl.) ἐπιρρεῖ (3sg.) “sundry and different waters flow”. This peculiar agreement was first recognised by Schmidt (1889), but Stang interpreted it as a relic of the PIE fourth gender. This hypothetical fourth gender would have subsequently been reanalysed as a mere inflectional marker and would have given rise to the neuter plural ending in -a/-ā. It is generally assumed however that the collective was a category of number, rather than gender, reconstructing a four-way contrast for animate nouns (namely, singular, dual, count plural, and collective plural), while inanimate nouns completely lack count plural. See Harðarson (1987), Melchert (2000, 2011, 2014), Loporcaro & Paciaroni (2011: 391–397), Hackstein (2012), and Oettinger (2014).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

The Gender System of Tocharian

65

of the masculine and the feminine. In particular, they take masculine agreement in the singular and feminine agreement in the plural, so that the targets show only two distinct sets of forms, even though they stand in agreement with the same controller. I made a typological and cross-linguistic comparison with Romanian and Standard Italian in order to illustrate that the genus alternans must be regarded as a real category value (a non-canonical gender value). Furthermore, this investigation has raised some methodological points of interest in the domain of general and typological linguistic analysis to the notions of gender value, agreement class, and inquorate gender. I have also discussed terminological difficulties in the labelling of the third Tocharian gender, examining the difference between the alternating gender and the so-called “neuter” in Tocharian. We have then dealt with the gender assignment system of Tocharian, commenting on inflectional, derivational, and semantic patterns that allow us to infer the gender of Tocharian nouns. Finally, the last section of the chapter has been devoted to the reconstruction of the Proto-Indo-European gender system from a comparative perspective, highlighting that the feminine was a later innovation than the other two gender values.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Chapter 3

Gender in the Inflection of the Noun The present chapter aims to investigate the evolution of the gender system in the inflection of the Tocharian noun. The main focus is the origin and the development of the feminine and the alternating gender as well as their formal and functional differentiation with respect to the masculine. As a consequence, endings and forms of those inflectional types that may have been relevant in their evolution are considered. The masculine gender is treated in less detail since its development is generally better understood and its relevance to the evolution of the gender system mainly relates to its merger with the neuter. 3.1

Tocharian Nominal Categories

In the noun Tocharian distinguished three grammatical categories: case, number, and gender. These categories are typically expressed within a paradigm by means of endings which carry information about case and number, and sometimes gender. Like other ancient Indo-European languages, Tocharian has maintained three numbers: the singular, the plural, and the dual.1 As evidenced in the previous chapter, Tocharian has three gender values: the masculine, the feminine, and the alternating gender. The Tocharian masculine mostly continues the PIE masculine gender, the Tocharian feminine mostly continues the PIE feminine gender, and the Tocharian alternating gender mostly continues the PIE neuter gender. Nevertheless, the Tocharian genus alternans should be considered a separate category from the PIE neuter.

1 On the development of the endings and forms of the dual, see Hilmarsson (1989a) and Kim (2018a). Krause (1954, 1955: 23–24) claimed that two other values may be added to the number category, i.e. the “paral” (TB -ne, ta -ṃ) and the “plurative” (TB -aiwenta). He suggested that the paral served for natural pairs and the dual for accidental pairs (see also Carling 2000: 190). After the critical treatment of this analysis by Winter (1962a), it is now agreed that the paral is nothing but a dual marker, while the plurative, limited to just a few nouns, made countable and distributional plurals and cannot be considered as a “morphologically signaled category of inflection” (p. 117). For the decline of dual morphology in Tocharian B, see Kim (2018a: 15–25).

© Alessandro Del Tomba, 2023 | doi:10.1163/9789004532892_004 Alessandro

Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

67

Gender in the Inflection of the Noun

As compared to other Indo-European languages, one of the most striking peculiarities of Tocharian is the category of case. In both Tocharian A and B, the case system is structured in two tiers: a first level consists of the so-called “primary cases”, largely inherited from Proto-Indo-European; the second level consists of the so-called “secondary cases”, whose origin is still disputed.2 A scheme of the Tocharian nominal cases is as follows: Table 8

cases

Case system of Tocharian

Tocharian A and B

Tocharian A Tocharian B

Primary nominative, oblique, genitive(-dative) — Secondary locative, perlative, allative, instrumental comitative, ablative

[vocative] causal

For the most part, morphological factors determine the division into these two tiers: while the primary cases are fusional, the secondary cases are agglutinative. The secondary case suffixes are attached to the oblique form of nouns inflected for singular, plural, or dual, while the suffixes themselves are number-indifferent. The equivalent of the PIE accusative is usually termed oblique in Tocharian. Syntactically, it functions as the accusative in many other Indo-European languages; morphologically, it is the stem on which the secondary cases are built. Furthermore, Tocharian is renowned for the morphosyntactic phenomenon known as “Gruppenflexion”: in noun phrases, secondary case markers (and, to a lesser degree, the genitive marker) are usually added to the last nominal in the chain, while all the preceding ones are inflected in the oblique. The secondary case suffixes are assumed to be of late origin. Some can be traced back to Proto-Tocharian (i.e. locative, perlative, and allative), while others are independent innovations of each Tocharian language. Carling (2000, 2008, 2012, 2017: 1354–1355) has dealt thoroughly with their morphological structure, functions, and evolution. The secondary cases will not be treated 2 It is usually claimed that the origin of the secondary cases is to be ascribed to substratum influence of non-Indo-European languages (see mainly Krause 1951; K. H. Schmidt 1987 and 1990; Thomas 1994; Barbera 2000: 29–31; Peyrot 2019b). For a diametrically opposite proposal see Carling (2012), who has highlighted similarities between the evolution of the case system of Tocharian and Romani. She argues that the almost completely parallel formation of their case systems may serve as an argument in support of an internal development, without invoking a foreign influence.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

68

Chapter 3

in this book. Instead, I will focus on those case endings that prove relevant to the diachrony of gender. For this reason, I will only consider cases inherited from Proto-Indo-European, i.e. the nominative, the oblique, and the genitive(-dative). 3.2

Tocharian Noun Declension

As noted in the previous chapter, Wolfgang Krause and Werner Thomas, the authors of the Tocharisches Elementarbuch (teb), selected the plural morpheme as the criterion by which classify Tocharian substantives. This led to the identification of seven classes. Nonetheless, if we regard both the singular and the plural paradigm and all minor differences in inflection, the number of inflectional classes increases and around thirty types can be identified. This should not come as a surprise; if we considered all minor inflectional differences in the three declensions of Ancient Greek (Attic), we would get a number of inflectional types very close to that of Tocharian. Thus each class identified by Krause and Thomas can be divided into several subclasses that in turn comprise the Tocharian inflectional types. In the first three classes, we find nouns that mostly build their plurals by means of a suffix marker. See the following synchronic scheme:3 Table 9

teb classes i, ii, iii

Tocharian B

class i

class ii class iii

Tocharian A

pl. ending

example

pl. ending

example

-a -sa -wa -na -nma -nta —

cmel : cmela luwo : lwāsa ost : ostwa ñem : ñemna teki : tekanma āke : akenta —

-ā — -wā, -u -äṃ -mnā-nt -ntu

lu : lwā — cmol : cmolu ysār : ysāräṃ arkämnāyärk : yärkant tiri : tirintu

3 A slightly revised version of teb classes has been proposed by Hartmann (2013: 63–71). For an introductory diachronic account of these classes, see Pinault (2008: 491–497).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

69

Class i is poorly represented in both Tocharian A and B. It forms a closed category. The plural ending -sa is only found in three Tocharian B nouns (lwāsa ‘animals’, piltāsa ‘petals, leaves’, lyyāsa ‘limbs’) and it has no formal correspondent in Tocharian A. Note that very often a noun does not belong to the same class or subclass in Tocharian A and B (cf. TB pl. lwāsa vs. TA pl. lwā). The ending TB -wa, ta -wā, -u is more productive than TB -a, ta -ā. Indeed, loanwords are occasionally inserted into this class. Examples include: TB kottär (pl. kottarwa) ‘family’ from Skt. gotra- ‘family, clan’ and TB tsain (pl. tsainwa) ‘arrow’ from OSIr. *dzai̯nu- ‘weapon’ (cf. Av. zaēnuš ‘baldric’, Av. zaēna‘weapon’). TB kottär /kóttər/ has been added to this class because of its formal resemblance to other members of the wa-class, like TB āmpär* ‘limb’,4 TB kwarsär, ta kursär ‘league’, TB tsaṅkär, ta tsäṅkär ‘summit, top’, etc., all ending in final -är /-ər/ (see § 3.6.3.2). The ending TB -nma (class ii.2) is productive.5 It derives from PIE *-mn-h2 by regular metathesis of *-mn- to -nm- in Tocharian B (Pinault 2008: 449). It is the plural marker of both inherited nouns and loanwords of Indo-Aryan (cf. kālp ‘eon’ from Skt. kalpa-), Iranian (cf. sāñ ‘skill, means’ from Khot. saña- ‘expedient, means’), and Chinese origin (cf. cāk ‘hundred quarts [dry measure]’ from MChin. *dzyek > shí 石 ‘stone; dry measure’).6 In Tocharian A, the expected ending *-mnā has been preserved in the adjective TA arkämnāṣi ‘pertaining to the burial places’ from *arkämnā ‘burial places, cemeteries’ (cf. TB erkenma ‘id.’). Indeed, nouns that are expected to show this ending have usually added the marker TA -n-t(u) to the plural form (cf. TB nakanma, ta nākmant ‘faults, errors’; TB wakanma, ta wākmant ‘distinctions, superiorities’, Pinault 2008: 495). On the other hand, the Tocharian B ending -na is usually considered to be matched in Tocharian A by -äṃ (class ii.1). However, the Tocharian B counterparts of nouns with the plural ending TA -äṃ belong to different classes. Furthermore, nouns with pl. -na form two well-differentiated subclasses in Tocharian B: (1) alternating nouns with no differentiation between nominative and oblique in the singular; (2) feminine nouns with differentiated nominative and oblique in the singular. The Tocharian A equivalents of subclass (2) belong to other inflectional classes with differentiated nominative and oblique plural.7 4 The singular of this noun seems not to be attested. The plural amparwa /ampǝ́rwa/ is attested twice in NS32 b1 and b2 (edition of the manuscript by Pinault 2012b). 5 See Hilmarsson (1991b: 152–155) for a subdivision of the Tocharian B class with plural -nma. 6 See Adams (dtb 164) on TB kālp, Del Tomba & Maggi (2021) on TB sāñ, and Lubotsky & Starostin (2003: 264) and Blažek & Schwarz (2017: 37) on TB cāk. The Middle Chinese reconstruction follows Baxter & Sagart (2014), with dzy- [dʑ]. 7 For both a synchronic and a diachronic discussion of TB -na, ta -äṃ, see the relevant paragraphs in section § 3.6.3.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

70

Chapter 3

In both Tocharian A and B, class iii is the most productive, assimilating most loanwords of Indo-Aryan origin. Krause & Thomas (teb  §§  167–173) divided it into subgroups on the basis of the vowel preceding the plural ending. Thus, we have: TB -enta, ta -ant; TB -onta, ta -ant; TB -ānta /-ánta/; TB -anta /-ə́ nta/, -änta /-ənta/, ta -äntu; TB -inta, ta -intu; TB -unta. In synchronic terms, the difference between these endings is easy to explain: the plural -nta is directly attached to the basic stem of the singular form, which can in turn end with all the aforementioned vowels. It follows that the singular has a zero morpheme, and the plural ending is just -nta. In Tocharian A, we find -ntu as an extended variant of TA -nt. It has become the most common plural ending for alternating nouns and has no formal counterpart in Tocharian B. As pointed out by Sieg, Siegling & Schulze (tg §§ 134–136; cf. also Pinault 2008: 497), the plurals in TA -nt have an allomorph -ntw- when constructed with suffixes of the secondary cases (cf. klopant ‘pains, sorrows’, loc.pl. klopäṃtwaṃ ‘in pains’), but not in the instrumental, where the nt-stem is maintained (cf. klopantyo ‘by pains’). The origin of the nt-plural is debated.8 Furthermore, in each of the classes outlined so far, we occasionally find nouns inflected only in the plural (pluralia tantum), e.g. mīsa ‘flesh’, ersna ‘form’. The remaining classes differentiate the nominative from the oblique in the plural. Table 10

teb classes iv, v, vi, vii

Tocharian B

class iv (class i) class v class vi class vii

Tocharian A

pl. ending

example

pl. ending

example

-ñ | -ṃ -a — -i | -ṃ -ñ | -ṃ -ñc | -ntäṃ

pātärñ mācera — yakwi, -eṃ riñ, -iṃ lāñc, -ntäṃ

-i | -äs — -e | -es -i | -äs -ñ | -s -ṃś | -ñcäs

pācri, -äs — pracre, -es akṣari, -äs riñ, -is lāṃś, -ñcäs

8 See Melchert (2000), who compared it with the “individualising” Anatolian suffix -ant-.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

71

Class iv consists of kinship terms that are regularly derived from PIE r-stems. They include: TB pācer, ta pācar ‘father’; TB mācer, ta mācar; TB tkācer, ta ckācar ‘daughter’; TB procer, ta pracar ‘brother’; TB ṣer, ta ṣar ‘sister’. The expected continuant of the nom.pl. PIE *-es vanished, and it seems that the Tocharian languages independently marked this case again, with the abolition of the expected *pacərə < PIE *ph2téres, etc. In Tocharian B, we have variant forms, e.g. nom.pl. tkātärñ vs. tkacera or pātärñ vs. pacera. On the basis of the textual distribution of the forms and the phonological shape of the stem, Peyrot (2008: 112–114) demonstrated that the cera-plurals are the latest innovation. However, how exactly the ending -a was probably introduced after pl. säsuwa ‘sons’, klaina ‘women’, and other feminine kinship terms is still a matter of debate. TA nom.pl. -e, obl.pl. -es is only found in the word for ‘brother’. Class v can be divided into three major subclasses. The first and most productive contains Tocharian B e-stems (nom.obl.sg. -e, the yakwe-type). In Tocharian A, the final vowel was regularly dropped. It is generally agreed that these nouns continue the PIE masculine thematic inflection (i.e. the PIE *o-stems). The nom.pl. TB -i is indeed the regular outcome of PIE *-oi̯ (see § 4.8.1). In Tocharian A, the expected continuant of this ending (TA †-e) seems to have been replaced by -añ (cf. *h1éḱu̯ oi̯ > TB yakwi ‘horses’, but TA yukañ). The obl.pl. is -eṃ in Tocharian B and -as in Tocharian A. Loanwords referring to human (male) beings usually belong to this class (e.g. TB ar(a)hānte ‘arhat’ from Skt. arhant- ~ ar(a)hant(a)- [bhsd 67; swtf 1.150–151]; TB winasāre ‘expert in monastic discipline’ from a Middle Indic intermediary of Skt. vinayadhara- [Pinault 1987: 143; von Hinüber 2001: 153; bhsd 489]). Another subclass inflects in a slightly different way: in Tocharian B, its members have a zero-marked oblique singular, palatalisation of the stem throughout the inflection, and obl.pl. TB -äṃ (cf. TB meñe, ta mañ ‘moon, month’, obl.sg. TB meñ, ta mañ, nom.pl. TB meñi, ta mañi, obl.pl. TB meñäṃ, ta mañäs). In addition, a group of Tocharian B nouns inflects like the previous subclass, but the obl.pl. TB -äṃ is not palatalising (cf. TB āśce ‘head’, nom.pl. āści, obl.pl. āstäṃ). A final class also has palatalising nom.pl. -i and non-palatalising obl.pl. TB -äṃ, ta -äs, but its members end with a non-palatalised consonant in the singular (cf. nom.obl.sg. TB kauṃ, ta koṃ ‘sun, day’, nom.pl. TB kauñi, ta koñi, obl.pl. TB kaunäṃ, ta konäs). Class vi is very productive and can be divided into a large number of subclasses. In Tocharian B, we find the following types:

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

72 Table 11

Chapter 3 Inflectional types with nom.pl. -ñ in Tocharian B

kantwo-type okso-type arṣāklo-type ymiye-type kälymiye-type wertsiya-type śamaśke-type saswe-type prāri-type

nom. sg.

obl. sg.

nom. pl.

obl. pl.

-o -o -o -iye -iye -ya -e -e -i

-a -ai -ai -ai -i -yai -e(ṃ) -e(ṃ) -i

-āñ -aiñ -añ -aiñ -iñ -yañ -añ -eñ -oñ

-aṃ -aiṃ -aṃ -aiṃ -iṃ -yaṃ -aṃ -e(nä)ṃ -oṃ

In light of the many similar endings and forms, it is reasonable to assume that some nouns shifted between these subclasses during the development of the nominal declensions, both in the prehistory of Tocharian B and in Proto-Tocharian. In Tocharian A, the identification of the inflectional classes is easier. We basically find the following plural forms: (1) nom. -añ, obl. -as; (2) nom. -āñ, obl. -ās; (3) nom. -iñ, obl. -is; (4) nom. -eñ, obl. -es. A convenient synchronic mechanism identified by Sieg, Siegling & Schulze (tg  §  146) highlights the fact that when a given noun ends with a vowel in the singular, the plural form quite often repeats that final vowel (cf. TA ri ‘city’, nom.pl. riñ; TA poke ‘arm’, nom.pl. pokeñ); on the other hand, when a given noun ends with a consonant in the singular, the vowel in the plural form varies (cf. TA olar ‘fellow, companion’, nom.pl. olariñ), although it usually belongs to those types with plural -añ or -āñ. From a diachronic perspective, the first type (nom.pl. -añ, obl.pl. -as) usually corresponds to the Tocharian B e-stems (teb class v.1); the second type (nom.pl. -āñ, obl.pl. -ās) corresponds to nouns belonging to class vi in Tocharian B (cf. TA oṅkaläm ‘elephant’, nom.pl. oṅkälmāñ vs. TB oṅkolmo, nom.pl. oṅkolmañ). However, there are significant exceptions. Indeed, it is important to note that feminine nouns referring to female entities always belong to this subtype with nom.pl. -āñ, obl.pl. -ās (with the exception of TA lānts ‘queen’, whose plural varies; see § 3.4.1.2). The Tocharian B equivalents of these feminine nouns belong to class ii (pl. -na).9 9 For a detailed overview of the plural ending -ñ and its various inflectional types in Tocharian A, see tg §§ 146–156 and §§ 226–240.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

73

Lastly, we have class vii, which is the least productive. The most prominent member is TB walo, ta wäl ‘king’ (pl. TB nom. lāñc, obl. lāntäṃ, ta nom. lāṃś, obl. lāñcäs). In Tocharian A, this inflectional class is apparently limited to this noun. In this book I will not deal with all of these classes; rather, only with those relevant to the diachronic analysis of the gender system. They are outlined in the following paragraphs. 3.3

Aim and Structure of the Chapter

The three pivotal questions this chapter addresses are (1) how the PIE feminine gender evolved in the inflection of the Tocharian noun, (2) how the PIE neuter gender evolved in the inflection of the Tocharian noun, and (3) whether the PIE neuter gender is continued as the Tocharian genus alternans. These three questions lead to other minor issues about the marking of alternating and feminine nouns from both a synchronic and a diachronic perspective, and, in general, about the consequences of the morphophonological mergers of the three inherited genders in the system of the noun. This last issue and the overall development of the Tocharian gender system will also be investigated in the next chapter. In the present chapter, we mostly concentrate on specific inflectional and derivational problems linked to the marking of alternating and feminine nouns from the perspective of comparative Indo-European linguistics. In order to understand how the PIE feminine gender evolved in Tocharian nouns, I investigate the Tocharian inflectional classes that may continue the four PIE types that are significant to the evolution of the feminine gender in Indo-European languages: (1) the non-ablauting *eh2-type; (2) the proterodynamic *h2-type; (3) the hysterodynamic *h2-type; (4) the ablauting *ih2-type ́ (the so-called devī-type); (5) the non-ablauting *ih2-type (the so-called ́ vr̥kī-type).10 In order to understand whether the PIE neuter is continued as the Tocharian alternating gender, I identify potential reflexes of the PIE thematic neuter and clarify how this reconstructed class has developed in Tocharian. Continuants of the athematic type also constitute the subject of my investigation, though I will limit my attention to those types whose origin may have been overlooked

10

For an introduction to these inflectional types from the perspective of reconstructed Proto-Indo-European, see Beekes (2011: 199–201, 204), UrIG § 1b. p. 2, § 3b. pp. 8–10, § 3c. pp. 10–11, § 3e. pp. 13–15, § 3g. pp. 16–17, § 3h. 17–19, and the relevant sections below.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

74

Chapter 3

but that have nevertheless played an important role in the evolution of the gender system. Among the teb inflectional types outlined above, there are some that are more relevant than others when carrying out an investigation into the development of the Tocharian gender system. They are the subject of this chapter. The Tocharian A classes are simpler, because the Proto-Tocharian word-final vowels *-a, *-e, and *-o have been lost in this language. For this reason, I mostly refer to Tocharian B when individuating and naming these types. Nonetheless, evidence from Tocharian A is consistently considered and analysed in tandem with that of Tocharian B. To keep track of the various types we will deal with in this chapter, below I give a reference matrix of the inflectional classes on which we will mostly — but not exclusively — concentrate our attention. The declensions below are simplified and are limited to the nominative and oblique singular and plural: Table 12

Reference list of the inflectional classes of Tocharian under investigation

śana-type aśiya-type kantwo-type okso-type arṣāklo-type wertsiya-type yakwe-type āke-type palsko-type

TB TA TB TA TB TA TB TA TA TB TA TB TA TB TA TB TA TB TA

nom.sg.

obl.sg.

nom.pl.

obl.pl.

-a -∅ -(y)a -(y)i ~ -∅ -o -∅ -o, -iye -∅ -e -o -∅ -ya, -yo -(y)i ~ -∅ -e -∅ °e -∅ °o -∅

-o -∅ ~ -āṃ -(y)ai -(y)āṃ -a -∅ -ai -∅ -e -ai -∅ -yai -(y)i ~ -∅ -e(ṃ) -∅ °e -∅ °o -∅

-ona -āñ ~ -añ -(y)ana -(y)āñ -āñ -āñ -aiñ -āñ -eñ -añ -āñ -yañ -yāñ -i -añ °enta -ant °onta -ant

-ona -ās ~ -as -(y)ana -(y)ās -aṃ -ās -aiṃ -ās -es -aṃ -ās -yaṃ -yās -eṃ -as °enta -ant °onta -ant

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

75

In addition, all nouns with nom.obl.pl. TB -na, ta -äṃ will be analysed (class ii.1). To achieve the aims of this chapter, etymological analyses are needed to understand the origin of the inflectional endings of the various nouns considered. Furthermore, in order to apply the comparative method, it is crucial to exclude borrowings from our comparisons and reconstructions. For this reason, the various etymological sections contained in this chapter will also highlight whether certain nouns belonging to the selected inflectional classes are either inherited or borrowed. If a large number of loanwords are attested into a given inflectional class this will also be used as an indicator of the productivity of the relevant inflectional class within a given timeframe (both prehistoric, i.e. reconstructed, or historical, i.e. attested). For each of the inflectional classes identified, I analyse the paradigm of the singular and the plural in both Tocharian languages in order to verify where the comparison between Tocharian A and B allows us to reconstruct Proto-Tocharian structures straightforwardly, and where they do not match. In this latter case, new problems will of course come to light and an explanation will be attempted for each. It will then become clear that some of these inflectional types have similar or equivalent characteristics, since they attest nominative and/or oblique endings that are often the same. As a consequence, it may be assumed that some of these classes influenced each other over the prehistory of the two Tocharian languages, i.e. in a Proto-Tocharian phase and then independently in Tocharian A and B. Although synchronic analyses are sometimes necessary, the main approach of the investigation is diachronic. The chapter can be divided into two parts: in the first part (§ 3.4 and § 3.5) I deal chiefly with the development of the feminine gender, while in the second part I investigate the development of the neuter gender (§ 3.6). In § 3.4, the evolution of feminine nouns denoting female referents is investigated (śana-type and aśiya-type). Some of the inflectional types from class vi are analysed in § 3.5 (kantwo-type, okso-type, arṣāklo-type, wertsiya-type). Each one of these types contributes to a better understanding of the evolution of the feminine gender in Tocharian. In the first sections of the second part (§  3.6.1, 3.6.2), an overview of the development of the thematic neuter is offered; furthermore, morphophonological confluences between the masculine, the feminine, and the neuter are investigated. In § 3.6.3, I discuss the origin of the plural endings TB -na and TA -äṃ, which played an important role in the evolution of both the feminine and the alternating gender. A short summary of the main findings concludes the chapter (§ 3.7).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

76 3.4

Chapter 3

Nouns Denoting Female Entities

This section aims to trace the development of two closely related inflectional classes of feminine substantives, whose plural formation ends in TB -na, as well as their Tocharian A matching nouns and forms. I will discuss problems about their inflection and highlight their central role in the evolution of the Tocharian feminine gender. All these grammatically feminine nouns share a core semantic feature: they denote female referents. From the point of view of their paradigm, they can be grouped into two main classes: (1) the śana-type, with the following inflection (exemplified with TB śana, ta śäṃ ‘wife’ and TA lānts ‘queen’): Table 13

Inflection of the śana-type

singular

nom. obl. gen.

plural

TB

TA

TB

TA

-a śana -o śano -oy śnoy

-∅ lānts -∅ ~ -āṃ lānts ~ lāntsāṃ -e lāntse

-ona śnona -ona śnona —

-āñ ~ -añ śnāñ* ~ lāntsañ -ās ~ -as śnās ~ lāntsas —

(2) the aśiya-type, with the following inflection (exemplified with TB aśiya ‘nun’, ta aśi ‘id.’):11

11

Note that TA -śś- is an inner-Tocharian A development of -śy- between vowels (cf. also the obl.sg.f. variants -ṣṣāṃ ~ -ṣyāṃ in the inflection of Tocharian A ṣi-adjectives, see § 4.7.1.3 and § 4.7.1.4).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

77

Gender in the Inflection of the Noun Table 14

Inflection of the aśiya-type

singular

nom. obl. gen.

plural

TB

TA

TB

TA

-(y)a aśiya -(y)ai aśiyai -(y)antse aśiyantse

-(y)i, -∅ aśi -(y)āṃ aśśāṃ* -(y)e aśśe

-(y)ana aśiyana -(y)ana aśiyana -(y)anaṃts aśiyanaṃts

-(y)āñ aśśāñ -(y)ās aśśās* -(y)āśśi aśśāśśi

Another feminine noun with the na-plural in Tocharian B is the word for ‘woman’, TB kliye, ta kuli. This noun forms a separate inflectional class by itself. Its paradigm is irregular and has several variant forms in some cases: nom.sg. TB kliye ~ klyiye, ta kuli, obl.sg. TB klaiṃ ~ klai ~ klaiñ, ta kule, nom.obl.pl. TB klaina, nom.pl. ta kulewāñ, obl.pl. kulewās. The etymological and morphological difficulties connected to this word have been the subject of a very long debate. Proposals about its origin have been made by several scholars (Pedersen 1925; Schmidt 1980: 409–410; Kortlandt 1988b; Hilmarsson 1996: 157–159; Blažek 2005; Pinault 2005; Adams dtb 242–243). However, none of the etymologies proposed is in my opinion fully conclusive and I myself am yet to identify a convincing source and derivation. The reader is referred to Peyrot (2008: 106–107) for the explanation of most of the variant forms, and to Pinault (2005) and Kortlandt (1988b) for some etymological proposals, the last one ultimately based on Schmidt (1980). As can be seen from the tables above, the Tocharian A nouns do not share the same inflection as that of Tocharian B. This mismatch is peculiar and requires an explanation. For this reason, in the following paragraphs I discuss the endings of the primary cases of these classes in order to outline their historical evolution from Proto-Indo-European to Tocharian. 3.4.1 The śana-type: TB Nouns with nom.sg. -a, obl.sg. -o and their TA Correspondents The analysis of the śana-type has proved to be a controversial topic. As we will see, the debate has focused in general on the paradigm of the singular, and in particular on the opposition between nom.sg. -a and obl.sg. -o. My aim is to

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

78

Chapter 3

understand if these nouns inherited their paradigm from Proto-Indo-European or if some analogical developments need to be postulated. Before going into these diachronic matters, however, some preliminary synchronic remarks will be made. 3.4.1.1 Members of the śana-type and Synchronic Problems The śana-type is not a productive class as it includes only three nouns: TB śana, ta śäṃ ‘woman, wife’, TB lāntsa, ta lānts ‘queen’, and TB ṣarya ‘(beloved) lady’. Inflected forms of the first two substantives are well attested; the latter has no equivalent in Tocharian A and it is attested only in the vocative and in the nominative singular in Tocharian B. However, on the basis of the comitative form TB ṣaryompa, attested once in B496 a3–4, we can infer the obl.sg. ṣaryo*. Tocharian B śana and lāntsa are matched in Tocharian A by śäṃ and lānts (frequently spelled lāṃts, as in e.g. A324 b4, YQ III.7 a8). Tocharian A lānts has two allomorphs of the oblique singular: besides the common lāntsāṃ (e.g. lāṃtsāṃ in YQ III.5 b8, perl.sg. lāntsānā in A78 b1, all.sg. lāntsānac in A78 a6), we find isolated cases of obl.sg. lānts (e.g. lā(ṃ)ts in A94 a5 and abl.sg. lāntsac in A319 b7). Since TA -āṃ represents the ubiquitous feminine oblique marker in both nouns and adjectives, it is reasonable to assume that TA lānts is the archaic form (cf. obl.sg. TB lāntso). In addition, the obl.sg. lāntsāṃ should be regarded as a later form because it does not show vowel weakening, which, in turn, has probably affected the plural paradigm of this noun: nom.pl. lāntsañ, obl.pl. lāṃtsas may derive from older nom.pl. *lāntsāñ, obl.pl. *lāntsās (see below). The plural paradigm of TA śäṃ presents a special problem. Indeed, besides the expected obl.pl. śnās, this noun is supposed to have a plural śnu. Since Sieg, Siegling & Schulze (tg § 179.c), this TA śnu is unanimously interpreted as a nominative plural. Winter (1985a: 262) argues that TA śäṃ had two parallel plural paradigms: (1) TA śnu (nom. = obl.) < *śənwa- had a collective meaning, while (2) TA nom. śnāñ*, obl. śnās was the regular “countable” plural. This is an ad hoc explanation, however. One could imagine śnu having been analogically created after the plurals TA sewāñ ‘sons’ (cf. TB säsuwa ‘id.’) and kulewāñ ‘women’, but the absence (or loss) of final -āñ in the nominative plural is still problematic.12

12

One might think that this śnu maintained the original situation prior to the addition of final -ñ (cf. TB säsuwa vs. ta sewāñ) but see the main text below. We cannot accept Čop’s (1975: 4) interpretation of final -u in TA śnu as the regular outcome of PIE *-ās.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

79

I have found only two attestations of TA śnu, and both are from passages with considerable problems of interpretation.13 The first is in A299 b2 ///p prāmne śnu ᛫ brahmavatiṣiṃ śriññäktes kātsaṃ cmolu nutässi cmol eṃtsäṣtär || “… the śnu of a brahmin. In order to make the rebirths disappear, he takes birth in the womb of the Śrīdeva of a Brahmāvatī” (cf. Peyrot & Semet 2016: 367). This leaf preserves the end of the 10th act of the Tocharian A Maitreyasamitināṭaka, which was translated into Old Uyghur as the Matrisimit. Unfortunately a Uyghur parallel of the Tocharian A passage is missing, and we therefore lack any external clues that could help to properly translate TA śnu. If śnu is a nominative, its position at the end of the sentence, before the punctuation, is surprising and invites caution. Furthermore, compositions in other languages dealing with the legend of the Buddha Maitreya do not mention that Brahmāyu (or Subrāhmaṇa), the father of Maitreya, has more than one wife.14 Reference is made only to Maitreya’s divine mother Brahmāvatī. A second attestation is in A86 a4, which is very fragmentary: ///tvāp śnu mā tās·āṃ ///. The restorations of the gen.sg. (bodhisa)tvāp at the beginning of the line and tās(km)āṃ ‘like, as’ at the end are quite certain. However, the translation of the line is still puzzling (/// (bodhisat)tvāp śnu mā tās(km)āṃ /// “… not like the śnu of the Bodhisattva …” (?)). Thus the contexts do not necessarily indicate that śnu is a nominative plural. No nominal modifiers or inflected verbs agree with this form. Alternative hypotheses can be advanced, though they are still inconclusive.15 I therefore believe there is no secure evidence for considering TA śnu an inflected form of TA śäṃ ‘wife’. Before proceeding further with the historical analysis of these nouns, let us return to Tocharian B, since there remains another controversial substantive which is said to be a member of the śana-type. It is a famous hapax legomenon 13

According to tg § 164 and Itkin (2019: 196), there is a third attestation of this form in THT 1560 b3, though the word division in uncertain. 14 Cf. e.g. the Maitreyavyākaraṇa as well as the Maitreyāvadāna in the Divyāvadāna. Cf. also the Khotanese version of the Maitreyasamiti preserved in chapter 22 of the Book of Zambasta: Z 22.164 subrahmä nāma hämäte brraṃmani päte mättrai balysä brahmānä māñäte dätena brrahmāvata nāmai māta “Subrahma by name, a brahmin, will be father of the Buddha Maitreya. He will resemble Brahma in appearance. Brahmāvatī by name will be his mother” (Emmerick 1968a: 314–315). See also the unpublished conference paper by Kumamoto (2009). 15 One could indeed claim that TA śnu is the nom.sg. of a u- or nu-adjective (e.g. yäslu ‘enemy’, lukśanu ‘shining’), or an inflected form of the otherwise dual śanweṃ ‘(two) cheeks’, from PIE *ǵenu-. However, both solutions are very tentative. Note that the a-vocalism of śanweṃ for expected **ś(ä)nweṃ is probably due to analogy after kanweṃ ‘knees’ (as Michaël Peyrot p.c. pointed out to me).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

80

Chapter 3

attested as an apparent oblique singular in the archaic document B275. The traditional reading of line b4, where the noun is attested, is as follows: tkātre petso aiṃ-ñ cai śāmnā (Peyrot 2008: 98; Kim 2009b: 113 fn. 6; Hartmann 2013: 161). According to this reading and division, the passage contains two hapax legomena: the first is our noun TB petso (equated with TA pats ‘husband’); the second is tkātre, a morphological hapax, if an archaic genitive singular of TB tkācer ‘daughter’ derived from PIE *dhugh2tr-ós (Gk. θυγατρός, Skt. duhitúḥ, OLith. dukterès). The genitive singular of this noun is indeed expected to have been tkātri* (cf. gen.sg. pātri from TB pācer ‘father’, gen.sg. mātri from TB mācer ‘mother’, gen.sg. protri from TB procer ‘brother’). A new look at this passage has been offered by Pinault (2010), who divided the sequence tkātre petso as tkātr epetso, with tkātr as a sandhi-variant of the obl.sg. tkātär, and epetso as the obl.sg. of an unattested noun TB epetsa* ‘fiancée’ (cf. also Pinault 2019: 97). The entire passage would be translated as follows: “The people will give their daughter as a fiancée’”. This reading has two important advantages: first, it does not suppose the irregular gen.sg. †tkātre; second, it makes the translation of the document more coherent with the Old Khotanese parallel passage in the Book of Zambasta.16 Z 22. 123 haṣṭātä ysāre salī jsīna uysorāṇu ttu kālu paṃjsa-satä-saluvo anū—ḍo māta päte kṣundai heḍä “The life of beings at that time will be eighty thousand years. A mother, father will give to a husband their five-hundred-year-old daughter as yet unmatured.” (cf. Emmerick 1968a: 307) Pinault’s analysis of TB epetso as the obl.sg. of epetsa* received broad consensus (cf. Malzahn 2011: 89–90 fn. 14; Fellner 2014a: 8; Hackstein 2017: 1320; Weiss 2018: 375). Although the new reading of the passage is certainly correct, I think that the hapax legomenon TB epetso should be considered as an adverb with the meaning of ‘in marriage’ (see Peyrot 2013a: 663 fn. 45), which has been built on the original oblique singular of the equivalent of TA pats ‘husband’ (< PT *petsǝ, cf. Skt. páti- ‘lord, master; husband’, Lat. potis ‘able, capable’, Gk. πόσις ‘husband’). According to this analysis, the final -o of epetso is due to the so-called mobile -o (“bewegliches o”), which is fairly common in metrical passages (cf. śauwlo for TB śaul ‘life’ at the same line of epetso, and nom.pl.m. poñco for poñc ‘all’ at line b5). Although deriving adverbs from substantives is not a productive development in the historical phase of the Tocharian languages, there is good evidence that it was in Proto-Tocharian (Adams 2015: 16

See already Peyrot (2013a: 663 fn. 45). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

81

172). Furthermore, very often a new adverb is formed with a prefix e(n)-, as in this case, which could have had either an intensive or a locative value. In this case, the adverb would mean ‘in husband’ → ‘in marriage’ (cf. TB elauke ‘far’, from e(n)- + lauke ‘remote, far’; TB eweta ‘in conflict (with)’, from e(n)- + weta ‘battle’; TB eṣe ‘together’, from e(n)- + ṣe ‘one’), and the expression TB epets* ayshould be translated as ‘to give [someone] to a husband; to give [someone] in marriage’ (cf. Khot. kṣundai hor- ‘to give to a husband’ in the aforementioned parallel from the Book of Zambasta). I have therefore not included this word in the śana-type. 3.4.1.2 Diachronic Analysis of the śana-type In this section I deal with the etymologies of each noun of the śana-type. I then analyse their problematic endings and forms in order to trace their history and derivation from Tocharian to Proto-Indo-European. TB śana, ta śäṃ ‘wife’ TB śana and TA śäṃ are the most prominent members of this class. They evidently go back to the PIE word for woman, *gwénh2 / *gwn-éh2-. This noun originally adhered to the proterodynamic inflection: Table 15

Proterodynamic paradigm of PIE *gwénh2-

case

R

S

E

‘wife’

nom.sg. gen.sg. acc.sg.

é — é

— é —

— — —

*gwénh2 *gwn-éh2-s *gwénh2-m

Leaving aside for the moment the outcome of this noun in Tocharian and looking at the other Indo-European languages, this noun exhibits three trends in its later development, as summarised below: (1) conservation of the PIE paradigm, as in OIrish bé ‘woman’ < *gwenh2-, gen.sg. mná < *gwneh2-s and Arm. kin, instr.sg. knaw.17 In Indo-Iranian the two PIE stems split into doublets, cf. the i-stem Ved. jáni- ‘wife, woman’, OAv jə̄ni- (YAv. jaini-) < *gwenh2-, and the ā-stem Ved. gnā ́- ‘wife, goddess’, OAv. gənā- ‘(heavenly) woman’ 17 It seems probable that OIrish bé is actually from *gwēn, while the feminine OIrish ben reflects a remarked nom.sg. PCelt. *benā > OIrish ben (Jasanoff 1989; Zair 2012: 223–224). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

82

Chapter 3

(YAv. ɣənā-) < *gwneh2- (Harðarson 1987: 130; EWAia 1.503–504 and 569–570; AiGr 3.113 and 137; Hoffmann & Forssman 2004: 97, 123); (2) generalisation of one of the two stems, as in Greek, cf. γυνή, Dor. γυνά, Beot. βανά (cf. possibly Myc. dat.pl. ku-na-ki-si and the derived adjective Myc. ku-na-ja /gunaiā/ ‘feminine’, a Pylos hapax found in the description of a vessel) < *gwn̥ eh2- (gew 1.334–335; Chantraine 1999: 242–243; Beekes 2010: 291–292);18 (3) generalisation of the full grade in both the stem and the suffix, as in OCS žena, OPrussian genno. In Germanic, *kwenō < *gwenā is the basis of the n-stem *kwenō(n) (cf. Goth. qino, OHG quena). For Tocharian, two elements are relevant: (1) the consonant ś- as the outcome of a palatalised (labio)velar; (2) the endings nom.sg. śan-a, obl.sg. śan-o, and the plural stem śno-. TB śan- and TA śäṃ point to PT *śən-, which in turn is the regular outcome of PIE *gwen- (strong stem). This means that some analogical levelling of the root took place in the prehistory of this word, since we do not have any alternation between palatalised labiovelar (*śən- < *gwen-) and unpalatalised labiovelar (*kwən- < *gwn-) in Tocharian. However, it is not entirely clear if this generalisation took place in a Proto-Tocharian phase or if it should be reconstructed for an earlier stage. If we opt for the second hypothesis, then the development of TB śana, ta śäṃ could have been parallel to that seen in Slavic, where the full grade was generalised, and the word became a non-ablauting *ā-stem (e.g. OCS žena < *gwenā). This analysis is supported by some scholars, including Winter (1981: 938), Ringe (1996: 94–97), Adams (dtb 677), and Kim (2009a: 78). Accordingly, the diachronic evolution of the singular paradigm would have been as follows: nom.sg. *gwenh2 >> *gweneh2 > PT *śəna > TB śana, ta śäṃ; acc.sg. *gwenh2-m >> *gweneh2-m > PT *śəno > TB śano, ta śäṃ. The problem with such an analysis is twofold. On the one hand, no other Tocharian continuant of an *ā-stem has a singular inflection with nom.sg. -a, obl.sg. -o, adjectival and pronominal inflections included.19 On the other hand, 18

The inflection of Gk. γυνή shows allomorphy. The stem γυνή(-) is attested only in the nominative and in the vocative, and the stem γυναικ- in all other cases (though a number of variant forms exist, including acc.sg. γυνήν, nom.pl. γυναί, acc.pl. γυνάς). The origin of the κ-stem is debated. The common view involves a comparison between Gk. γυναικ-, OPhr. nom.sg. knais, nom.pl. knạỵḳẹ[s] (dubious), NPhr. acc.sg. κ̣ναικαν, Arm. nom.pl. kanay-kʿ, and Messapian gunakhai (cf. Olsen 1999: 172). See the detailed study by Opfermann (2017). Cf. also Szemerényi (1960), who reconstructs an original adjective *γυναικός. 19 As I will show later (§ 4.4), the nom.sg. -a in such forms as alyāk ‘other’ (obl.sg. allok) and sana ‘one’ (obl.sg. somo) is secondary.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

83

Gender in the Inflection of the Noun

the fact that *-eh2 regularly yielded PT *-o even in word-final position is corroborated by other inflectional types (see § 3.5.1.2, § 3.5.2.4, § 3.6.2, § 4.8.2.2). As a consequence, a better explanation of the nom.sg. TB -a starts from PIE *gwénh2, which regularly evolved into TB śana (Pinault 1989a: 59). A special issue relates to the obl.sg. śano, because it cannot go back to the accusative singular PIE *gwénh2-m. After the loss of final *-m in Proto-Tocharian, this form should have yielded * śəna, and nominative and oblique would have become perfectly homophonous. In order to disambiguate these core cases, Tocharian generalised the stem of the weak cases PT *-o- < PIE *-eh2- in the oblique singular. This analysis is supported by Pinault (2008: 486) and is further corroborated by evidence that will be treated below and in the following sections. In particular, in certain other nominal classes, Tocharian seems to have continued the stem of the weak cases (e.g. the dative or the genitive singular) as the oblique, in order to differentiate nominative and oblique in a Proto-Tocharian phase. As for the palatalised consonant of the stem, it can be explained by analogical levelling based on the strong cases. This implies that a stem with palatalised consonant *śən- became the standard stem before the breakup of Proto-Tocharian. Another ending that needs to be discussed is the genitive singular TB -oy, ta -e. Following a suggestion by Cowgill, Ringe (1996: 54–55, 59–60) claims that TB -oy is the regular outcome of the genitive PIE *-éh2s, which yielded PT *-oy and then TB -oy, ta -e (cf. also Katz 1997: 61–64). The peculiar development of PIE *-s > PT *-y would be a specific Auslaut sound law that operated in monosyllables. However, the Tocharian B diphthong -oy- usually originated from a contraction over two syllables. Examples from verbal morphology include the optative allomorph -oy-, which only occurs in those subjunctive stems ending in PT *-a- plus the optative suffix *-əy- < *-ih1- (Hackstein 1995: 300 fn. 32; Pinault 2008: 440–442; Malzahn 2010: 348–349). The verbal root TB soy- ‘to be satiated’ can be traced back to the full-grade *seh2- (cf. Hitt. šāḫ- ‘to stuff up’, Gk. ἄεται ‘is satisfied’, Lat. satis ‘enough, sufficient’) plus the present formant suffix *-i̯e/o- (Peyrot 2013: 472). This example may show that only inherited sequences of the type *-eh2iV(-) turned into *-āyV(-) > PT *-oyV(-), while *-eh2i(C-) yielded PT *-ay(C-).20 Examples of TB -oy- from nominal morphology include: (1) TB poyśi ‘omniscient’ from po ‘all’ + aiśi ‘knowing’; (2) TB soy, ta se ‘son’, which is from PIE *suHi̯u-, cf. Gk. υἱύς ‘id.’ (Winter 1985a; Chantraine 1999: 1154). Following Winter (1999: 254–257) and Pinault (2008: 441), we can analyse the gen.sg. TB -oy /-oy(ə)/ (?) as PT *-o- + *-ǝy, where PT *-o was the regular 20

See further Hackstein (1995: 299–300), Pinault (2008: 442), and Malzahn (2010: 955).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

84

Chapter 3

oblique singular and PT *-ǝy was secondarily taken from the gen.sg. -i of the kinship terms and the demonstratives. The other examples provided by Ringe in support of a sound law PIE *-s > PT *-y in monosyllables can now be reconsidered: (1) the nom.pl.f. TB toy ‘those’ is not from *téh2-es > *tās; rather, it acquired final -y from the masculine inflection (pace Ringe 1996: 59 and 95; cf. nom.pl.m. cey and the TA counterpart nom.pl.f. to-, § 4.3.3, § 4.3.4); (2) TB trey, ta tre ‘3’ needs not to go back to PIE *tréi̯es > *trēs > PT *trey directly (pace Ringe 1996: 54–55), but PT *tŕe (< *trēs) more probably acquired final *-y either from the feminine PT *tərya (as per Pinault 2008: 554), or from the nominative plural ending (as per Michaël Peyrot p.c.).21 All things considered, the evolution of the singular paradigm of TB śana, ta śäṃ can be schematised as follows: Table 16

PIE

Evolution of the singular paradigm of TB śana, TA śäṃ

PT

TB

> *kwena > *śəna > *śəna > nom. śana nom. *gwénh2 *śəno > obl. śano acc. *gwénh2-m > *kwena(m) > *śəna gen. *gwnéh2-s > *kwnā(s) >> *śəno >> *śəno-yə > gen. śnoy

TA śäṃ śäṃ śne*

TB lāntsa, ta lānts ‘queen’ The second noun to be discussed is TB lāntsa, ta lānts ‘queen’, which is to be linked to TB walo /wə́ lo/, ta wäl ‘king’ (obl.sg. TAB lānt). The formal match between Tocharian A and Tocharian B and the unproductive inflectional class to which the noun belongs confirm its archaic formation. The morphological and semantic masculine counterpart TB walo, ta wäl is a substantivised participle from the PIE verbal root *u̯ elH- ‘to control’ (Lubotsky 1994a; liv 2 676).22 Although the feminine noun is evidently of pre-Proto-Tocharian origin, it is at first sight unclear whether it is a derivative of the masculine noun, or the substantivised outcome of the feminine participle. However, if we consider that both Tocharian nouns are members of an unproductive class and that feminine

21 The gender-neutral TB wi ‘2’ may be either from the neuter *d(u)u̯ oih1, as Kim (2018a: 71) suggests, or it may have analogically taken over the -i of the i-duals and antapi ‘both’ (Del Tomba 2020b: 28). 22 For possible reflexes of this root in Hittite, see Melchert (2022).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

85

Gender in the Inflection of the Noun

nouns deriving from masculine almost always belong to the aśiya-type, the derivation of PT *lantsa from an old participle seems more probable.23 Further evidence for this historical analysis comes from the reconstructed inflection of PT *lantsa. Indeed, it is usually assumed that PT *lantsa took the inflection after the model of PT *śəna (Pinault 2008: 486; Malzahn 2013a: 110). However, if TB lāntsa and TA lānts can be ultimately traced back to a substantivised feminine participle, it can be claimed that they inherited their inflection directly from Proto-Indo-European. Indeed, the feminine participle inflected ́ as a devī-type in the proto-language, with a proterodynamic inflection parallel to PIE *gwénh2-.24 Table 17

Participle of PIE *u̯ élH-

PIE

nom. acc. gen.

masculine

feminine

*u̯ l ̥H-ōn(t-s) *u̯ lH̥ -nt-m̥ * u̯ lH̥ -nt-os

*u̯ lH̥ -nt-ih2 *u̯ lH̥ -nt-ih2-m *u̯ lH̥ -nt-i̯eh2-s

> > >

masculine

feminine

*wəlōn *wlant *wlantos

*wlantya *wlantyam *wlantyās

In the feminine, a length-differentiated contrast *-a- vs. *-ā- between the strong and the weak cases can indeed be reconstructed for the antecedent of PT *lantsa. This contrast is expected to have yielded *-a- vs. *-o- in Proto-Tocharian. As a consequence, there is no need to reconstruct analogical developments in order to explain the singular paradigm nom. -a, obl. -o of TB lāntsa: in a Proto-Tocharian phase, the weak stem *lantso has been reanalysed as the Tocharian oblique. Thus we can schematise the following development: 23

24

As per Malzahn (2013a: 110): “The latter [scil. lāntsa] started out as a feminine formation in *-nt-ih2 […], which was based on the masculine form of a participle […] in -nt-”, emphasis by the author. Another word which seems to have continued the original feminine form of an athematic nt-participle is nom.sg. TB preṃtsa ‘pregnant’ (IT306 b5 preṃtsa, W33 a6 pre(t)sa, B505 a2 and b4 pretsa, IT932 a1 preṃtsa). For a new translation of IT306 b5, see Friss (2021: 14). The occurrence in IT932 a1 kliye preṃtsa “pregnant woman” would appear to be decisive for the interpretation of this word as an adjective. As recently confirmed by Friis (2021: 12–14), preṃtsa must continue PIE *bhérontih2 (f.) ‘carrying’ → ‘(child-)bearing’ → ‘pregnant’. The table is based on Lubotsky (1994a: 70) and Pinault (2008: 511–513). If the acc.sg. PIE *-ih2-m̥ underwent Stang’s Law, yielding *-īm, then the acc.sg. pre-PT *-yăm was reintroduced after other case forms. On Stang’s Law, see Pronk (2016: 23) and § 4.8.2.2.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

86 Table 18

Chapter 3 Evolution of the singular paradigm of TB lāntsa, TA lānts

PT nom. acc. gen.

* lantsa * lantsa * lantso

> >>

*lantsa > * lantso > *lantso-yə >

nom. obl. gen.

TB

TA

lāntsa lāntso lāntsoy

lānts lānts >> lāntsāṃ lāntse

As far as the plural inflection is concerned, one may be tempted to analyse the vowel -a- in the Tocharian A nom. lāntsañ, obl. lāntsas as the direct outcome of PT *-o- (cf. TB lāntso-na). However, it is more likely that TA -a- originated after vowel weakening and that the original vowel was *-ā-. Indeed, the oblique plural of TA śäṃ ‘wife’ is śnās, where vowel weakening could not have occurred. This means that the plural paradigm of both TA śäṃ and TA lānts was just the same as the plural paradigm of the aśiya-type. At this point one may legitimately wonder whether the original vowel in the plural was retained either by Tocharian A or by Tocharian B (cf. śn-ā-s vs. śn-o-na). Following Brugmann (Grundriss ii.2 [1911], pp. 205, 214, 226), Melchert (1994b: 241–242) has cast some doubt on the reconstruction of ́ the ablaut *-ih2-/-i̯eh2- in the plural paradigm of the devī-type, claiming that the nom.pl. may have not been *-i̯eh2-es, but in fact *-ih2-es. More recently, Lindner (UrIG § 3g., p. 16) also reconstructs *-ih2-es for the nom.pl. and further indicates *-i̯eh2-es as an innovation arisen in the individual branches (cf. Goth. -jōs, Lith. -ios). We have already seen that the same proterodynamic paradigm can also be reconstructed for PIE *gwénh2-/-éh2- ‘wife, woman’. If this is correct, then it may be claimed that Tocharian A has retained the original situation and that the Proto-Tocharian paradigm of the śana-type originally had a vowel *-a- < PIE *-ih2-es in the plural. Since in Tocharian B we notice a general tendency to replicate the oblique singular form as the stem of the plural, TB śno-na /ś(ə)nóna/ may have been influenced by the obl.sg. śano /śə́ no/. However, it cannot be excluded that the development described here operated in reverse: that in Proto-Tocharian the plural was based on a stem ending with -o- and Tocharian A likely aligned this relic class to the productive aśiya-type. TB ṣarya ‘(beloved) lady’ The last noun to be discussed is TB ṣarya (without equivalent in Tocharian A), whose etymology has caused much debate among scholars. This noun is usually translated as ‘beloved, dear (woman)’ (e.g. dtb 713; Broomhead 1962:

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

87

2.247), ‘Geliebte’ (Sieg & Siegling 1949: 180; Otto 2007), ‘female lover, concubine’ (Winter 1981: 938, 2003a: 205), ‘chérie, bien-aimée’ (Pinault 2008: 486).25 In recent years, a new interpretation has been proposed by Kim (2009b), who claims that TB ṣarya had the general meaning of ‘lady, mistress’. Kim largely bases his analysis on B33. The passage in question is as follows: B33 a426 (Udānālaṅkāra) saswe ṣarya sompastär te ¦ lord(m).nom.sg (f).nom.sg take.away.3sg.prs det retke yāmträ were te ¦ army(m).nom.sg do.3sg.sbj smell(m).obl.sg det pūwar tsakṣäṃ war paräṃ ᛬ fire(a).nom.sg burn.3sg.prs water(a).nom.sg bear.3sg.prs “The lord [or] the ṣarya takes this, the army may reduce [lit. make] that to a scent, [or] fire burns it, [or] water carries it off”. (cf. Peyrot 2013a: 705) Kim argues that the sequence saswe ṣarya has a sort of official meaning, and thus translates it as “lord and lady” (see also Otto 2007: 114). Pinault (2013a: 241–242 fn. 3) is against this new interpretation. He points out that the passage in question actually represents a common topos of Buddhist literature that deals with the impermanence of mundane goods, by enumerating all entities that cause the ruin of humans. Female characters are often included in this enumeration. Accordingly, Pinault claims that ṣarya in B33 a4 means ‘harlot, courtesan’, as the “darling by profession”. However, in other Buddhist maxims it is not necessarily harlots that are said to cause the ruin, but women in general, 25

In his dictionary, Adams (dtb 713) claims that TB ṣarya can be both a noun and an adjective referring to either masculine or feminine nouns. If so, it would be a sort of synonym of TB lare ‘dear’. However, we have no clear evidence that ṣarya can be used as an adjective, nor that it can refer to both male and female human beings or deities (Kim 2009b: 112; Otto 2007: 111). Adams mainly bases his analysis on a passage in B85 a2 from the Araṇemijātaka: ṣarya ammakki poññ āppai mā ñiś cempaṃts rakṣatsents aiṣṣäṃ “beloved mother, tell father not to give me to these rakṣas (sic)” (translation by Adams). However, as pointed out by Otto (2007) the fact that one can translate TB ṣarya as an adjective does not mean that it was an adjective in Tocharian B. Indeed, this term occurs in other passages without any other noun with which it can agree. Instead, then, two nouns are used in apposition. A more literal translation would be: “O lady! Mummy! Tell daddy that he mustn’t give me to those rākṣasas!” (cf. Couvreur 1964: 240; Schmidt 2001: 314). Furthermore, we have several examples of double appositional nouns in similar constructions as in line a1 of the same document: || tumeṃ uttare m(ñcu)ṣk(e) wcukaisa mātär lāntso eṅku “Thereupon prince Uttara while grasping [his] mother, the queen, by the cheek …”. 26 Verse: metre 8 ¦ 7 ¦ 7 (4+4 ¦ 4+3 ¦ 4+3).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

88

Chapter 3

as those who link men to mundanity and, in inspiring love and affection, cause the perpetuation of men in the saṃsāra.27 Furthermore, in other Tocharian passages, TB ṣarya, always refers to respectable and virtuous women, such as queens and princesses (e.g. the Buddha’s wife Yaśodharā and the wife of king Araṇemi). Thus the passage in B33 a4 does not necessarily imply that ṣarya means ‘harlot’. Let us consider all the attested forms of this noun: it is inflected eight times as a vocative (IT111 b3–4, AS15C b4, NS18 b1, NS699 b4, B85 a2, B91 a6, B516 b6), twice as a nominative (NS49 b5, B33 a4), and once as a comitative (AS15 b4). Other occurrences are unclear (THT1173 b3, THT 1262 a4). Starting with the vocative, in AS15C someone talks with Yaśodharā and informs her that she had been sent a gift: AS15C b4 (jātaka/avadāna) yaśodharañ suknaṃ weṣṣan-neś ṣarya Yaśodhara(f).gen.sg bring.3sg.prs speak.3sg.prs-3sg.suff (f).voc.sg ce hār saswe epiyacäññe this.m.obl.sg necklace(a).obl.sg lord(m).nom.sg memento(a).obl.sg lywā-c send.3sg.prt-2sg.suff “He gives (it) to Yaśodhara (and) speaks to her: « O ṣarya, the lord sent this necklace to you as a memento »”. (cf. Pinault 1989b: 189) In the passage above, the servant Chandaka delivered the necklace to Yaśodharā on behalf of the lord, and thus TB ṣarya should be translated with an official or reverential value. The translation ‘lady’ thus fits well here. Likewise, in NS18 a maidservant (TB ceṭika) asks a female character (probably princess TB Mitragāmini in line a2) the following question: NS18 b1 (Candraprabhāvadāna) ṣarya candraprabheṃ mäñcuṣkemeṃ (f).voc.sg Candraprabha(m).obl.sg prince(m).abl.sg kekamus(a) come.prt.pct.f.nom.sg “O ṣarya, did you come from prince Candraprabha?”. (cf. Georges-Jean Pinault apud CEToM) 27

Several Khotanese passages about the wiles of women can be compared, e.g. chapter 19 of the Book of Zambasta (the so-called straiya-parivāra ‘chapter concerning woman’), 23. 172–173 of the same text, the so-called lyrical poem (Kumamoto 2000), the tales of the animals in the Rāma Story (133–149), passages in the Book of Vimalakīrti (218), etc. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

89

Gender in the Inflection of the Noun

The translation of ṣarya as ‘lady’ is also preferable in NS18. The passage in B516 (literary drama) is difficult. We find two characters, Yaśodharā and a female door warden named Priyaśāriṇi, but it is unclear who the speaking character is: B516 b6 lyelyakormeṃ weṣṣäṃ ṣarya – – yaśodhara lāntsa memīyus(a) /// “After having seen (this), she speaks: « O ṣarya […], queen Yaśodharā deceived (by)… »”). What is clear is that in all aforementioned passages, the voc.sg. ṣarya is always used by servants when referring to princesses or queens. All other vocative forms come from the Araṇemijātaka. In one passage (B85 a2 and NS699 b4, which both contain the same portion of text), the young prince Uttara speaks to his mother (B85 a2 ṣarya ammakki poññ āppai mā ñiś cempaṃts rakṣatsents aiṣṣäṃ “O ṣarya! Mummy! Tell dad that he mustn’t give me to those rākṣasas!”), while in another passage (B91) king Araṇemi speaks to his wife (a6 ṣarya kauṃ (s)ū (pe)rn(e)w t(a)kā-ñ “O ṣarya! This day has become a glorious one for me”). In these texts, TB ṣarya can be translated with ‘beloved one, dear one’, although a meaning ‘O lady’ is possible too. Apart from B33, TB ṣarya is probably also attested as a nominative also in NS40 b5, where it can be translated both as “lady” and “beloved woman” (/// m(akā-yk)ne tarśauna pälwāmane ṣarya /// “… lamenting the deceptions of many sorts, the ṣarya …”, cf. Pinault 2015c: 153–154). In IT111, TB ṣarya refers to a queen, but the document is very fragmentary, the character that is speaking is ambiguous, and thus so too is the translation of our noun (b3 /// maimañcu ṣarya oro(tsa ?) /// “… O excellent one! The ṣarya … great …”; b4: ///ritse ṣarya kre(nt ?) /// “… ṣarya … good …”; for the edition, see Peyrot 2007: n° 111). Finally, the comitative is attested once: B496 a3-4 (literary, profane) sanai ṣaryo[a4]mpa śāyau karttse(ś) śaulu-wärñai one.f.obl.sg. (f).com.sg live.1sg.prs good.all.sg life-long “With a single ṣarya I live for the lifelong good”. Even though this leaf refers to a love story, both ‘lady’ and ‘lover’ fit well in the context. One may therefore wonder whether the basic meaning of TB ṣarya is ‘lady, mistress’, and that ‘beloved woman’ is a later meaning (Kim 2009b: 112), perhaps influenced by the fact that this noun is mostly attested in the vocative, which gives it a sort of affective pragmatic nuance.28 We now turn to the etymology of TB ṣarya. Attempts have been made to link this noun to TB ṣar ‘hand’ by postulating a substantivised possessive adjective 28

For the sake of comparison, the Tocharian A word for ‘lady’, TA nāśi (without equivalent in Tocharian B), is mostly attested with vocative value (cf. A106 a6, A149 a3 and b4, probably YQ III.5 a7, and A160 a6). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

90

Chapter 3

(see Van Windekens 1976: 449; Hilmarsson 1987b: 88). This etymology is still accepted by Adams (dtb 713), who reconstructs PIE *ǵheser-iHeh2- ‘(one) at hand’ → ‘the beloved’ (cf. Gk. χείρος ‘under control’). However, Icelandic hand-genginn ‘favourite’ is too meagre a parallel to support this hypothesis. More recently, Otto (2007) has argued that the noun is a derivative in *-ih2 from the verbal root PIE *ser- ‘to attach, connect’ (liv 2 534–535, cf. Lat. serō ‘to link, join’, Gk. εἴρω ‘to knit together’). The semantic evolution would have been ‘the one who is (physically/mentally) attached’ → ‘the one who is beloved’, via the metaphor of love as a physical/mental attachment (see also Willi 2010: 252–257). Phonologically this analysis is unproblematic, but from a semantic point of view it is less straightforward. Indeed, there is no clear evidence that Tocharian speakers could have considered the physical closeness to both a mother and a lover as aspects of one and the same notion (cf. Kim 2009b: 113). Furthermore, we have no other clear continuants of the PIE root *ser- ‘to attach, connect’ in Tocharian. The inflectional class to which TB ṣarya belongs suggests that we are dealing with a very old derivative or at least with a “noun belong[ing] to the oldest layer of the Tocharian lexicon inherited from Proto-Indo-European” (Kim 2009b: 114). Only two scholars have taken this important piece of evidence into consideration in their etymological discussions, namely Pinault (1989a: 58, 2008: 486) and Kim (2009b). For this reason, I present their proposals in more detail. ́ Pinault argues that TB ṣarya is a devī-derivative of the PIE word for ‘sister’, thus PIE *su̯ é-sr-ih2 > *sẃəsrya (palatalisation) > *ṣəṣərya (assimilation) > TB ṣarya /ṣə́ rya/ (simplification). At first sight, this development seems problematic as it supposes certain irregular changes. However, the fact that PIE *su̯ ésor‘sister’ is continued in Tocharian as TB ṣer, ta ṣar, i.e. with the same assimilation and syllabic simplification, may be used in support (Pinault 1989a: 58). From a historical point of view, PIE *su̯ ésor- can be analysed as an original compound of the reflexive pronoun *su̯ é- and the noun *ser-/sor- ‘woman’. The latter is in turn a good candidate for our Tocharian B noun. This analysis has been proposed by Kim (2009b), who claims that TB ṣarya is the regular outcome of PIE *ser-ih2 (*h1-ser-ih2 in his notation). In most Indo-European languages, the noun *ser-/sor- is attested either as the second member of compounds or as a grammaticalised suffix. Besides PIE *su̯ ésor- ‘sister’, examples include: the feminine numerals for ‘three’ and ‘four’ in Indo-Iranian and Celtic (cf. OIrish téoir, cethéoir, Ved. tisrás, cátasras < *tis(o)r-, *kwetes(o)r-) and the Hittite feminine suffix -(š)šara- (cf. Hitt. išḫa-ššara- ‘mistress’ from išḫa- ‘master’).29 If it is 29

See Sihler (1977) and recently Gąsiorowski (2017) for possible continuants of *(-)sr-ih2 in Germanic. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

91

an original compound, Lat. uxor ‘wife’ probably belongs here too.30 However, other Indo-European languages show continuants of *ser-/sor- as a free word. These include the thematised Cuneiform Luw. *ašra/i- ‘woman’, inferred on the basis of ašrul(i)- ‘female’, ašrulāḫit- ‘womanhood, femininity’ and ašraḫit‘id.’ (cf. Pinault 2013a: 246–247; Harðarson 2014: 38–41; Dunkel 2016: 40–41); the theonym Gk. Ἥρᾱ < *Sērā < *sēreh2 (Willi 2010); YAv. hāirišī- ‘woman’ < *sēr-is-ih2; and possibly YAv. ā̊ŋhairī ‘id.’ potentially from *(h1)-eh1-ser-ih2 ‘belonging to woman’ (as per Harðarson 2014: 41–42).31 According to Kim, TB ṣarya may be added to this list. A further objection put forward by Pinault (2013a) is that a recharacterisá tion of a feminine word by means of the feminine devī-suffix is redundant.32 However, the forms just discussed point to the reconstruction of a root noun *sor-/ser- that lost its autonomy as a free word soon after PIE, since it became a feminine suffix, a second member of compounds, or was recharacterised with other suffixes. As a consequence, the claim by Kim (2009b) that PIE *ser-/sorhas been enlarged with the productive and highly transparent derivational suffix *-ih2 cannot be so easily discarded, although the lack of exact morphological matches of *ser-ih2 in other Indo-European languages invites caution.33 To conclude, whatever ultimately its root, TB ṣarya is derived with the ́ ablauting feminine suffix *-ih2/-i̯éh2- (of the devī-type). Thus the proto-form from which this noun derives must have had the same inflection as TB lāntsa: nom.sg. -ya is the outcome of nom.sg. *-ih2 > *-yă > PT *-ya, while the obl.sg. -yo is from the weak stem *-i̯eh2- > *-yā- > PT *-yo. 3.4.2 The aśiya-type: TB Nouns with nom.sg. -(y)a, obl.sg. -(y)ai and their TA Correspondents The nouns belonging to the aśiya-type are grammatically feminine and denote natural female referents. This is therefore a feature that the aśiya-type and the śana-type have in common. On the other hand, these two inflectional classes are clearly distinct as regards their inflection and productivity. The two major 30 See Ernout & Meillet (1951: 1341), Luján (1996), and Harðarson (2014: 32–35). Cf. Pinault (2013a: 248–250) with references. 31 Following an old theory by Pott (1835: 554) and Sturtevant (1949), Kim (2005, 2009b) proposes to add Ved. strī-,́ YAv. strī-, Khot. strīyā-, Oss. Iron syl, Digor silæ to this list. This proposal has also been endorsed, with modifications, by Dunkel (2016); it has been rejected by Pinault (2013a: 242). 32 Tocharian already has two terms for ‘woman’ (the generic TB kliye ‘woman, female’ and the specific TB śana ‘wife’). However, it is hardly surprising that the lexicon of a given language has cases of quasi-synonymy. Indeed, a good example of such is Tocharian A, which, besides TA kuli ‘woman’ and TA śäṃ ‘wife’, has a third noun that precisely means ‘lady’, i.e. TA nāśi (on which see § 3.4.2 below). 33 See Dunkel (2016: 37). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

92

Chapter 3

inflectional characteristics distinguishing their paradigms are the oblique singular and the stem forming the derivatives and the plural: in the aśiya-type, the former ends in -(y)ai- and the latter ends in -(y)a-. Furthermore, the great majority of these substantives show palatalisation of the stem-final consonant in both the singular and the plural inflection. The paradigm of the Tocharian A equivalents displays different but uniform inflectional patterns: a usually unmarked nominative singular, obl.sg. -āṃ, gen.sg. -e, and the differentiated plural nom. -āñ, obl. -ās.34 The Tocharian paradigms are therefore identical to the feminine adjectival type ending in pl. TB -ana, TA nom. -āñ, obl. -ās (see § 4.7.1). From a synchronic point of view, the aśiya-type is very productive: if a new feminine noun with a female referent needs to be created, it is always added to this class. Furthermore, several feminine literary and non-literary proper names belong to this class; they were mostly borrowed from Sanskrit or Old Uyghur (e.g. the girl TB Cañca, obl. Cañcai; the princess TB Nānda; the queen TB Yaśodhara, obl. Yaśodharai; the queen obl. TA Kṣemāṃ; the Uyghur proper name TA Kutluk, obl. Kutlukāṃ).35 In Tocharian B, these loanwords are sometimes extended with a suffix, such as -śka or -kka (TB Lariśka, Priśka, Räknāśka, etc.). The most representative member of this class, i.e. TB aśiya, ta aśi ‘nun’, is also a loanword from either OKhot. aśiā- ‘id.’ or, in my view more plausibly, a Prākrit variety.36 The derivational processes involved have been described in the previous chapter (§  2.4.2) and thoroughly analysed by Malzahn (2013a). Here I focus on major derivational and etymological patterns that these nouns have in common. It is worth highlighting that no nouns directly inherited from Proto-Indo-European seem to belong to the aśiya-type. Indeed, inherited nouns that figure in this class have all undergone some derivational process. Examples include: the substantivised adjective TB eṣerña* ‘besistered 34

In Tocharian A, nouns of the aśiya-type usually end in a consonant or in -i in the nominative singular. Sporadic cases of final -ī and -ā are attested, but they are loanwords from either Sanskrit or Tocharian B. 35 As Wilkens (2021: 431) clarifies, OUy. Kutlug (with variants) is found with both masculine and feminine proper names. 36 I believe it is still uncertain whether Tocharian borrowed this word from Khotanese or from Middle Indian. Indeed, the noun is neither of Tocharian nor of Khotanese origin, but may have been borrowed into both languages from a Middle Indic form linked to Skt. arya-/ārya- ‘noble’. The source from which the Khotanese word derives is usually reconstructed as (Gāndhārī?) *aź(i)yā- < ayyā (cf. Pāli ayya), in turn from (Buddhist) Skt. āryikā (Bailey 1967: 9; bhsd 105). This *aźyā may have been borrowed in Tocharian as *aśya-. For this Gāndhāri phonological development, see von Hinüber (2001: 174).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

93

Gender in the Inflection of the Noun

one; sister’ (attested only in the plural eṣerñāna in B107 a5, b3 and b6) < PT *e(n)-ṣer-ña (Pinault 2008: 129); TB ṣer-śka ‘little sister’, which displays the same base of the previous noun; TB śamñāṃśka ‘girl’ (on which see below). The feminine suffix -(ñ)ña is of adjectival origin: etymologically, it is the paradigmatic feminine form of TB -ññe (Van Windekens 1979: 105 and 123; Malzahn 2013a: 115–116; see § 4.7.1.3). It is also a productive suffix used for creating oppositional feminine nouns. In some cases we see the substantivisation of both masculine and feminine forms of a ññe-adjective, as in ostaññe ‘male householder’ vs. ostañña ‘female householder’ and riññe ‘male citizen’ vs. riñña ‘female citizen’. In some other cases, -ñña is clearly an independent morpheme. This implies that it has been grammaticalised as a feminine suffix in the history of Tocharian B.37 Examples are: – TB kaṭapūtanäñña ‘female demon’ (= Skt. kaṭapūtanā-) from TB kaṭapūtane* ← Skt. kaṭapūtana- (mw 243; swtf 2.8); – TB kinnarña* (cf. TA kinnarñā*) ‘female Kiṃnara’ (= Skt. kiṃnarī-) from TB kinnare ‘Kiṃnara’ ← Skt. kiṃnara- (mw 283; swtf 2.75);38 – TB ñäkteñña ‘goddess’ from ñakte ‘god’; – TB °pläṅkṣiñña* ‘female seller’ from TB °pläṅkṣi ‘seller’; – TB yakṣañña ‘female yakṣa’ (= Skt. yakṣī-) from TB yākṣe ‘male yakṣa’ ← Skt. yakṣa- (mw 838; swtf 4.4). There is no corresponding suffix in Tocharian A. Indeed, all nouns formed with TA -ññā are loanwords from Tocharian B (e.g. TA ñäkteññā from TB ñäkteñña, cf. § 2.4.2). Another frequent morphological process used to create oppositional feminine nouns in Tocharian B provides for the substitution of the final vowel of the masculine noun with TB -a, as in oṅkolma ‘she-elephant’ from oṅkolmo ‘elephant’, mañiya ‘female servant’ from mañiye ‘male servant’ (borrowed from Iranian *māni̯a- ‘servant’, Tremblay 2005: 435), and mcuṣka ~ mñcuṣka ‘princess’ from TB mcuṣke ~ mñcuṣke ‘prince’. All other suffixes, including TB -śka and -kka, have been borrowed from Iranian (Klingenschmitt 1975: 149–150), the most common being TB -āñca, ta -āñc (Müller 1908: 47; Gershevitch 1961: 158). They are often used to form feminine nouns from Indian loanwords. Examples of the suffix TB -āñca, ta -āñc, particularly productive in Tocharian A, are:

37 On the value of TB -ññe and its grammaticalisation as a feminine suffix, see § 4.7.1.3. 38 Cf. nom.sg. kindarña in THT1245 a2.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

94

Chapter 3

– TA ārāntāñc* ‘female arhat’ from TA ārānt ‘arhat’ ← Skt. arhant- (swtf 1.150; bhsd 67); – TB upāsakāñca, ta wāskāñc ‘female lay-disciple’ (= Skt. upāsikā-) from TB upāsake ‘(male) lay-disciple’ ← Skt. upāsaka- (mw 215; swtf 1.411); – TA karmavāckāñc* ‘female Karmavācaka’ from TA karmavācak* ← Skt. *karmavācaka- (dta 105); – TA kānikāñc ‘girl, virgin’, an extended form to be linked to Skt. kanyakā‘girl, maiden, virgin’ (mw 249; swtf 2.19; dta 112); – TA kränolāñc ‘adopted girl’ (dta 172); – TB parivrājakāñca* ‘female mendicant’ from the unattested loanword from Skt. parivrājaka- ‘a wandering religious mendicant’ (mw 602; swtf 3.96); – TA pravārāpakāñc ‘?’;39 – TA pretāñc ‘female Preta’ (= Skt. pretī-, cf. OUy. pretanč) from pret ‘Preta’ ← Skt. preta- (mw 711; swtf 3.239); – TB brahmaṇāñca (attested once in IT956 a2), ta brāmnāñc ‘female brahmin’ (= Skt. brāhmaṇī-) from TB brāhmaṇe, ta brāmaṃ ← Skt. brāhmaṇa- (mw 741; swtf 3.269–270); – TA mäśkitāñc ‘princess’ from mäśkit ‘prince’;40 – TA ṣāmnerāñc ‘feminine novice’ (= Skt. śrāmaṇerī- ~ śrāmaṇerikā-) from ṣāmner ‘novice monk’ ← Gāndh. ṣamanera (cf. Skt. śrāmaṇera-, Pāli sāmaṇera). There are two Tocharian A members of the aśiya-type whose origin warrants a closer examination, as they are considered to preserve a more archaic formation. These are TA śomiṃ ‘girl’ and TA nāśi ‘lady’. The first noun is usually interpreted as a derivative of the masculine śom* ‘boy’ (attested once in A63 a2 as an oblique TA śomäṃ), by means of the suffix TA -iṃ, which is equated with the feminine suffix TB -(ñ)ña by Poucha (1955: 327) and Klingenschmitt (1994: 368). However, I have found no other feminine nouns derived with the feminine suffix TA -iṃ; I therefore see no reason to equate TB -ñña to TA -iṃ in śomiṃ ‘girl’. On the other hand, Peyrot (2012a: 193) links TA śomiṃ ‘girl’ to the adjective TB śāmña, which is the feminine form of śāmñe ‘human’. Although the derivational process involved is obscure (dtb 682), TB śāmñe seems to be a 39 Cf. Tamai (2014: 391 fn. 88). For the meaning, see Itkin (2019: 154). 40 As pointed out by Pinault (2015a: 173–174), TA mäśkit can be used with both masculine and feminine referents. The feminine mäśkitāñc is probably a secondary form, which corresponds to TB mcuṣka ~ mñcuṣka.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

95

secondary relational adjective in -ññe from TB śaumo ‘human being’ (cf. TB śay- ~ śaw- ‘to live’, Gk. ζώω, Ved. jīvati, YAv. juuaiti < PIE *gwih3-u̯ e/o- ‘to live’), with the reduction *-au- > -a- before a consonant cluster. The derivation of TA śomiṃ from PT *śawməñña is phonologically unproblematic, but the fact that Tocharian A does not show any continuant of the correspondent masculine *śawməññe is suspicious. The masculine TA śom* ‘boy’ has long been equated with TB śaumo ‘human being’ (Pinault 2008: 520). Both Tocharian forms derive from PT *śawmo, an original adjectival derivative in *-mo(n) from PT *śaw- ‘to live’. Now, since TA śomiṃ inflects as the feminine counterpart of an adjective in TB -mo, ta -m (of the klyomo-type, cf. nom.sg.f. TB klyomña, ta klyomiṃ; see § 4.7.2), I believe that śomiṃ and śom* belonged to the same adjectival paradigm in Proto-Tocharian, which can be reconstructed as follows: nom.sg.m. *śawmo, obl.sg.m. *śawmon; nom.sg.f. *śawməñña, obl.sg.f. *śawməñña (similarly Pinault 2008: 520). In Tocharian A, both the masculine and feminine have been substantivised with the meaning of ‘boy’ and ‘girl’, while in Tocharian B only the masculine survived with the generic meaning of ‘people, man’ (but with the deviant plural TB śāmna, on which see § 3.6.3.3). The expected Tocharian B counterpart of TA śomiṃ is probably attested in the problematic form TB śamñāṃ-śka ‘girl’. Adams (dtb 678) segmented this noun as śamñ-āṃśka, claiming that TB -āṃśka “denotes females”. However, this hypothetical suffix is not attested elsewhere. Rather, TB śamñā° is to be linked etymologically with TA śomiṃ as the regular outcome of PT *śawməñña. The final nasal in the Tocharian B stem śamñāṃ- |śamñán-| may have been taken from soṃśke ‘dear son’ (cf. also ylaṃśke ‘young gazelle’, wlaṃśke ‘soft, pliable’).41 The second noun, TA nāśi ‘lady, mistress’, is probably the feminine counterpart of TA nātäk ‘lord, master’. These two words were previously taken to be the equivalents of Greek ἄναξ, -κτος ‘lord, ruler’ (cf. Mycenaean wa-na-ka, Beotian Ϝάναξ, etc., and also OPhr. vanak, if not borrowed from Greek) and ἄνασσα, -ης 41

Even in these forms the origin of the nasal is debated. Klingenschmitt (1975: 150–153) and Winter (1985a) argued that -ṃ- /-n-/ has been analogically extended after the accusative singular of the n-stems. This analysis is convincing in the case of soṃśke. Klingenschmitt (1975: 154) seems to go further: he argues that the -ṃ- in śamñāṃśka is to be interpreted as an archaic residue of the Proto-Tocharian state of affairs, where the accusative *-an and the (dative-)locative *-ay were still formally and functionally distinguished. Afterwards, Tocharian B extended *-ay as the oblique, while Tocharian A has further reanalysed the locative as a genitive *-ay > -e. However, the obl. sg. -ṃ is limited only to masculine nouns in Tocharian B, and its spread to the feminine in Tocharian A seems to be a recent and independent development. Furthermore, the origin of TB śamñāṃśka seems to be quite recent, probably of the pre-Tocharian B stage, as we have no Tocharian A equivalent of the suffixes TB -śke and TB -śka.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

96

Chapter 3

‘lady, queen’. Winter (1970: 53) was the first proponent of this lexical isogloss, which is still supported by Adams (2017: 1376). However, there are two serious problems with this etymology: (1) the mismatching order of the consonants -t- and -k- in the masculine noun and (2) the loss of initial *u̯ - in Tocharian. For Gk. ἄναξ, Szemerényi (1979: 217) reconstructs an endocentric determinative compound PIE *u̯ n̥-h2eǵ-t- ‘one who led the tribe’, whose first member was PIE *u̯ en- ‘kin, tribe’, and the second *h2eǵ- ‘to lead’.42 If one wanted to link TA nātäk to this reconstructed form, a metathesis *kt > *tk should be postulated, which is however without parallel.43 Furthermore, the loss of the semivowel in such a phonetic environment is also unexpected. All these phonological difficulties invalidate the etymological link between Tocharian and Greek: their formal resemblance is totally accidental. Van Windekens (1976: 313) connected TA nātäk to the verb TA nätk- ‘to hold off, push away’, but this proposal has formal and semantic flaws. On the formal side, we should rather expect a very old derivative built on a lengthened *o-grade of the root (cf. the τόμος-derivatives, TB snai-netke ‘unprompted’, TA natäk ‘urge, pressure’, Malzahn 2013b: 167). As for the meaning, a semantic development ‘the one who pushes away’ → ‘the lord’ seems far-fetched.44 Since TA nātäk cannot be derived from any internal source, I have sought a foreign origin. It is tempting to link TA nātäk ‘lord’ to Skt. nāthá- (m.) ‘protector, possessor, lord’ (mw 534; swtf 3.15; Pisani 1941–1942), which can also be found in Pāli nātha, Pkt. ṇāha, and Gāndh. nasa. In Sanskrit, this noun is frequently attested in apposition to gods and men, cf. Skt. govinda-nātha- name of Saṃkara’s teachers (mw 367), nāka-nātha- ‘sky-lord’ (mw 532), loka-nātha- ‘saviour of the world (epithet of the Buddha)’ (mw 906, bhsd 464). Furthermore, ka-extended variants of Skt. °nātha- are also attested: Skt. nāka-nāthaka‘sky-lord’ (epithet of Indra), gaṇa-nāthaka- ‘leader of the attendants of any god; head of an assemblage corporation’ (used as epithet of Śiva and Gaṇeśa), vr̥kṣa-nāthaka- ‘lord of trees’, etc. Therefore, it may be hypothesised that TA 42 The proto-form from which the Greek word derives is a puzzle and etymological dictionaries of Greek raise the possibility of a loanword from a non-Indo-European language (gew 1.102; Chantraine 1999: 84; Beekes 2010: 98–99). Szemerényi’s reconstruction is also followed by Hajnal (1998: 66). See Willms (2010) for a slightly different reconstruction, which however does not invalidate the morphemic segmentation. 43 In order to avoid this problem, Winter (1970: 53–54) reconstructed PIE *wnatk- and further assumed a metathesis of the cluster *-tk- > *-kt- in Greek (like *τίτκω > τίκτω). However, he did not give an etymological segmentation of the proto-form. 44 Following Thomas (1964: 110), Van Windekens erroneously translated TA nätk-, tb nətkaas “soutenir, appuyer”, and thus claimed that TA nātäk originally meant “qui soutient, puissant”. See Jasanoff (1978: 39) for the correct meaning of the verb.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

97

nātäk ‘lord’ is actually a loanword from an intermediary language (potentially Sogdian) of Skt. nātha(ka)- ‘protector, possessor, lord’ (mw 534).45 Indeed, if directly from Skt. nātha(ka)-, a form TA **nātak would probably be expected (cf. kāccap* ‘turtle’ from Skt. kacchapa-, see Schwartz 1974: 406). As for TA nāśi ‘lady’, this noun is usually considered to be the feminine counterpart of TA nātäk. If this is correct, there is an additional problem related to this form. Indeed, evidence for the palatalised variant of the cluster -tk- is extremely meagre in Tocharian. In the verbal roots in -tk-, only the -t- gets palatalised, yielding -ck- (cf. TA the gerundive kāckäl from TA kātk-).46 The same kind of palatalisation also occurs in TA nācki ‘lords’, the nom.pl. of nātäk. This nom.pl. (only used with vocative value) is problematic since it is limited to this noun.47 Hackstein (2004: 175, 2017: 1328) suggests that *-y- palatalised the cluster *-tk- differently, yielding pre- TA *-śś-. If so, *natkya > *naśśi > TA nāśi. A last possibility is reconstructing a loanword from a Middle Indic continuant of *nāthyā- ‘the one related to the lord’, if this form evolved similarly to TA āmāś ‘minister’, ultimately from Skt. amātya- ‘a companion of a king’ (mw 81, dta 40). All these are, however, ad hoc possibilities.48 To sum up, we have seen that not a single member of the aśiya-type can be traced back to Proto-Indo-European, since nouns belonging to this inflectional class are mostly of late origin. It could therefore be concluded that the aśiya-type became a productive class of feminine nouns only in a relatively recent Proto-Tocharian period. Indeed, given the fact that we have clear examples of nouns with the same origin and matching inflections in both Tocharian languages, the origin of this inflectional class must be sought in a 45

For the adaptation, cf. TA kātak* or kātäk*, TB kattāke ‘householder’ ← a Gāndhāri or Middle Iranian extended variant of Buddhist Skt. gahastha- (bhsd 211) (< Skt. gr̥hástha-), cf. OKhot. ggāṭhaa-, BSogd. kʾrtʾk ~ kʾrtk; TA sāṃtäk, tb sāṃtke ‘medicine, remedy’ ← a Middle Iranian intermediary of Skt. śāntaka-. TA kātak* ~ kātäk* ‘master of the house, householder’ is only attested in the plural, where the reduced forms nom.pl. kātkāñ, gen.pl. kātkāśśi may be due to Vokalbalance (as per dta 110). Note, however, that the stem kātk- is also attested in derivatives, e.g. kātkeṃ ‘pertaining to a householder’ and kātkune ‘householdership’. On the origin of TB kattāke, ta kātak*, see recently Dragoni (2022: 75–76). 46 See Burlak (2000: 128), Malzahn (2010: 460–461), and Peyrot (2013a: 76). 47 Pace teb § 181, the nom.pl. of TA ratäk ‘army’ is not racki, but probably rackiñ (THT1134 a3). The instr.pl. rackisyo is attested once in A183 a5. 48 Building on my previous account of TA nātäk (Del Tomba 2020a: 79–80), Bernard (forth.) thinks that this noun would be better understood as a loanword from a reconstructed *nāθaka of so-called Old Steppe Iranian. Bernard’s hypothesis is formally unproblematic, but evidence for the reconstruction of the Iranian counterpart of Indo-Aryan nāthá- is meagre.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

98

Chapter 3

Proto-Tocharian stage. Taking the theoretical antecedent of TB aśiya, ta aśi as an example, we can reconstruct the following paradigm:49 Table 19

Evolution of the Proto-Tocharian singular paradigm of the aśiya-type

PT nom.sg.

*aśǝya

obl.sg.

*aśǝya

gen.sg.

*aśǝyay

> > >> >> >> >

TB aśiya TA aśi TB aśiyai TA aśśāṃ* TB aśiyāntse TA aśśe

As can be seen, in a Proto-Tocharian stage nominative and oblique formally overlapped. As a remedy, in both Tocharian B and Tocharian A the oblique was recharacterised, but in different ways: Tocharian B reanalysed the gen.sg. *-ay as the oblique and further acquired the gen.sg. -ntse from the n-stems, while Tocharian A maintained the original PT *-ay > TA -e as the genitive and developed an obl.sg. -āṃ (see recently Peyrot 2012a). As we will see, this evolution is shared by the feminine of those adjectives that have the same inflection as the the aśiya-type in both Tocharian languages (see § 4.7.1). 3.4.3 Conclusion In this section we have seen that the inflection of the feminine substantives belonging to the śana-type has to be interpreted as the outcome of the archaic proterodynamic inflection in *-h2/-éh2- and *-ih2/-i̯éh2-. In a Proto-Tocharian stage, the weak stem (or possibly the genitive form) has been reinterpreted as the Tocharian oblique. It is easy to imagine how this reanalysis took place: after the apocope of final consonants in pre-Proto-Tocharian, the nominative and accusative merged formally. This interpretation has already been proposed to explain the inflection of TB śana, but as far as the two other nouns are concerned it was usually assumed that the PIE acc.sg. *-ih2-m > PT *-ya had been analogically modified to PT *-yo after the obl.sg. śano (cf. e.g. Winter 1981: 938; Pinault 2008: 486; Malzahn 2011: 89 fn. 14). However, the śana-type 49 The Proto-Tocharian paradigm of the aśiya-type follows the reconstruction of the ProtoTocharian paradigm of the feminine adjectival inflection (Peyrot 2012a: 200–204). For further remarks on this topic, see § 3.5.2.5, § 4.7.1.4.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

99

is not a productive inflectional class, since it is confined to isolated feminine substantives. If we assumed that TB śana is the only noun whose inflection is original, then TB lāntsa and TB ṣarya are not expected to be included in this class by analogy, but rather in the aśiya-type, which is a productive class of feminine nouns referring to female entities. Analogical extension to this inflectional type would have been facilitated by the fact that the majority of the aśiya-nouns have (suffixal) -y- or display palatalisation/assibilation of the stem final consonant, just like TB lāntsa, ta lānts (assibilation) and TB ṣarya. As a consequence, the inflection of TB lāntsa and TB ṣarya must be original. In conclusion, all nouns of the śana-type have continued the archaic inflection inherited from Proto-Indo-European: the contrast between nom.sg. -(y)a vs. obl.sg. -(y)o mirrors the ablauting alternation between the zero and the full grade of the suffix *-(i)h2,*-(i̯)eh2-, where the original genitive singular has been reanalysed as the Tocharian oblique. As for the plural inflection, one may wonder whether the vowel -o- in Tocharian B śnona and lantsona should be regarded as a post-Proto-Tocharian replacement based on the obl.sg. of older *śəna° and *lantsa°, as testified by the Tocharian A plural forms obl. śnās, nom. lāntsañ (< *-āñ), obl. lāntsas (< *-ās). If so, Tocharian A would have retained the outcome of the zero grade of the suffix in the plural, cf. nom.pl. PIE *-(i)h2-es. However, it cannot be excluded — and it is probably even more likely — that the generalisation of -o- to the plural inflection was a Proto-Tocharian development which has been maintained in Tocharian B, while Tocharian A has analogically eliminated the śana-type in favour of the aśiya-type. The latter inflectional type was very productive in the historical phase of both Tocharian languages, even though it was probably formed in Proto-Tocharian following the model of the adjectival inflection. The plural inflection of both inflectional types poses a special problem because the comparison between the two Tocharian languages invalidates a direct Proto-Tocharian reconstruction. Indeed, Tocharian A and B diverge in the formation of the plural paradigm of these classes, since Tocharian B attests -ona and -(y)ana (nom. = obl.), while Tocharian A has differentiated markers in the nominative and in the oblique, i.e. TA -añ | -as and -āñ | -ās. In this case, the comparison between the two languages invalidates a direct Proto-Tocharian reconstruction. An important question is therefore which of the two languages maintained the older situation. There are two opposite ways to explain this mismatch: (1) Tocharian B maintained the older state of affairs, and thus Proto-Tocharian had *-na as the plural marker of these classes; (2) Tocharian A maintained the older state of affairs, and thus we have to reconstruct the situation of Proto-Tocharian as different from that of Tocharian B. Both hypotheses have advantages and disadvantages.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

100

Chapter 3

The former implies that Tocharian B maintained the Proto-Tocharian state of affairs unaltered, but also leads us to ask why Tocharian A lost the expected outcome of *-na and, more generally, how this ending came out in Proto-Tocharian. The second hypothesis suggests that Proto-Tocharian had formally differentiated nominative and oblique plural forms. This should also have been the situation in Proto-Indo-European. But why would Tocharian B lose such a differentiated paradigm? Yet, a third possibility is that Proto-Tocharian marked the plural paradigm differently to both Tocharian A and B. This problem cannot be addressed without considering evidence from adjectival and pronominal inflections. Indeed, in the continuant of the PIE thematic type we find a clear contrast between adjectives with f.pl. TB -ona, ta -aṃ and adjectives with f.pl. TB -ana, TA -āñ | -ās. Again, the comparison between Tocharian B and A is straightforward for the former type, while it is not for the latter, which in turn strongly resembles what we find in the noun inflection. Given the fact that these plural markers are characteristic of both nouns and adjectives, I will investigate the origin and the development of the feminine plural ending -na once I have dealt with the synchronic distribution and the diachronic evolution of the endings TB -na and TA -äṃ in the alternating nouns (§ 3.6.3) and considered data from the adjectival inflection (§ 4.7.1, § 4.7.3). 3.5

Grammatical Gender and the Development of Inflectional Classes from class vi (nom.pl. -ñ)

So far, I have investigated the evolution of the PIE feminine gender in a group of nouns which synchronically belong to class ii of Tocharian B. These nouns have been consistently compared with their Tocharian A equivalents in order to clarify the diachronic evolution of their endings and forms. Following the same method, I will in the following paragraphs deal with the historical evolution of selected inflectional types, which synchronically belong to class vi (nom.pl. -ñ).50 The aim is to understand how (1) the non-ablauting *eh2-type (i.e. the *ā-inflection), (2) the hysterodynamic *(e)h2-type (i.e. the ́ ́ *ā/ă-inflection), and (3) the *ih2-type (of both the devī-type and vr̥kī-type) evolved in the inflection of the Tocharian noun.

50

For an overview of the diachronic problems presented by this class and early bibliography on the topic, see Hartmann (2013: 413–424).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

101

This section is divided into three central parts. I will first investigate nouns with nom.sg. -o, obl.sg. -a, which can be grouped under two different types on the basis of their plural inflection: (1) feminine or masculine nouns with differentiated nominative and oblique in the plural (nom.pl. TAB -ñ, obl.pl. TB -ṃ, ta -s) and (2) alternating nouns with undifferentiated nominative and oblique in the plural (§ 3.5.1). Next, I will deal with two closely related inflectional classes, the so-called okso-type and arṣāklo-type, which both end in nom. -o, obl. -ai in the paradigm of the singular (§ 3.5.2). In the third part, I will investigate the origin of the wertsiya-type, whose members have a palatalised stem -ya- throughout the inflection of both the singular and the plural (§ 3.5.3). 3.5.1 The kantwo-type: TB Nouns with nom.sg. -o, obl.sg. -a and their TA Correspondents In this section, I investigate the diachronic evolution of a class of nouns, which I label the kantwo-type for convenience. Some preliminary remarks on the identification of each substantive are made (§ 3.5.1.1). These entail a revision of the list of the members usually proposed. Thereafter, I discuss the etymology of the nouns identified and examine the evolution of their inflected forms. I also discuss the synchronic gender of problematic nouns in order to establish a solid basis for their diachronic investigation (§ 3.5.1.2). One of the most recent and detailed works about the nouns of the kantwotype (nom.sg. -o, obl.sg. -a) is Malzahn (2011). In the specialised literature on Tocharian nominal morphology, this inflectional class has over the years become one of the most debated types, since the great majority of its members are supposed to go back to the PIE type in *-eh2 > *-ā. Nevertheless, there did not exist a general discussion of the problems presented by this class until Malzahn’s article, which is, as far as I know, the only work that has considered these nouns all at once. Most notably, she analysed both the synchronic attestations and the diachronic interpretations of each substantive of the kantwo-type. Given the wide amount of data she collected and the interpretations she proposed, in this paragraph I frequently refer to her article, though differing interpretations are sometimes proposed. From a synchronic point of view, only a few Tocharian B substantives pertain to this inflectional class. Their main characteristic is that they have a nominative singular -o and an oblique singular -a. As for the plural, the great majority of them fall into class vi.3 (nom. pl. -āñ, obl. pl. -aṃ, see below), while, for some others, no plural forms are attested. Furthermore, two alternating substantives, TB luwo ‘animal’ and TB āyo ‘bone’, can be included in a class in some way parallel to the kantwo-type: these

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

102

Chapter 3

words have nom.sg. -o, obl.sg. -a, but also attest the deviant plurals TB lwāsa and TB āsta (with no formal difference between nominative and oblique). Two other nouns with sa-plural are TB lyyāsa ‘limbs’ (TA lyiyā ~ lyā) and TB piltāsa ‘petals’ (TA pältwā), but the reconstruction of their singular paradigm is debated (see below). In Tocharian A, the few corresponding nouns show unmarked nominative and oblique singular forms. Judging by the comparison with Tocharian B and certain Tocharian A plural and derived forms (cf. instr.pl. käntwās-yo ‘with tongues’, °käntwāṣi ‘related to the tongue’, kātsaṣi* ‘pertaining to the belly’ < *kātsāṣi with vowel weakening), these belong to class vi.3 as well. The Tocharian A equivalents of TB luwo and TB āyo are TA lu and TA āy. As in Tocharian B, these nouns show no difference between nominative and oblique plural in Tocharian A (TA lwā and TA āyäntu). 3.5.1.1 The Members of the kantwo-type and Related Nouns Krause and Thomas (teb  §§  145, 159, 194) list six members: (1) TB kantwo, ta käntu ‘tongue, language’, obl.sg. kantwa; (2) TB kāswo ‘skin disease’, obl.sg. kāswa; (3) TB kātso, ta kāts ‘belly, abdomen, womb’, obl.sg. kātsa;51 (4) TB tāno52 ‘grain, seed’, obl.sg. tāna; (5) TB tsāro ‘monastery’, obl.sg. tsāra; (6) TB luwo, ta lu ‘animal’, obl.sg. luwa.53 In addition, at least three other nouns belong to this class: (1) TB āyo, ta āy ‘bone’, obl.sg. TB āya; (2) TB suwo ‘pig’, obl.sg. suwa; (3) TB maiyya ~ maiyyo ‘power, strength’, obl.sg. maiyya. Somewhat problematic and not listed by Malzahn is TB kāwo ‘desire’, which, according to Adams (dtb 164), has an obl.sg. kāwa. While the nominative singular is clearly attested (e.g. in NS39 b1 and in B588 b4), to my knowledge, no oblique singular form has been identified yet. However, the allomorph of the oblique singular stem can be easily inferred from secondary 51 Cf. also the compound war-katsa /wə́ rkatsa/ ‘water-belly; dropsy’ attested once in IT305 b4 (Filliozat 1948: 56 and 60). The word must be a calque from Skt. udakodara-, dakodara‘water-belly; dropsy’, a compound of (u)daka° ‘water’ and °udara- ‘belly’. Cf. also W42 a4–5 wra[a5]ntse kātsane “in (case of) water-belly (= dropsy)”. 52 Schmidt (apud EWAia 1.787) mentions a hypothetical TA tāṃ ‘grain’ without giving, however, the attestation (see also Malzahn 2011: 84 fn. 3). As pointed out by Peyrot (2018a: 258 fn. 16), this tāṃ may be an overlooked form of the homonymous obl.sg.f. of the demonstrative of remote deixis TA saṃ ‘that’. 53 As correctly pointed out by Malzahn (2011: 83 fn. 1), an obl.sg. †māskwa of TB māskwo ‘hindrance’ is never attested. The forms of the secondary cases (e.g. abl.sg. māswkameṃ [sic] ~ māskumeṃ) and the derivative maskwatstsai /maskwə́ tstsay/ speak in favour of an obl.sg. māskwä /máskwə/ (not †māskwa /máskwa/). Furthermore, this noun has a plural in -nta (cf. the derived adjective maskwantaññeṣṣe in B291 b6), which would be unexpected for a noun of the kantwo-type (Peyrot 2011: 151).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

103

cases and derivatives. Indeed, the causal kawāñ ‘out of desire’ — to be analysed as /kawáñə/54 — allows us to reconstruct the obl.sg. as kāwa* (cf. AS7L b3 [Karmavibhaṅga] läks ra misāṃts kawāñ nakṣäṃ l(āre śaul) “like the fish loses [its] dear life out of desire for meat”). A confirmation of this analysis can be found in the derivative kawātse ‘desirous’ (B516 b4),55 regularly based on the oblique singular form. In addition, one cannot claim that kāwo is a member of the okso-type (nom.sg. -o, obl.sg. -ai, stem -ai-), because a stem **kawai- should then be expected. Accordingly, TB kāwo must be assigned to the kantwo-type.56 Another noun that has not been considered by Malzahn is so far only attested in the plural in Tocharian B, nom.pl. käryāñ, obl.pl. k(a)ryaṃ ‘viscera (?)’: IT1 a4 ṣemeṃts käryāñ pruknānträ “The käryāñ of some are bounding” (cf. Broomhead 1962: 1.143–146; Wilkens & Peyrot 2017: 694) and AS7M b5 arañcä k(a)ryaṃ kerä rän(·)i///.57 This plural paradigm allows us to reconstruct the nom.sg. as karyo* /kǝ́ ryo/. The Tocharian A equivalent is TA kri ‘will, desire’ with nom.pl. käryāñ, obl.pl. käryās (dta 172, cf. TA käryāñ präṅki-ñi “[my] desires are restrained” in A115 a4). However, a translation ‘wills, desires’ for TB käryāñ does not make sense in the text and it should rather be translated ‘viscera, guts’, as Wilkens & Peyrot (2017: 693 and fn. 29) pointed out on the basis of the Old Uyghur translation ičägüläri ‘their entrails’. On etymological grounds (cf. Gk. κραδίη ‘heart’), Hilmarsson (1996: 100), followed by Adams (dtb 175), proposes a meaning ‘heart’, even though the regular word for ‘heart’, TB arañce, occurs in the same text (line a1). Therefore, it is tempting to analyse the original contrast between TB karyo* and TB arañce in light of similar pairs of synonyms referring to the heart as “the source of emotion”, on the one hand, and “the material organ”, on the other, as found in other Indo-European languages (cf. Bolelli 1948 for an analysis of ἧτορ, κῆρ, and κραδίη in Homer). 54 Cf. the similar accent position in läkleñ /ləkléñǝ/ ‘because of the suffering’. See Pinault (2008: 400 and 465). 55 This fragment is admittedly difficult to translate: the form aukatsāmat (in b4 weṣṣäṃ aukatsāmat ra māka no kawātse, cf. Sieg & Siegling 1953: 319–320) is hard to analyse and the word division is uncertain. Sieg & Siegling (1953: 320 fn. 8) propose aukat tsāmat “you will grow and increase” (cf. Adams dtb 136), but both Malzahn (2010: 547) and Peyrot (2013a: 843 fn. 1029) reject this division. For discussions, see Hackstein (1995: 338) and Malzahn (2010: 547 and 985). 56 In Tocharian A, we find an obl.pl. kāwas (A429 b7) and a derived adjective kāwaṣi (459 a3). From a formal point of view, TA kāw* may be the correspondent of TB kāwo. Carling (dta 150), however, tentatively translates this noun as ‘spring, fountain’. 57 The Tocharian verb pruknānträ in IT1 a4 corresponds to OUy. sekriyü sučıyu “springen” in the parallel passage. See Wilkens & Peyrot (2017: 685, 688 and 692) and Wilkens (2021: 602). For a new edition of AS7M, where the inflected form k(a)ryaṃ is attested in line b5, see Huard (2022: 171).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

104

Chapter 3

Also problematic is the alleged obl.sg. TB ekita ~ ekīta ‘help’ (dtb 80). No evidence of the nominative singular has been found so far, as it is only attested in the expression ekita yam- ‘to help’ (Meunier 2013: 173–174), and in some derived forms (cf. ekītatstse ‘helpful, helper’ and ekītatsñe ‘assistance’). From a derivational point of view, one might claim that it contains the suffix -ito, which also occurs in TB laukīto ‘stranger’ (to be linked with lauke ‘far’). If so, it might be assumed that the nominative singular of obl.sg. ekīta was ekīto* (cf. nom.sg. laukīto) and that the oblique singular of nom.sg. laukīto was laukīta* (cf. obl.sg. (?) ekīta).58 However, Dragoni (2022: 93–98) has recently proposed a foreign origin for ekīta. He reconstructs an adverbial formation construed with the prefix TB e(n)- and derived from a stem *kita-, the obl.sg. of a noun *kito ‘help’. According to him, this noun should in turn be interpreted as a loanword from pre-Khotanese *gīθa- ‘help’ (OKhot. gīha- ‘help’, ggīhaa- ‘helper’, Khot. ggīh- : ggīsta- ‘to help’ [Emmerick 1968b: 28–29]; cf. Av. gaēθā- ‘Wesen, Lebewesen, Welt’ [AirW 476–479]).59 Therefore, TB ekita should be considered an adverb (cf. Meunier 2013: 173), although it may have been derived from a noun ultimately belonging to the kantwo-type. On the other hand, another noun may share the same formation of TB laukīto. It has been read by Sieg & Siegling (1953: 333) as TB tekīta, a hapax legomenon attested in B530 b4 ///d vā ᛫ tekīta taśi wat « ᛫ » ya///. This fragment is a bilingual list of Sanskrit terms translated into Tocharian. Unfortunately, the Sanskrit counterpart of TB tekīta is missing, because the document is torn on both sides. Adams (dtb 322) connects this word to the action noun teki ‘disease’ and thus translates tekīta as ‘sufferer, sick person’, an oblique singular. Although this analysis is certainly possible from a linguistic point of view, the line in question may be read differently. A common difficulty in Tocharian paleography is how the signs ⟨na⟩ and ⟨ta⟩ are written and differentiated. Sieg & Siegling read three t-signs in the line, but it seems to me that the shape of the second differs from that of the other two. We therefore must decide whether the sequence should be read tekīna taśi or nekīta naśi. Before looking at the morphology of these forms, I checked how ⟨ta⟩ and ⟨na⟩ are written in the manuscript to which B530 belongs (Couvreur 1968). It seems to me that ⟨ta⟩ is usually written like our second akṣara, while 58 For a slightly different proposal, see Peyrot (2012a: 194). Cf. also Pinault (2015a: 176 fn. 39). 59 Van Windekens (1976: 176) claims that a noun TB ek* is inferable after ekaññi ‘possession’, and that this hypothetical word is a loanword from TA ek ‘fodder’. However, this hypothesis is highly improbable as it proposes a loanword from Tocharian A and as it is semantically problematic. Furthermore, TB ekaññi is related to TA akäṃtsune ‘possession’, as Carling (dta 2) and Adams (dtb 79) demonstrate.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

105

⟨na⟩ is written like the first (i.e. ⟨ne⟩) and the third (i.e. ⟨na⟩). The verbal form naśi is the expected 3sg.opt. of the verbal root TB nǝk- ‘to destroy, lose’ (Malzahn 2010: 324–326 and 681). On the other hand, TB nekīta could be the obl.sg. of a noun nekīto*. This would derive from an unattested action noun neki* ‘destruction’, regularly built on the subjunctive stem of nǝk-. If this analysis is correct, we must interpret the entire phrase as a figura etymologica with the meaning of “(s)he would destroy the destroyer”, or the like. However, the copyist of the manuscript under consideration sometimes writes ⟨ne⟩ for ⟨te⟩. Consequently, a reading {t}ekīta naśi cannot be excluded, which could be translated as “(s)he would destroy the infector/sick person” (cf. also tekanma nakṣeñca “destroyer of all diseases” in the Yogaśataka fragment AS2B a2). Therefore, both TB laukīto, ta lokit and TB nekīto*/tekīto* will be treated as members of the kantwo-type. According to Adams (dtb 141), TB auso*, a verbal noun built on the past participle of wǝs- ‘to wear, don’, appears to fit into this inflectional class. The supposed attested forms are: oblique ausa in THT1859 a1 and THT1105 b3, and locative ausane in AS4A a2. As regards the locative (listed also by Hartmann 2013: 326), the reading †ausane (AS4A a2) must now be corrected in aisene ‘in the cauldron’ (Pinault 2015b: 197). The other putative attestations of TB ausa are more difficult to analyse in both meaning and form. In particular, the reading of line a1 in the archaic fragment THT1859 is debated to such an extent that I cannot consider it a certain attestation of the noun.60 Much more certain is the reading ausa in THT1105 b3 (Karmavācanā) makā-yäkne ausa aṣitaṃ pār pitsamonta wasātai61 “you wore in many kinds, clothing (?), fur (?), plumage (?), scales (?)” (edited by Schmidt 2018: 51 and 98; cf. Tamai 2014: 369–370). All nouns attested (i.e. ausa, aṣitaṃ, pār, pitsamonta62 [?]) can be interpreted as oblique forms, but their exact meaning is uncertain, considering that they are hapax legomenona. Apparently these terms denote different kinds of human and animal hides, representing the previous lives of the character in the tale. Hannes A. Fellner (apud CEToM s. THT1105) considers our ausa the obl.sg. of

60 Adams (dtb 141) reads the line as ausa snai parnnā yāntaite and translates the sentence as “they exchanged clothes voluntarily”. This interpretation is rejected by Ogihara (2015: 106–107), who claims that the correct reading is ausa snai pernne ayāttaite. According to him, ayātaitte ‘untamed, untamable’ is the nominative singular of a te-adjective that must be linked to some other attested forms (e.g. obl.sg.m. ayātaicce, obl.pl.m. ayātaicceṃ), while ausa is a nominative singular of uncertain meaning. 61 TB wasātai is to be corrected in wäs(s)ātai (cf. lines a3 and a4 of the same text). 62 As observed by Hannes A. Fellner & Theresa Illés (apud CEToM s. THT 1105), the separation of pitsamonta into individual words is uncertain (possibly pitsa monta (?)).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

106

Chapter 3

the preterite participle of wəs- ‘to wear, don’. We will thus abandon the designation of this noun as belonging to the kantwo-type.63 Ogihara (2009: 426–427, 2011: 135 fn. 33) discovered the new inflected form mālo (in THT2382.1 b2), which appears to be the nom.sg. of the already attested obl.sg. māla, a kind of inebriating drink (translating Skt. maireya- in THT1103 b1; cf. also the derivative mālatsai ‘± drunkenness, related to mālo’ in B241 a3 [arch.]).64 This is a feminine noun (Hartmann 2013: 236). Dragoni (2022: 105–108) has also argued that the thus far overlooked nom.sg. keto ‘property, estate’ is actually attested in the magical fragment AS8B a2. Together with the obl.sg. keta and the gen.sg. ketāntse (Ot 19.1 a3), this newly identified form makes the singular paradigm of the word complete. Thus both mālo and keto belong to the kantwo-type. Another noun that can be included into the kantwo-type is TB eñcuwo ~ iñcuwo, ta añcu* ‘iron’ (see Peyrot 2008: 60 on the Tocharian B variants). In Tocharian B, this noun is attested twice at the nom.sg. (eñcuwo B255 b2, iñcwo B520 b6) and more frequently in the derived adjective eñcuwaññe ~ iñcuwaññe ‘made of iron’ (cf. Adams dtb 84). In Tocharian A, only the derived adjective añcwāṣi is attested. Based on this evidence we cannot determine whether this noun was inflected according to the kantwo-type or to the arṣāklo-type (nom.sg. -o, obl.sg. -ai, stem ending with -a-, see § 3.5.2.2).65 Nonetheless, a form eñcuwa ‘iron’, which clearly suggests that the noun is a member of the kantwo-type, is attested in THT1543.d b2 /// (tusa sū) śamaśke läkle lkāṣäṃ mäkte eñcuwa pilke wat “Because of that, the boy sees sorrow, [just] as iron or copper …”. Although the beginning of line b3 of the fragment is missing, the fact that TB eñcuwa should be interpreted as an oblique singular can be morphologically inferred on the basis of the other attested forms.66 In the list made by Malzahn (2011: 88), she includes two other substantives, TB śaro* ‘adult man, elder’ and TB ñāsso ‘part, portion’. As regards the first noun, she agrees with Peters (2004: 267 fn. 5) in reconstructing a nom.sg. śaro*, obl.sg. śara* for the attested plural paradigm nom.pl.

63 Hartmann (2013: 326), following Adams (dtb 114), interprets this noun as masculine (or alternating) on the basis of the ghost attestation in AS4A a2 (see the main text above). 64 See dtb 482. Cf. already Klaus T. Schmidt apud Tremblay (2005: 436). 65 In Del Tomba (2020a: 146–147), I included TB eñcuwo into the arṣāklo-type on the basis of the number of syllables of the word. 66 Note that THT1543.a and THT1543.d cannot belong to the same leaf. Rather, THT1543.e is the right part of a leaf beginning with THT1543.a, of which the recto and the verso need to be inverted.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

107

śrāy < *śráñǝ (?), obl.pl. śrānäṃ.67 This interpretation is in my opinion unconvincing. The oblique plural of this noun clearly shows a nasal as part of the stem that does not fit well with the other nouns of the kantwo-type: indeed, while the latter attests a plural nom. -āñ, obl. -aṃ, the reconstructed plural of śrāy may have been nom.pl. *-añǝ, obl.pl. *-anən. If, as Peters argues, this word went back to an extended *nt-stem, i.e. PIE *ǵerh2-nt-s, we should still see the outcome of *-t- somewhere in the paradigm, as in the case of TB walo ‘king’, nom.pl. lāñc < *ulH-nt-es, obl.pl. lāntäṃ < *ulH-nt-n̥ s (Lubotsky 1994a). As a consequence, Pinault’s diachronic interpretation is preferable, as he postulates a Proto-Tocharian stem *śəran- (Pinault 2008: 484–485). Furthermore, given the fact that no singular forms are attested and that the plural nom. śrāy, obl. śrānäṃ has no immediate parallels in Tocharian, I believe that the singular of this word cannot be established with any certainty.68 The classification of ñāsso ‘part, portion’ is also doubtful. According to Malzahn (2007), this word is attested in two documents: once in B547 a2 as a nom.sg. TB ñasso (with -a- /ə́ /?), and twice in THT1168 b4 as an obl. sg. TB ñāssa. The first fragment is a bilingual word-by-word translation of doctrinal Sanskrit expressions, in which TB s(e) ñasso would be the translation of Skt. y[o](ṃ)śaś, a sandhi-variant of yaḥ aṃśaḥ (Sieg & Siegling 1953: 342 fn. 13). Malzahn interprets nom.sg. ñasso ‘share, part, portion’ as an orthographic error for ñāsso and further links this word to TB ñāssa (THT1168 b4), analysing it as an oblique singular of ñāsso. The fragment THT1168 is part of an avadāna that tells the Buddhist story of the merchant Anāthapiṇḍika. According to Ogihara’s edition (2012a: 177–180), THT1168 should be joined with THT3034, with remnants of line b1 preserved in THT3034 only. Therefore, ñāssa is attested on line b5 of the leaf, which should be read as follows: THT1168+THT3034 b5 /// ⋅pä kete pelkiñ ñāssa uppāl ñaskeṃ po ñāssa sanai tinārrs(a) up(pāl)/// (Ogihara 2012a: 178). Malzahn’s translation of the first portion of the line is “… on his behalf, they demand blue lotus as a share (ñāssa). The entire share (ñāssa) of one coin (obl.) this one (nom.sg.fem.) …”.

67 On *-áñǝ# > *-áyǝ#, see Carling (2003: 93), Pinault (2008: 485), Peyrot (2012a: 185) and §  3.5.2.5. Adams (dtb 705) suggests that TB śrāy is from nom.pl. *ǵerh2-u̯ es, an ablaut variant of Gk. γραῦς ‘old woman’. However, Gk. γραῦς is rather from *ǵreh2-i̯u- (gew 1.324; Beekes 2010: 285), and Adams’ acc.pl. *ǵerh2-u̯ n̥s cannot be the ancestor of the Tocharian obl.pl. śrānäṃ. 68 Peters (2004: 267) proposes that TB pānto ‘support, companion’ should be included in the kantwo-type. On this noun and the problematic nom.pl. pantañ, see Malzahn (2011: 95 fn. 31) and Dragoni (2022: 140–144).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

108

Chapter 3

In defense of her analysis, Malzahn points out that THT1168 is more carefully written than B547,69 and therefore argues that ñasso is a mistaken form to be corrected in ñāsso. Although a wrongly spelled vowel is possible in itself, Peyrot (2008: 63–63) and Ogihara (2012a: 179) have shown that TB ñāssa in THT1168 should be interpreted as the perlative singular of TB ñyās ‘desire’ (with a development of ñy- > ñ- in initial position, otherwise attested in some other Classical Tocharian B documents). Contrary to Malzahn (2007: 242 fn. 22), who claims that it is unlikely that ñāssa is a perlative of ñyās because this document does not show “any eastern TB language features”, Ogihara (2012a: 179) points out that the copyist of the manuscript to which THT1168 belongs was a Classical-Late Tocharian B speaker. Furthermore, the common figura etymologica ñyāssa ñäsk- ‘to seek with desire’ attested also in THT1168+THT3034 b5 confirms this analysis. As a consequence, the entire line should be translated as follows: “… for whose sake they wish with desire a lotus. With all desire a lotus for one denar …” (translation adapted from Ogihara 2012a: 180). Therefore, TB ñasso in B547 (not †ñāsso) is to be considered a hapax legomenon. Finally, a last noun that can be included in the kantwo-type is TB patso ‘pollen, stigma’.70 It is a problematic word. From a synchronic point of view, it is attested several times in the nom.pl. ptsāñ (spelled once as pätsāñä in W38 a5): it occurs twice in the Berlin collection (B497 b8 (ptsā)ñä; B498 a8 ptsā(ñ)), twice in the Paris collection (AS3B a3 and b5 ptsāñ), and eleven times in the Weber series (W4 b1; W7 b4; W19 b2 and b5; W20 a5; W21 b4; W28 a6; W29 b1; W32 a5; W38 a5; W39 a3). Quite remarkably, TB ptsāñ is only found in nominal phrases with the adjective kurkamäṣṣi ‘pertaining to saffron (pl.)’. The rest of the paradigm is less easily established because all other inflected forms are found in fragmentary contexts. The nominative singular may be attested in the archaic document IT881 b2 /// pätso śkwäśko ma/// “… pollen, barberry …” (cf. Michaël Peyrot apud CEToM), while the oblique singular can be probably found in IT244 a3 ///kektseṃtsa || patsa tā///, “… on the body || pollen this …”.71 In addition, a form patsaṃ is attested in IT305 a4 and a5, which Sieg (1955: 70), Thomas (1964: 217), and Broomhead (1962: 1.20) interpret as the obl.pl. of TB patso ‘pollen, stigma’. Adams (dtb 388) objects that “the difficulty of 69 70 71

Other misspellings are attested in this document, e.g. monophthongisation of au into o, cf. onästrä for aunasträ in B547 a1. See Peyrot (2008: 53–54). See also Hartmann (2013: 70–71). On the basis of the prevalence of TB patso in agreement with the nom.pl.m. kurkamäṣṣi (Hartmann 2013: 215), the supposed obl.sg.f. tā cannot agree with patsa in IT244 a3, also because the word order clearly suggests that the demonstrative refers to a following word. Moreover, tā/// occurs at the beginning of a broken line, where it may stand for obl.sg. tā(na) ‘seed of grain’, among many other words.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

109

associating patsaṃ [obl.pl.] with ptsāñ [nom.pl.] in a single paradigm argues against the equation [of patsaṃ as an inflected form of patso]”. In other words, the claim by Adams is that we would expect ptsāṃ* /p(ə)tsán/ as the obl.pl. of a noun of the kantwo-type. However, there are parallels contradicting this claim. Indeed, nouns of the kantwo-type seem to display a contrast between nom.pl. -āñ /-áñ(ǝ)/ and obl.pl. -aṃ /-an/ in Tocharian B. An example is TB tāno ‘seed’, which has nom.pl. tanāñ /tanáñ(ǝ)/ (cf. tanāñä IT305 b3; tanāñä W11 a6), obl.pl. tānaṃ /tánan/ (PK DA M 5067.37 and .36 a36, a40) and further nom.pl. käryāñ /kəryáñ(ǝ)/ vs. obl.pl. karyaṃ /kə́ ryan/. Following Filliozat (1948), however, Adams (dtb 388) is in my opinion correct in not interpreting our patsaṃ as the obl.pl. of patso ‘pollen, stigma’ but as a separate word meaning ‘radish’ (= Skt. mūlaka-). The medical fragment IT305 deals with the incompatibility of food and drink. Lines a4–b1 specifically deal with the incompatibility with milk:72 IT305 a3–b1 ᛫ klyemoṃ warne lwasāts misampa [a4] mit panit wirot ᛫ malkwer patsaṃ uppāläṣṣana witsakaṃmpa kärkoṣ śātrempa mā swālle ᛫ kuse śuwaṃ patko mä[a5]sketär ᛫73 tranto naitwe korne karkar yamaṣäṃ ᛫ patsaṃ śemesteṃ kwrarāk arkwañai śeśuwermeṃ mā malkwer yokalle ᛫ kuse yokäṃ [a6] kaswā­tse mäsketär ᛫ ekṣinekäṃñana misa akauwse ᛫ t· y· k(u)lāst74 tesa wärñai āmpäl śuke salyiṃ ṣpä malkwermpa[b1]eweta || “Honey (madhu-) (or) jaggery (guḍa-) (is) incompatible with the flesh (piśitāni) of animals in marshy water (ānūpa-udaka-). Milk (payas-) (or) radish (mūlaka-) is not to be eaten with lotus (bisa-) roots/filaments [or] with germinated grains (virūḍhadhānya-). Whoever eats (it): the patko-disease occurs. [It] causes tranto-disease, naitwe-disease, [and] cancer in the throat (galagaṇga-). After having eaten radish (mūlaka-), garlic (laśuna-), Hyperanthera Moringa (kr̥ ṣṇagandhā-75), [and] shrubby 72 73 74

75

In the translation I give possible Sanskrit equivalents of technical terms in so far as bilingual evidence is available. It is possible that the copyist forgot to write a section (cf. the Sanskrit parallel below). The above restoration allows us to interpret TB kulāst ‘horse gram’ as a hypersanskritisation of Skt. kulattha-, as Skt. -st- regularly evolved -tth- in Middle Indic (cf. von Hinüber 2001: 181). It is curious that the same phenomenon can also be found in LKhot. kūlasta‘horse gram’, attested three times in the Khotanese Siddhasāra (Si §§ 1.55, 3.12, 22.19) as the regular translation of Skt. kulattha-. The exact identification of this plant name is uncertain. The translation follows mw 307. Sharma (2014: 189) interprets kr̥ ṣṇagandhā- as a synonym of śigru-, which Emmerick (1982) translates as ‘horse-radish’ in his translation of the Tibetan Siddhasāra.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

110

Chapter 3

basil (arjaka-), milk (payas-) is not to be drunk. Whoever drinks [it] becomes affected by leprosy (kuṣthābādha-). Meat of ekṣinek, akauwse … horse gram (kulattha-), etc. sour tastes (amla-phala-?) and salts (lavaṇa-?) [are] in conflict with milk (payas- or kṣīra-)76”. Filliozat (1948: 58–59) identified several medical texts that match some of the incompatibilities listed in the Tocharian fragment. Those which more closely resemble the portion of the Tocharian text are Carakasaṃhitā (Ca.), and Vāgbhaṭa’s Aṣṭāṅgahr̥dayasaṃhitā (AH) and Aṣṭāṅgasaṃgraha (AS), from which the Sanskrit equivalents in the Tocharian translation have been taken. The entire passages where the Sanskrit equivalents of Tocharian patsaṃ occurs are the following:77 Ca. Sū. 26.84 (2)78 grāmyānūpa-udaka-piśitāni ca madhu-tila-guḍa-payomāṣa-mūlaka-bisair virūḍha-dhānyair-vā naika-dhyamadyāt tanmūlaṃ hi bādhiryāndhyavepathu-jāḍya-kalamūka-tāmaiṇmiṇyamathavā maraṇamāpnoti “Meat (piśitāni) of domestic (grāmya-), marshy (ānūpa-) and aquatic (udaka-) animals should not be taken mixed with honey (madhu-), sesamum (tila-), jaggery (guḍa-), milk (payas-), black gram (māṣa-), radish (mūlaka-), lotus stalk (bisair) or germinated grains (virūḍha-dhānyair) because it causes deafness, blindness, tremors, coldness, indistinct voice, nasal voice or death”. AS Sū. 9.279 = AH Sū. 7.5080 grāmyānūpa-udaka-piśitāni madhu-guḍa-tila-payomāṣa-mūlaka-bisair virūḍha-dhānyaiś81 ca naika-dhyamadyāt “Meat of domestic, marshy, and aquatic animals should not be taken mixed with honey, jaggery, sesamum, milk, black gram, radish, lotus stalk or germinated grains”. 76

The last portion finds a closer match in Siddhasāra 1.56 śākāmla-phala-piṇyāka-kulatthalavaṇaiḥ saha karīra-dadhi-matsyaiś ca prāyaḥ kṣīraṃ virudhyate “For the greater part (prāyaḥ) milk (kṣīraṃ) does not agree (virudhyate) with vegetables (śākā-), sour fruits (amla-phala-), oil-cakes (piṇyāka-), horse gram (kulattha-), salt (lavaṇaiḥ) and together with bamboo shoots (karīra-), sour milk (dadhi-) and fish (matsyaiś)”. 77 In the Sanskrit translation I mark in italics those technical terms that have an equivalent in the Tocharian text. 78 Edition and translation by Sharma (2014). 79 Edition by Āṭhavale (1980). 80 Edition by Das & Emmerick (1998). 81 AH has madhu-guḍa-tila-payomāṣa-mūlaka-bisa-virūḍhakaiś.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

111

Gender in the Inflection of the Noun

Ca. Sū., 26.84 (4) na mūlaka-laśuna-kṛṣṇagandhārjaka-sumukha-surasādīni bhakṣayitvā payaḥ sevyaṃ kuṣṭhābādha-bhayāt “After eating radish (mūlaka-), garlic (laśuna-), śigru (kṛṣṇa-), arjaka, tulasī etc. one should not take milk because of the risk of leprosy”. In line a5 of IT305, TB patsaṃ clearly translates Skt. mūlaka- ‘radish, Raphanus sativus’. On the other hand, in a4 one might think that TB patsaṃ corresponds to Skt. māṣa- ‘bean, black gram’ and that TB uppāläṣṣana witsakaṃmpa translates Skt. mūlakabisair. However, it is more likely that TB witsakaṃ ‘roots’ is a Tocharian addition, in an attempt to clarify which part of the lotus is not compatible with milk. Furthermore, one cannot rule out the possibility that witsakaṃ does not refer to the ‘roots’ of the lotus but to its filaments. As a parallel, the ingredient upālṣa witsako (B50o–502 b11) in the recipe for the medical ghee named mahāvaidehika-gr̥hta must correspond to Skt. padmakesara- ‘filament of the lotus’ (for manuscript padmakesarä [IOL Khot 91/3 a2–3]) and LKhot. padmakyesärä- ‘id.’ (IOL Khot 91/3 b4–5) in the parallel portion of Jīvakapustaka 11 (Maue 1990: 164). Thus patsaṃ is not the obl.pl. of patso ‘pollen, stigma’, but an independent word with the meaning ‘radish (Raphanus sativus Linn.)’, as suggested by Adams (dtb 388) and Filliozat (1948: 129). All in all, Malzahn’s list (2011: 88) can now be amended to contain the following nouns:82 Table 20

Nouns with nom.sg. -o, obl.sg. -a

TB nom.sg.

gender obl. sg. nom. pl.

obl. pl.

stem

TA

eñcuwo ‘iron’ kantwo ‘tongue’ karyo* ‘±viscera’

?

eñcuwa

—

—

eñcuwa-

añcwā-

m.

kantwa

käntwāñ*

kantwaṃ

käntwā-

käntu

?

—

käryāñ

k(a)ryaṃ

—

kri (m.)

82

The list could of course be expanded if for additional nouns the relevant forms are identified in the texts. Other nouns that are not listed, but which could probably be listed here too, will be discussed in the following paragraphs of this section. For the peculiar inflection of maiyya ~ -yo, see the relevant section below.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

112 Table 20

Chapter 3 Nouns with nom.sg. -o, obl.sg. -a (cont.)

TB nom.sg.

gender obl. sg. nom. pl.

obl. pl.

stem

TA

kāwo ‘desire’ kāswo ‘skin disease’ kātso ‘belly’ keto ‘estate’ tāno ‘seed of grain’ nekīto*/tekīto* cf. above patso ‘pollen, stigma’ mālo ‘± alcohol’ laukīto ‘stranger’ tsāro ‘monastery’

f.

kāwa

—

—

kawā-

—

f.

kāswa*

—

—

kaswā-

—

f.

kātsa

katsāñ

—

katsā-

kāts

?

keta

—

—

ketā-

—

f.

tāna

tanāñ

tānaṃ

tanā-

—

m.?

nekīta

—

—

—

—

m.

patsa

p(ä)tsāñ

—

—

—

f.

māla

—

—

malā-

—

m.?

laukīta*

—

—

—

lokit

f.

tsāra

—

—

—

—

suwo ‘pig’ luwo ‘animal’

m.?

suwa

—

—

s(u)wā-

—

alt.

luwa

lwāsa

lwāsa

lwā-

lu

āyo ‘bone’

alt.

āya

āsta

āsta

ayā-; astā- āy

maiyya ~ -yo ‘power’

f.

maiyya

maiyyāñ ~ maiyyaṃ ~ maiyyāmaiyyana ? maiyyana

—

In addition, there are also some nouns for which only the obl.sg. TB -a is attested. They are not numerous (Malzahn 2011; Pinault 2012a): obl.sg. TB weta ‘battle’ (fem.; cf. also obl.pl. wetaṃ), obl.sg. TB śarka ‘song, music’, obl.sg. TB

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

113

śāmpa ‘conceit; pride’. These nouns are usually treated as belonging to a separate inflectional class with identical nom.sg. and obl.sg. -a. However, given the fact their nom.sg. is not attested, one cannot exclude that they actually belong to the kantwo-type (with nom.sg. -o*, see Malzahn 2011). As we can see, neither obl.sg. weta nor obl.sg. keta show a-umlaut (Pinault 2012a: 197), and the Tocharian A equivalents sometimes have different suffixation and inflection (cf. TA wac ‘battle’ ≅ tb weta and TA tsärk ≅ tb śarka).83 3.5.1.2 Analysis of the Nouns This section represents the central part of my discussion on the kantwo-type, in which I deal with the diachronic evolution of some of the nouns identified in the previous paragraph. Because of its etymology, TB kantwo, ta käntu ‘tongue’ is the obvious choice to start our discussion. The outcome of *-eh2 in word-final and internal position will be outlined. I will then deal with nouns that supposedly go back to the PIE type in *-eh2 > *-ā and some related nouns. Following the most recent discussions about these nouns (Dragoni 2022), it will be highlighted that several members of the kantwo-type are actually loanwords from prehistoric phases of Khotanese. This will confirm the archaicity of the kantwo-type as an inflectional class. As for the inherited nouns, I will discuss whether they can go back to the same PIE inflectional type or if phonological and/or analogical changes have mixed up different inherited stem types. Afterwards, I will deal with the remaining substantives. I will first investigate the origin of TB suwo ‘pig’ and TB luwo, ta lu ‘animal’, while the last two nouns, TB āyo, ta āy ‘bone’ and TB maiyya ~ maiyya ‘power, strength’, will be treated separately. Some final remarks will conclude the section (§ 3.5.1.3). TB kantwo, ta käntu ‘tongue’ The Tocharian word for ‘tongue’ has attracted the interest of many scholars, since it is the only member of this class for which cognates are found in most Indo-European languages. Before proceeding to the discussion of its historical development, however, the gender of the noun in both Tocharian languages must be clarified. In the singular, TB kantwo is consistently found in agreement with a masculine modifier:

83 Dragoni (2022: 242–243) also includes TB uwātano*, ta wataṃ* ‘Khotanese’ and TB kāmarto*, ta kākmart ‘chief’ into the kantwo-type. On TB kāmarto*, ta kākmart, see further below.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

114

Chapter 3

B118 b7 (Rājavr̥tti, literary) ārkwi mäsketär-ne käṃtwo white.m.nom.sg become.3sg.prs-3sg.suff tongue(m).nom.sg “His tongue becomes white”. B408 a6 (literary) mā po yātalle śāmñe kantwasa neg all be.able.grnd.m.nom.sg human.m.obl.sg tongue(m).perl.sg weṃtsi speak.inf “Not all can be said with the human tongue”. As for the plural paradigm, the oblique kantwaṃ /kǝ́ ntwan/ is attested in THT 1363.e b4 as kaṃtwaṃ and can be easily inferred on the basis of the perlative plural attested in AS17H a3 (Supriyanāṭaka) ᛬ colormeṣṣeṃ kantwaṃtsa “with tongues of colorme”. Although the meaning of colormeṣṣeṃ is unknown, it can be formally analysed as the obl.pl.m. of an adjective TB colormeṣṣe* in agreement with the perl.pl. kantwaṃtsa (cf. also colormetse NS11 b1; colormecce IT823 a2; colormeṃtsa B355 b2). This plural concord is not listed in Hartmann (2013: 327) or in Adams (dtb 147), as they report the gender of the noun as masculine or alternating. However, even without the plural agreement in AS17H, TB kantwo could not have been interpreted as an alternating noun because, if it were, it would have identical nominative and oblique plural forms. The gender of the Tocharian A equivalent, TA käntu, is more difficult to establish. Hilmarsson (1996: 79) claims that we have only three agreement sets: TA käntu agrees twice with a masculine modifier (A300 a8 and YQ II.10 a8), and only once with a feminine modifier (A57 a2), both in the singular. These contradictory environments led scholars to lemmatise the noun as both masculine and feminine (e.g. dta 163; teb § 194). The occurrences in the singular are given below (Hartmann 2013: 309–310):84 A300 a8 (Maitreyasamiti-Nāṭaka) napeṃsinäṃ käntuyo human.m.obl.sg tongue(m).instr.sg “with human tongue …”.

84

An additional agreement may be found in A300 b1 pälkāc tāpärk ṣñi käntwis ñareṣinäṃ /// “Now look at your own tongue of hell …”. However, it cannot be excluded that the obl.sg.m. ñareṣinäṃ ‘pertaining to hell’ is in agreement with another nominal lost in the lacuna. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

115

Gender in the Inflection of the Noun

YQ II.10 a8 (Maitreyasamiti-Nāṭaka) wärts knuṃts käntu broad.m.nom.sg supple.m.nom.sg tongue(m).nom.sg “wide and supple (is) the tongue”. A57 a2 (Rūpyāvatāvadāna) opal-yokāṃ käntuyo lotus-coloured.f?.obl.sg tongue(?).instr.sg “with a lotus-coloured tongue” (cf. tg § 58 p.33) The only attested plural is the following: A356 b2 (literary) triśkās käntwāsyo ?? tongue(?).perl.pl Let us start with the plural form. Hartmann (2013: 310) hesitantly gives the instrumental plural käntwāsyo as agreeing with TA triśkās, which he interprets as a hapax legomenon of uncertain meaning and formation. However, another inflected form seems to be attested in a broken passage of A375 a2 as triśkaṃ which, if an adjective, could formally be a feminine plural in agreement with pātruk·/// ‘skull(s)’ (likewise tg § 174).85 Otherwise, TA triśk* may be a noun with plural TA -āñ | -ās and loc.sg. triśkaṃ.86 In view of these inconclusive data, I agree with Sieg, Siegling & Schulze (tg § 58, p. 33) that TA triśkās is too uncertain to be used for identifying the gender of käntu. Returning to the singular paradigm we see that, in the first two passages, TA käntu and käntuyo agree with the targets wärts ‘broad’, knuṃts ‘supple’, and napeṃsinäṃ ‘human’, three adjectives inflected as masculine. Based on this nominal agreement we should consider TA käntu a masculine noun. However, the problematic passage in A57 a2 seems to contradict this analysis, since oppal-yokāṃ ‘lotus-coloured’ is generally interpreted as a feminine oblique singular. Hartmann (2013: 99–100) has correctly questioned this analysis. He lists a range of cases where the adjectival compounds of the type oppal-yok take an obl.sg. TA -āṃ when they refer to both masculine and feminine nouns.87 This leads to the conclusion that they are not gender-differentiated. 85 86

On the correct meaning of pātruk, see Malzahn (2014a: 91–92). No certain etymology has been proposed for the two forms discussed. Poucha (1955: 133) alone suggests a link with the verb TA träyk- ‘be confused, faint’. 87 I do not think that compounds of the type oppal-yok* can be interpreted as a “Karmadhārayabildungen”, as Hartmann seems to argue. These compounds are evidently of the bahuvrīhi-type, as demonstrated by the most prominent member of this type of Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

116

Chapter 3

Since substantives with the Tocharian A plural paradigm nom. -āñ, obl. -ās (cf. also wäṣta-käntwāñ ‘forked tongues’) can hardly be interpreted as alternating, it follows that TA käntu is definitely a masculine noun. This fits the Indo-European comparative situation nicely: given the fact that Avestan, Balto-Slavic, and some Old Irish and Breton formations point to the reconstruction of the noun as masculine in Proto-Indo-European (cf. AiGr 2.492; EWAia 1.592), Tocharian seems to preserve the original state of affairs. After having determined that ‘tongue’ is masculine in both Tocharian languages, the historical evolution of the noun is to be discussed. TB kantwo, ta käntu can be traced back to the familiar PIE word for ‘tongue’, PIE *dn̥ ǵhu̯ eh2-, through metathesis of *dn̥ ǵh- > *ǵn̥ dh- (Ringe 1996: 45–46; Pinault 2008: 428). The singular paradigm nom. -o, obl. -a has given rise to debate as far as the outcome of *-(e)h2 is concerned. For this reason, it is best to start the diachronic analysis of the kantwo-type with this noun. I will first deal with the origin of the nom.sg. -o, and then with the obl.sg. -a. In order to explain the nominative singular -o, three different proposals can be found in the literature: (1) Asigmatic nominative singular, PIE *-eh2, which regularly yielded TB -o. Accordingly, the nom.sg. can be reconstructed as PIE *dn̥ ǵhu̯ eh2 > PT *kəntwo > TB kantwo (Hilmarsson 1986a: 18; Pinault 2008: 428); (2) Sigmatic nominative singular, so that TB -o is the outcome of a pre-PT form with final *-ās (< PIE *-eh2-s), which yielded *-o(s) before the loss of final *-s. Thus PIE *dn̥ ǵhu̯ eh2s > PT *kəntwo > TB kantwo (Peters 1991; Kim 2009a; Malzahn 2011), while PIE *-eh2 > PT *-a; (3) TB kantwo does not derive from PIE *dn̥ ǵhu̯ eh2 directly, but rather it originated from a nasal-extended variant. The new nominative singular *-ōn first became *-ō(n) and then TB -o (Adams 1988b: 13–14, 2015: 177; Pan 2020: 133–134). The reason why Adams reconstructs TB kantwo as an old ōn̆ -stem (hypothesis 3) is twofold. To begin with, he argues that PIE *eh2 first became PT *a and then TAB a, in both internal and final positions; however, if PIE *-eh2was in the proximity of an etymological nasal, the sequence *-eh2N(-) should have given PT *-oN(-), through rounding of the vowel (Adams 1988a: 20). As compounds, TB ysā-yok, TA wsā-yok ‘gold-coloured’, calque from the Sanskrit bahuvrīhi suvarṇa-rūpa- (Pinault 2008: 562).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

117

a consequence, reconstructing a nom.sg. PIE *dn̥ ǵhu̯ eh2, acc.sg. *dn̥ ǵhu̯ eh2-m as the ancestors of TB nom.sg. kantwo, obl.sg. kantwa would make no sense according to Adams’ assumptions, since a paradigm with nom.sg. **kantwa, obl.sg. **kantwo is expected (i.e. exactly the opposite of the attested forms). Secondly, he claims that, within Indo-European, Tocharian is most closely related to Germanic. One of the similarities singled out by Adams is the extension of n-stems in both these Indo-European branches (Adams 1984). The same extension would have also affected TB kantwo, ta käntu, which has a nom.pl. -ñ < PIE *-n-es. As a consequence, he claims that TB kantwo mirrors Goth. tuggo (< PGerm. *tungōn-, Ringe 2006: 81; Kroonen 2013: 526), as both reflecting PIE *dn̥ ǵhu̯ ōn or PIE *dn̥ ǵhu̯ eh2ōn (Adams 2015: 177). In my opinion, these reconstructions are questionable. Indeed, there is no evidence that Tocharian had a Germanic-like distinction between strong and weak inflection (Fellner 2013: 20; Jasanoff 2018). Furthermore, it is commonly agreed that PIE *-eh2- did not develop into PT *-a-, but rather into TB -o-, TA -a-, -o- (cf. e.g. PIE *bhréh2-tēr > TB procer, ta pracar ‘brother’; PIE *u̯ éh2stu- > TB ost, ta waṣt ‘house’). On the other hand, the development of *-eh2 in word-final position is still debated. The two remaining explanations for the origin of the nom.sg. -o relate to this development. In order to assess these opposing theories we now turn to the reconstructed inflection of this noun in Proto-Indo-European. As pointed out above, the word for ‘tongue’ is attested in several IndoEuropean languages, though it has undergone irregular and analogical changes: the initial l- in Lat. lingua (cf. also the expected OLat. dingua), Lith. liežùvis, and Arm. lezu has probably been influenced by the outcomes of the PIE root *lei̯ǵh- ‘to lick’ (liv 2 404; Olsen 1999: 67; de Vaan 2008: 343); in Sanskrit, we find a feminine ā-stem, Ved. jihvā́-, with -ā- extended throughout the whole paradigm, along with a feminine ū-stem juhū́- ‘tongue; ritual spoon’ (EWAia 1.591; Pisani 1954: 143–144); in Old Persian and Germanic, it became an n-stem, cf. OP hạzān-, acc.sg. hạzzānam (Skjærvø 2007: 886) and PGerm. *tungōn (Ringe 2006: 81–82; Kroonen 2013: 526–527); Old Prussian insuwis and OCS językъ display loss of initial *d- before syllabic nasal and resuffixation with *-kъ in Slavic (Derksen 2015: 285); finally, in Celtic this noun became a t-stem, PCelt. *tangwāt- (Matasović 2009: 368).88 Among all these cognate formations, only 88 The main work on the evolution of the Celtic word for ‘tongue’ is Widmer (1997). He shows that nouns that originally belonged to other stems adopted the inflectional pattern of the t-stems for different reasons. As far as the word for ‘tongue’ is concerned, he argues that PCelt. *tangu̯ ā- was remodelled as a t-stem (PCelt. *tangu̯ āt-) because the regular outcome of the paradigm of this hysterodynamic noun would have created a unique and isolated inflection in Celtic. Widmer’s theory implies that the original sigmatic nom.sg.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

118

Chapter 3

Av. hizuuā- ‘tongue’ helps us to reconstruct the PIE inflectional type of this word. It is therefore worth having a closer look at the attested paradigm of Av. hizuuā-:89 acc.sg. YAv. hizuuąm (< PIE *-u̯ éh2-m), gen.sg. OAv. hizuuō (< PIE *-uh2-és), instr.sg. OAv. hizuuā (YAv. hizuua) ‘with the tongue’ (< PIE *-uh2-éh1) (Beekes 1985: 39–42; EWAia 1. 591–592; Martínez & de Vaan 2014: 60).90 This paradigm points to the reconstruction of a hysterodynamic type for Proto-Indo-European, with ablauting suffix *-éh2-/*-h2- (Kuiper 1942: 15; Peters 1991: 242; UrIG § 3h.2. p. 18): Table 21

Hysterodynamic paradigm of pie *dn̥ ǵhu̯ éh2-

case

R

S

E

‘tongue’

nom.sg. gen.sg. acc.sg.

— — —

é — é

— é —

*dn̥ ǵhu̯ éh2(-) *dn̥ ǵhuh2-és *dn̥ ǵhu̯ éh2-m

The nominative singular of the Avestan word is more difficult to reconstruct since it is only attested in compounds. I found the following attestations:91 Av. hizuuā̊.uxδāiš ‘parole prononcée par la langue’ (Y. 18.1; Y. 47.2; Y. 51.3), OAv. *tangu̯ ās was analogically influenced by the nom.sg. *-V̄ -s of the t-stem (< PIE *-Vt-s). This view has been accepted by some scholars (e.g. Matasović 2009), but there may be some problems of relative chronology. Firstly, as we shall see, the reconstruction of a sigmatic nom.sg. for PIE *dn̥ ǵhu̯ eh2- is not certain. Secondly, in Proto-Celtic the t-stems were still not a productive morphological class (Vijūnas 2009). One could draw an optimistic picture according to which this trend of attracting nouns from various classes to t-stems was only occasional in Proto-Celtic, but became more productive later, especially in Irish. However, the list of t-stems with a long vowel before the consonant, i.e. with nom.sg. *-V̄ (t)-s, includes only few substantives, and for many of them a Proto-Celtic reconstruction is impossible. Indeed, they are not listed in Matasović’s dictionary (2009). As a consequence, the reanalysis of the PCelt. word for ‘tongue’ from an ā-stem to a t-stem happened at a stage in which the nouns with t-inflection were few, particularly those with nom. sg. *-V̄ s. 89 For the evolution of PIE *dn̥ ǵhu̯ eh2- in Indo-Iranian, see EWAia 1.591–593 and Lipp (2009: 1.188–189), who reconstructs the following transitional stages: IIr. *ǰiǰhu̯ aH- > PIr. *dzi­dzwā > *[zidzwā] (dissimilation) > *[sidzwā] > OAv. hizuuā-. See also de Vaan (2011: 6). 90 On Av. hizū- and the instrumental plural OAv. hizubīš, see also Benveniste (1954: 30–31), Kuiper (1942: 16, 1978), and Peters (1991: 243). 91 The translations given here follow Kellens & Pirart (1988–1994). On the compound hizuuārəna ‘by moving the tongue’ (Yt. 5.6), see Oettinger (1983: 187–188), who reconstructs *hizuuā-arnā- ‘by a tongue movement’.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

119

hizuuā̊.āuuərətō ‘prisonnier de la langue’ (Y. 45.1), YAv. hitō.hizuuā̊ ‘dont la langue est liée’92 (Y. 65.9). The interpretation of Av. hizuuā̊° as the first member of the compound is disputed in both the linguistic and the philological analysis. For this reason, the two modern editions of the Gāthās (Kellens & Pirart 1988–1991 and Humbach 1991) have different readings: on the one hand, Kellens & Pirart have hizuuā̊° because it is “massivement imposé par la tradition manuscrite”; on the other hand, Humbach does not analyse the Old Avestan forms as compounds, emending hizuuā as an instrumental singular from hizū-. Humbach argues that the variant hizuuā̊ uxδāiš “by thought (voiced) by one’s tongue” is due to corruption, because the final -ā of hizuuā would have been assimilated to the initial u- of uxδāiš, due to the oral transmission of the text. Similarly, the vast majority of the manuscripts read hizuuā̊ for the sequence drəguuā̊ hizuuā̊ āuuərətō “the deceitful one, invited by one’s tongue” (Y 45.1), which, according to Humbach (1991: 165), normalised the orthographic variant with -uuā̊. Although Kellens & Pirart maintain the reading with hizuuā̊, they state that -ā̊ is an “absurd terminaison”, explaining the final vowel as a peculiarity of this word in the internal compound boundary.93 As a matter of fact, hizuuā° and hizuuā̊° alternate frequently in the manuscripts, but the variant hizuuā̊° is considered a bizarre form by almost all experts of Avestan.94 The nom.sg. YAv. hitō.hizuuā̊ (Y. 65. 9) is even more difficult to analyse. On the basis of this form, Peters (1991), Widmer (1997), and Malzahn (2011) reconstruct a sigmatic nominative singular PIE *dn̥ ǵhu̯ éh2s: indeed, from a diachronic point of view, only a final sequence PIr. *-ās (< *-eh2-s) turned into Av. -ā̊, while PIr. *-ā (< *-eh2) yielded Av. -ā.95 However, I believe that YAv. hitō.hizuuā̊ comprises insufficient evidence for arguing that the PIE nominative singular was sigmatic, because the nominative singular of hizuuā- never occurs as an

92

This compound is usually translated as ‘having a bound tongue’. For the new translation of hitō.hizuuā-, see Kellens (2009: 333). 93 For alternative proposals on the interpretation of hizuuā̊.āuuərəta-, see Kuiper (1978: 12ff.), Kellens & Pirart (1991: 187–188), and Kellens (1994: 60–61). 94 Cf. Kuiper (1978: 16), who argued that readings with hizuuā̊° must be corruptions for hizuuā°. For the distribution of the variants, see Pirart (1986: 188) Cf. also Skjærvø (2007: 886), who puts a question mark after a hypothetical nominative singular attestation of hizuuā-. 95 It seems that the supporters of the reconstruction of a sigmatic nom.sg. come from the School of Vienna, where they certainly attained Professor Jochem Schindler’s classes. In EWAia under the etymological discussion of Skt. jihvā́-, Mayrhofer refers to Schindler’s reconstruction of Av. hizuuā̊ < PIE *dn̥ ǵhu̯ éh2s. However, as far as I know, Schindler has never discussed this reconstructed form in his publications.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

120

Chapter 3

independent word and is only attested in compounds.96 Furthermore, in the Frahang ī oīm the gloss of Phl. uzwān ⟨ʾwzwʾn⟩ (cf. ManMP ⟨ʿzwʾn⟩ ‘tongue’) is Av. hizuua (nom.sg.), not hizuuā̊ (EWAia 1.591; Reichelt 1900: 187; Klingenschmitt 1968: 53). However, the glossary entry cannot be considered as probative evidence since it could have been based on other inflected forms. Given the above, I do not believe that we have sufficient evidence in support of the reconstruction of a sigmatic nom.sg. for the PIE word *dn̥ ǵhu̯ éh2-, nor do we have strong comparative evidence for claiming that the nom.sg. -o of TB kantwo can be traced back to a sigmatic nom.sg. *-eh2-s (cf. also Hilmarsson 1986a; Pinault 2008: 428 and 486, 2012: 189 fn. 48).97 In any case, I assume that both PIE *-eh2 and *-eh2-s would have turned into -o in Tocharian B (see § 3.5.2.4, § 4.8). Like the nominative, the oblique singular TB -a has also given rise to controversy. Scholars usually argued that the obl.sg. -a was influenced by the *ōn̆ -stems, so that TB -a would be the outcome of either the obl.sg. PT *-an < acc. sg. *-ōn-m (Adams 1988b: 13–14; Hilmarsson 1986a: 18) or the late gen.sg. PT *-anse, resegmented as *-a-nse (Pinault 2008: 486–487). On this issue, Malzahn (2011: 96–97) has now proposed a different explanation. Following the teachings of the late Jochem Schindler, she reconstructs the accusative singular of the PIE word for ‘tongue’ as *-eh2-m̥ (with syllabic nasal), 96 If one compares Av. hizuuā- with Ved. jihvā́-, some issues related to both the inflection and the gender of the IIr. noun come to light. Indeed, Av. hizuuā- is masculine, while Ved. jihvā́- is feminine. The relevant problems that the comparison highlights are: (1) the mismatching gender of the nouns; (2) the different shape of the nominative singular. Lipp (2009: 1.188–190) reconstructs a masculine noun with asigmatic nominative singular PIE *dn̥ ǵhu̯ éh2, which yielded IIr. *ǰiǰhuā. In Indo-Aryan, the word has been reinterpreted as a feminine ā-stem since members of the ā-inflection were only feminine from the Vedic period (Lazzeroni 1997: 193–205). On the other hand, if final -ā̊ in YAv. °hizuuā̊ is not due to compounding, one may wonder whether the masculine gender of the noun has hindered its inclusion in the feminine ā-stems, while the nom.sg. has become sigmatic under the pressure of original root nouns ending with a vowel in the nominative singular, like xā̊ f. ‘well’, °stā̊ ‘standing’, paṇtā̊ m. ‘path’, mazdā̊ m. ‘Mazdā’ (Skt. medhā- f. ‘wisdom’ < IIr. *mas-dhaH- < PIE *mn̥ s-dheh1-). The alternation between -uuā and -uuå in the manuscripts may partly mirror this development, though this may be speculative. 97 Malzahn (2011: 89) seeks to derive the nom.sg. -a of the Tocharian B feminine “thematic” adjectives from a non-ablauting PIE *eh2-stem. However, the ending of these Tocharian adjectives is not -a, but rather -ya, which cannot be reconciled with *-eh2 > *-ā. Indeed, following Malzahn’s explanation, the expected Tocharian B outcome of the PIE adjective in *-reh2 should have been TB **-ra, but the attested form is rather TB -rya. Her claim cannot therefore be considered as a real counterargument against the evolution of PIE *-eh2 > TB -o. I will discuss the evolution of PIE *-eh2 in word-final position more thoroughly in later sections of this chapter. For a discussion of the evolution of the feminine inflection in the Tocharian adjective, see § 4.8.2.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

121

and therefore suggests a sound law “Very Early pre-PT *-ah2m̥ > Later pre-PT *-ăm”. I find this sound law hard to accept. Firstly, it is not falsifiable and there are no other Tocharian forms that testify it. Secondly, even if we reconstructed a syllabic acc.sg. *-m̥ , I do not understand what the exact phonetic condition was for causing the loss of the laryngeal in the sequence *-eh2-m > *-aH-m̥ (perhaps through *-aH-ǝm?) > *-ăm. Klingenschmitt (1994: 393), followed by Kim (2009a: 79), argues that the obl.sg. -a is from the zero grade *-h2-, which was not characteristic of the accusative singular in the hysterodynamic type. This implies that the obl.sg. -a is to be traced back to the weak stem. Pinault (2008: 483–484) questioned this reconstruction, since it would not be coherent with the general development of the Tocharian oblique, which mostly mirrors the PIE accusative. He points out that, in several Tocharian inflectional types, the nominative and the accusative must have coalesced in the singular “en raison des lois phonétiques des finales” (Pinault 2008: 483). The same development must also be assumed for the paradigm of kantwo: both nominative and accusative should have merged as *kəntwo < *ǵn̥ dhu̯ éh2(m) in Proto-Tocharian, while the gen.sg. *ǵn̥ dhu̯ h2-és should have yielded *kəntwa. This *kəntwa may be the direct ancestor of the obl.sg. TB kantwa /kə́ ntwa/, ta käntu. As a matter of fact, this is not an isolated trend of development, since there are other Tocharian oblique singular endings that cannot go back to the PIE accusative. As pointed out in § 3.4.1.2, the contrast between nom.sg. -(y)a vs. obl.sg. -(y)o in the śana-type mirrors the ablauting alternation between strong and weak stem of the suffix *-(i)h2, *-(i̯)eh2-. Furthermore, Peyrot (2012a) has recently identified strong correspondences between the TB obl.sg. -ai and the TA gen.(-dat.) sg. -e and has highlighted the fact that Tocharian B feminine adjectives (with obl.sg. -ai) do not attest genitive singular forms. This clear piece of evidence allows us to support the reconstruction of a dative (or locative) PIE *-(e)h2-(e)i as the ancestor of the obl.sg. TB -ai (Pedersen 1941: 53, see further § 3.5.2.5 and § 4.8.2.4). Furthermore, the obl.sg. forms of the kinship terms in PIE *-ter- of the type TB pātär ‘father’, mātär ‘mother’, protär ‘brother’ cannot derive from the acc.sg. PIE *-tér-m̥ , which was expected to have yielded **-cär, but is instead the outcome of the zero grade stem of the gen.sg. *-tr-és > PT *tŕǝ > TB -trä ~ -tär (cf. Lat. patrem vs. Gk. πατέρα). Returning to the obl.sg. TB kantwa, I believe that, after the formal confusion between the nominative and the oblique in the paradigm of the singular (both resulting in *-o), Tocharian acquired a new obl.sg. *-a, which is itself the regular outcome of the weak stem of the hysterodynamic paradigm (probably of the gen.sg. PIE *-h2-és). The reconstruction of an obl.sg. *-a for Proto-Tocharian finds confirmation in the Tocharian A derived adjective °käntwā-ṣi ‘related to

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

122

Chapter 3

the tongue or language’ (dta 163; cf. TB käntwāṣṣe), which was regularly constructed on the PT obl.sg. *kəntwa. All things considered, the diachronic evolution of the paradigm of TB kantwo, ta käntu ‘tongue’ can be schematised as follows:98 Table 22

PIE

Evolution of the singular paradigm of TB kantwo, TA käntu

PT

TB

TA

> * kəntwo > *kəntwo > kantwo käntu nom. *dn̥ ǵhu̯ éh2 > *ǵn̥ dhu̯ ā acc. *dn̥ ǵhu̯ éh2-m > *ǵn̥ dhu̯ ām > *kəntwo *kəntwa > kantwa käntu gen. *dn̥ ǵhuh2-és > *ǵn̥ dhu̯ as > *kəntwa >> *kəntwanse (?) > käntwāntse käntwis

TB karyo* ‘±viscera’, ta kri ‘will, desire’ Beside TB kantwo, another noun with etymological comparanda is TB karyo* ‘±viscera’, ta kri ‘will, desire’. Since Sapir (1936: 263), TA kri has been connected to the familiar PIE word for ‘heart’, as represented by e.g. Skt. hr̥d́ -, ΟAv. zərəd-, Gk. κῆρ, Lat. cor (gen. cordis), etc. In fact, a Proto-Tocharian singular paradigm nom.sg. *kəryo, obl.sg. *kərya would fit well from both a Tocharian and an Indo-European comparative perspective. Following Hilmarsson (1996: 100), we can therefore posit PIE *ḱr̥di̯eh2 as the ancestor of TB karyo*, ta kri (cf. Gk. καρδίᾱ, Hom. Gk. κραδίη but also the stem Hitt. kard(i)-, OIrish cride and Skt. hr̥d́ aya-, Av. zərədaiia-).

98

There is some hesitation over the gen.sg. of Tocharian A. Carling (dta 130) indicates two variants, TA käntwis and TA käntwes, both attested in A300 (at lines b1 and b3 respectively). This fragment is part of the Maitreyasamiti. Parallels from the Old Uyghur Maitrisimit can be identified: A300 a5 may match Hami 21.5v9–12 (Geng et al. 1998: 33 and 90; Michaël Peyrot p.c.), while A300 a7 may match Mainz 973.r2–4 (Tekin 1980: 179–180). See Laut & Wilkens (2017: 184–185 and 385). These documents belong to Chapter 21 of the Maitrisimit. As far I can see, a Uyghur parallel of line b3 is missing. However, the fact that two variants of a genitive form are attested in the same fragment is very suspicious, and TA käntwes is actually written kätwes: b3 ṣñi kätwes mätkont prakte ypamtär kārūṇik. This line may refer to the tortures and penances that the penitents suffered in one of the eight hells. Hilmarsson (1996: 114) interpreted TA kätwes as an obl.pl. of kätwe*, which he translated as ‘deception’ or ‘sin’. See further Malzahn (2010: 553).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

123

TB kāwo ‘desire’ and TB tsāro ‘monastery’ The derivation of TB tsāro ‘monastery’ has been thoroughly investigated by Malzahn (2011: 98–99). We can extend her analysis to account for the evolution of kāwo. Following Krause (1952: 51), she links TB tsāro ‘monastery’ to the verb tsǝr‘be separated, separate’. However, the derivation of the noun from the verb raises some difficulties: (1) the non-productivity of the kantwo-type as a class of abstract derivatives;99 (2) a deverbal noun from tsǝr- is expected to show root-vowel -a- /ə́ /, instead of -ā- /á/ (cf. TB palsko ‘thought’ from plǝska- ‘to think’, TB traṅko ‘sin’ from trǝnk- ‘to lament’). In order to solve these problems, she claims that TB tsāro is a very archaic derivative of the Indo-European root from which the verb TB tsǝr- also derives, i.e. PIE *der- ‘to split’. She further reconstructs a derived abstract in *-eh2, i.e. PIE *dōreh2.100 Similarly, TB kāwo ‘desire’ is usually regarded as a deverbal noun from kawa‘to crave’ (dtb 164–165). If so, it is a very archaic derivative from the same PIE root *k(u̯ )ap- ‘well up’ from which the verb TB kawa- also derives (Malzahn 2010: 563; but liv 2 does not reconstruct such a verbal root). However, the matter is more complex than it seems. The problems involved are: (1) the lenition -p- > -w- in both the noun and the verb; (2) TA kāpā- ‘to surge up; be greedy’ as the apparent cognate of TB kawa- ‘to crave’; (3) alternation of -p- and -w- in the inflection of the Tocharian B verb.101 The formal match between TB kawa- and TA kāpā- is an issue on which scholars strongly disagree: on the one hand, Malzahn (2010: 563) reconstructs PT *kapa-, implying that Tocharian A attests the original form; on the other hand, Peyrot (2013a: 729) takes an opposite view, claiming that the Proto-Tocharian form was *kawa-. In fact, the only TB attestation of a p-form from kawa- is the isolated prt.ptc. kakāpau (adduced by Saito 2006: 301), which is not easy to interpret or translate (B66 a8).102 Since we do not have a parallel to account for the consonant mismatch between Tocharian B and A (that is, PT *-p- > TB -wor PT *-w- > TA -p-),103 I propose reconstructing different proto-forms for the 99

Among the inflectional classes with nom.sg. -o, the alternating members of the palsko-type are verbal abstracts (with nom.obl.sg. -o, nom.obl.pl. -o-nta). On this class, see § 3.6.2.1. 100 In fact, Malzahn claims that TB tsāro is the outcome of a plural *dōreh2-es. Kortlandt (2020: 116) has recently derived this noun from *dēr(e)h2 (cf. Gk. δῆρις ‘battle, strife’), but I expect *-ē- to have yielded TB -e-. The claim by Kortlandt that pre-PT *-ē- evolved into TA -a-, TB /a/ in open syllable is contradicted by e.g. TA mañ, TB meñe ‘moon, month’ < PT *ḿeñe(s) < *mēnēs < *méh1nēs (Normier 1980: 254; Adams dtb 503; Pinault 2013b: 346). 101 See Malzahn (2010: 562–563) and Peyrot (2013a: 729) for further details. 102 On TB kakāpoṣ, see Malzahn (2010: 563). 103 The evolution -p- > -w- is only attested in Late Tocharian B (Peyrot 2008: 88–90).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

124

Chapter 3

two Tocharian languages. Indeed, while TA kāpā- may be the outcome of PIE *k(u̯ )ap- ‘to well up’, the root from which the Tocharian B verb comes from may be PIE *geh2u̯ - ‘to rejoice’ (> Gk. γηθέω ‘to rejoice’, γάνυμαι ‘to be glad’, Lat. gaudeō ‘id.’), which resulted quite regularly in TB kawa-.104 The reconstruction of two different roots for the Tocharian A and B verbs could also explain the fact that in Tocharian A the verb is intransitive, while in Tocharian B it is transitive. Furthermore, it seems to me that TB kawa- and TA kāpā- also differ markedly in the meaning. In Tocharian A, this verbal root is attested in the following forms (Malzahn 2010: 562–563): 3sg.subj.act. omäl ysār ṣuṅkac kāpaṣ-äṃ “hot blood will rise to his throat” (YQ I.7 b1, cf. Ji 1998: 51) and 3sg.prt.act. śwātsiṣy ākālyo kāpar ymār “in their wish for food they soon became fully impatient” (A340 a3, cf. Schmidt 1974: 146 fn.1). As a consequence, the meaning of TA kāpā- is ‘to surge up, be impatient’, while TB kawa- means specifically ‘to crave’ (Peyrot 2013a: 729; cf. also the derivatives TA kāplune* ‘boiling’ vs. TB kāwalyñe ‘desire, craving’). TB kāwo ‘desire’ would be an old derivative to this root (perhaps of the τομή-type?): PIE *g(o)h2u̯ -eh2 > *gō/ău̯ ā > PT *kawo > TB kāwo, with the meaning ‘what makes someone glad’ → ‘what someone desires’.105 The suffix TB -to, obl.sg. -ta, TA -t We have seen that TB laukīto ‘stranger’ and TB nekīto* ‘± destroyer’ or tekīto* ‘± sick person’ may belong to the kantwo-type. The problem here is the origin of the suffix -(i)to, which is an unproductive derivational morpheme in Tocharian. The only match between Tocharian A and B is TB laukito : TA lokit ‘guest, stranger’, with regular monophthongisation *aw > o in Tocharian A (cf. the gen.sg. TA lo« ki »tāp-äk in A6 a4). In Tocharian A, we also find TA mäśkit ‘prince’, which corresponds to Tocharian B mcuṣke ‘id.’. As a consequence, the suffix TB -ito, TA -it only appears in four nouns, two in Tocharian B (laukīto and n/tekīto*) and two in Tocharian A (lokit and mäśkit). It seems that TB -ito, TA -it is the result of reanalysis, since the vowel -i- cannot synchronically belong to the stem (cf. laukaññe ‘for a long time’ /laukə́ ññe/). Pinault (2015a: 176) has recently dealt with the origin of this suffix. He reconstructs PT *-ǝy-to (*-itå in his notation), which in turn may have had two 104 On TB katk-, ta kātk- ‘to be glad’, see dtb 159 and Hackstein (2002: 8). 105 It is a still matter of debate whether the paradigm of TB kawa- began as a denominative to kāwo. For discussions, see Hilmarsson (1991c: 80–81) and Malzahn (2010: 563). Another noun that might be analysed as TB tsāro and kāwo is weto* ‘battle’, which, per Adams (dtb 661), is to be linked to the verbal root wəta- ‘to fight’ (cf. Lat. pūgna ‘fight’ ← pungō ‘to pierce, sting’). However, the consonantism of TA wac is troublesome. See further Pinault (2012a: 197–198).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

125

possible Indo-European sources: (1) *-ǝy- was part of the stem and PT *-to is from the “individualising” suffix PIE *-teh2 > *-tā; (2) PT *- ǝyto reflects a second compound member PIE *-Hi-t-eh2 > *-itā, from the verbal root *h1ei̯- ‘to go’ (cf. the type of Lat. comes, comitis ‘companion’, and Hom. Gk. περικτίτης ‘neighbor’ etc.).106 Pinault seems to favour the latter hypothesis, so that the meaning of PT *lawk-ǝyto would have been ‘coming from afar’. The original value of the second member *-ǝyto would have then become obsolete and it would have been employed in the derivation of few other nouns. However, if we assume that PIE *i palatalises neither velar nor labiovelar stops in Tocharian,107 I would expect PT *lawkəto > TB **laukato, ta **laukät as the outcome of a (virtual) compound *lou̯ k-Hi-teh2.108 Following the first hypothesis, one could posit an abstract noun *lawkǝy as the origin of TB laukito, as suggested by Pinault himself.109 If so, the original suffix was *-to, which would have been reanalysed as *-ǝyto via resegmentation of *lawkǝy-to as *lawk-ǝyto.110 The only problem with this analysis is that final TB -i is usually matched by TA -e in these abstract nouns (cf. TB telki ‘sacrifice’ : ta talke ‘id.’; TB leki ‘bed’ : ta lake ‘id’. etc.). As a consequence, one should assume 106 See Leukart (1994: 66–67). 107 Word-initially, PIE *i (*Hi) evolved into PT *yǝ > TA yä-, tb /yǝ-/, while it became PT *-ǝ- > TA -ä-, tb /-ǝ-/ in internal position. The palatalising effect of PIE *i is debated. Palatalisation seems to be regular in front of *-l- and dental stops, cf. *limn̥ ‘bay, like’ > PT *ĺǝmǝ > TA lyäm, tb lyam ‘lake’; PIE *-nti (3pl.) > PT *-ñcǝ > TA -ñc. Conversely, it is clear that *i did not palatalise labiovelars (e.g. *kwi-so- ‘who’ > PT *kwǝse > TB kuse, ta kus; PIE *du̯ itó- ‘second’ > PT *wǝte > TA wät, tb wate ‘id.’). Pinault (2008: 433) assumes that PIE *i did not palatalise labials, velars, labiovelars, or *s. 108 On the other hand, if laryngeal metathesis must be reconstructed, I would expect that a hypothetical pre-PT *ī would have palatalised the velar. 109 This *lauki may have derived from the verbal a-root TB lǝwka-, on which see Adams (2012) and Peyrot (2013a: 811). Cf. further the adverb laukar ‘afar’ (AS6A a5, a6, b7). 110 A similar type of reanalysis also characterised certain Ancient Greek nouns in -ῑτης. On several occasions, Van Windekens (1942: 295, 1944: 132, 1976: 176 and 266) equated this suffix with TB -ito, ta -it as both reflecting PIE *-īteh2 (cf. also Hirt 1912, 1927: 228). However, the Greek suffix can be easily explained as an indigenous formation through the same reanalysis that hypothetically characterised PT *-əyto. Indeed, as pointed out by Redard (1949) and (in part) Leukart (1994: 187–197), Gk. -ῑτης is a back-formation from πολῑτ́ ης ‘citizen’ (regularly from πόλῑς ‘city’), on the basis of which the -ῑ- has been reanalysed as part of the suffix and then generalised to form other common and proper nouns (e.g. Hom. Gk. ὁδῑτ́ ης ‘traveller’ ← ὁδός ‘road’; Gk. ὁπλῑτ́ ης ‘hoplites’ ← ὄπλον ‘tool, weapon’; Att. Gk. ἐρημῑτ́ ης ‘hermit’ ← ἔρημος ‘lonely, solitary’; Hom. Gk. Θερσῑτ́ ης ‘Thersites’, the antihero of the Iliad). This new suffix became increasingly productive in the history of Greek (with its feminine counterpart -ῑτις), especially from the Hellenistic period on, when it started to form technical terms, as well as ethnic designations and Biblical tribal names.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

126

Chapter 3

that the expected **loket became lokit under the influence of Tocharian B; however, this is speculative. A final hypothesis involves reconstructing a derivative PT *lawk(ǝ)ye ‘far; distance’ (cf. TB werpi-śke, ta warpiśke ‘little garden’, based on TB werpye*, ta warpi ‘garden’, etc.) from which an agent noun in *-to is derived. This reconstructed noun is expected to have evolved into TB laukīto, ta lokit. Be that as it may, TB laukīto, ta lokit is clearly related to the adverb TB lauke, ta lok ‘far, remote, away’. The second Tocharian B noun built with the suffix -(i)to is TB nekīto* or tekīto*. As we have already seen, both nouns would be hapax legomenona and their precise meaning cannot be identified. However, if TB -(i)to has a sort of agentive value, attested nekīta could mean ‘± destroyer’ (perhaps from an abstract *neki ‘destruction’), while {t}ekīta could mean ‘± sick person’ (dtb 322). So far, we have seen that the suffix TB -ito, ta -it must be historically segmented as TB -i-to (obl.sg. -i-ta), ta -i-t. This should be traced back to the agentive suffix PIE *-teh2 of the type Lat. nauta ‘sailor’, Myc. e-re-ta ἐρέτας ‘rower’, Gk. ἱππότης (Hom ἱππότᾰ) ‘horseman’, etc. (Pinault 2015a: 176; Adams 2015: 180). The reconstruction of the obl.sg. *-ta for this type of derivatives allows us to consider other agent nouns which seem to have been formed (or reanalysed) with the same suffix in Proto-Tocharian. The nouns in question are: (1) TB käryorttau, ta kuryart ‘merchant’; (2) TB olyitau ‘boatman’; (3) TB pälkostau* ‘spy’; (4) the loanword TB */kamarta-/ ‘chief, ruler’ (cf. kamartāññe ‘rulership’), ta kākmart ‘master, sovereign’.111 Pinault (2015a: 161–162) claims that the suffix -tau was abstracted from the noun TB käryorttau ‘merchant’, which is the most prominent and attested member of this class of derivatives. He analyses TB käryorttau as a compound of TB karyor°, ta kuryar° ‘trade’ and °ttau, an agent noun based on the verbal root PT *tətta- ‘to put’. He reconstructs a compound because TB käryorttau is very often spelled with geminated -tt-. According to Pinault (2015a: 162), once “the original meaning of the root of the second member vanishes”, the formation was reanalysed, and the suffix was abstracted. 111 TB */kamarta-/ and TA kākmart ‘ruler’ are considered to be loanwords from Bactrian καμιρδο ‘head, chief (god)’. Cf. also TB kamartike, ta kākmärtik ‘ruler, master’ from suffixed *καμιρδιγο, according to Pinault (2002: 262). As pointed out by Adams (dtb 149), however, the difficulty with a Bactrian derivation is the vowel of the second syllable, as Tocharian /ə/ (and not the attested /a/) would be expected from Bactrian ι. For this reason, Dragoni (2022: 77–78) has recently suggested that the word is a loanword from Proto-Khotanese-Tumshuqese *kamarda- (> OKhot. kamala- ‘head’). On TB mlyokotau, a kind of seed used to make a fuel (?), see Ching (2014: 45). A possible additional noun to be considered is kärccitake*, on which see Peyrot, Pinault & Wilkens (2019: 75–76).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

127

Conversely, I think that the gemination of TB -t- in the cluster -rt- > -rtt- is a phonetic change, as the following examples show: warto ~ wartto ‘forest’; kartse ~ karttse ‘good’; akarte ~ akartte ‘near’; gen.sg. udāvarttäntse (← Skt. udāvarta‘disease, ileus’); kerte ~ kertte ‘sword’; kamartāññe ~ kamarttāññe ‘rulership’; kamarttīke ~ kamartīke ‘ruler’, etc.112 Furthermore, TA kuryart, with stem kuryartā-, points to the reconstruction of a noun with nom.sg. *kw(ə)ryor-to, obl.sg. *kw(ə)ryor-ta for Proto-Tocharian, which would also explain the derivative TB käryortaññe, the name of a metre. The same analysis can also account for other nouns from this class, such as TB olyita-u ‘boatman’ from olyi (obl.) ‘boat’. In my view, the final -u must have been adopted from other nomina agentis, like yenmeu ‘gatekeeper’ (from yenme ‘portal’), TB yotkolau ‘controller, director [of a monastery attendants]’ (from *yotkol ‘order’), TB wetāu ‘warrior’, ta waco (from TB weta, ta wac ‘battle’),113 and from the adjectival type TB tallāu, ta tālo ‘miserable’, TB maiyyāu ‘powerful, strong’ (cf. Van Windekens 1979: 98–99). These formations contain the outcome of the PIE possessive suffix *-u̯ ent-. To conclude, we have seen that Tocharian inherited the suffix *-teh2, which can be compared with the Greek type ναύτης ‘sailor’. The Proto-Tocharian outcome of this suffix was used to derive agent nouns from nominal bases. The paradigm of the singular was nom.sg. *-to, obl.sg. *-ta. This paradigm has been maintained in isolated words, like TB laukīto, ta lokit (stem TA lokitā-) ‘stranger’, TB n/tekīto* (obl.sg. n/tekīta), and TA kuryart ‘merchant’. In Tocharian B, there is a general tendency to turn all these nouns into wənt-stems, of which the majority can be traced back to the possessive formations in *-u̯ ent-. This suffix formed denominal adjectives but in Proto-Tocharian it started to be reanalysed as an agentive suffix, cf. pre-TB *weta ‘battle’ → *wetaw ‘combating, warlike’ > TB wetāu ‘soldier, warrior’ (cf. TA waco). Tocharian B has therefore started to level all the original formations in nom.sg. *-to, obl.sg. *-ta with the existing wənt-stems. The result of this process is the attested conglomerate suffix *-taw, which regularly follows the nt-inflection.

112 Example of non-geminated -t- can be found in AS13I b2 käryortantäṃ, IT8 b1 käryor­ tantäṃne, NS73 a3 käryortau, B239 b3 käryortantäṃts, and frequently in the derived käryortaññe(ne), the name of a metre (cf. IT887 a2; AS17I a5; NS58 b3; B350 b3). 113 The formation of TB saṃtkīnau, ta sāṃtkenu ‘physician, doctor’ has not yet been explained, since we would rather expect TB -itau, ta -it. It is evidently derived from TB sāṃtke, ta sāṃtäk ‘medicine’ (← Middle Indic intermediary of Skt. śāntaka- ‘allaying’), but the two Tocharian languages do not match phonologically and the suffix TB -(i)nau, ta -(e)nu is not attested elsewhere.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

128

Chapter 3



TB tāno ‘seed of grain’, TB mālo ‘± inebriating drink’, TB eñcuwo, ta añcu* ‘iron’ and the Iranian loanwords Another noun of the kantwo-type is usually considered to go back to the same PIE inflectional type of kantwo, namely TB tāno ‘seed of grain’. Two different etymological analyses have been proposed so far: (1) TB tāno goes back to PIE *dhoH-neh2 ‘grain’ (> the plurale tantum Skt. dhānā́ḥ ‘grain’, Khot. dānā- ‘id.’, ManSogd. δʾn ‘id.’, Lith. dúona ‘bread’, Latv. duõna ‘slice of bread, heel of a loaf’)114 or (2) it is a loanword from either Indo-Aryan or Iranian.115 The former hypothesis has no problems from a phonological point of view; it is sustained by e.g. Adams (dtb 303) and Pinault (2008: 486).116 Peyrot (2018a: 258–260) supports the latter hypothesis, claiming that TB tāno was borrowed from Iranian *dānā-. There are two indications that may substantiate this analysis. On the one hand, the semantic resemblance between TB tāno and Khot. dānā- as both referring to single seeds that may be counted is remarkable;117 on the other hand, Peyrot reveals that the Baltic forms have some semantic problems if derived from PIE *dhoH-neh2 ‘grain’.118 If Baltic is removed from the list of comparanda, the peculiar distribution of the term in Indo-Iranian and Tocharian alone strongly suggests that Tocharian borrowed from Iranian. In past decades, it was assumed that the inflectional class to which tāno belongs was not productive and could not adapt loanwords. Peyrot (2018a: 258) adduces TB tvāṅkaro ‘ginger’ (← OKhot. *tvā̆ṃggara- or *tvā̆ṃggaraa-, with accented first syllable) as an example of an Iranian loanword inserted into a genuine Tocharian inflectional class (the so-called arṣāklo-type). This class is certainly more productive than the kantwo-type though, and its productivity can be easily reconstructed for pre-Tocharian B as well (see § 3.5.2.2). 114 Cf. the Young Avestan compound dānō.karš(a)- ‘grain-carrying’, where the ō-vocalism of dānō° does not necessarily indicate that it is a masculine a-stem (Malandra 2002: 229–230; EWAia 1.787). Note that Kortlandt (2013: 96) suggests a derivation from the zero grade *dhh3nā, with vocalisation of the laryngeal. 115 See Evangelisti (1990[1965]: 130) and Klingenschmitt (1994: 394 fn. 136). 116 There is no reason for claiming that the final -o in TB tāno should reflect an original plural *-eh2-es (pace Peters 1991: 243 followed by Malzahn 2011: 98). 117 A derived word tanākko apparently means ‘grain; little spot’ (cf. B580 b2 śwālyai (pai)yyeññe mokkocintse āntene tuciyai tanākkai lkāṣṣäṃ “he sees a yellow grain on the surface of the big toe of the left foot”, b3 rtarya tanākko āntene “a red grain on the forehead”, b3 arkwañña tanākko saiwai (paiy)y(ene?) “a white grain on the left foot”). In medical texts, the obl.sg. tanākkai (AS2A b3) and the perl.sg. tanākkaisa (AS2a a2) are used in the function of an adverb with the meaning ‘gradually, bit by bit’. 118 See Peyrot (2018a: 259–260) for an overview of the Baltic problems and for etymological suggestions.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

129

However, after the study of Dragoni (2022), the kantwo-type has been proven to be an inflectional class into which old loanwords from Iranian were adapted (particularly from Khotanese). In addition to TB tāno ‘grain’ and possibly *kito ‘help’, he suggests that TB keto ‘property, estate’, śāmpo* ‘haughtiness, pride’ (cf. TA śāmpāṃ*), and TB śarko* ‘song, singing’ (TA tsärk) are loanwords from prehistorical stages of Khotanese (both Proto-Tumshuquese-Khotanese and pre-Khotanese).119 Another complex word that can ultimately be connected to an Iranian antecedent is TB eñcuwo, iñcuwo, ta añcu* ‘iron’. Schwartz (1974: 409) was the first who suggested a relation with Khwar. hnčw ‘steel’, though he claimed that both Khwarezmian and Tocharian borrowed from a third language. Tremblay (2005: 424–425), on the other hand, reconstructed (in his notation) *ačú̯ an- > Khot. hīśśana- ‘iron’ (with irregular *ač-́ > hīśś-) and further claimed that a nasalised variant of *ačú̯ an- was borrowed into Tocharian. However, the development *ačú̯ an- > Khot. hīśśana- ‘iron’ is not expected: the initial Khot. h- is unetymological (see Maggi 2016: 76–77 with references), but the palatalisation *-a> Khot. -ī- could be from a secondary added suffix *-i̯a- (see below). Adams (2004, dtb 85) has put forward a different analysis. After collecting a number of etymologically related Iranian words meaning ‘iron’, he ultimately posited a Proto-Iranian ancestor *anćuwan- (in his notation). According to Adams, however, this proto-form would be etymologically unanalysable in Iranian terms. For this reason, he claimed that Tocharian was actually the source language, and that Iranian borrowed from Tocharian, which had in turn inherited this form from PIE *h1n̥ -ǵheu̯ eh2- ‘what is poured in’ → ‘cast iron’.120 I cannot agree with this analysis. Though TB eñcuwo could potentially be the outcome of a formation PIE *h1n̥ -ǵheu̯ eh2- from a formal point of view, it is improbable to me that this word spread from Tocharian to practically every Iranian language. Indeed, we find continuants of a proto-form *aću̯ an- in several Eastern and Western Iranian languages (Sadovski 2017: 572): *aću̯ ana- > Oss. æfsæn; *aću̯ ani̯a- > Khwar. ʾspny, Khot. hīśśana-, Shughni sipin, Waxi (y)išn; *āću̯ ani̯ā- > MP āsen ⟨ʾsyn⟩; *āćuna- > Parth. āsun ⟨ʾswn⟩, MP āhun ⟨ʾhwn⟩, NP āhan. I therefore remain unconvinced by Adams’ proposal. Recently, Peyrot, Dragoni & Bernard (2022) have revived and further refined Tremblay’s hypothesis, solving the formal problems linked to a Khotanese origin. The authors traced TB eñcuwo ~ iñcuwo, ta añcu* back to a Proto-Khotanese-Tumshuquese 119 See Dragoni (2022: s.v. and 242–243). Other possible loanwords from Khotanese will be discussed in the following section. 120 This Indo-European etymology has recently been accepted by Hackstein, Habata & Bross (2015: 103).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

130

Chapter 3

antecedent *henśwanya- which displays the addition of a suffix *-i̯a- that caused the umlauted vowel attested in Khot. hīśśana-. In Tocharian, this noun would have first been reanalysed as an adjective *eñcuwaññe and then TB eñcuwo, ta añcu* would have been back-formed from it. At least one further Iranian loanword can be included into the kantwo-type. This is TB mālo ‘± alcohol, spirit’ (obl.sg. -a, see Ogihara 2011: 135). Since Bailey (1959: 131), a foreign origin has been suggested for this term, as it has been connected with YAv. maδu- ‘Beerenwein’, Khot. mau- ‘intoxicant drink’, Christian Sogd. mwd[‘wine’ (cf. Skt. mádhu- ‘sweet; sweet drink’, cf. EWAia 2.302–303).121 To be more precise, TB mālo must derive from an Iranian variety where *-dbecame -l-. Therefore, Winter (1971: 152) connects this word with Bctr. μολο ‘wine’ < *malu- < *madu-. As one can see, however, the vocalism of Bctr. μολο /mol/ deviates from that of TB mālo /málo/. The Bactrian vowel is the outcome of u-affection of an original *-a- (in a labial environment), which results in a rounded back vowel (Gholami 2014: 65). Since the class to which TB mālo belongs testifies its early acquisition, one may claim that Tocharian borrowed this word before u-umlaut took place in Bactrian. Thus the loanword also gives important evidence to the relative chronology of Proto-Bactrian phonological developments.122 For the relevance of the adaptation of these loanwords into the kantwo-type, see § 3.5.1.3. TB kāswo ‘leprosy, skin disease’ and TB kātso, ta kāts ‘abdomen, belly’ As regards TB kāswo and TB kātso, ta kāts, no certain etymology has been proposed so far. TB kāswo is attested four times: twice as a nominative TB kāswo (IT305 b5; THT1111 b3), once as a perlative TB kāswasā (B282 a4 [arch.]), and once in the derivate kaswātse ‘leprous’ (IT305 a6). According to Filliozat (1948: 56–63), the fragment IT305 is a Tocharian reworking of passages from the Carakasaṃhitā. At line b5, TB kāswo matches Skt. kuṣṭha-, the Sanskrit technical term referring to skin disease in general, and to leprosy in particular (Emmerick 1984: 96–98). Moreover, the derived adjective TB kaswātse is the translation of Skt. kuṣṭhin‘suffering from kuṣṭha, leprous, a leper’. Conversely B282 is not a medical 121 The Sogdian word is attested in the Gospel Lectionary E5 (see Barbati 2016: 111). 122 According to Schwartz (1974: 411), another example of a Tocharian loanword from ProtoBactrian is TB spaktanīke, ta spaktānik ‘servant’ from *spaxtanīkə (cf. Bactrian σπαχνιιο ‘subject to service’). In addition, the hapax legomenon TB loc.sg. waṣākane /waṣákane/ ‘in terror’ (Peyrot, Pinault & Wilkens 2019: 84) has also been interpreted by Dragoni (2022: 185–86) as a loanword from Bctr. βιζαγο (MBctr. βyźg) ‘bad’.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

131

fragment but a poetic composition (kāvya) in which we find the following passage: a4 śaiṣṣe se kleśanmaṣṣai wämyu räskre kāswasā, “this world is roughly covered by the leprosy [?] of kleśas” (cf. Adams dtb 766). The last document to be discussed (THT1111) confirms the translation of TB kāswo ‘leprosy, skin disease’ and may suggest new etymological arguments. The passage in question is from the Tocharian Karmavācanā, of which several fragments are bilingual in Sanskrit and Tocharian.123 At line b3, we find a list of diseases: no eṅ(k)wetse toṃ te y(äknetsana teka)nm(a) koṣṭä kāswo piśträ kṣai apasmār “now there are such diseases of a man: koṣṭä, kāswo, piśträ, kṣai, apasmār” (cf. Schmidt 2018: 74; Tamai 2014: 378). Although an internal Sanskrit parallel for this passage is lacking,124 TB koṣṭ is clearly borrowed from Skt. kuṣṭha- ‘leprosy’. One may therefore wonder whether we have a sequence of apparent synonyms, i.e. koṣṭä and kāswo. However, following Schmidt (1986a: 68–70, 2018: 74), we can interpret these two terms as referring to different types of leprosy: the former would be the ‘black disease’, while the latter would be the ‘white disease’, a distinction that mirrors the modern one between lepromatous (black) and tuberculoid (white) leprosy. This identification is further confirmed by a specific section of the Sanskrit Karmavācanā that concerns the rite of ascetic vetting by which a candidate enters the community (Skt. upasaṃpadā ‘ordination’) and in particular with the so-called Befragung im Geheimen (Härtel 1956: 78–83). In this section, the rahoʾnuśāsaka ‘instructor in private’ (bhsd 454) explains one of the obstacles that may prevent the admission of the candidate: disease. Those which occur in both Sanskrit and Tocharian are (Schmidt 2018: 103): epilepsy (Skt. apasmāra- = TB apasmār), tuberculosis (Skt. kṣaya- = TB kṣai), goitre (Skt. gaṇḍa- = TB piśträ?), leprosy (Skt. kuṣṭha- = TB koṣṭä and Skt. kilāsa- = TB kāswo). According to Sāyaṇa, a medieval commentator of the Āyurveda, Skt. kilāsa- is ‘white leprosy’. This view is partly shared by Emmerick (1984: 96), who concludes that kilāsa- must have meant a “disorder of the coloration of the skin characterised by whiteness”, although it is unclear whether it denoted the same skin disease as that referenced in the Āyurvedic medicine. Now, given the fact that TB koṣṭ can correspond to ‘black leprosy’ and TB kāswo to ‘white leprosy’, the etymological connection proposed by Hilmarsson 123 For an overall overview of all known Tocharian Karmavācanā materials, see Ogihara (2013a: 325–326). For the edition and the translation of the text, see Schmidt (1986a; 2018), Tamai (2014), and Ogihara (2013b). 124 The Sanskrit parallel of THT1111 is attested in THT1116, a fragmentarily preserved document in which the list of diseases is missing, due to the damaged condition of the fragment.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

132

Chapter 3

(1996: 107) with PIE *ḱas- ‘grey, whiteness’ might be renewed.125 The derivational and semantic developments are as follows: *ḱas-u̯ o- ‘having whiteness’ → *ḱas-u̯ e-h2 ‘whiteness’ > PT *kaswo > TB kāswo ‘white leprosy; skin disease’.126 Although I think that Hilmarsson’s hypothesis is still a valid option to explain the etymology of the word, a foreign origin of TB kāswo has also recently been proposed. In particular, Bernard (forth.) and Dragoni (2022: 89–93) have renewed Lidén’s hypothesis (1916) that TB kāswo is actually a loanword from Iranian *kasū- (cf. Av. kasuuiš-).127 Dragoni (2022: 90) suggests a loanword from pre-Khot. *kasūwa-, itself derived from Proto-Tumshuqese-Khotanese *kasū-ka- ← OIr. *kasū-. The reason Dragoni proposes such an old loanword is that in Khotanese we only find LKhot. kasaa- ‘fever’ as the alleged continuant of *kasū-, which in turn cannot be the source of TB kāswo. Although the Iranian origin of the word as suggested by Bernard and Dragoni is in my view possible, the fact that attested continuants of *kasū- are quite isolated within Iranian and that the exact Iranian antecedent of TB kāswo is completely reconstructed urge caution. The etymology of TB kātso, ta kāts ‘stomach, belly; womb’128 is equally disputed. Pinault (1991: 186) suggests a connection with Gk. κατά ‘down’, Hitt. kattan ‘below’, and further argues that the Tocharian word is the outcome of an animate derivative of the adverb *kati, PIE *kati̯-eh2. The semantic evolution would have been ‘below’ → ‘what is below’ → ‘stomach’. Adams (dtb 165) puts forward a different hypothesis, connecting the Tocharian word with PIE *gwōt- ‘belly’, with possible cognates in Germanic (e.g. Goth. qiþus ‘stomach, belly’) and Latin (Lat. botulus ‘sausage’). This form would be suffixed in *-i̯ōn or in *-i̯eh2. In iew 481, PIE *gwet- is said to mean ‘swelling, 125 Cf. PGerm. *haswa- ‘grey’ (cf. ON hǫss, OE haso, MHG heswe ‘pale, dull’; cf. further PGerm. *hasan-, *hazan- > ON heri ‘hare’, OE hara ‘id.’, OHG haso ‘id.’, MDu. has ‘id.’) ultimately from PIE *ḱas- (or * ḱh1-es-) ‘grey; hare’ (cf. also Lat. cānus ‘grey, ashen, old’ < *ḱas-no-; Ved. śaśá- ‘rabbit, hare’, Khot. saha- ‘id’ < *ḱas-o-, etc.). 126 Cf. also Hackstein (2003: 84). This etymology seems to be accepted also by Malzahn (2011: 99), who writes that the Tocharian word may go back to an old plural form denoting ‘the grey ones’. Another possibility is to connect kāswo with PIE *ḱes-/*ḱseu̯ - ‘to comb, scratch’, but the vocalism of the root and the lack of palatalisation in Tocharian would be difficult to explain. 127 For the etymology of the Avestan term and dubious Indo-European cognate forms, see Kellens (1974: 367–368), Humbach (1974: 92), and Bernard (forth.). 128 TA kāts is mostly attested with the meaning ‘womb’ (Peyrot 2012a: 207 fn. 32). If correct, we have a case of asymmetry in overt marking: ‘womb’ is expressed by an overtly marked term on the basis of ‘belly, stomach’, but not vice versa. This assumption may be supported by the semantic evolution of terms of the same meaning in other languages. For example, Ved. udára- ‘belly’ > Old Gujarātī loc.sg. ūyari ‘womb’; Pkt. peṭṭa-, piṭṭa- ‘belly’ > Sindhī peṭu ‘belly, womb, foetus’.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

133

rotundity’, but from the perspective of lexical typology the meaning ‘stomach, belly’ is preferable in the proto-language. Indeed, in diachronic lexical typology we see a general trend from concrete to abstract meaning. Furthermore, the continuants of this root mean precisely ‘stomach, belly’, e.g. PGerm. *kwiþu> Goth. qiþus ‘stomach, womb’, OIcelandic kviðr ‘belly, womb’ (and kviðugr ‘pregnant’), OE cwiða ‘womb’, OHG quiti ‘vulva’, etc. From a formal perspective, Adams reconstructs the proto-form from which TB kātso, ta kāts derives with lengthened o-grade of the root. The o-grade is perhaps also attested in Lat. botu­lus ‘cumb, sausage’ (a loanword from an Italic language, where the PIE labiovelars developed into labial stops, Weiss 2020: 473 fn. 48), so that one could also say that PIE *gwōt- derives from PIE *gwet- through both qualitative and quantitative ablaut. However, many of the details remain unclear.129 Finally, Hilmarsson (1996: 112) connects TB kātso, ta kāts ‘belly, womb’ with the hapax legomenon TA kāc* ‘skin’ (A147 b4 y(p)es(uṃts)enāṃ kācyo epunt yṣitstseyaṃ lmont “sitting on a couch, covered with the skin of a leopardess”, cf. dta 109), with possible cognates in Lat. cutis ‘skin’, ON húð ‘id.’, OHG hūt ‘id.’, Lith. kiáutas ‘shell, rind, peel’, etc. If so, TB kātso, ta kāts could be from PIE *kuH-ti̯-eh2 > *kwatso > PT *katso, with delabialisation of *kwa- > *ka- before a consonant (Hilmarsson 1985b; Kim 1999: 158 fn. 42).130 Alternatively, Dragoni (2022: 90–93) has recently proposed a Khotanese origin for this word. He claims that TB kātso and TA kāts are independent back-formations from pre-Khot. *khādsāña, the original locative singular of the word attested in Late Khotanese as khāysāna- ‘stomach’. In particular, the Tocharians would have reinterpreted the locative singular *khādsāña as a nom.pl. *katsañə, which was then provided with a singular paradigm nom. kātso, obl. kātsa. Cases of back-formations in borrowed words are not very frequent in Tocharian (see Dragoni 2022: 92) and the pre-Khotanese locative singular form can only be reconstructed with final °ña and not °ñə, which would have made the reanalysis phenomenon more straightforward. Furthermore, from the point of view of the typology of language contact, body parts belong to the core vocabulary, which is not easily borrowed (Haspelmath & Tadmor 2009).131 Of course, social factors may promote borrowing irrespective of the semantic field, in particular if the borrowed item entered in a specific social 129 According to nil 185 and 187, the Germanic term is to be connected with *gwi̯eh3- ‘to live’. Kroonen (2013: 319) reconstructs PGerm *kweþu-, considering the derivation from PIE *gwi̯h3-i- conjectural. See also Mallory & Adams (2006: 185–186). 130 I thank Alexander Lubotsky (p.c.) for reminding me of Hilmarsson’s etymology. 131 According to Dragoni (2022: 261), linguistic contact between Tocharian and Khotanese should be placed at the boundary between the first and the second category in the borrowing scale elaborated by Thomas & Kaufmann (1988: 74–76).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

134

Chapter 3

environment, as supposed by Dragoni for TB kātso, ta kāts which would have been borrowed from the medical jargon. However, TB kātso, ta kāts is not used exclusively as a medical term, nor does it have a particular semantic connotation. It is the standard word for ‘stomach, belly, abdomen, womb’ in both Tocharian languages.132 Therefore, while not excluding a Khotanese origin for TB kātso, ta kāts, I remain cautious for the time being. TB suwo ‘pig’ and TB luwo, ta lu ‘animal’ Two fauna words display nom.sg. -o, obl.sg. -a: TB suwo ‘pig’,133 of unknown gender, and TB luwo, ta lu ‘animal’, an alternating noun with the rare plural morpheme TB -sa. The PIE source of the first term is *suH- ‘pig, swine’ (> Lat. sūs, Gk. ὗς, YAv. hū-, etc.), but the Tocharian paradigm is problematic since from PIE *suH-s we would expect a nom.sg. *swa/suwa, not the attested suwo (B549 a6, cf. Katz 1997: 79–80). For this reason, a proto-form enlarged with a nasal suffix is usually reconstructed, i.e. PIE *suHōn/*suHn- (Winter 1965: 192; Hilmarsson 1988a: 507–508; dtb 763). Peters (1991), Kim (2009a), and Malzahn (2011) are of a different opinion: they all claim that nom.sg. *suHs, acc.sg. *suHm yielded nom.sg. *səwas, obl.sg. *səwam in a pre-Proto-Tocharian period. However, the expected paradigm should have had final -a in both the nominative and the oblique singular. In order to explain the nom.sg. -o, Peters (1991: 243) and Kim (2009a: 79) argue that an analogical replacement of *-as by *-ās affected the nominative singular. On the other hand, Malzahn postulates a sound law pre-PT *-ăs > PT *-o > TB -o, so that the nom.sg. suwo would directly mirror PIE *suHs. Before commenting on this sound law, let us introduce the paradigm of TB luwo ‘animal’, clarifying its etymology and derivation. Two etymological proposals have been advanced so far:134 (1) TB luwo is from PIE *luHs- ‘± louse’ (cf. OHG lūs, OIcelandic lús, OE lows; MW lleu, MBret. lou [collective], etc.); (2) TB luwo is to be linked with the verbal root PIE *leu̯ H- ‘to separate, cut off’ (PSlavic *lȏvъ ‘hunting’ (?), Lat. luō ‘to suffer’ (?), Skt. lunā́ti ‘cuts off’) or *leu̯ ‘beschmutzen’ (liv 2 414, cf. also Gk. λῦμα ‘filth, garbage’ < *lus-mn).135 From a formal point of view, both Germanic and Tocharian point to PIE *luHs-, which 132 See Carling (2000: 212–214). 133 A plural form of TB suwo is perhaps to be restored in THT2701 line 4 ///teṃ yiknesa ṣkas ssuw/// “In this manner six pigs (?)” (Ching 2010: 307). 134 The two etymologies were first proposed by Pedersen (1941: 72) and Van Windekens (1976: 268) respectively, but the formulations presented here are from Hilmarsson (1988a: 507–508) and Adams (dtb 607; differently in Adams 1988a: 129). 135 According to Adams (dtb 607), Gk. λέων ‘lion’ can be interpreted as a nominal derivative from PIE *leu̯ H-, i.e. *leu̯ H-ōn ‘the hunter, predator’. However, several details are still Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

135

can be interpreted as a neuter s-stem built on the zero grade of the root PIE *leu̯ H-. The reconstruction of a neuter s-stem for the Tocharian word is suggested by the plural formation TB lwāsa, which displays an “s-Erweiterung”. This plural morpheme is extremely rare, since it is further attested in piltāsa ‘leaves’ (teb § 159) and lyyāsa ‘limbs’ alone.136 Therefore, there is no doubt that it is an archaism rather than a secondary “s-Erweiterung”.137 For this reason, I cannot agree with Adams (dtb 607) in arguing that the Tocharian B plural -sa in luwo “may result from a cross of this etymon with a PT *tsäuwā ‘animal’, reflecting PIE *dhéuhxōs ‘animal’”. This hypothesis faces two problems. On the one hand, no other Indo-European language points to a collective s-stem *dhéuhxōs, but rather to a thematic formation (e.g. Goth. dius ‘wild animal’, OE dēor ‘id.’ are from PGerm. *deuza- ‘beast’ < *dheu̯ só-, see Kroonen 2013: 94–95). On the other hand, we have no Tocharian continuants of Adams’ *dhéuhxōs. I therefore believe one must reconstruct a PIE s-stem for both the singular and the plural inflection of TB luwo, ta lu. We can now finally discuss the sound law proposed by Malzahn (2011: 94–95). As mentioned above, she believes that pre-PT *-ās and *-ăs resulted in PT *-o > TB -o. This sound law is aimed at explaining the singular paradigm of both suwo and luwo. However, while nom.sg. *suH-s (> *suwas), acc.sg. *suH-m (> *suwam) could theoretically underlie nom.sg. suwo, obl.sg. suwa, a sound law *-ăs > PT *-o could not account for the singular paradigm of luwo, because it comes from a neuter s-stem with both nominative and accusative reconstructed as *luHs. In accordance with Malzahn’s sound law, we would expect TB luwo both in the nominative and in the oblique singular and further reconstruct analogy after obl.sg. suwa to explain the obl.sg. luwa. Since this sound law does not solve all problems linked to the paradigm of TB suwo and luwo and, above all, because it is based only on these two nouns, I cannot accept it.138 unclear, and scholars still prefer a non-Indo-European (Semitic) source for Gk. λέων ‘lion’ (see Beekes 2010: 854; gew 2.113). 136 Winter (2003a: 117–118) reconstructs a nom.sg. lyiyo*, obl.sg. lyiya*. For an etymological proposal, see Van Windekens (1976: 567). For further details on the plural form, see Pinault (2008: 467), Schmidt (2008: 326–327), Malzahn (2010: 851). 137 The corresponding Tocharian A forms show a different development, since the plural of pält ‘leaf’ is pältwā, and the plural of lu ‘animal’ is lwā. See Winter (1965: 122–124) for further details. Kortlandt (2020: 116) reconstructs *l(e)uh2, acc. *lueh2m and explain the s-plural as a secondary collective formation caused by the homonymy of gen.sg. and nom.pl. *luh2es. I find this solution very hard to accept. 138 Malzahn (2011: 94) claims that a sound law *-ăs > o could account for members of the palsko-type (nom.obl.sg. -o, nom.obl.pl. -onta) as the descendants of an inflectional type in PIE *-h2s-, cognate with the so-called Greek κρέας-type. Meissner (2005: 122–128) has Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

136

Chapter 3

We are left with Winter’s PIE *suHōn (1965: 192), which would yield the attested TB suwo quite regularly. Since analogical influence between the Proto-Tocharian paradigms of suwo ‘pig’ and luwo ‘animal’ may have occurred, one may wonder whether the -o in TB luwo would have been taken from the word for ‘pig’ (Hilmarsson 1988a). There is, however, a problem with the reconstruction of PIE *suHōn itself. Indeed, no other IE language points to such a proto-form, and this isolation within the Indo-European domain is suspicious. As a consequence, I believe Tocharian inherited PIE *suH- and *luHsdirectly. For a certain stage, a paradigm with an undifferentiated nom.obl.sg. *sǝwa and *lǝwa is to be reconstructed. Then, a secondary distinction took place between the nominative and the oblique through the introduction of the forms *sǝwo and *lǝwo in the nominative. This final PT *-o > TB -o was plausibly introduced after other fauna terms that synchronically belong to either the okso- or the arṣāklo-type (both with a secondary obl.sg. -ai, cf. § 3.5.2), like okso ‘ox, cow’, arṣāklo ‘snake’, kercapo ‘donkey’, mewiyo ‘tiger’, oṅkolmo ‘elephant’, kraṅko ‘cock’, etc. As we will see, the singular paradigm of these nouns can be reconstructed as nom. *-o, obl. *-a for a certain stage of Proto-Tocharian (§ 3.5.2.4). As a consequence, both the singular inflection and the semantics of these nouns favoured the generalisation of the ending nom.sg. *-o to the otherwise undifferentiated singular paradigm of PT *sǝwa and *lǝwa. It is probable that TB suwo retained an obl.sg. suwa and did not develop an obl.sg. **suwai, because of its formal resemblance with TB luwo, obl. luwa. On the other hand, the plural PIE *luHs-h2 regularly yields the attested TB pl. lwāsa, while, in Tocharian A, it was expected to yield *lwās (nom. = obl.). This isolated plural form could have easily been remade into the attested lwā.139 There are two other nouns that have the rare plural TB -sa, namely piltāsa ‘leaves, petals’ and lyyāsa ‘limbs’ (see fn. 146 below). Winter (1962a: 112) and Schmidt (1982: 363) suggest that the paradigm of the word for ‘leaf, petal’ was parallel to that of TB luwo, positing a nom.sg. TB pilto*. The same reconstruction was recently advocated by Malzahn (2011: 86–87 fn. 10). On the other hand, Krause & Thomas (teb § 159, 2), Adams (dtb 415), and Pinault (2008: 205) give a singular pilta (nom.=obl.). I believe that the latter paradigm is correct. Indeed, the form pilta, attested in B622 b4 /// uppālṣe pilta nest /// “you are a lotus.m petal”, can hardly be interpreted as something other than a nominative. This clarified that this type is a recessive category in Greek (with less than thirty nouns) and that comparative evidence to reconstruct neuter nouns in *-h2s is very meagre. On the origin and the evolution of the palsko-type, which, in my view, is quite different, see § 3.6.2.1. 139 A form TB luwāñ seems to be attested in IT395 b2, which is very fragmentary. Formally, this luwāñ might be either a secondary nominative plural or a causal singular of luwo.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

137

makes the paradigm of TB pilta and TB luwo synchronically different. However, since the nom.sg. -o in luwo has been explained as secondary, their paradigms were probably identical at an unattested stage of Tocharian. This allows us to reconstruct an old neuter s-stem for the antecedent of TB pilta: the singular paradigm goes back to PIE *-Hs, while the plural paradigm is from PIE *-Hs-h2. The word may come from PIE *pelth2- (Pinault 2008: 205). The dual piltāṣ ‘two petals’ may also be traced back to PIE *pelth2-s-ih1 (Kim 2018: 73). TB āyo, ta āy ‘bone’ There is just one other alternating noun that has nom.sg. -o, obl.sg. -a: TB āyo, ta āy ‘bone’. In the previous edition of his Tocharian B dictionary, Adams (1999: 45) provided a list of variants for the singular paradigm of this noun: nom.sg. āy ~ āyo, obl.sg. āy ~ āya, with (synchronically) suppletive plural āsta. Pinault (2008: 333) argues that the singular is āy < *ayə, and further analyses āyo as a poetic form and āya as a new plural formation. However, as correctly pointed out by Peyrot (2008: 111–112), a hypothetical TB †āy is never attested in the entire corpus of Tocharian B. The singular forms are the following (Peyrot 2008: 111): (1) nom.sg. in W20 b3 oṅkolmaiññe āyo –(–)⋅ī[l]e “elephant’s bone is to be … ed”.140 Unfortunately, the correct reading of the line is obscured by ink stains from another leaf that was laid over it. However, Peyrot is certainly right in reading the final part of a gerundive at the end of the line. This gerundive is inflected as a nom.sg.m. in agreement with āyo ‘bone’. As for the internal coherence of the text, an elephant bone (i.e. its tusk?) that must be treated in some way would fit well in a medical context; (2) nom.sg. in B514 a2 (archaic) /// vidūṣake weṣṣäṃ ᛫ ayo mā ⋅i/// “the jester speaks: « a bone … not … »”; (3) obl.sg. in AS4A b1 (= NS27 a2) tsirauwñeṣṣe kauṣn āya ompalskoṣṣe mrestīwe pakṣäṃ “He breaks the bone of energy [and] he cooks the marrow of meditation” (cf. Meunier 2015a: 169). The fact that TB āya must be analysed as a singular is confirmed by the agreement with a modifier inflected as a masculine singular (tsirauwñeṣṣe ‘pertaining to energy’).141 140 The transcription partially follows Peyrot (2008: 111). Filliozat’s oṅko(lma)ññe āy (1948: 72) is based on an inaccurate fac-simile by Hoernle (1902), as Filliozat himself wrote (p. 64). The manuscript clearly reads oṅkolmaiññe for expected oṅkolmaññe, with ai for a probably due to the following palatal consonant (Peyrot 2008: 54). 141 Cf. the translation of the passage by Georges-Jean Pinault apud CEToM, where āya is translated as a plural form.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

138

Chapter 3

Furthermore, the derived adjective ayāṣṣe /ayáṣṣe/ corroborates this analysis, since it is regularly based on the oblique singular:142 IT246 b4 (BhikṣuPrātimokṣasūtra) se ṣamāne ayāṣṣe kemeṣṣe sucīkār yamastär pāyti “Whichever monk makes a needle-box (Skt. sūcīgharaka) out of bone [or] ivory, [he commits] an offence (Skt. pātaya(n)tikā)” (cf. Broomhead 1962: 1.80). Other fragments where one could read independent sequences of āyo or āya are broken or severely damaged, especially at the end of the line, where unfortunately these words are mostly attested. For many of them, the restoration of TB āyor ‘gift’ is preferable. Other probable, but not certain, readings of TB āyo are in IT826 b5 and THT1324.b a2. The former is a small fragment, but it seems to deal with medical practices; in the latter, the reading TB āyo may be supported by the attestation of the plural āsta ‘bones’ in line b1. Dealing with the paradigm of this word, Hartmann (2013: 267–268) proposes a new interpretation that seems to give credit to the variant forms given by Adams (1999: 45–46). After introducing and commenting on Peyrot’s analysis of the singular paradigm, Hartmann argues that the annexation of TB āyo to the kantwo-type is likely but not entirely convincing. Crucial to his argumentation is the hypothetical attestation of TB ay in B284 b2 (arch.), which he interprets as an oblique singular of āyo: apsāl śakattai ṣäp ay ṣesa pyākälyñe “striking with sword, club together with bone” (translation by Adams 1999: 619). Since the obl.sg. āya is attested in a classical document with some late forms, while the alleged obl.sg. ay occurs in an archaic one, Hartmann concludes that TB āy is the old and regular form. In the history of Tocharian B, a new nom. sg. āyo would then have resulted through reanalysis of a form with o-mobile and, later, the obl. sg. āya would have been analogically created after the paradigm of TB luwo ‘animal’. There may be some problems with this theory, however. If it is true that the spelling ay might be an archaic spelling variant of TB āy /áy/ (cf. ayo in the archaic fragment B514), the syntax of the sentence in B284 b2 is odd and Adams’ translation is puzzling. From a morphosyntactic perspective, the verb TB pyak- is never combined with ṣesa, a postposition usually constructed with the comitative. A form *ayämpa (or the like) would therefore be expected. I am further hesitant to assume a new singular paradigm nom. āyo, obl. āya originating from the creation of the new nom.sg. āyo from *ayǝ with o-mobile. Indeed, the Tocharian B phenomenon named o-mobile (“bewegliches o”) usually presupposes that an original final -ä /-ǝ/ is replaced by final -o in metrical (mostly archaic) texts in pāda- or colon-final position (Pinault 2008: 404–405; Malzahn 2012a). To my knowledge, the variant with o-mobile has never been 142 Cf. Peyrot (2008: 111); see Pinault (2008: 333) for a different analysis.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

139

reinterpreted as a new inflected form replacing the original one with final -ä. Perhaps the only exception could be the plural of the imperative active, where the variant with -o is not confined to the usual contexts (Malzahn 2010: 42). Alternatively, one might say that TB āyo is itself an example of o-mobile of a regular āyä* (as per Pinault 2008: 333). However, as acknowledged above, a clear occurrence of TB āyo is attested in a non-metrical text, i.e. a collection of medical recipes. To sum up, the correct paradigm of the word for ‘bone’ in Tocharian B is: nom.sg. āyo, obl.sg. āya, nom.obl.pl. (suppletive) āsta.143 The Tocharian A paradigm is: nom.obl.sg. āy, nom.obl.pl. āyäntu (dta 53–54). As far as the etymology of the term is concerned, we would ideally derive the Tocharian noun directly from the familiar PIE word for ‘bone’, namely PIE *h2óst- / *h2ést- (or *h3ést-). The plural TB āsta has evolved quite regularly. Pinault (2008: 428) outlines the following development: PIE *h2óst-h2 > *esta > PT *asta (through a-umlaut) > TB āsta.144 The origin of the singular TB āyo (obl.sg. āya) and TA āy (pl. āyäntu) is more difficult. Hartmann (2013: 448–453) and Adams (dtb 48–50) have recently summarised and commented on previous proposals. Van Windekens’ (1976: 173) derivation from PIE *h2éi̯u- ‘life-force’ (cf. Skt. ā́yu-) is phonologically attractive, but semantically difficult (cf. dtb 49–50). Hilmarsson’s *h2ei̯d-i-h2 ‘swelling’ (cf. Arm. ayt ‘cheek’) is also difficult, from both a morphological and a semantic point of view. Katz (1997: 73–77) takes *ay as the regular outcome of PIE *h2ést- > *ast > *as > PT *ay by sound law of pre-PT *-s > -y in monosyllables. Such a sound law, however, has no clear parallels in Tocharian (see § 3.4.1.2) and the word TB †āy no longer exists. As a matter of fact, the situation of this word is quite peculiar, because it is an accented monosyllable. The expected outcome of PIE *h2ést would have been PT *a after regular loss of final consonants.145 One may wonder whether this PT *a ‘bone’ (?) was reshaped to *ayo (obl.sg. *aya) after the paradigm of the word for ‘member’, TB lyiyo* /ĺǝ́yo/, pl. lyyāsa /ĺyása/ (TA pl. lyiyā ~ lyā), or other nouns of the kantwo-type referring to body parts. However, TB lyiyo*

143 See also Adams (dtb 48–49) and Malzahn (2011: 99). 144 Contrary to Van Windekens (1976: 172–173), who proposes a loanword from Khot. āstaa‘bone’ < *astaka- (with pl. āste), see Isebaert (1980: 190). 145 Adams (dtb 49) suggests that this PT *a was perceived as overly short by Tocharian speakers and it was extended in some way, perhaps by the outcome of the PIE suffix *-i̯o/-i̯eh2-, also attested in other body-part terms in some other Indo-European languages (e.g. Skt. āsya- ‘mouth’ alongside ās- ‘id.’).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

140

Chapter 3

is not attested in the singular, either in Tocharian B or in Tocharian A, and its etymology is equally uncertain.146 TB maiyya, -yo ‘force, strength’ The last substantive of the kantwo-type is the abstract noun TB maiyya, -yo ‘force, strength’. Here we discuss issues relating to both the inflection and the ultimate origin of this noun. As far as its paradigm is concerned, Adams (dtb 508) gives the following forms: nom.sg. maiyya ~ maiyyo, obl.sg. maiyya (archaic meyyā, maiyyā), nom.pl. maiyyāñ, obl.pl. maiyyaṃ ~ maiyyana. Hartmann (2013: 237) follows Krause & Thomas (teb  §  163.2, p. 121) in considering the nom.obl.pl. maiyyana the only regular form as he includes TB maiyya under class ii.1. This paradigm is bizarre for two reasons: (1) there are no other feminine nouns in Tocharian B in which nom.sg. = obl.sg.; (2) the plural ending -ana is confined to nouns of the aśiya-type which have a different inflection and different semantic peculiarities (see § 3.4.2). It is therefore worth having a closer look at the relevant attested forms. As far as the singular inflection is concerned, it is generally assumed that the original nom.sg. of this word was maiyya with maiyyo as a later variant (cf. dtb 508). Peyrot (2003: 62–64, 2008: 99–101) demonstrates that many substantives of the wertsiya-type (nom. sg. -ya ~ -yo, obl. sg. -yai) show a general trend to shift to a subtype with nom.sg. -o between the classical and late stages, while they consistently attest a nom.sg. -a in archaic documents. I have therefore checked the occurrences of TB maiyyo in the texts. They are all from classical and late texts, thus apparently confirming Peyrot’s distribution of the variants: maiyyo (NS103 a1 [class.], B21 b5 [class.-late], B231 b5 [class.-late], B278 b2 [class.], B371 146 For the identification of the word, see Pinault (2008: 146–147) with references. Blažek (2012: 16) has connected TB lyiyo* with Hitt. ḫalii̯e/a-zi ‘to kneel down’, which has been traced back to PIE *h2l-oi̯-/ *h2l-i- by Kloekhorst (2008: 273–274) (cf. the reduplicated halihla/i- ‘to genuflect’ < *h2li-h2l(o)i-). For Tocharian, Blažek reconstructs *h2li-h2i-, without clarifying how this proto-form could have evolved into TB lyiyo*. Witczak (2017) reconstructs an s-stem noun referring to fleshy parts of the body, which he derives from the PIE root *leh1- ‘±smooth’ (cf. Gk. λεῖος ‘level, smooth’, Lat. lēvis ‘id.’ < *leh1i-u̯ - (?), Gk. λῑς́ ‘smooth’ < *lih1-t-). According to him, evidence for this s-stem comes from OE līra ‘muscle, soft part of the body’, MLG liese ‘thin skin’, Lith. líesas ‘lean, thin’, Latv. liẽss ‘id.’, and Hitt. lēši, līšši ‘liver’. If Proto-Indo-European had such a neuter s-stem noun, a paradigm sg. *l(e)h1i-s, pl. *l(e)h1i-sh2 ‘soft part of the body’ could have evolved into sg. *ĺǝy, pl. *ĺǝysa (or sg. *ĺey, pl. *ĺeysa). This paradigm may then have been remade into nom.sg. *ĺǝyo, obl.sg. *ĺǝya, pl. *ĺǝyasa after other body part terms that belong to the kantwo-type. Otherwise, one may link TB lyiyo* with the PIE root *lei̯- (cf. Goth. liþus ‘member, body part’, ON liðr ‘joint’, OHG lid ‘joint, articulation’, Du. lid ‘id.’ < *liþu-, ON limr ‘limb’, E limb ‘id.’ < *limu-, Kroonen 2013: 338 and 340). See also Pan (2021: 91). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

141

b2 [class.], THT1131.i [class.-late?]; (mai)yyo (IT27 a1 [class.]); mai(yy)o (B17 b8 [class.-late]); maiyo (AS8B a4 [class.-late]). However, forms in nom.sg. -a not occurring in compounds can equally be found in classical texts, though such instances are rare (cf. IT178 b7 [class.] maiyya räddhiṣṣa “r̥ddhi-power”, B15 b6 [class.] po tu maiyya ṣamāneṃts “all this (is) the power of the monks” [vs. maiyyo in the parallel B17 b8]).147 Conversely, a nom.sg. °maiyya frequently occurs in compounds of the bahuvrīhi-type, which display a slightly different inflection with nom.sg. -a, obl.sg. -ai.148 I therefore wonder whether the original nom.sg. of maiyya as a free word was actually maiyyo, with the sporadic forms with -ya as the outcome of an analogical generalisation of the common nom.sg. °maiyya found in compounds. This explanation, tentative as it is, would have the advantage of inserting the noun maiyyo ~ maiyya into a regular inflectional class.149 Turning to the plural paradigm, the following lists contain all attestations of the noun’s plural forms: (1) IT96 a5 snai-maiyyañ; IT36 b1 ///maiyyañ; NS56 b2 śak maiyyaṃ ‘ten powers’; B211 a2 śak maiyyaṃ ////; B303.d. b1 ///maiyyaṃ; NS49A b5 pudñ(ä)kti maiyyan akṣāre150 ‘the Buddhas announce the powers’; B621 b3 /// ⋅tār maiyyaṃ akṣāre ‘they announce the powers’; (2) B31 a1–2 (mai)yyana; B553 b4 mai(yyana). All these documents are written in classical Tocharian B, with archaic forms in B211 and B553, and late forms in NS56. Note that TB śak maiyyaṃ in NS56 and B211 may not be a compound; it may instead be a noun phrase with the meaning ‘ten powers (= Skt. daśa balāni)’. The form ///maiyyañ in IT36 b1 may show the regular accentuation in compounds (cf. snai-maiyyañ in IT96 a5). 147 See also Peyrot (2008: 100 fn. 137). The older variant meyyā, meyya is attested in archaic texts, either as an oblique or with secondary cases (e.g. IT22 a3 meyyaṣṣī; B248 a1 meyyā; B274 b1 and b2 meyyāsā; B394 b7 mey(y)a///). See Peyrot (2008: 58–59) and Pinault (2008: 275). 148 Cf. e.g. nom.sg. śka-maiyya ‘having the ten powers’ (IT99 b4, NS32 a1, B5 a7, B 18 b6), obl.sg. śka-maiyyai (IT207 b2, B7 b7–8, B109 b3 [with śika-], B252 a3 [with śkä-], B252 a4 [with śkä-], B405b1, B583b1, THT1677.a a4), śka-maiyyantse (AS7J a1, B14 b2, B33 b6, B375 b3), nom.sg. snai-maiyya ‘having no power(s)’ (B46 a1), obl.sg. snai-maiyyai (THT1543.e a4). 149 Malzahn (2011: 93 fn. 25) seems to suggest that the noun was originally of the wertsiya-type, with nom.sg. -a, obl.sg. -ai, with earlier dissimilation of obl.sg. *maiyyai > maiyya. However, the obl.sg. -a is attested since the archaic stage, where an obl.sg. **meyyai would have hardly evolved into meyya after dissimilation. 150 This form is ambiguous, because it could theoretically be a sandhi variant of maiyyana (maiyyan= akṣāre). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

142

Chapter 3

On the other hand, since the form ///maiyyaṃ in B303.d. is the first discernible word at the beginning of a broken line, we cannot tell whether it was in composition or not. At this point, one may wonder whether the noun maiyyo ~ maiyya ‘power, strength’ took the ending -ana when inflected as a free word and the plural paradigm nom. -añ, obl. -aṃ when in composition. However, a closer look at the forms with the ending -ana shows that they are both partially restored. The restoration in B553 is by Claus-Peter Schmidt (1972: 55), who read the right portion of line b4 as ᛫ piṃś indriṣi mai(yyana)/// “the five indriya-powers”.151 Fragment B553 is in verse and, according to Sieg and Siegling (1953: 348), the metre is 4×7¦7. A plural mai(yyana) would thus have the advantage of restoring the first hemistich of the verse. However, this is not the only possible restoration, as a nominative plural mai(yyāño) (with o-mobile) or mai(yyāñä) could equally fit the lacuna. We are left with line a1–2 of B31, which according to Sieg & Siegling’s edition and emendation (1949: 1.52, 2.51) runs as follows: alloṅkna ṅke ś(āmna)ts (mai)yyana “However, other powers of (humans)” (translation by Hannes A. Fellner apud CEToM). Since the original manuscript of THT31 (= B31) is missing and since maiyyana is also the outcome of a restoration in this case, I would rather agree with Hilmarsson (1989b: 87) in being cautious in reconstructing maiyyana as the plural of TB maiyyo ~ maiyya. Like its inflection, the etymology of this noun has also proven controversial. Adams (dtb 509) links TB maiyyo ~ maiyya to the adjective TB maiwe ‘young’, which he traces back to PIE *moHi-u̯ o-. He reconstructs an old abstract in *-i̯eh2 derived from this adjective, which would have evolved into our TB maiyya. If the inflection given by Krause & Thomas is correct (with nom.sg. -a, pl. -ana), the only explanation I can think of to explain this deviating paradigm is to reconstruct TB maiyya as a substantivised adjective of the original feminine form of TB maiwe ‘young’, which does not, in fact, attest a feminine inflection. If so, both the endings nom.obl.sg. -a and -ana are of adjectival origin, as at a certain pre-stage of Tocharian the singular paradigm of feminine adjectives did not differentiate the nominative from the oblique. They both ended in *-ya (see § 4.3.4.6).152 Also, Isebaert (1980: 113, 184–185) suggests that maiyyo is an indigenous word, which he derives from PIE *modi̯eh2 (to the root *med- ‘to 151 The apparent masculine agreement with a feminine noun is not infrequent in verse. See § 4.6.1. ́ 152 Another possibility is to interpret TB maiyya as a vr̥kī-derivative of maiwe (see § 3.5.3.3). From a semantic point of view this reconstruction works fine, because the original meaning of PT *meyẃa would have been ‘pertaining to the youth’ and then ‘force, strength’. If so, however, the deviating plural maiyyana would be hard to explain.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

143

measure’ [liv 2 423]) with a similar development to that allegedly attested in TB paiyye ‘foot’ < *podi̯o- (but cf. dtb 432). On the other hand, Van Windekens (1976: 629) has proposed that this word was borrowed from Sanskrit māyā́- ‘art, wisdom, extraordinary power’ (mw 811). This proposal is unproblematic from a semantic point of view, but formally we would rather expect †māy(ä) (dtb 508, cf. TB saṃjñä ← Skt. saṃjñā-). In his forthcoming doctoral thesis, Chams Bernard proposes that TB maiyyo ~ maiyya is a direct loanword from so-called Old Steppe Iranian *mai̯(i̯)ā-, that is, the reconstructed cognate of Skt. māyā́-, OAv. māiiā- ‘craft, strength, joy’, YAv. maiiā-. The ultimate origin of the Indo-Iranian word is unclear and a BMAC origin has been proposed (Lubotsky 2001: 314; Bernard forth.). Accordingly, although an Indo-European etymology of TB maiyyo ~ maiyya could still be a possibility, an Iranian (or BMAC) etymology by way of a borrowing may equally — if not more easily — underlie the origin of this complex Tocharian B word.153 3.5.1.3 Conclusion In this section, I have analysed a group of nouns with nom.sg. -o, obl.sg. -a. This is a very heterogenous class continuing both inherited and borrowed nouns. Among the inherited words, the only two alternating nouns belonging to an inflectional type in some way parallel to the kantwo-type can be traced back to old s-stems (TB luwo, ta lu ‘animal’) or to PIE root nouns (TB āyo, ta āy ‘bone’). I have shown that these nouns have only secondarily been included into this class. As for TB suwo ‘pig’, we do not have any attestation of the plural paradigm and its inflectional type remains unknown. However, it may mirror its PIE reconstructed ancestor, with some motivated analogical adjustments to account for its nominative singular ending. Among the inherited words which distinguish the nominative from the oblique in the plural, I have confirmed that the only noun that has a certain etymology and exact Indo-European correspondences is TB kantwo, ta käntu ‘tongue’, whose paradigm continues the hysterodynamic type of PIE, with ablauting suffix *-éh2-/-h2-. To explain the contrast between nom.sg. -o, obl.sg -a there is no need to reconstruct either a sigmatic nom.sg. PIE *-eh2s or 153 Since the borrowing would have taken place in a stage predating the separation of Tocharian B and Tocharian A from Proto-Tocharian (Bernard forth.), I would not expect this noun to have originally been borrowed as meyyā ~ maiyya (nom.=obl., with pl. maiyyana) and then shift to the kantwo-type in the historical period. Indeed, all other prehistoric loanwords in Tocharian were adapted into an inherited Tocharian inflectional class and none developed unique paradigms. Therefore, the borrowing hypothesis may give credit to the reconstruction of maiyyo as an original member of the kantwo-type.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

144

Chapter 3

a nasal extension. Indeed, the nom.sg. most clearly continues the nom.sg. *-eh2 of PIE, while the obl.sg. has generalised the stem of the weak cases *-h2-. The stock of the other inherited nouns is made up of: (1) words with certain etymologies and exact Indo-European correspondences which can be traced back to a PIE type in *-(e)h2 (TB karyo* ‘viscera’, ta kri ‘will, desire’); (2) abstract nouns that are very old Tocharian formations (TB tsāro ‘monastery, nunnery’; TB kāwo ‘desire’); (3) nouns built with the suffix PT *-(i)to (TB laukīto, ta lokit ‘stranger’; TB nekīto*/tekīto*), which can be traced back to a PIE type in *-(e)h2. Thus, one could be tempted to reconstruct an ablauting paradigm for the ancestors of all these nouns, so that they inherited (or generalised) the full grade in the nominative (*-éh2-) and the zero grade in the weak cases (*-h2-). If so, there is no strict historical link between gender and the original inflectional type of these nouns, and this is the reason we also find some masculine nouns in the kantwo-type. Kortlandt (2013: 95–96) reconstructs a PIE hysterodynamic type with full grade in the nom.sg. and zero grade in the other cases for some of the members of the kantwo-type.154 This reconstruction is possible, although not entirely provable. Nonetheless, I will favour this hypothesis, as one can also argue that some old derivatives in *-ā < *-eh2 developed an ablauting paradigm in a pre-Proto-Tocharian period. This is an issue we will return to in the following sections, where I will argue that it is more economical to assume that Tocharian inherited and generalised the hysterodynamic type in *-h2- in the older stage of the pre-Proto-Tocharian nominal inflection. Despite the fact that the kantwo-type comprises both masculine and feminine nouns, there is evidence that shows that the bulk of this type comprised feminine nouns. Indeed, if we consider the Iranian loanwords found into this class, we notice that, unless the word denotes a male person, there is a strong preference for them to change from masculine to feminine (Dragoni 2022: 247). Thus, if kāswo and kātso have been correctly interpreted by Dragoni (2022) as Khotanese loanwords (see the relevant section above), they have become feminine, while the source word was masculine in Khotanese. Similarly, TB mālo, which is to be traced back to the Proto-Bactrian antecedent of Bctr. μολο /mol/, is also feminine, while it should have been either masculine or neuter in the source language (cf. Av. maδu- nt., Ved. mádhu- nt., Khot. mau- m., EWAia 2.302). This may show that the default gender assigned to this class was the feminine (cf. also TB tāno f. ‘seed’ ← Proto-Tumshuqese-Khotanese or pre-Khot. dānā- f.).

154 Note, however, that I follow the Erlangen reconstruction of the nominative singular with full grade in the suffix, thus *-éh2 (vs. Beekes and Kortlandt’s reconstruction of the ending with a zero grade *-h2). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

145

Gender in the Inflection of the Noun

A further conclusion we can draw by combining the evidence from inherited and borrowed words is that the kantwo-type is an archaic inflectional class which is synchronically unproductive. Indeed, only loanwords from the oldest stages of the Iranian languages with which Tocharian has been in contact can be found into this class, be they from Old Steppe Iranian (perhaps TB maiyyo ~ maiyya ‘power, strenght’), Proto-Tumshuqese-Khotanese (e.g. TB eñcuwo, ta añcu* ‘iron’, TB keto ‘property, estate’, TB tāno ‘seed’), pre-Khotanese (perhaps kāswo ‘leprosy, skin disease’, kātso, ta kāts ‘belly, stomach’), or Proto-Bactrian (mālo ‘alcohol’). No loanwords from the historical period have been adapted to this class. To conclude, the kantwo-type is an archaic inflectional class which was productive only in prehistoric times. The bulk of its inherited members have either inherited or secondarily acquired a hysterodynamic type in *-éh2-/-h2-, which has been continued in the contrast between nom.sg. -o, obl.sg. -a. As I will show in the following section, however, the kantwo-type must have originally included a larger number of nouns, some of which were secondarily transferred into the okso- and arṣāklo-classes. 3.5.2 The okso-type and the arṣāklo-type: TB Nouns with nom.sg. -o, obl.sg. -ai and their TA Correspondents The Tocharian B okso- and arṣāklo-types are two closely related inflectional classes. Since they have similar case endings, their paradigms seem to overlap at first sight. However, a closer look at their inflection and derivation reveals distinct differences. As can be seen from the table below, the inflection of these two types differs in the stem to which the case markers are attached: in the okso-type, all non-nom.sg. forms and derivatives are built on an ai-stem (cf. gen.sg. oksaintse* and the derived adjective oksaiññe ‘pertaining to the ox’), while in the arṣāklo-type they are built on an a-stem (gen.sg. arṣāklantse and the adjective arṣāklatstse* ‘± snake-infested’).155 Table 23

Inflection of the okso-type and the arṣāklo-type

okso-type arṣāklo-type

nom. sg.

obl. sg.

nom. pl.

obl. pl.

stem

okso arṣāklo

oksai-∅ arṣākla-i

oksai-ñ arṣākla-ñ

oksai-ṃ arṣākla-ṃ

oksaiarṣākla-

155 Cf. also the contrast between dual forms of the okso-type, e.g. TB oksai-ne ‘two oxen’, TB pokai-ne ‘(two) arms’, ta pokeṃ ‘id.’, and dual forms of the arṣāklo-type, e.g. TB yerkwanta-ne /yerkwǝ́ntane/ ‘two wheels’, wcūka-ne /wǝcǝ́wkane/ ‘(two) chins’. See Kim (2018a: 44–46). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

146

Chapter 3

The differences between the okso-type and the arṣāklo-type have caused some debate; a central question has been the origin of the ai-element.156 Winter (1989) was the first to deal with this problem in a systematic way. He showed that the two Tocharian B inflectional classes are in complementary distribution: all members of the okso-type are disyllabic, while all members of the arṣāklo-type are tri- or polysyllabic. As a consequence, Winter explains the contrast -ai- vs. -a- as depending on the position of the accent in the plural: the substantives of the okso-type were stressed on the last syllable, while the substantives of the arṣāklo-type were stressed on the penultimate syllable. Winter’s analysis is generally accepted today. However, the historical issues to which these classes give rise are by no means solved, to such an extent that there is little convergence of scholarly opinion. On the strength of views expressed by some scholars (most notably Pinault 2008: 483–485 and Peyrot 2012a), I will in this section deal with the origin of these classes and with the spread of TB -ai in the nominal declension. My final aim is to understand the role of these inflectional classes in the evolution of the *eh2-stems from PIE to Tocharian. However, before proceeding to this diachronic matter we begin with an introduction to the nouns of these classes as well as a discussion on some of the etymologies of their members. 3.5.2.1 The okso-type As stated above, the okso-type consists of disyllabic nouns, which build their plural and derivatives on a stem ending in -ai- (cf. oksaiññe ‘pertaining to an ox’). The nouns of this type are typically feminine, but we can also find masculine nouns (e.g. okso ‘ox’, pānto ‘support(er)’, naunto* ‘street, road’, Hilmarsson 1987a). The stem finals are rarely attached to a preceding palatalised consonant (e.g. swāñco ‘ray of light’). In addition, some nouns attest alternation between -o and -iye in the nominative singular, a phenomenon that is still being discussed by specialists of Tocharian. Hilmarsson (1987a: 44–45) argued that the nom.sg. -iye is the result of analogy after other inflectional classes, as he recognised the nom.sg. -o as the oldest variant. This analysis would be substantiated by matters of historical phonology. Indeed, the o-umlauted stem in some of the okso-nouns can be explained by reconstructing an older nom.sg. -o.157 156 See recently Hartmann (2013: 413–417). 157 The assumption that the nom.sg. -o has been replaced by -iye seems to pose no difficulties from the point of view of Tocharian A. An apparent counterexample could be TB prosko ‘fear’, whose Tocharian A counterpart is praski ‘id.’. However, TA praski (alt.) cannot be the morphological match of TB proskiye (f.) for formal reasons. Following Peyrot (2008: 103, 2012a: 211) and Pinault (2011a: 174), the possibility of an independent formation in the two Tocharian languages seems to be the best way to explain this mismatch. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

147

Taking into consideration the meaning of the nouns, we distinguish the following semantic groups: – words designating animals, e.g. TB okso, ta opäs* ‘cow’, TB koro*158 ‘mule’, TB kraṅko159 ‘chicken, cock’ (= Skt. kukkuṭa-), TB tsāktso* ‘± duck’ (hapax nom.pl. tsaktsaiṃ in AS16.8 a5); – words designating plants, e.g. TB pyāpyo, ta pyāpi ‘flower’; – abstract and action nouns, e.g. TB prosko (~ -iye) ‘fear’; TB ścono, ta śoṃ* ‘enmity’; TB yoko (~ -iye), ta yoke ‘thirst’; TB pānto ‘support’; – terms for body parts, e.g. TB pokai (obl.), ta poke ‘arm’; TB klautso ‘ear’; TB porsno ‘ankle’. In addition, there are some nouns without any common semantic feature, like TB kolmo*, ta koläm ‘pond’,160 TB naunto* ‘street’, TB koṣko (~ -iye) ‘pit, hole’, TB koto* ‘crevice, hole in the ground’,161 TB lyauto ‘opening’ (cf. TA lot ‘hole’ and TB laute ‘moment’, see Hilmarsson 1988c).162 This inflectional class is synchronically very productive. On the one hand, there are some nouns that developed new inflected forms with an ai-stem analogically, as in the case of the late obl.pl. eśaiṃ (IT85 b2) from TB ek ‘eye’, shaped after nouns for body parts of the okso-type.163 On the other hand, this 158 TB koro* is mostly attested in the plural in documents that deal with caravan-passes (korai PK Bois B18 a4, koraiṃ B577 b2; cf. also koraiśke (?) PK DA M 507.27 b2). Pinault (2008: 391–392; 2009c) suggests a likely derivation from a reconstructed BMAC word *khara (cf. Skt. khara-, Av. xara-, Khot. khara-). See also Emmerick apud Emmerick & Skjærvø (1982: 35–36). According to Adams (dtb 263), TB khare* ‘ass’ (hapax legomenon in B511 a1) is a later loanword from Sanskrit. 159 TB kraṅko ‘chicken, cock’ (cf. perl.pl. kräṅkaiṃtsa AS16.8 a4 and the adjective kräṅkaiññe W14 a5, THT1520 a3, etc.) needs to be related to the onomatopoeic PIE root *kerk-/krek‘make noise’ (cf. the nominal derivatives in Gk. κρέξ, κρεκός ‘ruff’, Skt. kr̥kara- a kind of partridge, kr̥kavā́ku- ‘chicken’, YAv. kahrka° in kahrkāsa- ‘vulture’, MIrish cercc ‘hen’). It seems that we have the outcome of a nasalised variant *krenk- in Tocharian, which is also attested in Germanic (cf. OE hringan ‘to sound, ring’, ON hrang ‘noise’). In Khotanese we find kr̥ ṅga- ‘fowl, (wild) cock (= Skt. kukkuṭa-)’ (Bailey 1967: 52, 1979: 64), which strikingly resembles the Tocharian noun. Since all other Iranian languages have continued the nasalless variant (cf. YAv. kahrka-tāt, NP kark, Oss. kark, etc., de Vaan 2000: 284), one may wonder whether Khotanese borrowed this term from Tocharian (or vice versa?). See Dragoni (2022: 118–121). 160 For the meaning, see Huard (2022: 411–412). 161 Dragoni (2022: 111–113) suggests that koto* may actually mean ‘excrement’. He interprets this word as a loanword from pre-Khot. *gūθa- ‘excrement, faeces’ (> Khot. gūha-). 162 On TB pīto ‘price’, see § 3.6.2.1. 163 The palatalisation of the stem in eśaiṃ (vs. non-palatalised sg. ek < PIE *h3ekw-) comes from the dual stem eś°, which is from *h3ekw-ih1 (Kim 2018a: 78). In addition, TB klautso ‘ear’ (ta klots, du. klośäṃ) has two different stems: the singular has nom.sg. klautso, Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

148

Chapter 3

class comprises some loanwords from Indo-Iranian. A clear example is TB pātro, ta pātär ‘alms bowl’, borrowed either from Skt. pātra- (nt.) or from an intermediate language.164 TB koṣko ‘pit, hole’ (not ‘hut’ as previously assumed) seems to be a loanword of Iranian origin for which Bernard & Chen (2022) reconstruct a borrowing from either an unknown Middle Iranian language or an unattested Proto-Bactrian word. It is generally assumed that the bulk of this class is to be traced back to two PIE stem types: stems in *-on and stems in *-eh2 (Hilmarsson 1987a: 44; Pinault 2008: 484). Recently, Kim (2018a: 94) and Korlandt (2020) have questioned this reconstruction, as they claim that nouns belonging to the okso-type do not clearly continue n-stems “despite numerous attempts to connect them with” this PIE stem-type (Kim 2018a: 94). They both reconstruct this class as exclusively continuing stems in *-eh2. However, it is well-known that Tocharian inherited and further generalised nasal stems and endings.165 This is, in my view, shown by TB okso ‘ox’, the etymologically more unambiguous member of this class (see immediately below). Therefore, before proceeding further, it is worth recalling and commenting on the etymology of this noun in more detail. The etymology of TB okso, ta opäs* has never been in doubt: it has been linked to the familiar PIE word for ‘ox’, continued by many Indo-European languages, e.g. Ved. ukṣán- ‘id.’, Av. uxšan- ‘id.’, Goth. auhsa ‘id.’, OE oxa ‘id.’, OHG ohso ‘id.’, MW ych ‘id.’, MIrish oss ‘deer’, etc. This word is usually reconstructed as a hysterodynamic n-stem *uks-én-, *uks-n- (Oettinger 1980: 46; EWAia 1.20).166 Accordingly, the nominative singular was PIE *uksēn. However, this reconstructed form cannot be the direct ancestor of nom.sg. TB okso for phonological reasons (final TB -o, lack of palatalisation, o-umlaut), and several of obl.sg klautsai, while the dual is constructed on a stem klauts° (cf. klautsane /klautsə́ ne/ ~ klautsne). I agree with Hilmarsson (1989a: 102–103) that the original forms must be sought in the dual, as reflecting an Indo-European *ti-stem, *ḱlou̯ ti- from PIE *ḱleu̯ - ‘to hear’. This noun originally had a ne-less form, as confirmed by the derivative klautsa-pälṣi /klautsǝ́ pǝlṣi/ ‘± pricking up the ears’ (IT246 a4; B162 b2 (?)). Also in this case, the singular paradigm must be analogical after body part nouns of the okso-type. 164 Chams Bernard (apud Dragoni 2022: 140) proposes that TB pātro is from Khot. pātrā̆(ultimately borrowed from Sanskrit or a Prākrit variety). In Old Khotanese, the word is attested with variants in the Book of Zambasta, cf. acc.sg. pāttäru (Z 2.78c), pātro (Z 2.170c), pāttro (Z 2.169b). 165 Cf. Pinault (2008: 476–481, 489, 499), Pronk (2015: 206–209), Jasanoff (2018, 2019), and Adams (2021: 269–271). 166 The PIE root is sometimes reconstructed with a labiovelar, but there is no secure evidence in support of this reconstruction. Höfler (2015: 232) favours the following PIE internal derivation: *h2eug- ‘to grow’ → *h2eug-es- ‘strength’ → *h2ug-s-ó- ‘having strength’ → *h2ug-s-on- ‘the strong one’.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

149

the Indo-European cognates just mentioned cannot continue a nom.sg. *uksēn either. Indeed, Tocharian, (West) Germanic (OHG ohso, OE oxa < *uhsan- < *-on-), and Celtic (MW ych, OBret. ohen < *uxsō) offer evidence for the reconstruction of a nom.sg. with o-vocalism in the suffix. This comparative evidence has led some scholars to reconstruct nom.sg. *-ō(n) for an older stage of Proto-Indo-European, by arguing that Celtic, Germanic, and Tocharian preserved the original form (Szemerényi 1989: 154; Peters 1993: 394–395; Höfler 2015: 231–232; Jasanoff 2018). The paradigm of TB okso presents additional problems. Indeed, the fate of PIE *-ō(n) in word-final position remains a debated issue in the historical phonology of Tocharian. Scholarly opinion can be divided into two trends of thoughts: Kortlandt (1988a: 84), Ringe (1996: 89–90), Pinault (2008: 421–422), and Kim (2018a: 101–102) have supported *-ō > *-ǝw > TB -u, while Hilmarsson (1988a), Fellner (2014c: 63), and Jasanoff (2018) have argued *-ō > *-o > TB -o.167 The supporters of the first hypothesis are certainly aware of the case of TB okso and they also agree that some of the members of the okso-type are from PIE *on-stems. As a consequence, Pinault (2008: 421–423, 2017b: 144–145) and Hajnal (2005: 228 fn. 27) claim that nom.sg. -o is the outcome of a secondary *-ōn, originated from the contraction of the inherited *-ō(n) with the so-called Hoffmann suffix PIE *-oHōn (cf. OAv. mąθrān- ‘knowing the mąθras’ < *mantra-Han-). The outcome of this conglomerate suffix would have been a Proto-Tocharian vowel with o-timbre, which yielded TB -o, ta -∅. However, Jasanoff (2018) points out that there is no evidence that the Hoffmann suffix was productive in Tocharian, nor that Proto-Tocharian developed a Germaniclike morphological distinction between weak and strong inflection. Two solutions can be put forward: (1) either the nom.sg. *-ō restored the final nasal after all other n-forms of the paradigm (Ringe 1996: 10–11), (2) or 167 Clear examples of Tocharian outcomes of PIE *-ō are (Ringe 1996: 89–90; Pinault 2008: 421–422; Kim 2018a: 101–102): (1) PIE *h3eḱtō ‘eight’ > TB okt, ta okät; (2) PIE -oH (1sg. thematic ending) > TB -u (1sg.subj.); (3) *ḱu̯ ṓ ‘dog’ > TAB ku; (4) PIE *du̯ óh1 ‘two’ > PT *wu > TA wu; (5) *-u̯ ṓs (prt.part.) > TAB -u; (6) *h2ent-bhoh1 > TA āmpuk ‘both’ (if final -uk is not analogical after TA puk; Kim 2018a: 85–86). In (3)-(4)-(5), PT *-ǝw may have resulted through affection by *-u̯ -, but for all others the situation is more complex. The reduction of PT *-u > *-ǝ in ‘eight’ may be analogical after *ṣǝptǝ ‘seven’ (Kim 2018a: 101). Jasanoff (2018) has recently questioned the sound law *-ō > PT *-u. However, I do not see any reasons for his reconstruction of a PIE dual *-ōu̯ for (1)-(4)-(6) (see also Hilmarsson 1989a: 9–10), and there are no parallels of a hypothetical long diphthong *-ōu̯ yielding TAB -u (on TB akrūna, ta ākrunt ‘tears’, see § 3.6.3.2). Furthermore, I see no reason either for his claim that the participles in TAB -u are the outcome of a re-formed neuter *-uu̯ us (see Peyrot 2010b: 79), or that the 1sg.prs. TB -u is from a lenited form of PIE *-mi (but see Malzahn 2010: 28–30).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

150

Chapter 3

Tocharian has simply preserved the original nom.sg. *-ōn. A clear parallel for this development is found in Greek, e.g. Gk. κύων ‘dog’, gen.sg. κυνός, Gk. ἄρσην ‘male’, gen.sg. -ενος, where final -n may represent either a preservation or a restoration (Chantraine 1933: 158–159; Mayrhofer 1986: 159; Byrd 2015: 21). These solutions would also explain other cases of Tocharian nominal n-stems with nom.sg. -o (e.g. the adjectives of the klyomo-type).168 To sum up, the nom.sg. PIE *-ōn of the n-stem was either preserved or it was remade in *-ōn very early in the pre-history of Tocharian through analogical levelling after other n-forms of the paradigm that caused the replacement of the inherited nominative case. Furthermore, there is strong evidence that, at the same stage, Tocharian generalised the vocalism of the suffix from the nominative throughout the rest of the paradigm. Indeed, the fact that we do not have any traces of a suffix *-en- in this type is confirmed by the lack of palatalisation. And yet, there are no traces of *-on-, either. If we, for instance, consider the nominative plural, it is expected to have evolved PIE *-on-es > PT *-eñə > TB **-eñ. As a consequence, we have to assume that the suffix was *-ōnin all case forms, and that it regularly yielded *-an- in all the non-nominative singular cases.169 This kind of generalisation is a common trend in Tocharian 168 Jasanoff (2018) rejects both solutions, since he believes that Tocharian shortened long vowels before final nasals. I cannot agree with this shortening, since all forms he adduces can be explained differently (e.g. the obl.sg. TB -a in the kantwo-type, on which see § 3.5.1.2). Pace Malzahn (2011: 94–95), there is no clear evidence that PIE *-ōn yields PT *-āy (see the main text above), nor that the PIE ending *-ō was enlarged in pre-Proto-Tocharian by *-s. 169 There is still some hesitation in the Tocharian development of PIE *-ō- in non-final position. The communis opinio is that *-ō- gives PT *-a-. However, the oft-cited PIE *dhoHneh2 ‘grain’ > *dhōnā > TB tāno is better explained as a loanword from Iranian (see Peyrot 2018a: 257–259 and further § 3.5.1.2). TB krāmär ‘weight, heaviness’ (cf. TA krāmärts, tb kramartse ‘heavy’) need not be the outcome of PIE *gwróh2-mr (as per e.g. Ringe 1996: 8; Pinault 2008: 424), since internal -ā- /-á-/ can regularly reflect a vocalised laryngeal (Hilmarsson 1996: 174–175; dtb 230–231). Also, TB āntse, ta es ‘shoulder’ is usually compared with Gk. ὦμος ‘id.’ (cf. e.g. Ringe 1996: 7; Pinault 2017c: 135), as both reflecting a lengthened grade PIE *ōmso-. However, other hypotheses have been formulated to account for the vocalism of both Tocharian and Greek (cf. gew 2.1148 and further Hilmarsson 1989a: 127–128; Hackstein 2002: 190–191; Kim 2018a: 81 fn. 205; the reconstruction of PIE *h2ems-is based on Tocharian, cf. Kloekhorst 2008: 178). There are, however, other examples that may prove an evolution PIE *-ō- > PT *-a-. The collective suffix PIE *-ōr always surfaces as -ar in both Tocharian languages, as in TA ytār, tb ytār-ye ‘road’ < PIE *h1itōr (cf. § 3.6.3.4), TA ymār ‘quickly’ < PIE *h1imōr (Van Windekens 1976: 592), PT *ẃǝsar- ‘spring’ (cf. TB ysāre ‘wheat’, ta wsār ‘grain’) < *u̯ esōr (Peyrot 2018a: 251–252; Pinault 2017c: 131). Theoretically, TB aknātsa, ta āknats ‘foolish’ may reflect either PIE *n̥ -ǵneh3-(n)t- or a zero grade *n̥ -ǵnh3-(n)ti- (see recently Friis 2021: 18–20). However, cognate formations from other Indo-European languages point to the former form (cf. Gk. ἄγνωτος ‘unknown’, Lat. ignōtus ‘ignorant; unknown’, Ved. ájñāta- ‘unknown’ and further Gk. ἀγνώς ‘unknown; ignorant’;

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

151

nominal morphology (Fellner 2014c: 63). We can draw a comparison with the situation we find in Latin and Greek, where on-stem nouns sometimes generalised the allomorph with -ō- throughout the paradigm (e.g. sermō, -ōnis ‘speech, talk’, praedō, -ōnis ‘brigand, pirate’, ἀγών, -ῶνος ‘gathering’, αἰών, -ῶνος ‘lifetime, generation’), while other kinds of nasal stems partially retained vestiges of the original ablauting paradigm, particularly in Greek (e.g. homō, -inis ‘person’, carō, -nis ‘meat’, δαίμων, -ονος ‘deity, spirit’, κύων, -νος ‘dog’). On the other hand, the spread of the ai-stem for the expected *a-stem should be interpreted as secondary. We will deal with this secondary replacement in the following paragraphs, where an overview of the previous interpretations will also be given. The Tocharian A equivalent of nom.sg. TB okso is reconstructed as opäs*, based on the hapax legomenon nom.pl. opsi in YQI.4 a4. This form has been analysed and discussed in-depth by Pinault (1999: 467–468, 2008: 457–458). He argues that TA opäs* attests a phonological development within Tocharian A, according to which the consonant cluster PT *-ks- developed into TA -ps- (cf. TB klayksa- vs. TA klāypsā- ~ klepsā- ‘to dry up’ < PT *klayksa-; TB ekṣalye vs. TA opṣäly ‘festive day; celebration’, see Pinault 2015e). What is unexpected is the nom.pl. TA -i, which cannot match the nom.pl. TB -aiñ. It follows that TB okso and TA opäs* synchronically belong to different inflectional classes. However, there is strong evidence to support that the nom.pl. -i of TA opsi is secondary. Leaving aside the Indo-European comparative evidence, nom.pl. TA -i (= TB -i) is usually accompanied by the palatalisation of the stem-final consonant (e.g. TA mañi, tb meñi ‘moons, months’). As a consequence, TA **opṣi would have been expected (Pinault 2008: 498). Following Pinault (1999: 468), it is possible to assume that the non-palatalising nom.pl. -i in the hapax legomenon TA opsi was introduced after TA kowi ‘cows’, which is attested just before opsi in YQI.4 a4–5 (Maitreyasamiti-Nāṭaka): (klankañ wā)mpuṣ yetwentuyo kowi opsi kayurṣāñ: ma(hirṣāñ), “(vehicles) adorned with decorations, cows, oxen, bulls, bu(ffaloes)” (cf. Ji et al. 1998: 37). To conclude, on the basis of this clear PIE etymology there is good reason to hypothesise that other nouns of the okso-type derive from ōn-stems, too. TB skiyo ‘shadow’ (dtb 773), which has no match in Tocharian A, is a noun usually considered to be the outcome of an *eh2-stem. This substantive has been the subject of several investigations from both Tocharian and the Indo-European comparative perspectives. Many problems are involved. TB skiyo has possible cognates in most Indo-European languages, including but Goth. unkunþs ‘unknown’ < *n̥ -ǵnh3-t-; see Pinault 2012a: 187–188 and Hackstein 2012: 156–157). Therefore, I still assume with PIE *-ō- > PT *-a- (cf. Pinault 2017c: 144).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

152

Chapter 3

Gk. σκιᾱ́ ‘shadow’; Ved. chāyā́- ‘shadow, reflection’, YAv. a-saiia- ‘throwing no shadow’ (de Vaan 2003: 120; Lubotsky 2001: 35), MP sāyag (Phl. ⟨sʾdk’⟩, ManMP ⟨sʾyg⟩) ‘shade’ < *sāya-ka-, ManSogd. syʾk, Christian Sogd. syʾq; Latv. seja ‘face’, OCS sěnь ‘shadow’; Alb. hije ‘id.’ < OAlb. hē (Demiraj 1997: 201; Matzinger 2006: 96). Despite these cognate forms, the precise identification of the PIE root and the type of suffixation involved are debated. Beekes (2010: 1350–1351) reconstructs an original ablauting formation ́ *skéh2-ih2, *skh2-i̯éh2-, i.e. a PIE root *skeh2- followed by the so-called devī-suffix (cf. also gew 2.731; the reconstruction of the root without a palatovelar follows Lubotsky 2001). Accordingly, Indo-Iranian and Slavic would have generalised the full grade of both the root and the suffix, while Greek would have generalised the allomorph of the weak cases (Lubotsky 2001: 35).170 However, there is evidence that the i-element found in all Indo-European descendants of this noun was part of the root (as also per Rasmussen 1989: 33; Ringe 1996: 18–19; Lubotsky 2001: 35) and that the laryngeal was *-h2-.171 Accordingly, Rasmussen (1989: 61) and Ringe (1996: 18–19) reconstruct *skéh2i-h2, *skh2i̯-éh2- (cf. also Lubotsky 2001: 35). In any case, the various Indo-European forms continue different apophonic grades from this paradigm. The full grade of both the root and the suffix was apparently maintained in Indo-Aryan, where the noun has become an ā-stem.172 Other languages, including Greek, selected the zero grade of the root and the full grade of the suffix. This probably evolved as follows *skHi̯eh2 > *ski̯eh2 (Pinault’s law) > *skii̯ā (Sievers’ Law).173 Returning to Tocharian, TB skiyo should be traced back to PT *skǝy(y)o. The final vowel PT *-o is the regular outcome of *-eh2. However, it is still not 170 Mayrhofer (EWAia 1.559) reconstructs the laryngeal as PIE *h1. The reason behind this reconstruction is the connection with the Slavic forms. Indeed, the vowel -ě- in OCS sěnь cannot be from PIE *-eh2- > *-ā-, though it could be from *-eh1- > *-ē-. However, the Slavic vowel may also be the outcome of a diphthong. 171 Outcomes of the verbal root PIE *skeh2i- ‘to shine’ may be traced in Goth. skeinan ‘id.’, Croat. sînēm < *skiH-n-, and OCS sijati (liv 2 546; see further Derksen 2008: 450–451; Kroonen 2013: 246–247). On the other hand, OCS sěnь ‘shadow’ could be the direct descendant of PIE *skeh2i- followed by an n-suffix, thus PIE *skeh2i-n- > OCS sěnь (with regular monophthongisation of *-ai̯- > -ě-; cf. also Derksen 2008: 447, 2014: 549). In addition, if Lat. scaevus ‘left, inauspicious’ and Gk. σκαιός ‘id.’ are independent derivatives from this root (de Vaan 2008: 541), they both presuppose a proto-form *skai-u̯ os (cf. *gwei̯h3- ‘to live’ → *gwih3-u̯ os ‘living, alive’ > Gk. ζῷον, Lat. vīvus, Lith. gývas, Latv. dzîvs, etc.). If this analysis is correct, then the suffix cannot have been *-ih2/-i̯eh2. According to Yakubovich (2002) and Hitch (2017: 518–519), Sogd. sy- ‘to seem, appear’ and Khot. se- ‘id.’ are from *skāi- < PIE *skeh2i̯-. 172 Neri (2003: 332) reconstructs *skh2oi̯-áh2 as the antecedent of the Vedic form. 173 In my opinion, it is not possible to reconstruct laryngeal metathesis here (*h2i > *ih2), since metathesis usually occurs between stops.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

153

clear why palatalisation of the stem did not take place. In fact, this is a debated topic of Tocharian historical phonology since the precise contexts where PIE *i failed to palatalise are still unclear. As a matter of fact, we have no other clear evidence of a PIE sequence *-K/Ḱi- continued in Tocharian, so it is difficult to verify whether the lack of palatalisation in TB skiyo is regular.174 On the other hand, we know that PIE *e also palatalised those consonants that PIE *i failed to palatalise, e.g. *kw and *u̯ (cf. *du̯ ito- ‘second’ > TB wate, ta wät and *kwi- > TB kuse, ta kus ‘who’ vs. *u̯ éghno- ‘cart’ > TB yakne, ta wkäṃ and *kwetu̯ ores ‘four’ > TB śtwer, ta śtwar).175 Be that as it may, the nom. sg. TB skiyo may show the outcome of *-eh2 > *-ā > PT *-o > TB -o. We can conclude that other nouns that synchronically belong to the okso-type were originally *(e)h2-formations in the proto-language, including some abstract nouns mentioned above.176 3.5.2.2 The arṣāklo-type As mentioned above, the arṣāklo-type differs from the okso-type in the formation of the genitive singular, the plural, and the derived forms, which are not built on a stem in -ai-, but on a stem in -a- (e.g. arṣāklatstse* ‘snake-infested’). This inflectional type includes both masculine and feminine nouns. As in the okso-type, we also find several words relating to flora and fauna in the arṣāklo-type, like TB oṅkolmo, ta oṅkaläm ‘elephant’, TB mewiyo ‘tiger’, TB 174 Normier (1980: 256) and Pinault (2008: 423) suggest that PIE *i does not palatalise bilabials, velars, labiovelars, and *s, while Burlak (2000: 122–123) and Hackstein (2017: 1312) claims that PIE *i does not cause palatalisation at all. Van Windekens (1976: 88–89) lists alleged examples of *k > ś before *i, but they are all uncertain. 175 Bateman (2011) argues that typologically velars are among the consonants most easily palatalised. Accordingly, Ringe (1996: 18–19) claims that palatalisation of the PIE velars in front of *i must have happened in Tocharian and thus that it should also have affected TB skiyo. He argues that the laryngeal in *sḱHi- was not lost and that it must have survived as a sort of non-front vowel until after palatalisation had run its course. The fragility of this reconstruction is acknowledged by the scholar himself. Admitting that *i palatalised, one may claim that Tocharian inherited the Indo-European paradigm of the word for ‘shadow’ still intact. This led to an opposition between non-palatalised *sḱeh2i- > PT *skai- and palatalised *sḱii̯- > PT *ṣəy- (or the like). This aberrant alternation in the paradigm might have been normalised: the resulting form would show the consonantal skeleton of the former, but the vocalism of the latter. But this solution is extremely questionable. A last possibility is to dismiss the etymological link of TB skiyo with Gk. σκιᾱ́ etc., and to rather support a derivation from the PIE root *skeu̯ H- ‘to cover’, with possible continuants in Germanic (cf. OHG scuwo ‘shadow’) and Latin (cf. Lat. obscūrus ‘dark’). 176 It could be claimed that some abstract nouns belonging to this inflectional type represent the outcome of the PIE τομή-type (Pinault 2011a: 174), cf. TB prosko ‘fear’ < PT *prosko (umlaut) < pre-PT *presko (PIE *perk- ‘to fear’ (?), cf. TB prǝska-, ta präskā- ‘to be afraid’, TA praskañi ‘fearful’, dtb 402; Hilmarsson 1987; iew 820).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

154

Chapter 3

kercapo ‘donkey’, TB arṣāklo, ta ārṣal ‘snake’, TB tanākko ‘grain seed; little mark’,177 TB tvāṅkaro ‘ginger’, possibly TB yerkwantalo* ‘leopard (?)’178 and śakāto* ‘stick, club’,179 as well as body part nouns, like TB pratsāko, ta pratsak ‘breast’, TB wcuko /w(ə)cǝ́wko/ ‘cheek, jaw’ (later wicuko), TB ckācko* /cǝkácko/ ‘leg’, TB du. tskertane ‘(two) calves’, TB du. samākane ‘(two) eyelids’ (Kim 2018a: 45). We also find several loanwords that may come from either Iranian or substratic BMAC language(s). Examples include: (1) TB witsako ‘root’, borrowed from an Iranian form related to Oss. Iron widag, Digor wedagæ, jedagæ (cf. Av. vaēti- ‘willow’, Winter 1971: 222; Tremblay 2005: 426); (2) TB mewiyo ‘tiger’ (= Skt. vyāgra-), probably to be linked with LKhot. muyi ~ mauya < OKhot. *mūya- < *mauya- (cf. ManSogd. myw ‘id.’);180 (3) tvāṅkaro ‘ginger’, loanword from OKhot. *tvā̆ṃggara- or *tvā̆ṃggaraa- ‘id.’ (see Bailey 1937: 913 and Dragoni 2021: 305–307). The inclusion of TB ampoño ‘rottenness, infection’ in the arṣāklo-type is less certain, as the nom.sg. form is only attested in the late text B510 b6. This ampoño might be from older ampoña* (Dragoni 2021: 307–308). In any case, it may be linked with a Middle Iranian form related to Khot. haṃbūta- ‘rotted, festering’. A more complex word is TB kercapo ‘donkey’. In past decades this noun was considered a loanword from the ancestor of Skt. gardabhá- ‘donkey’ < *gord(h)ebho- (Pisani 1942–1943: 25; Van Windekens 1976: 214; dtb 210). It has been assumed that the borrowing happened in an early Indo-Iranian period, taking place before the merger of non-high vowels in Indo-Iranian (Carling 2005: 54). However, this scenario is to be rejected, not only for chronological issues, but also because a hypothetical *gordebho- is expected to yield PT *kerts’əpe- > TB **kerśape or **kertsape, as Winter (1971: 222) and Pinault (2008: 393–394) have demonstrated. Even if this form were at a certain point transferred to the arṣāklo-type because of its meaning, there is no way to explain the unexpected outcome of *d. In his forthcoming dissertation, Bernard

177 See Chapter 3, fn. 117. 178 On TB yerkwantalo*, see recently Adams & Blažek (2021). 179 Cf. THT1285 + THT3054 a2–3 enmelyaṣṣana śakāta(ntsa) “with enmelya-sticks” and THT1507 + THT1680 + THT2981 a4 /// me śarāk śānmyarne auntsantene= ym(a)ṣṣi lalaṃṣkai kektseñ khaurcurṣṣana śakātantsa räsk(are) /// “they tied up (his) undergarment (and) they started to sharply (flog?) his tender body with branches of wild date palm” (edition and translation based on Ogihara 2012a: 155–162). 180 See Blažek & Schwartz (2017: 58–59) with references. However, a common onomatopoeic origin cannot be dismissed.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

155

investigates an original proposal by Anreiter (1982: 23–24) and convincingly argues that both TB kercapo and Skt. gardabhá- are loanwords from a BMAC etymon *gardepa-, with inner-Tocharian palatalisation. 3.5.2.3 On the Origin of the Inflection The diachronic evolution of the okso- and the arṣāklo-types has been one of the most debated topics within Tocharian nominal morphology. The most important and/or recent discussions are Adams (1980), Hilmarsson (1987a, 1989: 82–83), Winter (1989), Hajnal (2005), Kim (2007b, 2018: 67–68), Pinault (2008: 483–485), Peyrot (2012a), Hartmann (2013: 413–424), and Jasanoff (2018). Each one of these scholars has taken a step towards a clearer understating of the development of these inflectional classes. The pivotal question of this section is how the *(e)h2-type and the *ōn-type evolved into these Tocharian inflectional types, merging their inflection in Proto-Tocharian. This central question leads to a number of sub-issues: (1) the reconstruction of the Proto-Tocharian paradigm(s); (2) the origin of the contrast between ai- and the a-stems in Tocharian B and their historical relation with the ā-stems of Tocharian A; (3) the origin of the obl.sg. TB -ai. In this section I will address all these issues. Although the problems are clear, they are not easy to solve. Indeed, the data involved are difficult to analyse from a diachronic perspective since they require the reconstruction of intermediate and non-attested stages. It follows that my historical account of these inflectional types must be taken as an attempt to trace their evolution: my final results are admittedly not entirely new, nor fully conclusive. However, I hope they will be an impulse for further investigations into this important topic of Tocharian nominal morphology. The structure of the rest of the section is diachronically oriented. I first deal with the reconstruction of the Proto-Tocharian paradigm of the okso- and arṣāklo-types. I then deal with its evolution from Proto-Tocharian to Tocharian A and Tocharian B. I conclude by recapitulating my findings. 3.5.2.4 Reconstruction of the Proto-Tocharian Paradigm It is assumed that the okso-type and the arṣāklo-type must have descended from a common proto-type (cf. e.g. Winter 1989: 111–115; Hilmarsson 1989a: 82–83; Pinault 2008: 484–485; Kim 2018a: 67–68). This is certainly correct and substantiated by synchronic and diachronic evidence. Firstly, we have already seen that the difference between the two types is that the members of the okso-type are disyllabic, while the members of the arṣāklo-type are mostly trisyllabic, so that an accent-conditioned change may have caused the split of the common proto-type.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

156

Chapter 3

Secondly, apart from the highlighted similarities in their inflection, the members of both okso- and arṣāklo-types have many semantic features in common: names of animals and plants, terms for body parts, and abstract nouns are typical of both classes. The great majority of nouns belonging to both classes are feminine (Hartmann 2013: 80). Thirdly, from a derivational point of view we find e.g. derivatives in -nto in both types. Compare the following examples: disyllabic naunto* ‘road’ (obl.pl. nauntaiṃ) and pānto ‘support, companion’ (obl.pl. pantaiṃ) vs. trisyllabic auñento ‘start, beginning’ (obl.pl. auñentaṃ*, cf. TA oñant) and yerkwanto* ‘wheel’ (obl.pl. yerkwantaṃ, cf. TA wärkänt). Fourthly, loanwords from both pre-Khotanese and Old Khotanese can be found in both types, but they are adapted to one or the other type depending on the number of syllables in the borrowed items (Dragoni 2022: 243, 259–260). Therefore, there are good reasons for claiming that the okso-type and the arṣāklo-type descend from a common proto-type. However, we need to understand how this common proto-type was inflected and if its split should be reconstructed for pre-Tocharian B or for Proto-Tocharian. In order to answer this question we need to closely compare the Tocharian B data with that of Tocharian A. As Peyrot (2012a) points out, the formal differences between the Tocharian B okso-, arṣāklo-, and kantwo-types do not exist in Tocharian A. The great majority of Tocharian A nouns matching these Tocharian B inflectional types have an unmarked singular paradigm and nom.pl. -āñ, obl.pl. -ās. Some examples are as follows: TB pyāpyo vs. TA pyāpi (nom.pl. pyāpyāñ in e.g. A25 b1, A91 a4, pyāppyāñ in e.g. A68 a2, A320 b6; obl.pl. pyāpyās in e.g. A108 a1, pyāppyās in e.g. A312 b1); TB kolmo* vs. TA koläm; TB ścono vs. TA śoṃ*;181 TB arṣāklo vs. TA ārṣal (obl.pl. ārṣlās in e.g. A1 b3); TB oṅkolmo vs. TA oṅkaläm (nom.pl. oṅkälmāñ in e.g. A22 b6; obl.pl. oṅkälmās in e.g. A395 b3; cf. the derived adj. oṅkälmāṣi in A403 b6); TB yerkwanto* vs. TA wärkänt (obl.pl. wärkäntā(s)/// in e.g. A152 b1); TB kantwo vs. TA käntu (instr.pl. käntwāsyo in A356 b2); TB karyo* vs. TA kri (nom.pl. käryāñ in A115 a4, obl.pl. käryās (?) in THT2424 b2); TB kātso vs. TA kāts (cf. derived adj. kātsaṣi* in e.g. A68 a5). 181 See Georges-Jean Pinault apud Ogihara (2012b: 172).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

157

On the other hand, there are three cases where a Tocharian B noun with stem in -ai- corresponds to a Tocharian A noun with stem in -e-. These are: TB pokai (obl.) ‘arm’ : TA poke (du. pokeṃ, obl.pl. pokes), TB yoko (~ -iye) ‘thirst’ : TA yoke, and TB swāñco (~ -iye) ‘ray of light’ : TA swāñceṃ. On the basis of these word-equations it is sometimes assumed that *-ay already served as an oblique in Proto-Tocharian, since TA -e is the regular outcome of PT *-ay.182 However, Peyrot (2012a: 211–213) has claimed that none of these equations is probative. TA poke can be compared with other body parts nouns that also have an e-stem in Tocharian A, including pe ‘foot’, du. kanweṃ ‘knees’, du. śanweṃ ‘jaws’, lymeṃ ‘lips’, du. āleṃ ‘palms (of the hand) etc.183 It may be suggested that the e-stem in Tocharian A body part nouns actually represents an original dual form and that the rest of the inflection was created on the basis of the dual (on which see Kim 2018a: 76). As far as Tocharian A is concerned, the dual marker -e(ṃ) mostly continues PIE *-oih1 (Del Tomba 2020b: 26–28). TA yoke is compared by Peyrot with other abstract and action nouns ending in -e, while Jasanoff (2008) reconstructs an oi̯-stem *ēgwh-oi-.184 TA swāñceṃ cannot be the exact morphological match of TB swāñco ~ -iye, since the Tocharian A noun seems to be a derivative from the Proto-Tocharian ancestor of swāñco (cf. the nasal enlargement).185 Furthermore, a small class of Tocharian B nouns inflects exactly as the okso-type with the only exception of having a nom.sg. in -iye. As pointed out above (§ 3.5.2.1), in the history of Tocharian B some nouns of the okso-type were developing a parallel nom.sg. in -iye (Hilmarsson 1987a: 44–45; Peyrot 2008: 102–106). However, it seems that a class with nom.sg. -iye, obl.sg. -ai already existed in Proto-Tocharian, the so-called ymiye-type (Peyrot 2012a: 188). Only 182 Kim (2018a: 67–68) reconstructs both the okso-type and the arṣāklo-types as *ay-stems in Proto-Tocharian, with a subsequent reduction of posttonic *ay > *a in Tocharian B. The same reduction would have also occurred in Tocharian B adjectives with pl.f. -yana, which, according to Kim, goes back to PT *-yayna. However, the reconstruction of an *ay-stem for the Proto-Tocharian paradigm of the feminine adjectives is questionable (see § 4.7.1.4). On the alleged reduction of *ay > *a in pre-Tocharian B, see the main text below. 183 Winter argues that we must posit *pokiye and not *poko as the nom.sg. of obl.sg. TB pokai. If so, this word would have been a member of the ymiye-type and TA poke would regularly match TB *pokiye. Cf. also the irregular paradigm of TB paiyye ‘foot’ (nom.obl.sg. paiyye, du. pai-ne, nom.pl. pai-ñ, obl.pl. pai-ṃ), which is matched by TA pe ‘id.’ (du. pe-ṃ, nom.pl. pe-ñ*, obl.pl. pe-s, see Kim 2018a: 80 with references). 184 Note that the exact morphological formation of this word is unclear (Pinault 2008: 433; dtb 552–553). 185 One may wonder whether PT *swañca- (obl.) was resuffixed in *swañca-ññV with the following development in Tocharian A: *swañca-ññV > *swañcaiññV > *swañceñǝ > TA swāñceṃ. Cf. also Hilmarsson (1987c: 62–65).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

158

Chapter 3

five nouns can be considered to belong to this class: TB oskiye ‘habitation’, TB kaumiye ‘pond’, TB ymiye ‘path; station of the life’, TB ṣpakiye ‘pill, poultice’, and TB säly(i)ye ‘line’. The corresponding Tocharian A nouns usually end in -e in the singular: TA yme : TB ymiye; TA oṣke : TB oṣkiye. This correspondence is parallel to the type TB kälymiye : TA kälyme (with TB nom.sg. -iye, obl.sg. -i, nom.pl. -iñ, obl.pl. -iṃ; TA nom.obl.sg. -e, nom.pl. -eñ, obl.pl. -eṃ), where TB -iye /-ǝye/ phonologically corresponds to TA -e (Peyrot 2012a: 209, 2021: 458–459). The historical analysis of the ymiye-type is debated because its members are not attested in the archaic period of Tocharian B (with the exception of ymiye), and their paradigm seems to be nothing but a hybrid combination of the okso-type and the kälymiye-type. Peyrot (2012a) reconsidered his previous opinion on the secondary development of the ymiye-type (Peyrot 2008: 105–106) because the existence of this inflectional class in Proto-Tocharian is necessary to explain the acquisition of the ending -iye by the okso-type. To sum up, the Tocharian B okso-type (ai-stems) and arṣāklo-type (a-stems) are mostly matched in Tocharian A by an inflectional class with a zero ending in the singular and plural nom. -āñ, obl. -ās.186 Two scenarios can therefore be outlined: (1) Proto-Tocharian had both *ay- and *a-stems and Tocharian B preserves this situation unaltered; (2) Proto-Tocharian had only *a-stems and Tocharian B developed the ai-stems later. Three pieces of evidence speak in favour of the second hypothesis. Firstly, in the Tocharian A counterpart of Tocharian B okso- and arṣāklo-types we do not find any systematic counterpart of TB -ai. This may imply that there was no okso-like class in Proto-Tocharian, where *-ay did not serve as an oblique yet (see above). Indeed, Tocharian B nouns belonging to the okso-type that are matched in Tocharian A by e-stems display both nom.sg. -o and -iye or exclusively -iye. Secondly, in the feminine inflection of the adjectives, the obl.sg.f. TB -ai consistently matches with the gen.sg.f. TA -e, and not with the obl.sg.f. -āṃ; as we will see, the latter ending should be reconstructed as a pre-Tocharian A innovation, since certain adjectival classes point to the reconstruction of an unmarked singular ending *-a for both the nominative and the oblique of the Proto-Tocharian feminine paradigm (see § 4.7.1.4 and § 4.8.2.3). Thirdly, some other inflectional types that have TB -ai as the oblique singular are not matched by -e in any case form of Tocharian A (cf. e.g. 186 The Tocharian B ymiye- and kälymiye-types correspond to a Tocharian A inflectional class with a singular ending TA -e and a plural paradigm nom. TA -eñ, obl. TA -es (Peyrot 2012a: 210–211). If the inflection of the ymiye-type is old, Tocharian A -e can correspond to both nom.sg. TB -iye and TB -ai(-).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

159

the wertsiya-type  §  3.5.3, and the inflection of the nomina agentis of the aknātsa-type). It follows that Proto-Tocharian may have had only one inflectional type, and that the origin of the obl.sg. -ai and the ai-stems (i.e. the okso-type) is most likely to be sought in a pre-Tocharian B period (see below). One can further speculate, however, that at the moment of the separation of Tocharian A and B from Proto-Tocharian, the common language was starting to generalise the element -ai in the okso-type. In this case, only Tocharian B would have continued this trend of development. What is clear is that in early Proto-Tocharian most of the nouns that synchronically belong to the okso- and arṣāklo-types were not ai-stems. Now, since most of the Tocharian A nouns matching Tocharian B oksoand arṣāklo-types have a plural paradigm with nom. -āñ, obl. -ās, we need to understand whether the vowel stem TA -ā- < PT *-a- is to be interpreted as an innovation or an archaism. As a matter of fact, nom. -āñ, obl. -ās is a common plural set of Tocharian A, so one might think that the spread to this type is secondary. Furthermore, once we have understood that the obl.sg. *-ay should be interpreted as a later innovation, the next task is to establish what the oblique singular in Proto-Tocharian was. I agree with Peyrot (2012a) that just two endings can be postulated: either PT *-a (cf. TA ā-stems) or PT *-o (cf. TB nom.sg. -o). The Tocharian A zero-marked singular is ambiguous since it could go back to both these Proto-Tocharian endings. However, only the first would have yielded -ā- in the plural. Therefore, it is more economical to assume that both okso- and arṣāklo-types were *a-stems in Proto-Tocharian and that Tocharian A has preserved the original state of affairs. It is generally assumed that the conflation between the Proto-Tocharian outcomes of the PIE *ōn- and the *(e)h2-stems was caused by the homophonous nominative singular PT *-o: The reason for the wide-spread merger of ā- and ōn-stems in Tocharian is the identical outcome of *-ā and *-ōn in the nom. sg. (Hilmarsson 1986a: 18). Older feminine *-ā-stems have joined this class [i.e. nasal stems] due to the coincidence of the nom. sg. B -o (Pinault 2017a: 1339). The identity of the nom. sg. forms […] was the basis for the amalgamation of the two types (Jasanoff 2018: 77).187 187 Jasanoff thinks that the coalescence was between the outcome of “amphikinetic n-stems” and “amphikinetic i-stems” (see below).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

160

Chapter 3

Although formal identity of the nom.sg. forms is certainly reconstructable (cf., for instance, TB kantwo ‘tongue’ < PIE *dn̥ ǵhu̯ eh2 and TB okso ‘ox’ < PIE *uksōn) and the nominative singular is cross-linguistically a pivotal grammatical case, additional homophonous case forms may be reconstructed in order to historically account for the merger of these stem types. Indeed, if I am right in reconstructing the common antecedent of okso- and arṣāklo-types as inflecting according to the kantwo-type, it can be hypothesised that from this type the obl.sg. *-a also spread to the Proto-Tocharian outcome of the PIE *eh2-stems and that the *a-reflexes in the common proto-type of both okso- and arṣāklo-types did not develop from *-eh2- by sound law. Moreover, the Leiden School (see Beekes 1985: 20–36, 2011: 199–201) reconstructs the Proto-Indo-European *(e)h2-stems with an original ablauting suffix *-éh2/-h2. If we were to accept that Tocharian inherited and generalised this hysterodynamic ablaut throughout the inflection of the “ā-stems”, then the merger with the *ōn-stems would be perfectly understandable. Accordingly, the nom.sg. *-o may be historically interpreted as the outcome of the full grade PIE *-éh2 and the obl. sg. *-a as the outcome of the zero grade PIE *-h2-. These endings would have therefore merged with nom.sg. *-o < *-ōn and with obl.sg. -a < *-an < *-ōnm̥ (by a semiregular haplology (?), cf. Jasanoff 2018). 3.5.2.5 From Proto-Tocharian to Archaic Tocharian B We have so far dealt with the reconstruction of the Proto-Tocharian paradigm of the okso- and arṣāklo-types. We now turn to the source of the obl.sg. -ai and the origin of the contrast between ai- and a-stems in Tocharian B. These two problems are related. Indeed, the connection between the nouns with ai-stems (i.e. the okso-type) and nouns with a-stems (i.e. the arṣāklo-type) warrants a more extensive discussion of the origin of the obl.sg. -ai, which has been controversial since the beginning of the study of Tocharian nominal morphology. Let us start with the proposal by Winter (1989), who was the first to identify the two inflectional types under discussion. He reconstructs a sound law PT *-an > TB -ai, according to which the nasal was vocalised in Tocharian B, at least in morpheme-final position.188 He attributes the difference between obl.pl. oksaiṃ and arṣāklaṃ to a change from *oksan# to *oksai#, with the restoration of the nasal in e.g. the obl.pl. *oksan > *oksai >> oksai-ṃ. This sound law is accepted by some scholars (e.g. Hajnal 2005). Hilmarsson (1989a: 82–83) notes that this development was conditioned by the accent as follows: accented *-án- became TB -ai-, while unaccented *-an- yielded TB -an. I see two problems with this hypothesis. First, 188 See Hartmann (2013: 415–417).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

161

the obl.sg. of the okso-type is never accent-final (cf. pyāpyai /pyápyay/; Peyrot 2012a: 184). Second, I cannot find any phonetic motivation for the change *n > *y.189 Another theory holds that TB -ai may derive directly from ProtoIndo-European and that the source of this ending should be sought in the PIE amphidynamic i-stems. Thus acc.sg. PIE *-oi̯-m̥ > obl.sg. PT *-ey > TB -ai (Van Windekens 1979: 16 and 177; Marggraf 1975; Čop 1975: 11; Adams 1980: 440). A recent contribution by Jasanoff (2018) has brought this theory back to the attention of scholars. He claims that the amalgamation of the PIE *ōn-stems with the PIE *oi-stems (with the generalisation of the allomorph *-ōi̯- throughout the paradigm) was caused by the homophony of their nominative singular, both reconstructed as yielding pre-PT *-ō > PT *-o. In my view, a problem with this theory is that the amphidynamic i-inflection (if continued in Tocharian) was too small a category to account for the spread of the obl.sg. TB -ai (Hilmarsson 1987a; Pinault 2008: 483). Conversely, Jasanoff claims that this merger was favoured by the “immense productivity of the amphikinetic i-declension in Tocharian”, but he does not clarify which nouns he is referring to. As Pinault (2008: 483) rightly objects, if Tocharian inherited this class, it could not account for the proliferation of the obl.sg. -ai. In addition, if *ōnand *ōi-stems really merged in pre-Proto-Tocharian based on the identity of their nominative singular, I would expect the new conflated paradigm to be based on the most productive stem-type, which in the case under discussion is not the *oi-stem. However, I do not reject a priori the possibility that the outcome of some *oi-stems come to be part of this Tocharian inflectional class. Indeed, the reconstruction of *oi-stems might account for those e-stem nouns in Tocharian A whose Tocharian B correspondents belong to the okso-type. What seems impossible to me is that the primary origin of the obl.sg. -ai could 189 This sound law might have a parallel in the alleged lenition of PT *-m(-) > TB -w(-)/-u(-) (cf. 1sg.prs.act. TB -u/w, -m vs. TA -m), but the conditions for such a development have not been stated satisfactorily and alternative derivations can be proposed (see Peyrot 2013a: 410–411). In order to explain the inflection of the outcome of n-stems in Tocharian, one could be tempted to reconstruct a sound law *-V́ nə(-) > *-V́ yə(-) for Proto-Tocharian, which might account for such forms as obl.sg. TB oksai < PT *oksáyə < pre-PT *-a/enə(n) < PIE *-ō n̆ -m̥ , obl.pl. TB oksaiṃ < PT *oksáyəns < pre-PT *-a/enəns < PIE *-ō n̆ -m̥ s or TB kalymi < PT *kəĺḿə́yə < pre-PT *-’ənə(n) < PIE *-en-m̥ , obl. pl. TB kälymiṃ < PT *kəĺḿə́yəns < pre-PT *-’ənəns < PIE *-en-m̥ s. If so, the okso- and arṣāklo-types would have become ay-stems in an early Proto-Tocharian phase. Then in a later but still Proto-Tocharian phase the arṣāklo-type would have monophthongised *-ay when the accent moved back (cf. Kim 2007 and Peyrot 2012a), producing the a-stem form of the plural. However, counterexamples can be found, and I was unable to find convincing parallels in articulatory phonetics for this sound change. See further below.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

162

Chapter 3

be reconstructed on the basis of such a small category, particularly for the feminine adjectival inflection (cf. § 4.7.1). The two last theories that we must address are those of Pinault (2008) and Peyrot (2012a). Their results are greatly at odds, since Pinault argues that the origin of TB -ai should be sought in the *ōn-stems, while Peyrot sees it in the *(e)h2-stems. Rather than favouring one or the other, I instead intend to show that both theories are correct since they offer complementary explanations of the origin and the spread of TB -ai. We start the discussion with Peyrot’s hypothesis.190 Peyrot’s theory proposes that the ending TB -ai is of Proto-Indo-European origin: it is the outcome of PIE loc.sg. *-(e)h2-i and dat.sg. *-(e)h2-ei, reanalysed as the oblique in a pre-Tocharian B stage. In past decades, Pedersen (1941: 43), Lane (1976: 145–146), Klingenschmitt (1975: 153, 1994: 319–320), and Kim (2009a: 84 fn. 29) have proposed or supported the same Proto-Indo-European origin, basing their analyses on the formal aspects of this derivation. However, Peyrot makes this derivation clearer and more precise through closer innerTocharian correspondences. Indeed, he claims that the gen.sg. TA -e and the obl.sg. TB -ai must be analysed as the outcome of the same PIE form, namely the dative-locative, and that this marker served as a genitive-dative in ProtoTocharian. This claim receives strong support from a close comparison between Tocharian A and B. Peyrot’s evidence can be summarised as follows: (1) some of the inflectional classes with obl.sg. TB -ai have the corresponding Tocharian A nouns with gen.sg. -e (e.g. obl.sg. TB aśiyai : gen.sg. TA aśśe, from TB aśiya, ta aśi ‘nun’); (2) Tocharian B lacks any gen.sg.f. form in the adjectival inflection, while Tocharian A consistently attests a gen.sg.f. -e (e.g. obl.sg.f. TB klyomñai : gen.sg.f. TA klyomine from TB klyomo, ta klyom ‘noble’); (3) several adverbs synchronically end in TB -ai, ta -e (e.g. TB amāskai ‘with difficulty’, TB anaiśai ‘carefully’, TA kātse ‘close’, TA pre ‘outside’ etc.). All these correspondences lead to the reconstruction of TB -(y)ai, TA -(y)e < PT *-(y)ay < PIE dat.sg. *-(i̯e)h2-ei̯ and/or loc.sg. *-(i̯e)h2-i (cf. also Kim 2018a: 94). The possible reanalysis of a locative singular marker as an oblique might seem unexpected, but it may be accounted for as a case of exaptation (Lass 1990; Van de Velde & Norde 2016). In particular, when Tocharian developed (or calqued) the secondary case markers 190 For yet another proposal, see Hackstein (2012: 161), who seems to equate the TB obl.voc.sg. -ai found in the formation in TB -eñca with the vocative of the type γύναι. He concludes that the homophony between the vocative and the oblique singular is nothing but the preservation of an older state of affairs.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

163

in a Proto-Tocharian stage, the local cases were the first ones to be created, in particular the locative (TB -ne, ta -aṃ), the allative (TB -sa, TA -ā), and the perlative (TB -śc, ta -ac).191 Therefore, the locative marker PT *-ay started to lose its function, which was replaced by the agglutinative suffix. Then, in addition to being used to derive adverbs, it was refunctionalised to mark the oblique singular. That is, it retained its status as a cumulative morpheme albeit acquiring a new function. I thus reconstruct the singular paradigm of the pre-Tocharian B ancestor of both the okso- and arṣāklo-types as follows: nom.sg. PT *-o > pre-TB *-o; acc.sg. PT *-a >> pre-TB *-ay. This *-ay may still have served as a genitive-dative in Proto-Tocharian. When Tocharian B reanalysed it as the oblique, the gen.sg. was marked with the nasal genitive *-nse, which was attached to the new obl.sg. *-ay, thus *-ay-nse > TB -aintse. Unfortunately, we cannot be certain of the genitive singular in Tocharian A, because ā-stems matching TB okso- and arṣāklo-types usually do not attest genitive singular forms. A direct correspondence between gen.sg. TA -e : obl.sg. TB -ai can be observed in gen.sg. TA oṅkälme : obl.sg. TB oṅkolmai from TA oṅkaläm, tb oṅkolmo ‘elephant’ (Pinault 2009b), which may be used for reconstructing a gen.sg. *-ay for Proto-Tocharian. However, the isolation of this genitive form requires caution.192 Although Peyrot’s analysis can explain the source of most of the obl.sg. -ai, it can hardly account for the origin and the spread of the Tocharian B ai-stems (i.e. the okso-type). We should therefore determine whether additional sources of TB -ai- can be identified. At this point, Winter and Hilmarsson’s theory on the difference between the okso-type and the arṣāklo-type becomes relevant again. We have already seen that they explain these two classes by means of different outcomes of PT *-an- conditioned by the accent (Marggraf 1975): considering the oblique plural, on the one hand, the substantives of the okso-type were disyllabic and stressed on the last syllable (e.g. oksáiṃ); the substantives of the arṣāklo-type, on the other hand, were trisyllabic and stressed on the penultimate syllable (e.g. arṣā́klaṃ).193 As a result, in the okso-type the accent would have caused the diphthongisation: the two inflectional types would have the same origin, but the arṣāklo-type would preserve an older state of affairs. 191 For a diachronic stratification of the Tocharian case paradigm, see Carling (1999, 2008). 192 A further comparable item may be TA kātse ‘near, close’, which could be related to TA kāts ‘belly, womb’, tb kātso ‘id.’, a member of the kantwo-type. TA kātse is traced back to PT *katsay by Pinault (1991: 186) and Hilmarsson (1996: 112). See further Peyrot (2012a: 207). The gen.sg. TA käntwis from käntu ‘tongue’ must be secondary. On the form kätwes (A300 b3), cf. Hilmarsson (1996: 114) and Malzahn (2010: 553). 193 Cf. Winter (1987: 305–306).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

164

Chapter 3

Recently, Kim (2007b: 19–20, 2018: 44–46, 67–68) and Peyrot (2012a: 184–185) have called this development into question, claiming that the correspondence is to be interpreted the other way around. They argue that both okso- and arṣāklo-types were originally *ay-stems in a prehistoric stage of Tocharian B. The diphthong would have been maintained in accented position (i.e. in the okso-type), but monophthongised in posttonic position (i.e. in the arṣāklo-type).194 That is to say, after the breakup of Proto-Tocharian, the *a-stems first became *ay-stems, continued as the TB okso-type, and then a selection of these new *ay-stems turned into *a-stems, becoming the TB arṣāklo-type. The sound law underlying these developments can be schematised as follows (Peyrot 2012a: 189): * -́ ayn > * -́ an. However, there is no strong evidence to testify this sound law except for the alleged reduction of -oyto -o- in TB impf. and opt. forms of the type tākoṃ ‘may they be’ < *tákoy-ǝn (Kim 2007b: 19–20 fn. 32; Peyrot 2008: 142–144). A general fact in favour of Kim and Peyrot’s hypothesis is that stressed syllables are better maintained cross-linguistically. Although this is certainly true, it does not mean that they cannot undergo modifications but simply that they are louder and less likely to be dropped. Indeed, diphthongisation of stressed vowels can be found in the historical development of many languages. A good example in this sense is found in the phonetic evolution from Latin to Romance languages, where ́ cases of diphthongisation of stressed vowels are frequent (e.g. Lat. pĕdem > ́ ́ It. piede; Lat. pŏrtum > Sp. puerto; Lat. nŏvum > OFr. nuef, etc.). Furthermore, neither Kim, nor Peyrot clarify how these alleged *ay-stems would have come about in pre-Tocharian B (or Proto-Tocharian); I do not think that the generalisation of the new obl.sg. -ai is a sufficient explanation. By applying Occam’s Razor, the classical interpretation as formulated by Winter provides a solution in the context of this intricate set of developments. With the reconstruction of okso- and arṣāklo-types as both coming from *a-stems, Pinault’s theory on the origin of TB -ai becomes relevant (2008: 484–485). Pinault argues that the source of TB -ai should be sought in the nominative plural of the *ōn-stems, PIE *-ōn-es > PT *-añə. He claims that in final syllables an accented sequence PT *-áñǝ# would have regularly become *-áyǝ#. This sound law can be more clearly discerned in two isolated forms (Carling 2003: 92–93): (1) nom.pl. TB śrāy ‘elders’, whose obl.pl. śrānäṃ /śránǝn/ clearly speaks in favour of the reconstruction of a Proto-Tocharian nominative plural *śrañǝ; (2) TB ylai-ñäkte ‘Indra’ < *ylañ-ñəkte < *ylan-ñəkte (cf. TA wlāṃ-ñkät). According to Pinault, in an unattested phase of Tocharian B, all nouns of the okso-type spontaneously developed a nominative plural *oksay, and from this proto-form the element *-ay was extracted and then generalised as the oblique 194 Cf. also Chapter 3, fn. 189 above. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

165

Gender in the Inflection of the Noun

singular of some other inflectional classes. Soon after, the expected nom.pl. *oksay was replaced by TB oksai-ñ on the model of arṣāklañ. In broad terms, I agree with the sound law suggested by Pinault, although my proposal differs in certain details. First, the diphthong that arose in the nominative plural can hardly be the source of the oblique singular (see the criticism by Peyrot 2012a: 191). Second, if nom.pl. *-áñǝ really evolved into *-áy, I would expect to find more direct evidence of this ending. Still, I believe Pinault’s sound law can be slightly modified as follows: PIE *-ōn-es > PT *-áñǝ# > pre-TB > *-áyñǝ# > TB -áiñ, i.e. in stressed syllables a palatal nasal transferred the palatalisation to the preceding vowel, which thus became a diphthong. From a phonetic perspective, this development can be accounted for as a case of vowel breaking caused by anticipated palatal pronunciation (assimilation) of a vowel before a palatal consonant. It follows that, if a noun of the okso-type had an obl.sg. pre-TB *-ay as the outcome of the gen.(-dat.)sg. PT *-ay and a nom.pl. pre-TB *-ayñǝ as the outcome of the sound law just discussed, it may have generalised *-ay- as the basic stem of all other cases and derivatives, which were likewise stressed on the last syllable.195 A schematic summary of the final development of okso- and arṣāklo-types is as follows:196 Table 24

Evolution of okso- and arṣāklo-types from pre-Tocharian B to Archaic Tocharian B

okso-type pre-TB nom. sg. obl. sg. nom. pl. obl. pl.

* ́-o * -́ ay *-áñǝ *-án

> * -́ o > * -́ ay > *-áyñǝ > *-án

arch. TB

arṣāklo-type

pre-TB

arch. TB

> ́-o > -́ ai > -áiñ(ǝ) >> -áin

nom. sg. obl. sg. nom. pl. obl. pl.

* -́ o * -́ ay * -́ añǝ * -́ an

> * -́ o > * -́ ai > * -́ añ(ǝ) > * -́ an

195 A further objection to Pinault’s sound law may be that the residual kantwo-type could have had a nom.pl. *-áñǝ in Proto-Tocharian. However, this inflectional type shows many differences with respect to the okso-type. From a diachronic point of view, there are, apparently, no historical n-stems continued in the kantwo-type and the nom.pl. marker -ñ may have been added at a later stage. From a synchronic point of view, the kantwo-type shows a clear contrast between stressed nom.pl. -āñ and unstressed obl.pl. -aṃ (vs. -áiṃ of the okso-type), and an obl.sg. -a (vs. -ai of the okso-type). As a consequence, even if PT *-áñǝ# became pre-TB *-áyñǝ# also in this type, then analogical levelling from the rest of the paradigm could have easily changed it back to *-áñǝ. 196 As concerns śray ‘elders’ (attested only in classical and late texts), I would suggest the following development: *śráñǝ > *śráyñǝ > (*śráyn >) śrā́y. There are two reasons for which this development could have taken place: on the one hand, TB śray seems to be an accented monosyllable and the apocope of the final nasal could have happened earlier; on the other hand, dissimilation of the two palatals ś … ñ could have taken place. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

166

Chapter 3

A further indication of the phonetic change *-áñ > *-áyñ may be warranted by the fact that, apart from the kantwo-type, final -āñ /-áñ(ǝ)/ is extremely rare in Tocharian B.197 Furthermore, this modified version of Pinault’s sound law PT *-áñǝ# > pre-TB *-áyñǝ# > TB -áiñ partially resembles similar processes of assimilation in late and colloquial texts. All these developments involve assimilation of the pronunciation of a palatal consonant over a preceding (or following) vowel. Examples are: (1) a and a /ǝ́ / > ai (Peyrot 2008: 54–55), e.g. ravaiññe (G-Su 2 a1) for rapaññe /rapǝ́ ññe/ ‘pertaining to the last month of the year’, läksaiñe* (e.g. IT206 b1) for läkṣaññe /lǝkṣə́ ññe/ ‘pertaining to a fish’, oṅkolmaiññe (W20 b3) for oṅkolmaññe /onkólmaññe/ ‘pertaining to an elephant’; (2) non-accented ä /ǝ/ > i /ǝy/ (Peyrot 2008: 55–57), e.g. astariññe (B586 4) for astaräññe /astǝ́ rǝññe/ ‘purity’, miñcuṣka (B400 a5) for mäñcuṣka /mǝñcuṣka/ ‘princess’, bram-ñikte (e.g. B107 a8–b1) for brahm-ñäkte ‘god Brahma’;198 (3) the isolated word āñme > āyme ‘soul, self’. Furthermore, a comparable phenomenon occurred in the prehistory of Tocharian B, where the Proto-Tocharian palatalised labials *ṕ and *ḿ transferred their palatalisation to a following *ǝ colouring it to i (e.g. TB pilko, ta pälk ‘thought’ < PT *ṕǝlko; TB mit ‘honey’ < PT *ḿǝtǝ < PIE *médhu-). All these developments occurred in different chronological stages of Tocharian B and they are never the same development. Still, they are all similar and may comprise a kind of drift. 3.5.2.6 From Archaic Tocharian B to Late Tocharian B In his book on variation and change in Tocharian B, Peyrot (2008: 78–84) deals with variant forms that attest final -ñ alternating with final -ṃ. He collected and commented on a large amount of data, which appears however quite 197 I found the following forms: (1) TB kāñ (IT9 b1; B45 a2; THT1375.c a5) is a word of unknown meaning and etymology (dtb 158); (2) TB luwāñ is attested only once in IT395 b2 /// mā luwāñ śau///. If not an error for luwāññe ‘pertaining to animal’ (as it seems not to be, since it is written with final ä and virāma) this luwāñ could either be a late nom.pl. of luwo ‘animal’ (regular nom.obl.pl. lwāsa), which has been analogically created after nouns of the kantwo-type (both with nom.sg. -o, obl.sg. -a), or a causal singular; (3) TB sāñ ‘skill’ is a loanword from Khotanese saña- ‘expedient, skill’, see Del Tomba & Maggi (2021); (4) the hapax legomenon nom.pl. läkle-lyakāñ ‘seeing suffering’ (AS7E a6 [class.]; Sieg 1938: 22), a verbal governing compound of the rita-type, is not written as one would expect (cf. Malzahn 2012b: 114 “Widersprüchlich ist der Befund bei der Form B läkle-lyakāñ […]”; cf. further yikne-ritañ; yolo-ritañ; see also Fellner 2018); (5) TB yäkt-āñ (B351 a5) is a sandhi-variant of yäkt-āñm ‘feeble, weak’. Other instances of final -āñ are of no value (cf. the verbal forms with 1sg. ñ-enclitic, e.g. krasāñ in B400 b1 or nautāñ in B591 a7; nervvāñä/// B591 a3 is to be restored as nervvāñä(ṣṣe), cf. Peyrot 2013a: 323). 198 There may be a few examples of -a- /ǝ́ / > -i- /ǝ́y/, e.g. in lykiśke (B192 b3, class.) for lykaśke /ĺkǝ́ śke/ ‘small, little, fine’ (cf. Kim 2018a: 53; Hilmarsson 1989a: 85; Pinault 2011a: 182 fn. 41). On TB ñaś ~ ñiś ‘I, me’, cf. Peyrot (2008: 56) and Malzahn (2017).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

167

Gender in the Inflection of the Noun

inconsistent. Indeed, some cases offer support for a -ñ > -ṃ development, while some others do not. He concludes that a sound law -ñ > -ṃ should be postulated in any case, at least for the late stage of Tocharian B. The main reason Peyrot deals with this problem of Tocharian B historical phonology is the attestation of variant forms in the nominative plural of class vi (TB -ñ). Krause and Thomas (teb § 185) were the first to notice these variants, but they were not able to determine whether -ñ > -ṃ was due to sound change or analogy. Attestations of a nom.pl. -ṃ in place of the expected -ñ can also be found in the classes with pl. -añ, -aiñ, which are also the most common classes with nom.pl. -ñ. Table 25

Variant forms of the nominative plural in nouns with ai-stems

tocharian b stage

nom.pl. forms

archaic

-aiñ ñ-plural: kuṣaiñä (B275 b1, verse); nauntaiñä (B275 a3, verse); klyotaiñä (AS9B b7, prose); pyā­ ppyaiñ (B275 a2, verse); ṣconaiñ (THT 1281 a5)a ñ-plural: kotaiñ (AS7H b4, verse); pokaiyñ(o) -aiñ (B214 b3, verse)b ñ-plural: kaumaiño (B45 b7, verse) -aiṃ n-plural: oksaiṃ (AS15B b3); koraiṃ (B577 b2); nauntai(ṃ) (AS17J b1), nauntaino (AS16.4.1 b5); parśaiṃ (AS16.1 b3, AS17G b6); pyapyaiṃ (AS8C b6; IT14 b2; AS6D a3); swañcaiṃ (IT107 a2, NS37 a1); ṣitaiṃ (IT1094 b2); tsaktsaiṃ (AS16.8 a5?)c ∅-plural: swañcai (THT1455 a3, frgm.)

archaic-classical classical

dominant nom. pl.

a Cf. also nauntainä in B394 b3 (arch.), where the context is unclear. I have not included in this list the hapax legomenon TB kompaino (B588 a1), of uncertain meaning (dtb 216; Thomas 1997: 100; Dragoni 2022: 114–115). Malzahn (2012a: 62) interprets it as a nom.pl. with o-mobile. If this interpretation is correct, then TB kampaino should be considered a nom.pl. in -ain from an archaic text (Hilmarsson 1996: 166). b Cf. also nom.pl. yolaiñ from yolo ‘bad; bad action’ in AS7G b5 (arch.-class.), B526.a b3 (class.), etc. c I have omitted ṣpakaiṃ ‘pills, suppositories’, which is found several times in construction with the gerundive pl.f. yamaṣṣällona from yam- ‘to do’, though it seems to be inflected as a nom.pl., i.e. ṣpakaiṃ yamaṣṣällona “pills are to be made” (cf. dtb 729–730).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

168 Table 25

Chapter 3 Variant forms of the nominative plural in nouns with ai-stems (cont.)

tocharian b stage

nom.pl. forms

dominant nom. pl.

late colloquial

— n-plural: (o)ksaiṃ (PK Bois B30 a1), ok(s)ai(ṃ ?) (PK Bois B19 a5), oks(aiṃ ?) (PK Bois B104 a3); korraiṃ (PK LC 11 b4?) ∅-plural : (oks)ai (PK Bois B139 a2), oksai (PK reserve B 3.2. a1); korai (PK Bois B19 a4)

-aiṃ / -ai (?) -aiṃ / -ai

Though complex, it is possible to discern a clear development in the okso-type. Indeed, we systematically find -aiñ in archaic texts, -aiṃ in classical texts, and -ai in colloquial texts. Outside of archaic and archaic-classical texts, the only form with -aiñ is TB swañcaiñ (B108 [late]), which is however used as an oblique plural and “can only be explained as a hypercorrect spelling after the merger of -ṃ and -ñ” (Peyrot 2008: 80). Table 26

Variant forms of the nominative plural in the arṣāklo-type

tocharian b stage

nom.pl. forms

dominant nom. pl.

archaic

ñ-plural: kercapañ (B118 b3); tvāṅkarañ (AS9A b7); mokośwañ (B118 b3) ñ-plural: oṅkolmāñ (NS30 b3) ñ-plural: mewīyañä (IT195 a6); yerkwantalañ (IT195 a6); tanaulykañäa (IT96 b2) n-plural: arṣāklaṃ (IT199 b2, damaged) — n-plural: kercca(p)paṃ (PK Bois B104 a3?), kercapaṃ (PK réserve 1517 B3.1 a4,), kerccapaṃ (PK Bois B20 a3?)

-añ

archaic-classical classical

late colloquial

-añ -añ / -aṃ

(?) -aṃ

a Adams (dtb 295) interprets TB tanaulykaṃ in B48 a5 (Udānālaṅkāra) as a nominative plural: tanaulykaṃ ramt sektwetse pile ra ptark(aso) “leave the suppurating wound [which you are buzzing around] like flies” (translation by Adams). On the other hand, Sieg & Siegling (1949: 70–71) translates tanaulykaṃ as a regular oblique: “… also like the wound of pus [attracts] flies. Give up the …” (cf. Hannes A. Fellner apud CEToM s. THT 48).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

169

As far as the arṣāklo-type is concerned, the situation is clear. Except for the nom.pl. TB arṣāklaṃ in a classical fragment for which I do not have a clear explanation, we systematically find the plural form -añ in archaic, classical, and classical-late texts. The only attestations of a variant -aṃ are from colloquial texts. If we consider Tocharian B agent nouns of the aknātsa-type, which attest a plural paradigm identical to that of the arṣāklo-type, we find confirmation for this development, since nom.pl. -añ is attested in several classical texts: aknātsañ (B263 a4, [arch.]; SI B 121(2) a2 [class.]; B2 b1 [class.]; B31 a7 [class.]; B286 b2 [class.]) vs. aknātsaṃ (B23 b7 [class.]), aiśeñcañ (B11 b4 [class.]; B24 b3 [class.]; B31 a6–7 [class.]; B32 a1 [class.]), kauṣentañ (AS7H a6 [arch.-class.]; AS17J b6 [class.]), yokäntañ (B248 a3 [arch. with late form]), kälpaucañ (NS263 a1 [class.]), kärsaucañ (B597 a2 [class.]), yaṣṣūcañ (B78 a1 [class.]), ynūcañ (AS1A b1 [class.]; B45 a2 [class.]), kleñcaṃ (AS6A a2 [class.]), etc.199 This data confirms the sound law proposed by Peyrot, and moreover shows that -aiñ became -aiṃ earlier than -añ became -aṃ. The motivation that underlies this development is phonetic: the nasal in -aiñ lost its palatalisation earlier because it was in the proximity of a palatal (semi)vowel and, possibly, because final shwa was lost earlier in post-tonic position (/-áyñə/ > /-áyñ/). The causative pelkiñ ‘for the sake of’ > pelkiṃ (cf. also pelykiṃ in PK DA M 507.7 a6, LC 6 a1; SI B Toch 11 a5; B109 b4; B177 a6) confirms this assumption. Other examples of -iñ > -iṃ can be found in the nominative plural of the kälymiye-type (teb vi.1.), e.g. TB riṃ ‘cities’ for riñ* in THT1311 b6 (literary) kucesa plkāntär toṃ riṃ no/// “but how are these cities to be seen/visible” (cf. Malzahn 2010: 716); TB kälymiṃ ‘regions, directions’ for kälymiñ* in B108 b6 (literary) śwāra kälymiṃ po prautkar nermi(t)eṃ (p)oyśintasa “The four directions (became) filled up with artificial Buddhas” (cf. Meunier 2013: 156; but see also Peyrot’s translation 2008: 133–134 “they [i.e. the beams] filled all four cardinal points with artificial omniscient (Buddhas)”). Instances of a nom.pl. -iñ are rare in Tocharian B, e.g. kärtse-yamiñ in B81 b5 (class.).200 Furthermore, it cannot be excluded that depalatalisation of -ñ > -n only occurred in the proximity of a palatal vowel or semivowel, i.e. only before -ai- and -i-, and that the arṣāklo-type extended this new nom.pl. -ṃ [n] by analogy.201 199 Nom.pl. -aṃ for regular -añ is also found in the wertsiya-type (Peyrot 2008: 79–80): e.g. wertsiyaṃ in B221 (if a real nominative, as per Peyrot 2008: 79, but cf. also Thomas 1957: 172 who considered it as an oblique); ploriyaṃ in B289 a6. The nom.pl. TB säsuśkañ seems to be consistently written as such (e.g. in B25), while the voc.pl. is always säsuśkaṃ (B81 a1; B198 a5; B1573.a; B108 and probably THT3596 a4). Cf. also voc.pl. ṣaiyyiśkaṃ. 200 The form sākṣiñ in B623 a3 is a hapax of uncertain meaning (see dtb 744 for a suggestion). 201 Indeed, it should be note that -ñ > -ṃ in the nominative plural mostly occurred in those classes with obl.sg. -i or -ai. One may therefore wonder whether this development

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

170

Chapter 3

Another interesting fact that, to my knowledge, has not thus far been recognised is that certain nouns belonging to the okso-type are sometimes inflected as members of the arṣāklo-type in classical-late and late texts, and vice versa. Examples include: TB klyoto (nom.pl. klyotaiñä AS9B b7 [arch.] and klyotaiṣṣe* AS2A a5 [class. ~ late], AS3A a5 [class. ~ late] vs. klyotañ202 THT500–502 b9 [late]), TB wrāko* (obl.pl. wrakaiṃ in AS17K a3 [class.] vs. wrākaññeṃ in AS18A a2 [class.]), TB pānto (perl.pl. pantaintsā in B274b4 [arch.] and pantaitstse in AS7K b1 [class.] vs. nom.pl. pantañ B108 a6 [late]), TB kraṅko ‘chicken’ (kräṅkaiññe ‘stemming from chicken’ in W14 a5 [class.], kräṅkaiññi in THT1520 a3 [arch.] vs. kräṅkañe in AS3A b3 [class. ~ late]), TB pyāpyo ‘flower’ (gen.sg. pyapyaintse IT879 b3 [class.] vs. pyapyantse (?) W32 b2 [class.]203), and probably TB mantālo* ‘±malice’ (mantālaitse ‘evil-minded’ in IT51 b3 [class.?] and possibly IT262 a1 [class.?] for expected **mantālatstse).204 If not mere writing errors, these variants may prove that there was confusion between the inflection of the okso- and arṣāklo-types even in the historical phase of (late) Tocharian B. In addition, if we accept this sound law, then we can explain the irregular plurals klyotañ and pantañ in the late documents B500–502 and B108 a6 as hypercorrect forms, as pointed out by Malzahn (2011: 95 fn. 31) (cf. perhaps the perl.sg. klañtsa for klaiñtsa in B330 a3 [late] and wrākaññeṃ for wrākaiññeṃ* in AS18A a2 [class.]). The data just discussed may be summarised as follows:

originated in the okso-type and the kälymiye-type and then spread to other classes with obl.sg. in -a-i (e.g. arṣāklo-type, wertsiya-type, etc.). 202 The medical fragment THT 500–502 shows several grammatical and orthographic mistakes. Thus klyotañ may simply be an error by the scribe. 203 The fragment W32 is very faded and a reading pyapyaintse cannot be excluded. 204 Another case could be TB āppo* ‘dad’. The nominative singular of this noun is not attested so far, but all scholars agree in reconstructing it with final -o. Since it is disyllabic, we would expect it to be a member of the okso-type. However, a genitive singular āppantse (e.g. in B589 b4) and not *appaintse is attested. Since this noun is mostly attested in the vocative (cf. B83 a5 āppa ate yāmtsi päkn(ā)star-ñ “Daddy, do you intend to give me away?”), the gen.sg. āppantse may actually be based on the vocative form. Cf. also the derivative appakke ‘dear dad’, with -(k)ke (with its variants, on which see Malzahn 2013a: 112–14) forming hypocoristics. On this form and the alleged gen.sg. pyapyantse, cf. Hilmarsson (1996: 35).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

171

Gender in the Inflection of the Noun Table 27

Evolution of the nominative plural in the okso- and arṣāklo-types in Tocharian B

archaic arch.-class. classical late okso-type -aiñ arṣāklo-type -añ

-aiñ -añ

-aiṃ -añ / -aṃ

colloquial

-aiṃ (>> -añ) / -ai -ai -añ / -aṃ -aṃ

3.5.2.7 Conclusion To sum up the results of our investigation, we have seen that the okso- and arṣāklo-types are two closely related inflectional classes in Tocharian B. They have similar case markers, but the former includes ai-stems, the latter includes a-stems. Both types comprise masculine and feminine nouns, but, as in the kantwo-type, the default gender was the feminine (Hartmann 2013: 80). On the other hand, the majority of the nouns belonging to these classes of Tocharian B chiefly correspond to only one inflectional type in Tocharian A, which includes ā-stems. After having considered several hypotheses to explain this mismatch, we have seen that the most likely hypothesis is that Tocharian A maintained the archaic state of affairs, as only one class can be reconstructed for Proto-Tocharian. This proto-type inflected as an *a-stem, thus coinciding with the attested kantwo-type. Dragoni’s recent study on the Khotanese loanwords in Tocharian brings additional evidence to this reconstruction (2022: 242–243). He shows that the kantwo-type includes only loanwords from either Proto-Tumshuqese-Khotanese and pre-Khotanese, while no item from Proto-Tumshuqese-Khotanese can be found in the classes with obl.sg. -ai. Most of the Khotanese loanwords belonging to this class can be dated to the Old Khotanese period. This testify that the kantwo-type was productive in a more archaic (prehistoric) stage, while only at a later stage did the common prototype from which the okso- and arṣāklo-types derive become productive. Furthermore, the few Khotanese loanwords from the prehistoric period belonging to the class with obl.sg. -ai may have originally been inflected as members of the kantwo-type (with nom.sg. -o, obl.sg. -a) and only secondarily taken the ending obl.sg. -ai, thereby changing inflectional type and thus following an inner-Tocharian trend of development. I have subsequently investigated the development of the a-stems in Tocharian B, commenting on the origin of the ai-stems and the obl.sg. -ai. If this ending can be traced back to a dat.sg. *-(e)h2-ei (functioning as a genitive) and/or to a loc.sg. *-(e)h2-i (by exaptation), it probably began to serve as an oblique between the late Proto-Tocharian period and immediately after the dissolution of Proto-Tocharian particularly in Tocharian B. The generalisation

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

172

Chapter 3

of the ai-stem in Tocharian B is the outcome of an analogical levelling which originated not only in the oblique singular, but also in the nominative plural, which evolved by sound change as follows: PT *-áñǝ > pre-TB *-áyñǝ > Archaic TB -áiñ(ǝ) > Classical TB -aiṃ > Late TB -ai (Colloquial TB -ai). 3.5.3 The wertsiya-type: TB Nouns with nom.sg. -ya, obl. sg. -yai and their TA Correspondents The members of the wertsiya-type are all feminine in gender. A formal characteristic is that they have a palatalised stem or a palatal stem final. Their inflection for the archaic period of Tocharian B was as follows: Table 28

Inflection of the wertsiya-type

inflectional class

nom. sg.

obl. sg.

wertsiya-type

wertsiya-∅ wertsiyai

nom. pl.

obl. pl.

stem

wertsiyañ* wertsiyaṃ wertsiya-

If we compare this inflectional type with other classes examined so far, it is apparent that the wertsiya-type is halfway between the aśiya-type and the arṣāklo-type: the singular is the same as the former, while the plural is like the latter (apart from the palatalisation).205 The wertsiya-type has received little attention from a historical perspective. Even though several studies have referred to nouns from this class, there has not until now been a systematic investigation of their origin and evolution. Peyrot (2008: 101, 2012a: 189–190) divides this inflectional type into two subclasses: a class of disyllabic words (the so-called wṣeñña-type) and another of trisyllabic words (our wertsiya-type). This subdivision is functional to the diachronic analysis of Tocharian B. Indeed, from archaic to classical-late Tocharian B, the nouns of the wertsiya-type developed a new nom.sg. in -o, taken analogically after the arṣāklo-type, with which the wertsiya-type shares the following characteristics: (1) number of syllables; (2) stem in -a-; (3) case markers, except for the nominative singular (i.e. -a vs. -o).206 Peyrot (2008: 205 Winter (1989) grouped the arṣāklo-type and the wertsiya-type under a single inflectional class. See Peyrot (2012a: 190) for criticism. 206 The only substantive this explanation might not account for is TB peñiyo ~ -ya ‘splendor, beauty’, which, according to Peyrot (2008: 100), is attested in a fragmentary archaic text (AS12K b4) in the variant peñiyo. However, Peyrot and I now believe that a reading peñiyā (archaic form for classical peñiya) cannot be excluded, though the line is very faded. The apparent masculine agreement in B428 b4 śāsaṃṣṣe peñyai “splendor of teaching” does not seem probative to me (pace Thomas 1964: 212 and Hartmann 2013: 375–376). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

173

101) further claimed that disyllabic nouns of the wṣeñña-type could have had varying nominative singular forms, although they are not attested. He later changed his view, claiming that the wṣeñña-type did not develop a nom.sg. form in -o because it comprised disyllabic nouns (Peyrot 2012a). Although I agree with this modified view, it is for my investigation unnecessary to split the wertsiya-type into two subclasses: this distinction is secondary, and it is not always relevant to the reconstruction. Instead, I will analyse this inflectional type from an Indo-European comparative perspective with the aim of reconstructing the PIE source from which the nouns of this type derive. Therefore, I will refer to both Peyrot’s wertsiya- and wṣeñña-types with the single label “wertsiya-type”. 3.5.3.1 Members of the wertsiya-type Several nouns can be counted in this class. From a semantic point of view, they are mostly inanimate, either abstract or concrete. From a derivational point of view, they are formed by means of various suffixes and can be schematised as follows:207 (1) TB -lya, TA -lyi, e.g. TB emalya, ta omlyi ‘heat’, TB kaccalya* ‘joy’; (2) TB -yeñña, TA -yeṃ, e.g. TB weśeñña (~ -o), ta waśeṃ ‘voice’, TB wṣeñña ‘dwelling place’; (3) TB -oñña, e.g. TB ścmoñña ~ śmoñña, ta śmoññe ‘place’, TB ampoña* (~ -o) ‘rottenness’; (4) TB -auña, e.g. TB katkauña ‘joy’, TB läkutsauña ‘light’, TB wrauña ‘?’; (5) TB -ya /-(ǝ)ya/, TA -i, e.g. TB atiya*, ta āti ‘grass’ (?), TB arśakärśa ‘bat’, TB kremīya ‘?’, krorīya* ‘horn (?)’, TB newiya ‘canal’, TB peñiya, ta pañi ‘splendor’, TB poṣiya*, ta poṣi ‘wall, side’, TB ploriya* a wind instrument, TB preściya ‘time, occasion’, TB śkwarya ‘creeper’, TB yoñiya, ta yoñi ‘path, track’, TB wertsiya, ta wartsi ‘council, gathering, assembly’. Indeed, B428 shows several late forms, such as a3, a4 klāmtte for klāmte, a6 śaulanmas for śaulanmaṃts, a7 kātär for tkātär. Therefore the obl.sg. śāsaṃṣṣe may be either a grammatical mistake for obl.sg.f. śāsaṃṣṣai or an example of ai > e in late texts (Peyrot 2008: 59). 207 Pinault (2014b: 207–209) has recently attempted to take the hapax TB arśakärśa ‘bat’ ́ (= Skt. maṇḍilya for manthilya-, B549 a6) as a vr̥kī-derivative. This word looks like an indigenous Tocharian compound. Pinault interpreted the first component arśə° as a cognate of TB arkañe* ‘darkness’, as both referring to the notion of night (but cf. also Carling 2004 and Adams 2016b). The second component °kərśa can be historically analysed as ́ a vrkī-derivative of the thematic noun *kur-ko-, referring to young or small animals (cf. Hitt. kūrka- ‘colt, foal’, the Iranian nouns MP kurrag ⟨kwlkˈ⟩, NP kurra, Oss. kur < *kurna-, and probably some other derivatives in Nuristani languages, on which see Hegedűs 2002). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

174

Chapter 3

I will deal with each member of this class. First, I will consider the nouns of the first four groups, while those of group (5), which is also the most productive, will constitute a separate section. 3.5.3.2 Analysis of the Suffixed Nouns From a formal point of view, the derived abstract nouns ending in TB -lya, TA -lyi may be interpreted as substantivised feminine adjectives. A clear example would be TB emalya, ta omlyi ‘heat’ (Pinault 2017b). This abstract noun is the expected feminine form of the adjective TB emalle, ta omäl ‘hot’, for which a feminine inflection is not attested either in Tocharian A or in B. One can therefore reconstruct for Proto-Tocharian an adjective *eməlle (m.), *eməĺĺa (f.) ‘hot’, from which the feminine form has been substantivised as an abstract noun (Hilmarsson 1987b: 91). The noun TB kaccalya* ‘joy’ (attested twice as a perlative singular in AS16.5 a3 and B520 a3) can be analysed in the same way, if we reconstruct an adjective kaccalle* ‘joyful’ (gerundive of TB katk- ‘to rejoice, be glad’, cf. also ka(cca)lñ(e)ne in NS29 a3). Another possibility is that both abstract nouns were derived with the outcome of the PIE suffix *-ih2 of the ́ vr̥kī-type (see below).208 On the other hand, TB wṣeñña ‘dwelling place’ and TB weśeñña, ta waśeṃ ‘voice’ are abstract nouns from the verbal roots TB wǝs- ‘to dwell’ and from the noun TB wek, ta wak voice’ < PIE *u̯ okw- (cf. Lat. vōx, Skt. vā́c-, OAv. vāxš [nom.sg.], Gk. *ὄψ, etc.), respectively. The palatalisation of the stem in these derived forms is problematic. Pinault (2012a: 190) assumes that both substantives were originally feminine agent formations in *-ēn-ih2 > *-yeñña, reanalysed as abstract nouns under the influence of the common abstract suffix TB -(äñ)ñe. Otherwise, one can assume an old thematic derivative from which a ñe-adjective was built and then substantivised (Kim 2007b: 19 fn. 30), but then the origin of the palatalised stem would be left unexplained. Perhaps the palatalisation comes from the underlying verbs (cf. the agent nouns of the type TB aiśeñca from ayk- ‘to know’, TB keṣṣeñca from kəs- ‘to extinguish’, TB ñäṣṣeñca from ñəsk- ‘to desire’, etc.). As regards TB ścmoñña ‘place’, Pinault (2012a: 190) reconstructs a secondary derivative in -ñña from an action noun *ścəmo (< *stem-eh2?), itself derived from the verbal root TB stǝma- ‘to stand’ (see Winter 1962b: 27 for the reconstruction of the root). However, one may also reconstruct a derivative from 208 Kortlandt (2020: 118) attempts to trace these nouns back to the “proterodynamic iH1-stems”, which are reconstructed by the author as denoting verbal abstracts. The reconstruction of this type of derivative (cf. Lat. aciēs) has been questioned (see Piwowarczyk 2021).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

175

the preterite stem |ścǝma-|, thus *ścǝma-eñña > TB ścmoñña (Peyrot 2010b: 72). This last solution has the advantage of equating the suffix of ścmoñña to that of wṣeñña and weśeñña. According to Pinault (2012a: 190), TA śmoññe was remade from an older *ścmon. Another possibility is that śmoññe is actually from a generalised stem PT * ścmoññay, originally used as a locative singular (cf. Peyrot 2012a: 205–206). This last option might explain the preservation of internal -ññ- in Tocharian A. From the aforementioned root TB katk- ‘to rejoice’ we also have TB katkauña ‘joy’ (older katkewña), probably based on an unattested adjective kātke* ‘joyful’ or a derived noun PT *katkey (cf. TA kācke ‘joy, happiness’). The suffix -auña is merely a feminine variant of the well-attested abstract suffix -auñe (Pinault 2012a: 190).209 The second noun with the suffix -auña is TB läkutsauña ‘light, radiance, brilliance’ (older läkutsewña). It is matched in Tocharian A by lukśone ‘id.’ (probably reshaped for *luktsone after lukäśnu ‘shining’, as per Georges-Jean Pinault apud Malzahn & Fellner 2014a: 70 fn. 31). The basis of these nouns is the adjective TB lakutse ‘shining’ (cf. also the noun TA lkäś ‘light’).210 However, the abstract suffix TA -one is usually matched in Tocharian B by -auñe. There are two ways to account for this incongruity. One option is that PT *-awñe is the older form and Tocharian A has preserved the archaic situation. Alternatively, PT *-awña first developed to pre-TA *-on and was then recharacterised under the influence of TA -one.211 209 If of Tocharian origin, this suffix may derive from the weak stem of the heteroclitic PIE suffix *-ur/n (Pinault 2011c). As pointed out by Kim (2007b), in some Middle Iranian languages we find continuants of a similar suffix, cf. Sogd. -ōni-, Khot. -ūña-, -auña- (see Emmerick apud Emmerick & Skjærvø 1987: 16 and Degener 1989: 160). Since the Iranian suffixes share both formal and semantic similarities with the Tocharian one, it is probable that we have a case of loaned morpheme (Kim assumes that Tocharian borrowed from Iranian). 210 See Malzahn & Fellner (2015: 71). Apparently, TA lkäś ‘light’ is a hapax legomenon attested in A249 a2. As Michaël Peyrot (p.c.) pointed out to me, one cannot rule out the possibility that this lkäś is misspelled for lukäś (cf. TA lukäśnu ‘shining’ and the variants of TA pukis ~ pkis [gen.sg. of TA puk ‘all, every’]). The same phenomenon would also underlie TB läktsauña vs. TB läkutsauña. 211 Formally, TB wrauña may belong here. It is a hapax legomenon attested in B28 b4 (Udānālaṅkāra). Most Tocharian dictionaries and lexicons (e.g. dtb 673; Poucha 1955; Thomas 1964) assume that we are dealing with a sort of talking bird. This meaning was suggested by Sieg & Siegling, who were the first translators of the fragment. The first part of line b4 runs as follows: (k)u(se) parśine ksa tuk sū weṣy entwe mäkte ramt wrauña “Wer immer ihn fragen sollte, genau dasselbe sagte er dann, gleich wie eine Prediger-Krähe (?)” (Sieg & Siegling’s translation 1949: 47). The translation of wrauña as ‘Prediger-Krähe’ has been accepted by most scholars. Adams (dtb 673) goes a step further, as he proposes that TB wrauña means ‘myna (Acridotheres tristis)’. Krause (1951: 199) suggests an etymological connection with Balto-Slavic, cf. Lith. várna ‘crow’, OCS vrana, Russ. voróna, etc. However,

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

176

Chapter 3

As Peyrot (apud Dragoni 2021) observes, TB ampoño ‘rottenness, infection’ may belong to this class, because the nom.sg. -o is only found in a late text (B510 b6), where it may stand for older ampoña*. Dragoni (2021) has recently analysed the Tocharian B verb ampa- ‘to rot, decay’ as a loanword from LKhot. haṃbva- and our ampoña* as a possible Tocharian derivative in -eñña from this verbal root.212 3.5.3.3 Analysis of the Nouns in TB -iya, ta -i In this section I investigate the origin of the productive group of derivatives in TB -iya, ta -i. It is tempting to connect the form of the suffix with the PIE suffix *-ih2, and in what follows I will attempt to show that this connection is correct for some of the nouns belonging to the wertsiya-type. As is well known, however, two different formations in *-ih2 can be reconstructed for ́ ́ Proto-Indo-European: the devī-type and the vr̥kī-type.213 These two reconstructed formations shared certain formal and semantic features, but they also had several differences. It is therefore worth recalling their functions before proceeding further. ́ As already mentioned in § 3.4.1.2, the devī-type inflected proterodynamically, with an unmarked nominative singular. It was used to form feminine nouns ́ from athematic stems. For this reason, we find continuants of the devī-suffix in the feminine inflection of both *nt-participles and perfect participles in

there are problems with such a connection from both a phonological and a semantic perspective. The Tocharian word cannot be the exact match of the Balto-Slavic forms. See further Sergio Neri apud Hackstein, Habata & Bross (2019: 260). Van Windekens (1976: 583) suggests that the Proto-Tocharian outcome of a zero grade *u̯ r̥n- was suffixed by *-auña, with an evolution *wərn-auña > TB wrauña after dissimilation of the two nasals. I find this solution very improbable, especially since the suffix -auña is not productive and clearly forms abstract nouns. Furthermore, the reconstruction of a root *u̯ or- on the basis of the Balto-Slavic evidence has been dismissed by Kortlandt (1985b: 121) and Derksen (2015: 490–491). They believe that the PIE root *kor- (Gk. κόραξ ‘raven’, Lat. corvus) was replaced by *u̯ or- in Balto-Slavic. In addition, we lack direct evidence for translating TB wrauña properly and I therefore consider the meaning of the word not totally unsettled, just like the question of a possible etymological connection with the proper name TB Wrau attested several times in secular documents. 212 Following a suggestion by Michaël Peyrot (p.c.), TB oṅkarño /onkə́ rño/, TA oṅkriṃ ‘porridge’ may also display a similar suffix if it derives from older *oṅkarña. For the vowel assimilation oṅkarño > Late TB oṅkorño, see Peyrot (2020). 213 I agree with Pinault (2014a) in reconstructing both suffixes for Proto-Indo-European. On ́ the value of the laryngeal for the vr̥kī-type, see Pinault (2014a).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

177

several Indo-European languages, including Indo-Iranian and Greek. The main ́ functions of the devī-suffix are (Pinault 2014a; Fellner 2014a):214 (1) forming possessive derivatives (e.g. Gk. μέλισσα ‘bee’ ← ‘provided with honey’ from μέλι, gen. μέλιτος ‘honey; Gk. γλῶσσα ‘tongue, language’ ← ‘provided with a peak’ from γλῶχες ‘beard of a corn’); (2) deriving oppositional feminine nouns from masculine ones (e.g. Ved. jánitrī- ‘genitrix’ from jánitar- ‘genitor’); (3) forming verbal and nominal abstracts (e.g. Gk. φύζα ‘flight, panic’ from φεύγω ‘to flee, escape’). ́ The vr̥kī-type formed feminine nouns from both nominal and adjectival thematic stems. A few masculine nouns occurred as well (e.g. Ved. rathī-́ ‘charioteer’).215 Furthermore, it is reconstructed with no ablaut. As suggested by Lohmann (1932: 69), the original meaning of the suffix was affiliation (“Zugehörigkeit”).216 Its main functions are (Rau 2007; Fellner 2014a): (1) forming possessive exocentric derivatives, or “genitivals” as they are sometimes called (OCS sǫdii ‘judge’ ← ‘pertaining to the verdict’ from sǫdъ ‘verdict, court’; Ved. rathī-́ ‘charioteer’ ← ‘pertaining to the chariot’ from rátha- ‘chariot’); (2) deriving oppositional feminine nouns from masculine ones (Ved. vr̥kī-́ ‘she-wolf’ from vr̥ḱ a- ‘wolf’; Ved. arāyī-́ ‘evil (female) spirit’ from árāya- ‘evil spirit’); (3) individualising formations (things or entities with the characteristic of the basic form), mostly from thematic adjectives (typically from colour or material terms, e.g. Ved. kr̥ ṣṇī-́ ‘night’ from kr̥ ṣṇá- ‘black’; ON reyðr ‘rorqual’ from rauðr ‘red’). 214 Functions (2) and (3) can be interpreted as an extension of function (1), which is to be considered the original function. ́ 215 Pinault (2014a: 274) claims that the vr̥kī-derivatives do not show any specialisation of gender, except for the fact the derived nouns are animate. Although masculine nouns are ́ equally attested, the bulk of the vr̥kī-derivatives are of feminine gender (Macdonell 1910: 269 lists only 11 masculine nouns). ́ 216 It should be noted that the exact value of the laryngeal in the vr̥kī-suffix still represents a matter of debate (some scholars have recently reconstructed the suffix as *-ih1, see mainly ́ ́ Widmer 2005 with references). The relation between the vr̥kī-suffix and the devī-suffix has also been the topic of debate. Olsen (2000: 402) derives the former from the latter, while some other scholars, like Stempel (1994: 205), have suggested the opposite. I assume that both suffixes are of Proto-Indo-European origin (cf. the discussion in Pinault 2014a with references).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

178

Chapter 3

Let us now turn to Tocharian nouns of the wertsiya-type in light of the semantic patterns and derivational mechanisms of the two *-ih2 suffixes.217 ́ Two nouns may be analysed as old derivatives of the vr̥kī-type. They are: TB wertsiya, ta wartsi and TB preściya (cf. the underived noun TA praṣt). The comparison between TB wertsiya ‘council’ and TA wartsi ‘id.’ allows us to reconstruct a common ancestor PT *wertsəya. If Adams (dtb 665) is right in proposing a connection with PIE *u̯ e(H)rdh- ‘to grow, thrive’ (liv 2 228; Hackstein 2021: 194),218 then we may reconstruct a thematic deverbal noun, from which a derivative in *-ih2 ‘pertaining to growth’ → ‘mass’ would have evolved PT *werts(ə)ya ‘group, reunion’ > TB wertsiya, ta wartsi. ́ Another noun that may be traced back to the vr̥kī-type is TB preściya ‘time, occasion’. It has no direct correspondent in Tocharian A, where we find the underived noun TA praṣt ‘id.’ instead. These two words clearly differ in their derivation. The Tocharian A noun has been linked with Germanic, cf. OHG frist ‘period of time’, OE first ‘id.’, ON frest ‘delay’ (Pinault 2008: 203; Hartmann 2013: 465–466). It is possible that they come from the PIE root *steh2- ‘to stand’ (nil 638 and 646), prefixed with *pro- ‘beyond, forward’ (cf. OCS prostъ ‘simple, free, straight’ [Derksen 2008: 421–422], MIrish ros ‘wood, height’ < PCelt. *frosto- ‘elevated land’ [Matasović 2009: 142], Skt. pratiṣṭha- ‘standing firmly’, Skt pr̥ ṣṭhá- ‘back’). To this list we may add TA praṣt ‘time’, as if from PIE *pro-stH-o- > *prosto-, i.e. ‘what stand beyond’ → ‘time’ (cf. also Lat. postis ‘door-post’, OHG fast ‘firm, fixed’). A possible caveat of this etymology is that TA praṣt ‘time’ seems to be a feminine noun, even though it belongs to the alternating class iii. In any case, the Tocharian B term is clearly derived from this word by means of a suffix *-ya. This may have originally had a slightly different

217 I will not discuss nouns that are too uncertain or otherwise useless from a historical perspective. This is the case of TB newiya (a loanword from Iranian [dtb 364], cf. Av. nāuuiia- ‘pertaining to a channel, navigable’ [in collocation with ā̆p- ‘water; river’], OP nāviya- [in collocation with ā̆p-], Phl. *nāydāg ⟨nʾywtʾkˈ⟩ [in collocation with āb ‘water; river’ and rōd ‘river’]; see Filippone 2017 for the meaning of the Iranian words), TB atiya*, ta āti (because of the unexpected lack of palatalisation), TB śkwarya ‘creeper’ (etymology unknown), TB ṣaiweñña* (see Winter 2003b), and TB śantālya ‘shepherd (?)’ (unclear derivation; cf. Adams 2009b: 5–6; Ching 2015: 46). 218 The exact reconstruction of this root is difficult, as demonstrated by the following derivatives: Skt. ūrdhvá- ‘straight, upright’, YAv. ərəδβa- ‘raised up’, Gk. ὀρθός ‘straight, upright’, OCS rodъ ‘genus, birth’ (iew 1167; gew 2.415–416; EWAia 1.243). The main problem lies in the shape of the first part of the root, since some languages point to the reconstruction of an initial *u-, while some others of an initial laryngeal. See recently Barber (2014: 32–36) and Hackstein (2021). For other Tocharian continuants of this root, see Winter (1952: 191) and Hackstein (2021: 190, 195–196).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

179

meaning, probably ‘season’ (← ‘pertaining to time’, cf. e.g. smāyana preściyaṃts ‘of the summer seasons’ = Skt. grīṣmasya ‘of summer’, Ogihara 2011: 129). As pointed out by Hilmarsson (1987b: 86–91), the fate of the suffix *-ih2 was regulated by a “Tocharian version” of Sievers’ Law, which originally implied *-əya before heavy syllables, including those syllables ending with a consonant cluster or a long vowel (but cf. also § 3.5.3.4). TB peñiya ‘splendour, beauty’ and TA pañi ‘id.’ point to the reconstruction of PT *peñ(ə)ya. Possible Indo-European connections are difficult to find. Poucha (1955: 158) argued that PT *peñ(ə)ya could come from *(s)pen-d- ‘to shine, glitter’, a root otherwise attested only in Baltic, e.g. the verb Lith. spingėt́ i ‘id.’, the noun Lith. spindà ‘splendour’, etc. The same etymology is also supported by Adams (dtb 423). However, we lack cognates from other Indo-European languages, and the Baltic root is itself problematic (Derksen 2015: 421). Following an old suggestion by Duchesne-Guillemin (1941: 164), Beekes (2010: 1546) has linked the Tocharian words with the productive PIE root *bheh2- ‘to shine, appear’ (Gk. φαίνω ‘to make visible’, Skt. bhā́ti, etc.). From a comparative perspective, this root is well attested in nominal derivatives constructed with a nasal suffix, like Ved. bhānú- ‘beam of light’, YAv. bānu- ‘id.’ < *bheh2-nu- or Skt. bhāna- ‘appearance, evidence, perception’ < *bhaH-ana- (?), OIrish bán ‘white’ < *bhe/oh2-no-. However, these derived proto-forms cannot historically account for the Tocharian substantives. There is no way to reconcile the Tocharian root vowels with the PIE root *bheh2-, since TA -a-, TB -e- can only be the outcome of *-o- or *-ē-. In view of these formal problems, Pinault (2006b: 181–183) has reconstructed a BMAC word *paṇi ‘wealth’, of which TB peñiya, TA pañi ‘splendour, beauty’ would be derivatives. For some of the remaining nouns of the wertsiya-type, the underived base is still attested in Tocharian A. Examples include: TB kroriya* ‘horn (?)’ vs. TA kror ‘crescent of the moon’; TB poṣiya* ‘wall’, ta poṣi ‘side’ vs. TA posac and posaṃ ‘below, next to’; TB yoñiya, ta yoñi ‘path, zone’ vs. TA yoṃ ‘trace’. As we will see, these underived formations are of particular importance to the diachronic analysis of the nouns. The hapax TB kroriya* ‘horn (?)’ seems to be derived from the ProtoTocharian antecedent of TA kror ‘crescent of the moon’.219 So far, two 219 Note that the exact meaning of TB kroriya* is uncertain. It is attested once in B580 b4 tarnene krorīyai śuk(ly)ā///, which Hilmarsson (1996: 182) translates as “the horn on the top of the head …”. If the body of the Buddha is being described, this expression may refer to the uṣṇīṣa, one of the thirty-two major lakṣaṇas of the Buddha. However, the symbol of the crescent moon is also a recurring iconographic element of crowned Buddha figures from the Tarim Basin. In addition, the possibility cannot be ruled out that krorīyai is

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

180

Chapter 3

etymological proposals have been put forward.220 Hilmarsson (1985a) argues that TA kror is cognate to Arm. ełǰewr ‘horn’ and Hitt. karāu̯ ar ‘id.’, which are said to reflect PIE *ghreh1-ur. Although this derivation is formally unproblematic for Tocharian, it relies heavily on the etymological connection with Armenian and Anatolian. However, neither Arm. ełǰewr ‘horn’ nor Hitt. karāu̯ ar are self-evident continuants of PIE *ghréh1-ur. Indeed, the palatalised consonant -ǰ- in Arm. ełǰewr cannot be the outcome of velar *gh (Pisani 1950; Scala 2003), while for Hitt. karāu̯ ar an etymological connection with PIE *ghréh1-ur is openly rejected by several scholars (e.g. Rieken 1999: 349 fn. 1722; Kloekhorst 2008: 446–447). Thus, the reconstruction of a PIE noun*ghréh1-ur ‘horn’ is fragile. On the other hand, Adams (1991: 5–7) connects TA kror with the IndoEuropean word for ‘horn’, PIE *ḱer-. This root noun has been the subject of an exhaustive investigation by Nussbaum (1986), who discusses most of the derived Indo-European forms. One of these formations is a heteroclitic paradigm with nom.sg. *ḱr̥h2sr̥ . Starting with this reconstructed form, Adams claims that a final *-u has been added in Tocharian, which in turn caused u-umlaut of the root vowel PT *-a- < *-h2-, thus * ḱr̥h2sr-u > *k(ə)ra(s)ru > *kroru > TAB kror(°). However, there is no evidence that u-umlaut affected internal PT *-a(Hilmarsson 1986a: 21–22; Ringe 1996: 98–99). Thus other possibilities need to be investigated.221 Kloekhorst (2008: 446–447) argues that the heteroclitic paradigm of Hitt. karāu̯ ar, karaun- ‘horn, antler’ originated from the PIE stem *ḱer- ‘horn’ (Nussbaum 1986: 1–18). He therefore posits PIE *ḱr-ó-ur/n- for Hittite. As we will see below (§ 3.6.3.2, cf. Del Tomba 2020c), my reconstruction proposes that the Proto-Indo-European word-final sequence *-u̯ r̥ was reflected as a metathesised *-ru in Tocharian. Therefore, if Tocharian inherited the same heteroclitic paradigm reconstructed by Kloekhorst for Hitt. karāu̯ ar, then it should have regularly evolved as follows: *ḱr-ó-ur > *ḱr-ó-ru > *kreru and finally TAB kror(-) after u-umlaut and apocope of *-u. Otherwise, if Melchert (1994a: 86, 2014: 259–260) is correct in reconstructing PIE *-eh2-ur for the Hittite suffix -āu̯ ar, then TAB kror(-) might also be from PIE *ḱreh2-ur.222 actually a derived adjectival form (obl.sg.f.) in agreement with a noun lost in the lacuna, though the position of the accent as suggested by the spelling (/krorə́yay/) is problematic. 220 I attempted to account for TAB kror(-) in Del Tomba (2020c). In the main text below I provide only a summary. 221 See also Kim (2019b: 145 fn. 12) for additional criticism of Adams’ etymology. 222 TAB kror(-) cannot be from PIE *ḱrh2-u̯ or because the laryngeal was lost in this position (cf. *prH-u̯ ó- > PT *pərwe> TB parwe ‘(at) first’, ta pärwa-t ‘eldest’). On the correspondence TB -o- : TA -o-, see Burlak & Itkin (2003).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

181

A final possibility is to follow Van Windekens (1976: 236) in linking TA kror to the PIE root *k(w)reh1- ‘to become bigger, stronger’ (cf. Lat. crēscō ‘to be born, increase, grow’, de Vaan 2008: 142–144).223 At any rate, TB kroriya* can be derived from the ancestor of TA kror. If it is not an adjectival form, it may ultimately be analysed as the outcome of a ́ devī-derivative with an originally endocentric meaning. In addition to TB kroriya* there is another noun that may be derived from an original heteroclitic formation, namely TB ploriya*. According to Pinault (1994: 188–189), this noun refers to a kind of wind instrument, probably a flute (cf. also the derivative TB ploriyatstse* ‘musician, flutist’). The obl.pl. TB ploriyaṃ in THT1104 a4 (Karmavācanā) seems to correspond to Skt. vāditra ‘musical instrument; instrumental music’ (mw 940).224 Pinault (2008: 285 fn. 11) is surely correct in proposing a descendant of either PIE *bhelH- ‘to roar’ or *bhleh1- ‘to blow’. However, the type of derivation involved is unclear. Adams (dtb 463) works with the second root and posits *bhloH-ru-i̯eh2. In my opinion, this reconstructed form could not have evolved into ploriya, but would rather have given *plāriyo. On the other hand, Pinault claims that TB ploriya* represents “l’elargissement d’un nom d’action *plor ‘bruit, son’ < *plæwär ou *plåwär” (Pinault 2008: 385 fn. 11). Although PT *-w(ə)r is easily derivable from the PIE heteroclitic suffix *-ur/n, the first part of both forms is problematic. On the one hand, if PT *ple- is the outcome of *bhleh1- > *bhlē-, I would expect palatalisation of the lateral. On the other hand, I cannot envision any proto-form from which PT *plo- could have come. It is well known that pre-PT *-w- is usually lost between vowels. If we therefore reconstruct the pre-PT suffix as *-u̯ or, we could suggest that the vowel -o- in TB plor-iya originated from a contraction: *bhloH-u̯ or > *bhlōu̯ or > PT *plor > TB plor° (just like *kwrih2-u̯ or > *kwryawer > TB karyor ‘trade’). Otherwise, a final possibility involves the reconstruction of a metathesised proto-form *bhleh1-ur > *bhleh1-ru-, which yielded *pĺeru- > PT *pĺoru through u-umlaut, and then *pĺorəya >> TB ploriya. However, the unpalatalised initial cluster pl- makes the reconstruction of *bhleh1-ur suspect.225 From a semantic point of view, the noun *plor- (vel sim.) should have meant ‘sound’, while the derivative TB ploriya underwent the semantic evolution ‘having sound’ → ‘wind instrument, flute’ (Pinault 2008: 385). The selection of the r-stem instead of the n-stem in the derivational deveĺ opments discussed above closely resembles the pairs Ved. pīvarī-, Gk. πίειρα 223 See Del Tomba (2020c: 58–59). 224 See Schmidt (2018: 97). 225 Depalatalisation of *-ĺ- for dissimilation with *-y- (cf. TA klyokäś vs. TB klokaśce ‘pore; opening of the body’; see Hilmarsson 1996: 151)?.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

182

Chapter 3

́ ‘fat, fertile’ vs. Ved. pīvan‘fat’, Gk. πίων ‘id.’ as both derive from PIE *piH-ur, -uen- (cf. Gk. πῖαρ ‘fatness’). As pointed out by Fellner (2014a: 70–71), external derivatives usually select the strong stem, while internal derivatives usually select the weak stem. The next word to be discussed is TB poṣiya*, ta poṣi. These nouns differ slightly in their meaning. TB poṣiya usually means ‘wall’, while TA poṣi seems to have the more general meaning ‘side’ (Barbera 2000: 235–236), since it can refer to (1) the “sides” of human beings (e.g. A320 a3), (2) the “sides” of an animal (e.g. A12 b4), or (3) the “sides” of a house, i.e. its walls (e.g. A8 a3). These nouns must derive from a third noun. Evidence for this third noun comes from Tocharian A, where we find TA posaṃ and TA posac as postpositions governing both the genitive and the oblique (Meunier 2015a: 345–346). There are two clues that allow us to reconstruct these postpositions as original nominal inflected forms. First, the fact that they govern the genitive is unusual. Indeed, as Carling (2000: 368 and 399) notes, the genitive as a governed case usually refers to living beings or abstract concepts. Second, an isolated form TA posā is attested in A146 a5 (Nandagarbhāvakrāntinirdesa) kuli tāṣ śālyās posā “if a woman is at the right side”. If Winter (1985b: 584–585) is wrong in reconstructing haplology for poṣṣāsa, then TA posā can only be the perlative singular of an unattested noun TA pos*. One may therefore claim that inflected forms of TA pos* underwent a process of grammaticalisation, first becoming relational nominals and subsequently postpositions. Several etymological proposals have been made for TB poṣiya*, ta poṣi, and TA pos*.226 Fraenkel (1932: 229) connects the Tocharian words with Lith. pùsė, Latv. puse, OPrussian pausan, pauson ‘côte, moitié’. The common ancestor of these nouns is reconstructed with an ablauting paradigm *pou̯ s-, pus- ‘± half’ (Fraenkel 1962: 676). It could therefore be argued that Lithuanian and Latvian continue the zero grade, and Old Prussian and Tocharian the full grade. Otherwise, the Tocharian word could come from the PIE root *peh2- ‘to protect’ (liv 2 460) which is attested with an s-extension in several Indo-European languages (Hitt. paḫš- ‘to protect’, Lat. pāstor ‘to herd’, OCS pasti ‘to pasture’; cf. TB pask-, ta pās- ‘to protect’). The original meaning of TA pos* could have been ‘what protects’ → ‘wall’, and then the derivatives in -iya ‘pertaining to the wall’ → ‘side’. 226 See Couvreur (1947: 11 fn. 14) and Klingenschmitt (1994: 313) for further proposals, none of which are phonologically satisfactory. See also Tremblay (2005: 427), who assumes a loanword from OIr. *pāzu- ‘face’. In my view, if the word TB poṣiya*, ta poṣi is really a loanword from Iranian, it would be more promising to link it to the root *paus- ‘to dress, cloth, cover’ and the denominative Persian verb *pōš- (Cheung 2007: 303).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

183

There is another noun belonging to this class that attests Tocharian o-vocalism in its root; namely, TB yoñiya, ta yoñi ‘path, zone’. Again, Tocharian A shows continuants of the underived noun, TA yoṃ ‘trace, footprint’. Unless a loanword from Iranian (cf. Khot. gyūna- ‘gait, course, time’, see Isebaert 1980: 142), the most straightforward comparison would be with Lat. iānus ‘passage’ (old u-stem) and iānua ‘door’ (Van Windekens 1976: 604). Accordingly, Latin and Tocharian would both continue an n-derivative of the PIE root *i̯eh2- ‘to go’, still attested in TA yoṃ, while TB yoñiya and TA yoñi would be derivatives in -ya of this noun (with the meaning ‘having/leaving traces’ → ‘path, caravan’).227 The last noun to be discussed is TB kremīya, attested twice, apparently as a nominative singular. The meaning of this word is not easily deduced. For this reason, Filliozat (1948) and Broomhead (1962) do not translate it, while Adams (dtb 234) says that it designates a part of a plant. The noun is attested in IT601 a4 ///sti ᛫ kremiya/// (without context) and in W5 a6 ///re ᛫ erkäntse yasoñña kremīya ᛫ tsäṅkacca pyāpyo ᛫ śärt (new transcription by Michaël Peyrot, p.c.). This last fragment is very difficult to translate since it contains a number of hapax legomena. The only noun that can be translated with confidence is pyāpyo ‘flower’. TB tsäṅkacca228 is probably the name of a flower. On TB erkäntse and the possible reconstruction of a noun erk(ǝ)* ‘decoction (?)’, see Carling (2003: 89, 2004; contra dtb 100). Finally, TB yasoñña might be formally linked with the noun TB yāso ‘desire, passion’. Since the fragment contains a list of medical ingredients, TB kremīya may indeed designate some kind of plant, as Adams proposes (dtb 234). If we look at TB kremīya in light of the nouns discussed so far, we can safely leave the element -iya out of our historical discussion. This TB krem(e)° has no clear formal match in Tocharian.229 If it is not a loanword, the proto-forms from which TB krem° may derive can be summarised by the two following notations: *Krom-, *Kreh1m-, where *K may represent any velar stop. I have therefore checked for Indo-European 227 One might further claim that TA yoṃ ‘trace, footprint’ is the exact counterpart of Lat. iānus. 228 Broomhead (1962: 1.7) read tsäṅkana, which is impossible. On the other hand, Filliozat (1948: 66) read tsäṅkantä, which is the accepted reading (cf. dtb 803; Blažek & Schwartz 2017: 62; Ching 2016: 55). This form is usually interpreted as a variant plural of the regular tsäṅkana (see also Ching 2010: 384), but this tsäṅkantä would have been a mistake for tsäṅkanta. 229 A possible comparable item might be kremot, attested in W37 a3 tsikallona kremotsa āśne ya(maṣä)lle “… are to be shaped; it is to be applied to the head with kremot” (cf. Filliozat 1948: 87). Adams (dtb 234) analyses TB kremot as a compound of °mot ‘alcohol’ and kare° /kə́ re/, a term that is usually translated as ‘rank, dignity’ (Adams) or ‘good’ (Winter 1968: 61; Hilmarsson 1996: 84).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

184

Chapter 3

forms matching these proto-forms and I found a quite appealing correspondence in Gk. κρόμμυον ‘onion’, with variant forms κρόμυον (Hom.) and κρόμβυον (pap.) (Chantraine 1999: 586; gew 2.23–24), MIrish crem ‘(wild) garlic’, crim, W craf ‘id.’, OE hramsan ‘ramsons’ (pl.) (Kroonen 2013: 242–243), Lith. kermušė ̃ ‘wild garlic’, OCS črěmošь ‘ramsons’, Russ. čeremšá ‘Allium garlic’ (iew 80–81; Derksen 2015: 239–240), Yazghulami gamš ‘wild onion’, Tajik kamč ‘Allium rubiginosium’ (Steblin-Kamenskij 1982: 73). Greek and Germanic point to *krom-, while Celtic and Balto-Slavic point to *krem-. If Tocharian belongs here, TB krem° could be ranged under the first group, as continuing *krom-. The original Proto-Tocharian formation from which TB kremīya is derived is however unknown. Some of the forms just mentioned point to an extension in *-us-, while others do not attest any direct medial *-u-. As the word is also attested in Eastern Iranian languages, one cannot even exclude a loanword. However, since the meaning of kremīya cannot be established with confidence, the working analysis given above must be regarded as speculative. 3.5.3.4 Conclusion To sum up, we have seen that some nouns belonging to the wertsiya-type can ́ reflect formations in *-ih2 of both the devī-́ and the vr̥kī-types. Although the Indo-European comparison is either ambiguous or too meagre to ascertain the derivation of some of these nouns, we have been able to clarify the derivational process involved. The reconstruction of possible underlying underived formations has been central to the investigation of nouns of the wertsiya-type. In some cases, Tocharian A clearly attests the noun from which a ya-derivative has been formed. One can further assume that the formal and the semantic division between ́ the devī-́ and the vr̥kī-types became increasingly opaque in the history of Tocharian. The result of this process suggests that these inherited formations influenced each other before finally merging into a single category. This is not an isolated development, since the same formal and semantic merger of the ́ devī-́ and the vr̥kī-type can be discerned in several other Indo-European languages and branches.230 In order to explain the development *-ih2 > TB -(ə)ya, ta -i we have already seen that Hilmarsson (1987b) proposed the generalisation of a “long” Sievers’ variant of the suffix; I believe that this is still a possibility. This suffix regularly

230 See e.g. Cardona (2003: 161) and Gōto (2013: 21–23) for Indo-Aryan, Johnsen (2005) for Germanic, and Piwowarczyk (2021) for Latin.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

185

Gender in the Inflection of the Noun

yields PT *-(ə)ya in front of stem-final consonants that could not be palatalised, and the partial productivity of this allomorph may have been facilitated by a number of BMAC and Old Steppe Iranian loanwords with a suffix *-ii̯ā̆- → -(ə)ya that eventually entered into Tocharian in a prehistoric phase (e.g. newiya ‘canal’ ← OSIr. *nau̯ ii̯ā- ‘thing relative to navigation; canal’, see Bernard forth.). 3.6

The Evolution of the PIE Neuter in the Noun Inflection of Tocharian

This section aims to clarify how the PIE neuter gender evolved in Tocharian and to what extent it was continued as the Tocharian genus alternans in the inflection of the noun. Much attention will be paid to the development of the thematic neuter and to cases of gender fluctuation caused by morphophonological mergers with the masculine and the feminine. The section is divided into two parts: in the first part I investigate the evolution of the thematic neuter singular and the merger with the masculine (§ 3.6.1), and the evolution of the thematic neuter plural and the merger with the feminine (§ 3.6.2); in the second part I analyse the evolution of the athematic neuter in the n-stems and r/n-stems, and I deal with pluralia tantum in Tocharian B and the reanalysis of some neuter plural as singular in Tocharian A (§ 3.6.3 and § 3.6.4). I argue that Tocharian inherited thematic neuters from Proto-Indo-European as a productive category but that the outcomes of this stem type underwent a threefold evolution: while the majority of thematic neuters were regularly continued as alternating nouns, others were reinterpreted as masculine, and yet some others as feminine. To substantiate my claims, I compare the evolution of Tocharian with the situation we find in the gradual transition from Classical Latin to modern Romance languages, where the neuter gender has been lost but thematic neuter nouns have been frequently reinterpreted as masculine or feminine. 3.6.1 The Evolution of the Thematic Neuter Singular The classification of PIE neuter nouns and their endings is based on the shape of the stem. The stem could be thematic or athematic. Those stems that ended with *-o-, rarely alternating with *-e-, are thematic. From the inflectional point of view, it is well known that the neuter did not mark any difference between the nominative and the accusative. In the athematic inflection, they were zero-marked in the singular. In the plural, thematic and athematic types

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

186

Chapter 3

shared the same ending PIE *-h2. The two inflections are traditionally schematised as follows:231 Table 29

Nominative and accusative in the inflection of the PIE neuters

nom.acc.sg. nom.acc.pl.

athematic

thematic

*-∅ *-h2

*-o-m *-e-h2

From a diachronic point of view, an important difference between the athematic and thematic neuter is that the latter is chronologically more recent than the former. In fact, several athematic neuter nouns can be reconstructed for the proto-language. They are generally continued as alternating in Tocharian. The absence of formal differences between the nominative and the accusative in PIE is perfectly mirrored in Tocharian, since alternating nouns are limited to class i, ii, and iii (nom. = obl.). Examples are numerous (Pinault 2008: 491–497; Hartmann 2013: 523): – – – –

TB āsta ‘bones’ (pl.) < PIE *Host-h2 (YAv. asti, Lat. ossa); TA waṣt, tb ost ‘house’ < PIE *u̯ eh2stu- (Skt. vā́stu-, Gk. ἄστυ); TAB or ‘wood’ < PIE *doru- (Skt. dā́ru-, Gk. δόρυ, Hitt. tāru-); TA ysār, tb yasar ‘blood’ < PIE *h1esh2(ō)r (Hitt. ešḫar, Skt. ásr̥-k, Latv. asinis); – TA ṣtām, tb stām ‘tree’ < PIE *st(e)h2-mn- (MIrish taman, HG Stamm (m.), Lat. stāmen).

This short list is purely illustrative and could easily be extended. On the other hand, comparative evidence allows us to reconstruct only a restricted set of thematic neuters for Proto-Indo-European. Examples include: PIE *(H)i̯ugóm ‘yoke’ > Lat. iugum, Hitt. iuka-, Skt. yugám, Gk. ζυγόν; PIE *pedom ‘place, ground’ > Hitt. pēdan, Gk. πέδον, Skt. padám; PIE *u̯ (e)rdhom ‘word’ > Lat. 231 Cf. Melchert (2014), Steer (2014), Lundquist & Yates (2018: 2083–2085, 2093), Neri (2017: 85–86), UrIG § 1.c, p. 3. In Greek and Latin, the nom.acc.pl. ending surfaces with a short vowel -ă. On the basis of this evidence, Beekes (1994) reconstructs a single ending *-h2 for both thematic and athematic inflections, claiming that *-eh2 originated in pronouns and then spread to the thematic inflection in the individual Indo-European languages. It is generally assumed, however, that nt.pl. Gk -α, Lat -a are actually innovations (see e.g. Olander 2015: 234–236; Neri 2017: 86; Weiss 2020: 228). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

187

verbum, Goth. waúrd; nomina instrumenti in e.g. PIE *-trom, like *h2erh3-tro-m > Lith. árklas, Lat. arātrum (lengthening based on arāre ‘to plough’), Gk. ἄροτρον, Arm. arawr, OIrish arathar, and a few others.232 However, the paucity of reconstructable thematic neuter nouns is not surprising, as it is well known that the thematic type (in both nominal and verbal morphology) is posterior to the athematic type (Lazzeroni 2002) and that Indo-European languages have inherited and further generalised this category. A related question is where do we find Tocharian continuants of the PIE thematic neuter? If we approach this problem from a formal perspective, the nom.acc.sg. PIE *-om is expected to have yielded nom.obl.sg. PT *-e > TB -e, ta -∅. There are two classes with this singular paradigm: the masculine yakwe-type (nom.pl. -i, obl. -eṃ) and the alternating āke-type (nom.obl.pl. -e-nta).233 In the following paragraphs, I attempt to identify PIE neuter nouns in these inflectional types. 3.6.1.1 The yakwe-type The yakwe-type is a productive class of masculine nouns with a differentiated plural nom.pl. TB -i, ta -añ, obl.pl. TB -eṃ, ta -as (class v.1). The origin of this class is generally well understood (teb §§ 179–180; Adams 1988a: 112–113; Pinault 1989a: 78–81): the bulk can be traced back to the PIE *o-stems, as it is made clear by TB yakwe, ta yuk ‘horse’ < PIE *h1éḱu̯ o- (m./f.) itself (Skt. áśva-, YAv. aspa-, Gk. ἵππος, Myc. i-qo, Lat. equus, etc.). As Hackstein (2013) demonstrated, TB yakwe is attested with both masculine and feminine agreement in secular documents (Ching 2010) and thus represents an important relic of an older PIE bipartite system with sex-indifferent thematic inflection (cf. Gk. ἵππος ‘male horse mare’ [Laconian hίππος], Lat. equus ‘id.’, Ved. áśva- ‘id.’ vs. the remodelled feminine equa, Vedic áśvā-). A further exception appears to be the old thematic formation TB yente, ta want ‘wind; humour (wind)’, which is feminine. This is unexpected from both a synchronic and a diachronic point of view (cf. Skt. vā́ta-, Av. vāta-, Lat. ventus, 232 For one view on the instrument suffixes in Indo-European, see Olsen (1988). 233 In addition, there is another class with nom.obl.sg. TB -e, ta -∅. This class may be labelled the kante-type (TB: sg. -e, pl. -enma; TA: sg. -∅, pl. -ant/-antu), and it consists of a few numerals (teb class ii.2). The most prominent member is TB kante, ta känt ‘100’, which regularly comes from PIE *ḱm̥ tom (> Lat. centum, Gk. ἑκατόν, Skt. śatá- etc.). As pointed out by Winter (1992: 122), the plural formation of these numerals cannot be reconstructed for Proto-Tocharian with confidence, because it seems to be of recent origin in both Tocharian B and A. This is particularly clear in Tocharian B, since TB kante mechanically selected the plural marker on the basis of the number of the syllables of the word (Winter 1992: 120). TB yaltse, ta wälts ‘1,000’ and TB tmāne ~ tumane, ta tmāṃ ‘10,000’ behave like TB kante, ta känt. On the other hand, TB pkante, A pkänt ‘hindrance’ has been presumably included in this class based on its formal resemblance with TB kante, ta känt. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

188

Chapter 3

Goth. winds, OHG wint, all of masculine gender). In § 2.3.1 I have offered a tentative explanation of this problem of gender assignment. Another feminine noun belonging to this type has been recently discovered by Peyrot & Meng (2021), who convincingly derive the hapax legomenon TB santse ‘daughter-in-law’ (THT1253 + THT3056 a3) from PIE *snusó- (f.) ‘daughter-in-law’ (cf. Gk. νυός, Lat. nurus, Arm. now). As to the morphology of nouns of the yakwe-type, Malzahn (2013b) has shown that many of them can ultimately be analysed as derivatives of the τόμος-type (e.g. TB werke, ta wark ‘hunt, chase’ < *u̯ órǵ-o- ‘work’; TB kene, ta kaṃ < *ǵhu̯ ón-o- ‘sound’, etc.). There are, however, some other nouns that do not continue this productive type of masculine nouns (teb § 180 p. 130).234 Klingenschmitt (1994: 316) analyses TB yakne, ta wkäṃ ‘way, manner’ as the outcome of a neuter *u̯ eǵh-no(but cf. OIrish fén ‘wagon’ [m.], W gwain; see further OHG wagan < *u̯ oǵh-no-). It is likely that the Tocharian A form led him to this reconstruction. Indeed, TA wkäṃ does not belong to the yakwe-type, as it is an alternating noun with plural form wäknant (thus a member of the āke-type). It is peculiar that we find a similar variation in the plural paradigm of TB pelaikne ‘law (= Skt. dharma-)’, a compound of pele° ‘law’ and °yakne ‘manner’ (Pinault 2020a), which shows both nom.obl.pl. -enta (āke-type) and nom.pl. -i, gen.pl. -eṃts (yakwe-type).235 Therefore, this noun may originally have been a neuter (continued as alternating in Tocharian A and in the Tocharian B compound), which was reinterpreted as a masculine in Tocharian B as a result of the morphophonological merger of masculine and neuter in the thematic paradigm of the singular. A clear case of a PIE neuter noun reinterpreted as a Tocharian masculine is TB ere ‘appearance’ (obl.pl. ereṃ), if correctly identified as the outcome of a neuter s-stem *h3eros- (Gk. ὄρος ‘mountain’, see dtb 99 and § 3.6.3.1). Another peculiar case that evidences the reanalysis of an originally masculine noun is TB ṣpane (pl. not attested), ta ṣpäṃ ‘sleep’ < PIE *su̯ ep-no-. The Tocharian A noun has two plural variants: TA ṣäpnant (āke-type, alternating) and TA nom. ṣäpnañ*, obl. ṣäpnas (yakwe-type, masculine, cf. loc. pl. ṣäpnasaṃ ‘in the dreams’ in A78 a1 and A56 b3). Comparative evidence points to the reconstruction of a masculine noun (PIE *su̯ ep-no-, cf. Lat. somnus m. ‘sleep, dream’, Skt. svápna- m. ‘id.’, Av. xvafna- m. ‘id.’; Gk. ὕπνος m. ‘id.’, OCS 234 As pointed out by Nussbaum (2017), neuter forms of the type *R(ó)-o- are occasionally found (cf. Ved. rókam vs. rókaḥ ‘light’, etc.), but evidence from Indo-European languages is too meagre for comparison with the Tocharian data. 235 See Peyrot (2008: 115–116). The form peleykne given by Isebaert (1980: 113) seems not to be attested (Peyrot 2008: 59). It was possibly inferred by Isebaert on the basis of teyknesa ‘thus, such’ > taiknesa, compound of te and perl.sg. yaknesa (dtb 325–326).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

189

sъnъ m. ‘id.’ and Alb. gjumë m. ‘id.’ continue *sup-no-; cf. also OE swefn ‘dream’, which is neuter), which allows us to reconstruct the noun as masculine for pre-Proto-Tocharian.236 There is one further example that may prove the sporadic reinterpretation of thematic neuter nouns as masculine. It is TB twere* ‘door’ (pl. tweri), which must be related to the familiar PIE word for ‘door’, *dhu̯ ór-/dhur- (nil 130–135). This root noun has been extended with different suffixes in many Indo-European languages (nil 131; EWAia 1.764–765; Beekes 2010: 566). Among these derived forms, we find outcomes of a neuter o-stem *dhu̯ oro- in Skt. dvā́ra- ‘door, gate, passage’, OP duvara- ‘(palace) gate’, Lat. forum ‘market place, public space’, OCS dvorъ ‘courtyard’, Lith. dvãras ‘estate, village’ (nil 131).237 Indo-European languages do not attest a derivationally similar masculine stem. Based on this comparative evidence, we can therefore argue that TB twere* ‘door’ is the regular outcome of the neuter noun PIE *dhu̯ órom and that its masculine gender and inflection are secondary.238 In the following section we will verify whether inherited masculine nouns have been reanalysed as alternating. 3.6.1.2 The āke-type All nouns belonging to the āke-type are alternating and have a plural ending in TB -enta, ta -ant. Some of them have a clear etymology. TB āke, ta āk ‘end, tip’ is usually traced back to PIE *h2eḱ-os-, with a clear cognate in Lat. acus, aceris ‘husk, chaff’ (dtb 40). Some other nouns of this class are said to go back to PIE s-stems, such as TB ṣalype, ta ṣälyp (but pl. ṣälypas) ‘oil’, TB cake ‘river’, TB īke ‘place’, TB yarke, ta yärk ‘honour, veneration’.239 236 It is not uncommon to find doublets of this kind in Tocharian A. Another example may be TA pāk* ‘part’, which shows a plural pākäntu with feminine agreement (cf. A7 b5–6 ka(pśiṃ)ñāṣās pākäntu “body parts”) and a nom.pl. pākañ with masculine agreement (cf. A367b2 tre pākañ “three parts”). See Hartmann (2013: 399). 237 The Balto-Slavic forms are masculine, but the accentuation of the Slavic noun points to an old neuter (Illič-Svityč’s Law; see Derksen 2015: 148–149 and Matasović 2014: 62–73). 238 Cf. also TB yetwe (pl. yetwi) ‘ornament’, which is a derivative in *-u̯ o- of the verbal root TB yǝta- ‘to adorn; be decorated’. This noun was borrowed in Tocharian A as yetwe, but the gender and inflection of the nouns differ in the two Tocharian languages. Indeed, TA yetwe is alternating, while TB yetwe is masculine. Theoretically, pre-TB *yetwe could have been alternating (< neuter) at the time of borrowing, and TA could have maintained the gender of the borrowed word, while Tocharian B reinterpreted the noun as a masculine. However, since loanwords are typically inserted into class iii (alternating), it is more probable that the gender of TA yetwe is an innovation. 239 Another neuter s-stem continued in Tocharian is TB °kälywe /kəĺwe/, ta °klyu ‘fame’, attested only in the dvandva-compound TB ñem-kälywe, ta ñom-klyu ‘renown’ (← ‘name’

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

190

Chapter 3

TB ṣalype, ta ṣälyp (pl. TA ṣälypañ) is usually connected with Gk. ἔλπος ~ ἔλφος (Hesychius) and Alb. gjalpë ‘butter’ (nil 612–613).240 Beekes (2010: 415–416) notes that if ἔλπος is from *sélp-os- we should expect spiritus asper in Greek and no oscillation between internal -π- and -φ-. He accordingly suggests a Pre-Greek origin. However, Greek ἔλπος ~ ἔλφος may simply display a case of psilosis. The Sanskrit form sarpís- ‘molten butter, lard’ is equally derived from PIE *selp-.241 Germanic points to the reconstruction of a feminine *salbō‘ointment’ (cf. Goth. salba, OHG salba) and a neuter *salba- (cf. OHG salb).242 Theoretically, the Tocharian noun could have been an adjectival derivative of PIE *selp(h2)- ‘fat’, which was subsequently substantivised as a neuter (cf. the type of Lat. serum ‘whey’ from *ser-ó- ‘flowing’, Skt. punaḥ-sará- ‘running back’, beside *sor-ó- > Gk. ὀρός ‘whey’). As far as TB cake ‘river’ is concerned, the reconstruction of an s-stem from the verbal root *tekw- ‘to flow, run’ is formally possible, but it is not supported by comparative evidence. In Balto-Slavic we find OCS tokъ ‘current, course’, Lith. tãkas ‘(foot-)path’, Latv. taks ‘id.’ < PIE *tokw-o- (Derksen 2015: 457). Balto-Slavic seem to match YAv. taka- (m.) ‘flowing, course’ morphologically. For Tocharian, I see two options: TB cake is either the outcome of a thematic derivative *tekw-ó- ‘flowing’, substantivised as ‘river’ (cf. Lat. serum), or an original nt-participle from the same root, i.e. *tekw-ont- ‘that which flows (nt.)’ → ‘river’ (dtb 267). On the other hand, TB īke ‘place’ (pl. ikenta ~ ykenta)243 can unambiguously be compared with Lat. vīcus ‘village’ (OLat. veicus), apparently a masculine o-stem (cf. also Gk. (ϝ)οἶκος ‘house, household’, de Vaan 2008: 675). + ‘glory’). It is from *ḱleu̯ -os- (cf. Ved. śrávas-, Gk. κλέος, OIrish clú, etc.). See Höfler (2012: 132–138). 240 The shape of the Tocharian word is peculiar because of the palatalised -ly- /ĺ/. Mechanically, TB ṣalyp- /ṣǝ́ ĺp/, ta ṣälyp points to the reconstruction of *selep- or perhaps *selpi-, which are not found elsewhere in the Indo-European domain. However, on the basis of TB kǝĺp- ‘to steal’ < *kĺəp- < PIE *klep-, one may trace TB /ṣǝ́ ĺp-/, ta ṣälyp back to PT *ṣəĺp- < * ṣĺəp- < *slep-. 241 Following Nussbaum (1986: 145), Rieken (1999: 180) reconstructs an original h2-stem *selp(e)h2- secondarily enlarged with an “s-Formans”. She therefore equates Skt. sarpís- with Gk. ἔλφος: the former would be the outcome of *sélp-h2-s- and the latter of *sélph2-e/o-s-, with -φ- from *-pH-. The original *(e)h2-stem could have been retained in such forms as Gk. ὄλπη ‘oil flask’ and Germ. *salbō-. Beekes (2010: 416) apparently dismisses the etymological connection of ὄλπη with ἔλφος ~ ἔλπος. 242 Theoretically, it is possible that PGerm. *salbō < *selpeh2 is to be ultimately interpreted as the neuter plural of *salba < *selpom. 243 It seems that the distribution between the variants ike(°) and yke(°) is partially conditioned by the position of the stress, since the latter variant is only found in inflected or derived forms with more than two syllables, like in secondary case forms (e.g. ykene 90K-58F-01

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

191

TB yarke, ta yärk ‘honour, veneration’ has been related to Skt. arká- ‘ray, light, shine; song’, and Arm. erg ‘song, poem, playing’ as reflecting PIE *h1erkw-o- (m.). Schindler (1980: 84) questions this derivation, claiming that, if from a masculine *o-stem, we would not expect this noun to be alternating in Tocharian. He therefore argues that the Tocharian noun points to an *s-stem *h1erkw-os- (cf. also Hilmarsson 1986e; Ringe 1987: 102; Pinault 2008: 497), but this reconstruction cannot be substantiated from a comparative perspective. I would rather propose that the merger of the masculine with the neuter in the thematic inflection resulted in the reanalysis of old masculine nouns as alternating, since they both ended in PT *-e in the singular. As a consequence, TB yarke, ta yärk ‘honour, veneration’ can be traced back to a masculine thematic type which was transferred to the alternating class at a later stage (cf. below).244 However the reconstruction of a neuter s-stem cannot be excluded (cf. the e-grade of the root). A similar case is TB erepate (pl. -enta), ta arämpāt ‘shape (= Lat. forma)’, a compound of TB ere, ta aräṃ ‘appearance’ (see § 3.6.3.1) and PT *pate (pl. *patenta). Since Pisani (1942–1943: 28), PT *pate has been compared with Skt. bhāti- ‘splendor; evidence perception’, bhāta- ‘shining, appearing’ as derived from PIE *bheh2- ‘to shine’. Following Van Windekens (1976: 149), we may reconstruct a substantivised participle *bhh2-to- > *băto- ‘splendid, appeared’.245 a11, AS13C a2, AS17H b3 and b5, AS17I b2, NS36 and 20 b5, NS80.3 b3, B3 a6, B32 b6, B88 b2, B92 b3, B278 b1; ykemeṃ IT127 b1, B108 b2, B143 b2), in derived adjectives (ykeṣṣa B41 a3), in the plural (e.g. ykenta AS19.22 b3 and b5, SI B 121(2) b2, SI P 2 a3, B45 b3, B241 b4, THT3153 b2; ykenta/// THT614 a1; ykentane 90K-58F-01 a2; DA M 507.37 and .36 a54, B88 b2, B302 b3, B427 b5, B506 b3; ykentäne B545 b3; ykentameṃ IT127 b1; ykentaṣṣeṃ B213 a1), and in the compound yke-postäṃ ‘bit by bit’ (e.g. G-Qm 1 a2, IT55 b7, IT188 b3, IT271 b4, IT307 a3, IT723 a2, AS6C a1, AS7M b2, AS15A b4, AS17A a2, NS55 a4, SI P 2 b6, B10 b7, B45 b4, B46 b1 and b3, B99 b2, B107 b9, B205 b2, B270 b1, B271 a2, etc.). Similarly we have TB ore vs. pl. wrenta. 244 Pinault (2008: 30) reconstructs PIE *bhag-os- (nt.) as the ancestor of TB pāke, A pāk ‘part, portion’. However, other Indo-European languages point unambiguously to a thematic *o-stem (cf. Ved. bhága- ‘dispenser, gracious lord, patron; prosperity’ (mw 743), OAv. baga‘portion, distribution (?)’, YAv. baɣa- ‘lord, god; dispenser’, OP baga- ‘god’; Ved. bhāgá‘share, portion’, OAv. bāga- ‘id.’, see Lubotsky 1981). The Slavic noun *bogъ ‘god’ (cf. OCS bogъ ‘id.’, Russ. bog ‘id.’, etc.) is considered to be a loanword from Iranian (see Derksen 2008: 50). In my view, it is more likely that Tocharian also borrowed this word from Iranian, given the semantic and formal similarities of TB pāke, ta pāk with IIr. *bhāgaand the absence of strong comparative evidence outside Indo-Iranian (cf. further LKhot. bāga- ‘part, portion’, where the maintenance of internal -g- suggests a borrowing; see Van Windekens 1976: 636 and Tremblay 2005: 424). 245 Pinault (2020a: 474) instead reconstructs *bhuh2-to- as the antecedent of TB °pāte, ta °pāt.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

192

Chapter 3

A more complex case is TB śātre ‘grain’246 (pl. śatrenta in THT1349.i b1), which is usually taken as the outcome of an instrumental noun *gwi(e)h3u-o-trom ‘± Lebensmittel’ (dtb 682). Pinault (2008: 368–369) doubts this reconstruction, since the instrumental suffix *-tro- has not left unambiguous reflexes in Tocharian. He therefore reconstructs *gwi̯oh3-tu- (cf. OAv. jiiātu- ‘life’), which, however, would require a heavy remodelling of the expected outcome (Peyrot 2018a: 257). For this reason, I maintain that the classical etymology is to be preferred, despite the scarcity of the suffix *-tro- in Tocharian.247 Another noun that can be reconstructed as neuter is TB wase, ta wäs ‘poison’. Adams (dtb 634) gives no plural forms, but Thomas (1964: 239) suggests wsenta as the plural of TB wase (cf. also Van Windekens 1976: 563). In fact, this wsenta may be partially restored in B355 a5 (Supriyāvadāna) /// (mā) tarya wse(n)ta taśimme ᛬ “… I should (not) touch them, the three poisons”.248 According to the Sanskrit parallel, the poisons are probably those of an evil nāga, who is in possession of the amoghā herb that Supriya, on its way to reach Baradvīpa, needs to take in order to escape demon Nīlagrīva (Divyāvadāna VIII. 217–218 [Supriyāvadāna]): sā nāgaparigr̥hītā tiṣṭhati | sa khalu nāgo dr̥ ṣṭiviṣo ᾽pi śvāsaviṣo ᾽pi sparśaviṣo ᾽pi daṃṣṭrāviṣo ᾿pi “That [= the herb] is in the possession of a nāga. That nāga has a poisonous sight, a poisonous breath, a poisonous touch, and a poisonous bite”.249 This plural form allows us to reconstruct TB wase as an alternating noun. As far as the etymology of the noun is concerned, TB ̆ ‘venom, poison’ wase, ta wäs can be compared with Skt. viṣá- (nt.), Av. vīša(nt.) < PIE *u̯ isom, and, more distantly, to Lat. vīrus, Gk. ἰός.

246 For the meaning of the word, see Ching (2012: 308–309) and Peyrot (2018a: 255–256). Cf. Sieg (1954: 70–71). 247 Other highly hypothetical continuants of the PIE suffix *-tro- are TB enmetre (B500–502 b1) and TB tsarātre (W3 a5 and THT497 b8), two unclear components attested in medical formulas. For the former, see Maue (1990: 163) and Adams (dtb 92); for the latter, see Adams (dtb 799). 248 As in the corresponding Sanskrit passage of the Supriyāvadāna the deity gives instructions to Supriya in third person, one may wonder whether the verb taśimme is to be analysed as taśi-mme, a 3sg.opt. + the pronoun suffix with syntactic gemination (cf. aksau-mme in B108 b10): “… he should (not) touch them, the three poisons”. Another possible reading of the Tocharian line is /// tarya wse(n)ta naśimme ᛬ “… I should destroy them, the three poisons”, where naśim is 1sg.opt.act. to the verbal root nək- ‘to destroy’; a 3sg.opt.act. to the same root can, in my view, be read in B530 (see § 3.5.1.1). 249 See Cowell (1886: 105). Cf. Sieg & Siegling (1953: 232 fn. 5).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

193

There exist further examples that may point to old thematic neuters, but their etymology is either too uncertain or comparative evidence is weak250 (e.g. TB lakle ‘suffering, sorrow’ < PIE *lug-lo- [m. or nt. ?], cf. Gk. λευγαλέος ‘unhappy’, λυγρός ‘id.’, Lat. lugēre ‘be sad’).251 3.6.1.3 The Mainstream Development of the Thematic Neuters In the previous paragraphs, we have seen that Tocharian inherited a handful of thematic nouns that comparative evidence may allow us to reconstruct as neuter. For the most part, they have been continued as alternating: they converged into the āke-type and merged with the outcome of PIE *e/os-stems. However, a few of them were reassigned to the masculine gender, as they synchronically belong to the yakwe-type. The same kind of evolution can be seen in the history of Latin. Some examples include (Loporcaro 2018: 19; Rovai 2012): Lat. dorsus (m.) ‘back’ (Pl. Mil. 397) vs. regular dorsum, dorsī (nt.) Lat. corius (m.) ‘leather, skin’ (Pl. Poen. 139) vs. regular corium, coriī (nt.) Lat. lactem (m.) ‘milk’ (Petr. 7.1.1) vs. regular lac, lactis (nt.) Lat. vīnus (m.) ‘wine’ (Petr. 41. 12) vs. regular vīnum, vīnī (nt.) Lat. caelus (m.) ‘heaven’ (Enn. Ann. 559) vs. regular caelum, caelī (nt.) Lat. aevus (m.) ‘life(time)’ (Pl. Poen. 1187) vs. regular aevum, aevī (nt.)

250 According to Hilmarsson (1986c), TA klop (pl. -ant ~ -āntu) ‘misfortune’ (= Skt. duḥkha-) can be derived from nt. *ghlobom with a cognate in OIcel. glap ‘mistake, misfortune’. On the other hand, van Beek (2013: 319) hesitantly links TB yenme ‘gate, entry, portal’ with Gk. εὐνή ‘lair, bed’ (cf. also Hilmarsson 1986: 52–53). He reconstructs PIE *h3iebh-mn-odenoting ‘that into which one penetrates’ as the ancestor of the Tocharian word and PIE *h3ieumneh2- ‘cave lair’ as underlying Gk. εὐνή, with a special phonetic development of *h3iebh-mn- ‘to enter’ > *Hieu-mn-. However, all these explanations are problematic. On the paradigm of TB ore ‘?’, quoted by Krause & Thomas (teb § 167), see Winter (2003b). Adams (dtb 103–104) gives TB ewenta as the plural of TB ewe ~ iwe ‘inner skin, leather’, but I was unable to find this plural form. This alleged ewenta may actually be a misreading for the adverb eweta ‘in conflict (with)’. 251 One could also be tempted to interpret some adverbs ending in TB -e as frozen neuter forms. For instance, TB ñatke ‘urgently, quickly’ (linked to nǝtka- ‘to push away’) can be interpreted as an original *R(e)-(o)-derivative, which is reconstructed by Malzahn (2013b: 169) as *ñatke ‘pushing, holding off’ (cf. the derived adjective eñaktetse, on which see also Ogihara 2009: 396–398). Cf. also TB lauke ‘far’ from *lou̯ kó- ‘free, light space’ (Lith. laũkas, Skt. loká-). Similarly, TB ate ‘away’ was traced back to PIE *h2et-om by Hilmarsson (1996: 51).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

194

Chapter 3

Some “neuter-to-masculine” nouns may not reflect an early shift from neuter to masculine, but rather the fact that the original gender was actually masculine. Adams (2011: 271) gives the example of Lat. collum ‘neck’, which is neuter from the Classical period of Cicero onward, but Early Lat. collus (up to Varro) was originally masculine. Sporadic cases of the reverse development are equally attested. Examples are (Loporcaro 2018: 234–235; Lazzeroni 2002: 156): Lat. catilla (nt. pl.) ‘bowls’ (Petr. 50.6) vs. regular catillus (m.) Lat. nāsum (nt.) ‘nose’ (Pl. Am. 444; Mil. 1265) vs. regular nāsus (m.) Lat. uterum (nt.) ‘womb, belly’ (Pl. Aul. 591) vs. regular uterus (m.) Lat. puteum (nt.) ‘pit, well’ (Pompon. Dig. 19.1.14) vs. regular puteus (m.) Lat. cāseum (nt.) ‘cheese’ (Varro, Rust. 2.1.4.; Apul. Met. 1.5) vs. regular cāseus (m.) Lat. pāne (nt.) ‘bread’ (Pl. Cur. 367) vs. regular pānis (m.) Lat. sāl (nt.) ‘salt’ (Varro, Gram. 64; Lucr. 4.1162), sāle (Ennius, Ann. 386; Cato, Agr. 145, 162) vs. regular sāl, -is (m.) As this list shows, neuter variants of regular masculine nouns are attested from Archaic to Imperial Latin. However, real signs of the decline of the neuter begin to appear only at a later stage (with some earlier instances in Plautus and Petronius; Adams 2011: 271–273). Indeed, with the gradual depletion of the neuter gender, confusion between masculine and neuter gradually increased, causing a confusion of the two inflections. I suggest that the same kind of doublets can be reconstructed for an unattested phase of Tocharian. That it is to say, after the morphophonological merger between masculine and neuter in the singular, it is reasonable to assume that some nouns started to shift inflectional class and gender. The case of Tocharian is more difficult to evaluate because we do not have the attestation of this gender fluctuation and inflectional oscillation (see further § 4.8.1). Cases where original neuter nouns may have been reassigned to the masculine gender in Tocharian are: PIE *dhu̯ or-o- (nt.) > *twere (alt.) → (m.) > TB twere* ‘door’ (m.) PIE *u̯ éǵh-no- (nt. ?) > *ẃəkne (alt.) → (m.) > TB yakne ‘manner’ (m.), TA wkäṃ (alt.) PIE *h3er-os- (nt.) > *ere (alt.) → (m.) > TB ere ‘appearance’ (m.) The last example matches the Latin type corpus, corporis ‘body’ (nt. III decl.), reanalysed in Late Latin as a masculine II declension noun corpus, corpī. It is

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

195

probable that other cases like these still remain to be discovered in Tocharian. The idea that the masculine and the neuter fluctuated for a while may be supported by certain isolated forms. A good example is TA ṣpäṃ ‘sleep’, which attests a plural inflection of both the yakwe-type (obl.pl. ṣäpnas, masculine) and the āke-type (ṣäpnant, alternating). However, we also have apparent cases of the reverse development, i.e. masculine nouns reassigned to the alternating gender: PIE *h1erḱ-o- (m.?) > *yərke (m.) → (alt.) > TB yarke ‘honour’ (alt.), ta yärk (alt.) PIE *tekw-o(nt)- (m.?) > *cəke (m.) → (alt.) > TB cake ‘river’ (alt.) PIE *bhh2-to- (m.?) > *pate (m.) → (alt.) > TB °pāte (alt.), ta °pāt (alt.) These data are contradictory only in appearance. Indeed, they simply point to variation in the gender assignment of (pre-)Proto-Tocharian, showing that the system was flexible for a period of time before it was standardised and became more fixed. This development was caused by the formal merger of the masculine and the neuter in the thematic inflection of the singular. Another piece of evidence that may support a relation between the diachronic evolution of the gender systems of (pre-Proto-)Tocharian and Latin is that some old neuters are inserted into new inflectional types, whose plural morpheme is the outcome of a reanalysis of the final part of the stem as part of the ending. The Tocharian classes with pl. -wa < *-u-a, -na < *-n-a, -nma < *-mn-a, and -nta OIt. -ora, Rom. -uri). The source of this ending is to be sought in the morphological reanalysis of plurals of the type corpus : corpora ‘body(s)’, tempus : tempora ‘time(s)’, etc. This class became quite productive in the history of Old Italian, acquiring new members like OIt. cambio ‘exchange’ : cambiora from cambiare ‘to change’, OIt. campo ‘field’ : campora from Lat. campus, -ī m. ‘id.’, OIt. fuoco ‘fire’ : fuocora from Lat. focus, -ī ‘fireplace; heart’, OIt. prato ‘meadow’ : pratora from Lat. prātum, -ī ‘id.’ , OIt. orto ‘vegetable garden’ : ortora from Lat. hortus, -ī ‘garden’ (Loporcaro, Faraoni & Gardani 2013). Cf. also the productive neuter plural marker -er in German (old s-stems), as compared to the much rarer Dutch plurals in -eren (of the type been ‘bone’, pl. beenderen ‘bones, skeleton’, blad ‘leaf’, pl. bladeren, lied ‘song’, pl. liederen). The same happened in Tocharian, where the plural forms ending in *-C-a were reanalysed as *-Ca and then these new plural markers (particularly TB -nta, ta -nt, and TB -nma) were generalised to other formations that are etymologically unrelated to these endings.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

196

Chapter 3

To sum up, the developments described above were caused by morphophonological mergers between the three inherited genders. This produced fluctuation in the assignment of gender. In the noun inflection, the outcome of this development caused the shift of the lexical gender of some nouns. The origin of this evolution is clear, but from the data it is not easy to detect exactly how this gender reassignment operated. 3.6.2 The Evolution of the Thematic Neuter Plural So far we have focused on the development of the thematic neuter singular, investigating details of its formal and functional merger with the thematic masculine. In the following section, I will analyse the evolution of the neuter plural and its merger with the feminine. I suggest that evidence for this merger may be found in the so-called palsko-type, where old thematic plural forms may have been reanalysed as singular due to the formal merger of the neuter plural with the feminine singular. 3.6.2.1 The palsko-type From a synchronic perspective, the nouns belonging to the so-called palsko-type make up a coherent class. They are typically alternating and have no formal differentiation between nominative and oblique in the inflection of both the singular and the plural: nom.obl.sg. -o and nom.obl.pl. -onta. Since this is the only Tocharian B inflectional class with obl.sg. -o (with the exception of the unproductive and semantically marked śana-type), it follows that if a noun has an obl.sg. -o (or derived forms regularly based on the oblique stem) and does not refer to a female entity, it can be included in this class of alternating nouns. From a diachronic perspective, this type is problematic. As pointed out by Adams (2015: 179), in origin it is a heterogeneous group, which, for the most part, is represented by verbal nouns (both abstract nouns and nomina actionis). Synchronically, the nouns of the palsko-type can be divided into two groups: (1) nouns that have cognate verbs; (2) nouns without any cognate verb attested. In the following, I will deal with these two groups separately. Nouns with Cognate Verbs The mechanism by which the noun is derived from the verb is not uniform. Indeed, the root vowel of the noun does not always match the root vowel of the verb synchronically (Ringe 1987; Adams 2015: 179–180). See the following examples (Tocharian A loanwords from Tocharian B are given in square brackets):

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

197

Gender in the Inflection of the Noun Table 30

Nouns of the palsko-type derived from verbs

vowel correspondence

noun

verb

(1) a :: a

TB kāko /káko/ ‘invitation’ TB krāso /kráso/ ‘vexation’ TB plānto /plánto/ ‘joy’ [TA plānto ‘id.’] TB yāso /yáso/ ‘passion’a TB palsko /pə́ lsko/ ‘mind’ TA pältsäk ‘id’ TB raso /rə́ so/ ‘span’ TB traṅko /trə́ nko/ ‘sin’

TB kwa-, kaka- ‘to call, invite’ TB krasa- ‘to vex, annoy’ TB planta- ‘to rejoice, be glad’

(2) ə :: ə

(3) aw :: aw (4) ay :: ay (5) əy :: ə

(6) ’əy :: əy (7) yə :: wə (8) ay :: əy

TB pauto /páwto/ ‘flattery’ TA potoc ‘id.’ TB laiko /láyko/ ‘lotion (?)’d TB pilko /pə́ylko/ ‘insight’ TA pälk ‘id.’ TB pirko /pə́yrko/ ‘rise’ TA opärkā ‘at sunrise’e TB misko /mə́ysko/ ‘trade’ TB ṣiko /ṣə́yko/ ‘(foot)step’ TA ṣik ‘id.’ TB yarpo /yə́ rpo/ ‘merit’ TB traiwo /tráywo/ ‘mixture’

TB yasa- ‘be excited’ TB plǝska- ‘to think’ TA pälskā- ‘id.’ TB rəs- ‘to stretch’ TB trənk- ‘to lament’ or PT *trənk- ‘to cling’b TB pawta- ‘to flatter’ TA pawtā- ‘id.’ TB layka- ‘to wash’ TB pəlka- ‘to look’ TA pälkā- ‘id.’ TB pərka- ‘to rise, come up’ TA pärkā- ‘id.’ TB məsk- ‘to exchange’ TB səyka- ‘to take a step’ TA säykā- ‘to be flooded’ TB wərpa- ‘to enjoy’ TB trəywa- ‘to mix’

a Adams (dtb 533) glosses the word as feminine, but there are no agreement environments that allow us to establish the gender of the noun with certainty. Adams probably bases his analysis on B155 b4 cmelñeṣṣai katkauwñai y(āso)mpa ṣesa “together with joy and passion about the birth”. However, Hartmann (2013: 230) clarifies that in this case the adjective only agrees with the following noun, the obl.sg. katkauwñai ‘joy’. b See Adams (dtb 332). c See Pinault (2008: 434). d The meaning follows Filliozat (1948: 119) and Broomhead (1962: 2.113). Adams (dtb 610) proposes ‘bath, washing’. The word is only attested in the nominative singular. e This adverb, prefixed form of pärk*, is a hapax legomenon attested in A265 a3. The meaning follows Krause & Thomas’ “zur Morgenzeit” (teb § 286).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

198

Chapter 3

Let us ignore for a moment the origin of final TB -o and focus instead on the mismatching root vowel between the noun and the verb. As can be seen, the nouns grouped in (1)-(2)-(3)-(4) merely repeat the root vowel of the underlying verbs. The relevant issue here is whether the nouns are derived from the verbs or vice versa. For groups (1)-(2) the first solution seems generally more plausible,252 but the case of TB krasa- is problematic. Malzahn (2010: 613) analyses TB krasa- as a denominal verb from an unattested continuant of a PIE *o-stem (cf. also dtb 231 and Hilmarsson 1991a: 145–146). In fact, this verb may have originated from the noun TB krāso ‘vexation, torment’. This solution was recently confirmed by Dragoni (2022: 123–126), who demonstrated that TB krāso was borrowed from OKhot. graysa- ‘torment’ (attested in Late Khotanese) and then a denominal verb was created internally in Tocharian B. On the other hand, TB pawta- ‘to honour, flatter’ is derived from an abstract formation *bhou̯ dho- ‘listening, attention’, which, however, is not directly attested itself (Malzahn 2010: 730). Otherwise, one might say that TB paut-o is itself directly derived from PIE *bheu̯ dh- ‘to wake up, pay attention’ (Gk. πεύθομαι, Skt. bódhati, see liv 2 82–83) and that the verb is denominal after this attested substantive. The nouns in the other groups have different root vowels to the verbs to which they are related. Nouns in (5)-(6)-(7) are the continuants of the PIE *e-grade, while the underlying verbal roots go back to the zero grade (Winter 1988: 777–778). Indeed, labial consonants had a palatalised counterpart in Proto-Tocharian, which mostly resulted in the corresponding non-palatalised consonant with colouring of the following PT *ə to TB i.253 Therefore, the vowel 252 TB kwa-, kaka- ‘to call’ has been derived from PIE *ǵhuH- ‘to call’ (cf. Skt. hávate, see Van Windekens 1976: 192 and Hackstein 1995: 24). Hilmarsson (1996: 200–201) reconstructs *ǵhuH-kH- yielding *kwaka- > PT *kaka-, while a non-extended root *ǵhuH- should have developed TB kwa-. TB kāko ‘invitation’ is historically derived from the subjunctive stem |káka-| of TB kwa-. TB planta-, ta plantā- ‘to rejoice, be glad’ is considered a cognate of Lat. splendeō ‘to shine’. For the development PIE *-nd- > PT *-nt- (instead of *-nts-), see Malzahn (2010: 742) and dtb 459. TB yasa- ‘to be excited’ is an intransitive verb derived from TB yəsa- ‘to excite (sexually)’ < PIE *i̯es- ‘to seethe’ (Gk. ζέω ‘to boil, ferment’, Skt. yásati ‘to boil, become hot’, etc.). As for the verbs with ə-grade, TB pləska-, ta pläskā- ‘to think’ is generally derived from *bhl ̥g-sḱe/o- (cf. Lat. fulgō, Melchert 1978: 104), while for TB rəs-, ta räsā- ‘to stretch’ no clear etymology is forthcoming. 253 TB palsko, ta pältsäk (with t-epenthesis) ‘thought’ may have derived directly from PT *pəlsk- < *bhl ̥k-sḱe/o-. Otherwise, if from an original *e-grade, the noun was originally *pĺəsko (cf. 3sg.subj. plāskaṃ), which subsequently evolved into *pəĺsko > *pəlsko with regular depalatalisation of *ĺs > ls (cf. TB pəlka- ‘to see’ vs. TB pləska- ‘to think’ and TB kərsa- ‘to know’ [3sg.prt. śārsa] vs. TB krəsta- ‘to cut’ [3sg.prt. karsta]). See Kim (2007c) and Peyrot (2013a: 479–480). On the origin of TB pəlka-, ta pälkā- ‘to look’, see Norbruis (2020).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

199

mismatch between nouns and verbs in groups (5)-(6) is to be interpreted as an original paradigmatic opposition between the zero grade of the verb and the *e-grade of the derived noun, which in turn caused palatalisation of the preceding consonant. A confirmation of this analysis is offered by TB ṣik- ‘footstep’ < *sei̯k-, which shows palatalisation of the s- (cf. the underlying verb TB sǝyk- < *sik-). Furthermore, if TB yarpo /yə́ rpo/ (7) has been correctly identified as derived from wərpa- (Winter 1988: 777), we can account for the palatalisation of the initial *w- by postulating an e-grade of the root *u̯ erP-. The cases of TB traiwo ‘mixture’ (cf. the derived adjective traiwoṣṣe) and TB trəywa- are difficult because the etymology of the verb is debated. However, the type of vowel correspondence between the noun and the verb suggests that the former derives from a form with either *o-vocalism or *ē-vocalism in the root, while the underlying verb shows the outcome of the zero grade. If so, this noun might be interpreted as a derivative of the τομή-type (with lack of o-umlaut in roots with ai- or au-diphthongs, cf. Peyrot 2013a: 52; Pinault 2008: 433–438).254 The vowel mismatch described so far can be historically presented in the following terms:

254 The underlying verb TB trəywa- has been related to the PIE root *terH- ‘to drill, rub’ (Gk. τείρω ‘to exhaust, distress’, τρῑβ́ ω, Lat. terō ‘to rub’, Lith. tìrti ‘to investigate’, etc. liv 2 632), but the derivation and the ablaut grade from which it derives are unclear. On the basis of the alleged participle tattripu of Tocharian A, Adams (dtb 337) reconstructs PT *trəyp-, but Malzahn (2010: 671) claims that p for w could be secondary (likewise Peyrot 2013a: 759 fn. 322). It could be that the paradigm of the verb actually originated from the noun. Thus, TB traiwo could be the outcome of either PIE *treH-i- or *trHo-i- (cf. past ptc. Lat. trītum < *treh1-i-, de Vaan 2008: 616), enlarged with the resultative adjectival suffix *-u̯ o-, thus *treHi-u̯ o- or *trHoi-u̯ o- (cf. TB traiwe ‘mixture (?)’, Malzahn 2013b: 168). Our TB traiwo would be the original neuter plural reanalysed as a (collective) singular. Otherwise, TB tri-w- is from an athematic present PIE *trH-ei̯- (de Vaan 2008: 616) followed by -w-. If so, a derived noun based on the o-vocalism of the stem was created in pre-Proto-Tocharian; this is of course speculative. The adjective etriwaitstse* is based on a noun etriwo* (obl. sg. etriwai in AS16.8 b1), which seems to be derived directly from the verb. Compare also TB ṣǝrt-, ta ṣärttw- ‘to incite, instigate’ vs. the nouns TB ṣārtto* ‘encouragement (?)’ (obl. sg. ṣārttai), TB ṣertwe ‘instigation’ (τόμος-type), and TB spartta-, ta spartwā- ‘to turn’ vs. the nouns TB spārtto ‘discipline (?)’, TB spertte ‘behavior’, ta spartu. See Pinault (2008: 448) and Malzahn (2013b: 169).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

200

Chapter 3

Table 31

Vowel correspondence between nouns and underlying verbs

tocharian vowel correspondence

pie grade

’əy :: ə ’əy :: əy yə :: wə ai :: əy

*e :: *∅ *e :: *∅ *e :: *∅ *e(h1i) :: *∅(h1i) [*o(i̯) :: *∅(i)]

Thus among the nouns of the palsko-type derived from attested verbal forms, one group replicates the vowel timbre of the corresponding verb, while the other continues an original apophonic grade *-e-. It is possible that the former is the outcome of a derivation internal to Tocharian (i.e. synchronic derivation), while the latter represents a lexicalised outcome of a PIE derivational process (i.e. diachronic derivation). Origin of the palsko-type We now turn to the final vowel TB -o and to discuss its origin. We must first clarify how these nouns were inflected in Proto-Tocharian. The comparison between Tocharian A and B yields a clear picture, as illustrated by the following correspondences: TB palsko ‘mind, thought’, pl. pälskonta :: TA pältsäk, pl. pälskant TB wartto ‘forest’, pl. wärttonta :: TA wärt, pl. wärtant TB parso ‘letter’, pl. pärsonta :: TA pärs, pl. pärsant TB pilko ‘insight, view’, pl. pilkonta :: TA pälk, pl. pälkäntu*255 TB ṣiko ‘(foot)step’, pl. ṣikonta :: TA ṣik, obl.pl. ṣikās Only five nouns attest a plural paradigm in Tocharian A. Three of them match the inflection of Tocharian B. On the basis of this correspondence, it is safe to reconstruct a Proto-Tocharian inflectional class with nom.obl.sg. *-o > TB -o, ta -∅, and nom.obl.pl. *-onta > TB -onta, ta -ant. The only two nouns that may invalidate this reconstruction are TA pälk and TA ṣik. As for the first noun, the plural -äntu is very productive in Tocharian A, so it could be secondary.256 255 Cf. TA pälkäntwäṣ in A227–228 b7 and TA pälkäntwā-ṣi in A222 a1. 256 Tocharian B parso ‘letter’ is also assumed to have doublet plural forms. Indeed, a plural pärsanta ‘letters’ is supposedly attested once in B206 b2 (arch.-class.). One may claim,

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

201

The plural TA nom. ṣikāñ, obl. -ās is more problematic. I will return to this form below. As pointed out by Hilmarsson (1986: 19) and Adams (2015: 179), the nucleus of this class is to be sought in deverbal nouns derived with the PIE abstract suffix *-eh2. By assuming that the plural -nta is late, this reconstruction is phonologically unproblematic, since an original paradigm nom.sg. *-eh2, acc.sg. *-eh2-m would have yielded nom.obl.sg. PT *-o. But there are two additional problems to be solved: (1) why do these nouns not inflect as members of either the kantwo-type or the okso-type? (2) why are these nouns alternating and not feminine? I believe that these two questions are linked, and that a common answer can be offered. In my view, some of the nouns of the palsko-type are historically collective neuter forms in *-eh2 of thematic neuter formations in *-om of the following types: (1) PIE *i̯ugóm ‘yoke’, pl. *i̯ugéh2; (2) PIE *h2érh3-trom ‘plow’, pl. *h2érh3-treh2; (3) and perhaps *dheu̯ sóm, pl. *dheu̯ séh2, if this latter type is to be reconstructed for the proto-language (Goth. dius ‘wild animal’, ON dýr, OE dēor < *dheu̯ sóm, but cf. also OCS duxъ ‘breath, spirit’ < *dhou̯ sóm, Nussbaum 2017: 244–246; cf. also PIE *u̯ erdhom ‘word’, *u̯ eghi̯om ‘vehicle’).257 Furthermore, they may also be the outcome of neuter nouns of the R(ó)-o-type. We have seen that these derivatives are typically masculine. However, neuter forms are occasionally found in Indo-European languages. An example is Hitt. u̯ arpa- (nt.) ‘enclosure’, mostly used in the plural u̯ a-ar-pa (Melchert 2014; Nussbaum 2017: 234). This noun can be compared in both meaning and formation with PT *werpe (cf. A72 b2 loc.sg. tālont warpaṃ “in a miserable enclosure”), which was the base of TB werwiye ‘garden’ (colloquial spelling for werpiye*, cf. the derived adjective werpyeṣṣe*), TB werpiśke ‘garden’, ta warpi ‘garden’. This reconstructed noun seems to be also the source of the verb TB warpa-, ta wārpā- ‘to surround’ (dtb 637; Malzahn 2o12: 167). A cross-linguistic comparison with Latin and Romance languages becomes significant again. Indeed, in the gradual transition from Classical Latin to the Romance languages, several neuter plural forms became feminine singular, such as Classical Lat. arma, -ōrum ‘arms, weapons’ (nt.pl.) > Late Lat. arma ‘weapon’ (fem.sg.) > It. arma ‘weapon’ (fem.sg.), Sp. arma ‘id.’, Port. arma ‘id.’. however, that this is a case of scribal error. It is further unclear whether this pärsanta should be rather linked with the hapax nom.pl.f. pärsantaṃṣana ‘splendid (?)’ (THT1105 a3). A similar case is träṅkänta ‘sins, offence’ (from traṅko), which is attested twice in B334 a2, a7 (arch.-class.) for expected träṅkonta or träṅkoṃnta. 257 Cf. also Hilmarsson (1986c: 110): “Perhaps in this case the -o : -onta flexion is based on an old neuter collective plural in *-ā?”. For a theoretical framework of this merger based on Latin data, see Rovai (2012).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

202

Chapter 3

Another comparable type is Classical Lat. folium ‘leaf’ (nt.sg.), whose paradigm split into doublets: the original neuter singular folium was reinterpreted as masculine with the meaning ‘sheet of paper’ in Italian foglio, while the original neuter plural folia was continued as a feminine noun and maintained the original meaning of the Latin word in Italian foglia. In Old French, both feuil < folium and feuille < folia mean ‘leaf, foliage’, while Modern Fr. feuille (f.) means ‘leaf; sheet of paper’ as in Sp. hoja (Kahane & Kahane 1949; Giacalone Ramat 1998: 112–115; Loporcaro 2018: 67). It follows that the palsko-type can be traced back to either *eh2-formations or old neuter plural forms reinterpreted as singular.258 This development was caused by the morphophonological merger of the singular inflection of the feminine in *-eh2 and the plural inflection of the neuter thematic stem, both ending in *-o in Proto-Tocharian. This merger would have favoured the reanalysis of old neuter plural forms as singular. If this is indeed the case, we have to assume that words with *o-inflection (from both the feminine *-eh2 and the old neuter plural) displayed variation in the prehistory of the classes with pl. ending TB -añ/-aiñ, ta -āñ and the palsko-type for a while, ending ultimately in the selection of one paradigm over the other. Indirect evidence for such a reconstruction comes from other nouns with a formation parallel to that of the palsko-type but with different inflection, gender, and root grade. Some examples include (Adams 2017: 1374): TB prosko f. ‘fear’ (obl.sg. -ai) : TB prəska- ‘to be afraid’; TB yoko f. ‘thirst, desire’ (obl.sg. -ai) : TB yok- ‘to drink’; TB ṣārtto* (obl.sg. -ai) ‘encouragement (?)’ : TB ṣərtt- ‘to incite’; tsāro (obl.sg. -a) ‘monastery’ : TB tsər- ‘to separate’. The deviant plural TA nom. ṣikāñ, obl. -ās ‘(foot)steps’ for the expected **ṣikant may be now interpreted in the same light (as per Hackstein 2021: 195). Nouns Without Cognate Verbs According to Adams (dtb s.v.) the few nouns of the palsko-type for which no cognate verbs are attested are: (1) TB wartto, ta wärt ‘forest’; (2) TB miśo ‘urine’; 258 TB traiwo ‘mixture’ seems to be linked with traiwe* ‘id.’ (hapax legomenon in IT301 b3). Synchronically they are two different nouns, but they may originally have belonged to the same paradigm that split into doublets after the morphophonological merger of the neuter with both the masculine and the feminine (note that TB traiwe is masculine). Similar cases might be TB pilke ‘copper’ and TB pilko ‘insight’, both derived from the PIE root *bhleg- ‘to burn, shine’ (see also Malzahn 2013b: 170). I am unconvinced by Kortlandt’s recent argument (2020: 119) that the nouns of the palsko-type are to be traced back to the hysterodynamic i-stems, as this stem type can hardly account for the paradigm of the palsko-type. In addition, Kortlandt does not explain why these nouns are typically alternating, while the hysterodynamic i-stems are reconstructed as masculine or feminine in Proto-Indo-European.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

203

Gender in the Inflection of the Noun

(3) TB oko, ta oko ‘fruit’; (4) TB parso, ta pärs ‘letter’; (5) and perhaps TB to ‘(pubic) hair (?)’.259 Another noun should be added to this list; namely, TB pīto ‘price, cost’, a loanword from the pre-form of Khot. pīha- ‘price’ < *pīθa-.260 Adams (dtb 412) analyses the noun as masculine and gives the following paradigm: nom.sg. pīto, obl.sg. pīto, gen.sg. pītantse, obl.pl. pitaiṃ, with a derived adjective pitaitstse ‘± having a price’. This paradigm is unusual, since it makes TB pīto a concurrent member of three inflectional classes: the palsko-type (cf. nom.obl.sg. -o), the okso-type (cf. obl.pl. -aiṃ and the derived adjective), and the arṣaklo-type (cf. the gen.sg. -antse). In the following, I will show that TB pīto is a regular alternating noun of the palsko-type and that all the deviant forms need to be interpreted differently. Let us take a closer look at the number of occurrences that each stem has. I found the following attestations: Table 32

Occurrences of the inflected forms of TB pīto ‘price’

stem

occurrences

pito-/pīto-

nom.obl.sg. pito (IT574 b3; Ot 12 a14; AS7A a1; AS18A a4, b5; DA M 507.5 b2; DA M 507.23 a10; DA M 507.37-.36 a76; DA M 507.38 a54; DA M 507.42-.40 a4; LC 39 a2; B99 b3; B100 a1; B315 b3; B337 a2, b3; THT1107 a5; THT1548.a a3, a5) nom.obl.sg. pīto (IT105 b2; IT134 a1; IT222 b2; AS18A a5; NS95 b2; B516 a2) nom.obl.sg. p(i)t(o) (DA M 507.38 a52) all.sg. pitoś (DA M 507.34 a26; DA M 507.38 a69) perl.sg. pitosa (B203 b4; B204 a3; B1460.a a2) perl.sg. pītosa (IT159 b5; THT 1548.b b3) gen.sg. pīta(ntse) (B94 b2) obl.pl. pitaiṃ (IT255 a2; B211 b2) adj. pitaitse (B316 a1)

pītapitai-

259 An additional noun that could be added to this class is TB śāñcapo, whose meaning was proven to be ‘mustard (seed)’ by Chams Bernard and Ruixuan Chen at the “Tocharian in Progress” conference (8 December 2020). The authors convincingly analyse this word as a loanword from the pre-Khotanese form of historical śśaśvāna- ‘mustard’. TB śāñcapo is mostly attested in the nominative singular but a derived form śāñcapotse (apparently based on the obl.sg. śāñcapo) can be read in W26 b2 (dtb 681). 260 See Bailey (1967: 196–197, 1978: 242), Tremblay (2005: 428), and Dragoni (2022: 154–160).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

204

Chapter 3

As one can see, the stem pito- / pīto- is the standard variant. The oblique plural pitaiṃ ‘prices’ occurs only twice: (1) IT253 a2 /// śtwāra kälymiṃtsa yäkweceṃ pitaiṃ /// “In the four quarters (of the heavens) the purchase prices in horses [are]…” (transl. by Broomhead 1962: 1.262); (2) B211 b2 abhiṣekṣeṃ pitaiṃ /// “prices of the consecration …” (?). On the other hand, the derived adjective pitaitse ‘having price’ is only attested in B316 a1 snai preke pitaitse /// “without time having a purchase price” (literal translation; cf. Broomhead’s pitaitse ‘having a purchase price’, 1962: 2.179). Ogihara (2009, 2013a) discovered the word TB ṣito ‘messenger’ in the Berlin fragment B333.261 This noun is a member of the okso-type and thus has all its non-nominative forms regularly based on the stem ṣitai-. Given that the akṣaras ⟨pi⟩ and ⟨ṣi⟩ are very similar in the Tocharian Brāhmī, one may wonder whether all the pitai-forms could actually belong to the paradigm of ṣito ‘messenger’ (Ogihara 2013b: 207–208; Peyrot 2007: n° 253): IT253 a2 (literary) /// śtwāra kälymiṃtsa yäkweceṃ ṣitaiṃ /// “In the four directions, horsed messengers (obl.)…”; B211 b2 (Buddhastotra) /// abhiṣekṣeṃ ṣitaiṃ /// “consecrated messengers” (read so but emended to pitaiṃ by Sieg & Siegling 1953: 126); B316 a1 (Bhikṣu-Prātimokṣasūtra [Vinayavibhaṅga]) snai preke ṣitaitse (ṣarsa) /// “seasonably by the messenger” (= Skt. akāla dūtasya, cf. Ogihara 2009: 208–209).262 It follows that all the okso-like forms of TB pito ‘price, cost’ are ghosts. As far as the a-stem is concerned, it is apparently attested once in B93 b2 (Araṇemijātaka) /// śpālmeṃ tsaiñ(enta)sa pīta(ntse)/// “… with excellent ornaments of the price (of)…”. The gen.sg. pītantse is the outcome of a restoration by Schmidt (2001: 326) for the attested pīta///. This restoration was recently accepted by Tamai (2018: 389). However, it is untenable because TB -antse is the genitive singular of the arṣāklo-type, in which only nouns with more than two syllables are included. One would rather expect pitontse* (or at least *pitaintse) as the gen.sg. of pīto. Similarly we usually find the perl.sg. pitosa ‘with the cost of, at the price of’. In fact, these problems were solved by Hilmarsson (1991d: 76), who analysed TB pīta < *pǝ-yǝta as the imperative of TB yǝta- ‘to adorn’ (cf. Malzahn 2010: 792). The line should therefore be translated as follows: “… adorn with excellent ornaments …”. All things considered, we can conclude that TB pīto ‘price, cost’ is a regular alternating member of the palsko-type. 261 See Pinault (2017c: 138–148) for the etymology and the attested forms of TB ṣito. 262 Edition and translation of B316 a1 follow Pinault (2017c: 140–142). Since the spelling gen. sg. -tse for -ntse is usually confined to late and colloquial texts (Peyrot 2008: 69), while B316 is an archaic-classical fragment, TB ṣitaitse can also be interpreted as a derived tse-adjective.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

205

Returning to the other five nouns, the fact that, synchronically, no cognate verbs are attested does not imply that they never existed historically. In this respect, a clear case is TB wartto, ta wärt ‘forest’. Adams (dtb 630) assumes an etymological connection with Skt. vr̥ ti- ‘surrounding, covering, enclosure’ (< PIE *u̯ r̥ti-) and OE worþ ‘enclosed place, piece of land, farm’ (< *u̯ orto-), but it is difficult from both the phonological and the semantic point of view. On the semantic side, the development ‘enclosure’ → ‘sacred enclosure’ → ‘sacred grove’ → ‘forest’ is not straightforward; on the phonological side, from PIE *u̯ r̥ti- I would expect palatalisation or assibilation of PIE *-t-. An elegant solution has recently been proposed by Hackstein (2021). He argues that TB wartto, ta wärt is to be derived from the verbal root *u̯ re(H)dh- ‘to grow, be high’, through the resultative verbal adjective *-u̯ o-, thus *ur̥ (H)dh-u̯ o‘grown, upright/high’.263 This form would have been subsequently enlarged with the collective suffix *-eh2. The only problem with this etymology is that we should expect TA wärtu* instead of the attested TA wärt as the outcome of the Proto-Tocharian sequence *-wV (cf. TB spertte, ta spartu ‘behaviour’). I see two possible solutions to this problem. The first implies the reconstruction of the non-complex suffix *-eh2. If so, the cluster -tt- in Tocharian B could be explained by means of a secondary gemination of -t- in front of -r-, which is a common phonetic development (see § 3.5.1.2). However, Indo-European nominal derivatives of the verbal root *u̯ er(H)dh- are very frequently suffixed with *-u̯ o- (e.g. *(u̯ )r̥ (H)dh-u̯ -o- > Ved. ūrdhvá- ‘upright’, YAv. ərəduua-, ərəδβa‘id.’; *(u̯ )orHdh-u̯ o- > Gk. ὀρθός ‘standing’).264 The same type of suffixation is therefore expected for Tocharian too. A last possibility is to invoke some kind of contextual change, like the dissimilation of the sequence *w__w to *w__∅, thus *wərtwo > *wərto.265 Furthermore, the possibility cannot be ruled out that TB wartto, ta wärt is to be historically analysed as an original neuter plural form, according to the following development:

263 See Barber (2014: 32–36) for the problems related to the reconstruction of this root. 264 Cf. the Hsch. gloss βορσόν ̇ σταυρόν, Ἠλεῖοι, etc. (see Chantraine 1999: 818–819). See Hackstein (2021 188–191). 265 It is assumed that a dissimilatory reduction *u̯ __u̯ > *∅__u̯ may have occurred in the continuant of the underlined root in some Indo-European languages for phonetic/phonotactic reasons (see Hackstein 2021: 189–191). It is therefore possible that in Tocharian a progressive dissimilation took place, either in (pre-)Proto-Tocharian or in Tocharian A. Indeed, TB wartto could be the outcome of both PT *wərto (with gemination *-rt- > -rtt-) and *wərtwo (with assimilation *-rtw- > -rtt-).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

206

Chapter 3

(1) Resultative verbal adjective *u̯ r̥Hdh-u̯ o- ‘grown, upright/high’; (2) Substantivised thematic noun *u̯ r̥Hdhu̯ om (nt.) > *wərt(w)e ‘tree’, pl. *u̯ r̥Hdhu̯ eh2 > *wərt(w)o ‘mass of trees’ (see Winter 1972: 385–390); (3) Reanalysis of PT *wərto as a singular with subsequent specialisation of the meaning as ‘mass of trees’ → ‘forest’; (4) PT *wərto ‘forest’ > TB wartto /wǝ́ rtto/, ta wärt. Another noun with no attested cognate verb is TB miśo ‘urine’. This noun is to be linked to PIE *(h3)mei̯ǵh- ‘to urinate’ (Skt. méhati, Av. maēzaiti, Lat. meiō, perf. mīxī, Gk. ὀμείχω, ON míga-). This is a highly productive verbal root which generated derived nouns in several languages. It is however quite remarkable that none of them is formed through the suffix *-eh2. In view of the palatalised stems, Adams (dtb 497) suggests that the Tocharian word may have derived from an *i̯e/o-present (cf. Lat. meiō). However, since TB miśo is only attested as a technical word in medical texts, the possibility that it is a loanword should be seriously considered.266 Khotanese has a noun mīysaa- ‘urine’ (Bailey 1979: 332), attested once in the so-called Bodhisattva compendium (IOL Khot 150/3 a4). However, it is difficult to reconcile Khot. -ys- [z] with Tocharian -ś-. In view of the word’s meaning, it is also possible that it has undergone some kind of tabooistic distortion. The next noun to be discussed is TB oko, ta oko ‘fruit’. The formal resemblance between Tocharian A and B strongly suggests that one language borrowed from the other. Van Windekens (1976: 332) advocates that Tocharian A is the source language, but this is improbable. For this reason, any formal link with the PIE root *h2eu̯ g- ‘to grow’ is problematic because only in Tocharian A would PT *aw yield o. This root is continued in Tocharian as TB awks-, ta oks‘to grow, increase (Gk. αὔξω ‘to increase’, Malzahn 2010: 547, cf. also Kümmel apud liv 2 288–289, who sets up a PIE root variant with final *-s-). However, a last possibility that could connect TB oko with PIE *h2eu̯ g- is to start with a zero grade *h2ug-e-h2 (neuter plural or *eh2-derivative), which would have yielded *uko > *oko (umlaut) > oko (cf. PIE *uksōn > *ukso > okso) quite regularly. For the semantic development, cf. Latv. auglis ‘fruit’ ← augti ‘to grow’ (Derksen 2015: 68). Otherwise, Winter (2011: 229–230) suggests an etymological connection with OCS agoda ‘fruit’, Russ. jágoda ‘berry’, Lith. úoga ‘id.’ and Goth. akran ‘fruit’ (cf. also dtb 115). The origin of TB parso, ta pärs ‘letter’ is debated. Van Windekens (1976: 365–366) derives TB parso, ta pärs from TB pǝrsa-, ta präsā- ‘to sprinkle’, but 266 TB miśo is only attested twice: AS2B b5 [Yogaśataka] kewiye miśosa ‘with cow urine [= Skt. mūtra-], B497 a9 kewi(ye) (miś)o ‘cow urine’.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

207

this is semantically problematic. Tremblay (2005: 428) suggests a loanword from “Primitive Khotanese” *parsa-, which is said to be the ancestor of LKhot. pa‌ʾsa ‘messenger, emissary (?)’ (with variants pasa, paiʾsą, paisa). Bailey (1979: 224) claims that this word comes from PIE *pel-(ḱ)- ‘± to turn, wind’, but his reconstruction is doubtful because continuants of this verbal root are not attested in other Iranian languages and the Iranian origin of Arm. parsem ‘to throw (in a sling)’ is unproven and semantically problematic (Hübschmann 1897: 514). Furthermore, LKhot. pa‌ʾsa- is sporadically attested and only in late texts, where, moreover, a meaning ‘messenger’ does not always fit the context. From a formal point of view, TB parso, ta pärs may only derive from a word with internal -r-, but LKhot. pa‌ʾsa cannot be the outcome of Tremblay’s OKhot. *parsa-, as internal -r- is not expected to yield the subscript hook in Late Khotanese. Rather, LKhot. pa‌ʾsa and its variants point to an Old Khotanese form *palsa-. Indeed, OKhot. -l- tends to be replaced by a subscript hook before a consonant and to palatalise a previous -a- to -e- already in later Old Khotanese text (cf. e.g. beʾysa-, beysa-, ba‌ʾysa-, baysa- < OKhot. balysa‘Buddha’). Therefore, I think that an etymological link between TB parso, ta pärs ‘letter’ and an alleged OKhot. *palsa- (not *parsa-) is dubious.267 The last noun to be discussed, TB to, is very difficult to identify. Its meaning was established as ‘human body hair, pubic hair’ by Adams (1987 and dtb 327). This noun seems to be attested only once in AS8A b6: AS8A b6 ārtärne päknāträ klai ekalmī  yāmtsi Ārdrā.loc.sg intend.3sg.sbj woman(f).obl.sg make.subject.inf naine ysissi yoñyeṣṣe to pwarne ? touch.sexually.inf ? ? fire(a).loc.sg hom yamaṣäle sā ekalmī mäsketrä oblation.nom.sg do.ger.m.nom.sg this.f.nom.sg be.subject.3sg.prs.act This passage is particularly difficult to interpret and translate. It is a SanskritTocharian text, but the Tocharian part is not a translation of the Sanskrit; it is instead a detailed commentary on the practical aspects of a tantra, which, according to Filliozat’s interpretation (1948: 95–97), is named brahmadaṇḍa. As a consequence, the Sanskrit passage does not help us understand the content of the Tocharian passage, which explains how this spell should be cast, 267 Cf. Isebaert (1980: 104), who suggested a loanword from an Old Iranian form *pr̥ sā- ‘asking, questions’ (I thank Chams Bernard for reminding me of Isebaert’s etymology). See further Dragoni (2022: 146–150).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

208

Chapter 3

enumerating for each lunar mansion the ingredients and oblations that one should burn in order to take control of someone. In the Tocharian passage, a woman is to be subjected to someone and, apparently, a to must be placed into fire to achieve this goal. There are two further terms that are difficult to interpret. The first is naine. Filliozat (1948: 89–91) reads taine and interprets it as a locative plural of the demonstrative pronoun TB se ‘this’ (p. 143). Adams (1986: 339–340) initially included this form in the paradigm of to, but later changed his mind, analysing TB taine as a pronominal dual (dtb 327). Conversely, both Schmidt (1997: 256) and Pinault & Malzahn (apud CEToM s. PK AS 8A) read naine, but their interpretations are different: on the one hand, Schmidt connects this word to TA neyaṃ and thus translates ‘(female) bottom’, but there is no evidence in support of this meaning (cf. Tamai 2014: 392 who translates neyaṃ as ‘on the mat (?)’); on the other hand, Pinault & Malzahn link TB nai* to Chinese nǎi 奶 ‘women’s breast, nipple, milk’ (see Pulleyblank 1991: 221 for the Middle Chinese reconstruction), which would fit the context well (see also Kim 2018a: 52 and 62 fn. 161).268 The second problematic word is yoñyeṣṣe. Morphologically, it is clearly a derived ṣṣe-adjective, but the base yoñye° is not clear (cf. also the loc. sg. (?) yoñyene in AS8B a4). Adams analyses it as a new word with the meaning of ‘pubis’. This would be etymologically connected with Skt. yoni- ‘womb, vulva’. On the other hand, Sieg (1955: 78–80) interprets yoñyeṣṣe as a mistake for TB yoñyaiṣṣe ‘pertaining to the path, domain’. However, based on current knowledge of Tocharian this yoñyeṣṣe can be interpreted as a late form of the regular yoñyaiṣṣe (cf. Peyrot 2008: 59). Adams (1986: 240) objects that neither AS8A nor AS8B show confusion between -ai- and -e-, but this is not true since another clear example that can be adduced is TB ce for cai ‘these’ in AS8A b7. This text is not carefully written; it contains many misspellings, omissions of akṣaras, and colloquial forms. From a formal point of view, a form yoñyeṣṣe is therefore totally justified. However, some problems remain in terms of its meaning. Indeed, if derived from TB yoñiya ‘way, path, domain’, a meaning ‘pertaining to the way, domain’ apparently does not fit the context of the passage. The Tocharian text resembles the structure familiar to us from several tantras that deal with techniques for subjugation (Skt. vaśīkaraṇa) and burnt offerings (Skt. homa). However, I failed to find a closer text from which the Tocharian fragment could have been translated, though I observed that pubic hair is not usually used as a burnt offering. Filliozat (1942: 145) suggests that TB to is actually a scribal mistake for the demonstrative nom.acc.pl.f. toṃ ‘those’ 268 For yet another hypothesis, see Thomas (1991), who interprets naine as an adverb.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

209

Gender in the Inflection of the Noun

(cf. klai ‘woman’ for klaiṃ in AS8A b6 and b7). However, the meaning of several words in this passage is still obscure, and a more careful investigation, including a systematic search for Sanskrit equivalents, is required. Adams (dtb 237) gives a second attestation of a hypothetical word to ‘pubic hair’ in W2 a6, where Filliozat (1948: 65) reads the plural form tonta: W2 a6 weñ erkasenta lāni yamaṣṣälona kete269 ratre krāke tonta alā/// “erkasenta and lāni are to be made; to whomever the red dirt and the tonta … (?)” (cf. Adams dtb 237). Fragment W2 is very faded but, in my view, †tonta cannot be read because the line reads totka, as observed by Broomhead (1962: 1.4). I read line a6 as follows: /// – [⋅e]rkase[nta] l[ān]i [ya]maṣṣälona ᛫ kete ratre krāke totka a[lā](ṣṣäṃ) “… erkasenta lāni are to be made; to whom the red filth (scil. the menstrual blood (?)) is to such an extent ailing …” (?). Thus, it is better not to include the apparent hapax legomenon TB to in the palsko-type as this word either refers to a specific term involved in a tantra or should be regarded as a ghost word. To summirise, the members of the palsko-type can be historically analysed as verbal nouns. For some, the verb from which they derive is still attested. For the others, we have seen that a verbal root can be reconstructed by comparison with other Indo-European languages. The palsko-type can ultimately be traced back to the PIE type in *-eh2 and to old thematic collectives reinterpreted as singulars. Alternating Gender and the Origin of the Plural Endings TB -na and TA -äṃ In this section I discuss the synchronic distribution and diachronic evolution of the two plural endings TB -na and TA -äṃ, which played an important role in the development of both the alternating and the feminine gender. These endings are usually considered to be the outcome of the neuter plural of nasal stems, which underwent reanalysis: PIE *-n-h2 > *-n-ă > PT *-na > TB -na, ta -(ä)ṃ. Despite this alleged common origin, they have a different distribution: no Tocharian B nouns with plural in -na correspond to Tocharian A nouns with plural in -äṃ. Their productivity is different as well: TB -na is the plural marker of a fair number of nominals, while TA -äṃ is confined to only five substantives. A useful parameter to divide Tocharian B nouns with the plural ending -na is grammatical gender. We have seen that the members of the so-called śanaand aśiya-types are feminine (§ 3.4). With the exception of the masculine TB śaumo ‘man, person’ (§ 3.6.3.3), all other Tocharian B nouns with plural in -na 3.6.3

269 Filliozat (1948: 65) reads kene, but Broomhead’s kete (1962: 1.4) is to be preferred.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

210

Chapter 3

are alternating. This gender-based division mirrors a formal one: feminine nouns are differentiated for the nominative and the oblique singular, while alternating nouns have one form for both the nominative and the oblique in the singular. The feminine nouns have already been discussed above (§ 3.4). The alternating nouns will be examined in the following paragraphs. They can be grouped into various subclasses on the basis of three factors: the singular paradigm, the nominal stem, and the phoneme preceding the plural marker (teb §§ 162–164). Since the aim of this section is to trace the origin of the plural marker TB -na, it is more convenient to divide these nouns into two groups: (1) nouns that have the basic plural TB -na; (2) nouns that have a slightly different plural TB -una. These groups are scrutinised individually (§ 3.6.3.1 and § 3.6.3.2). The distribution and origin of TA -äṃ are investigated in the subsequent sections (§ 3.6.3.4 and § 3.6.3.5). 3.6.3.1 Alternating Nouns with the Plural Ending TB -na Although the inflectional class with pl. TB -na is more productive than the etymological correspondent TA -äṃ, it seems to represent a closed category in the historical phase of Tocharian B. In this respect, an important piece of evidence is that only a few loanwords are morphologically inserted into this class (e.g. TB tsäṅkana ‘naked barley’, if correctly identified as a loanword from Chin. qīng 青, an abbreviated form of qīngkē 青稞 ‘highland barley’,270 and possibly TB karāk ‘water pot’, on which see below).271 Most of the Tocharian B alternating nouns with plural in -na show etymological and derivational problems. In certain cases, this ending is to be interpreted as an innovation; in others, it can be traced back to Proto-Indo-European. The latter is the case for four nouns that together make up quite a coherent subclass. The members of this subclass are: (1) TB ṣarm (ta ṣurm) ‘motive, cause,

270 See Ching (2010: 384, 2016: 52–53). Lubotsky & Starostin (2003: 264) claim that Chin. qīng 青 ‘blue, green’ was also borrowed into Tocharian as the adjective TAB tseṃ ‘blue’ (see also dtb 810). See also the discussion in Blažek (2016: 232–233) and Blažek & Schwarz (2017: 62–63). 271 In his dictionary (dtb 678–679), Adams refers to a noun śaṃts ‘announcement’ (from Skt. śaṃsa- ‘recitation, invocation’), allegedly attested in the perlative plural in AS7H a6 śaṃtsnasa spärkālñe westrä “the dissolution is learned/spoken of by announcements” (ed. by Sieg 1938: 36 and transl. by Adams). However, the current reading of the line is rather pärnāññana (wäntarwa)ṃ(ts) ś(r)aṃts tūsa spärkālñe westrä “the dissolution is therefore said [to be] the removing of external (objects)” (cf. Georges-Jean Pinault apud CEToM). TB †śaṃts ‘announcement’ is therefore a ghost word.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

211

origin’, variant plurals ṣarmna, ṣärmanma, ṣärmana < PIE *ser-men-;272 (2) TB sārm ‘germinated seed’, pl. sārmna, sarmana;273 (3) TB ñem (ta ñom) ‘name’, pl. ñemna, from PIE *h1neh3-men- (or *h3neh3-men-); (4) TB stām (ta ṣtām) ‘tree’, with irregular pl. stāna from PIE *sth2-men-.274 Their derivation from PIE *men-stems is made evident by the final -m in the singular, which is from pre-PT *-mən < PIE *-mn̥ . The Tocharian A correspondents have the final -m as well, but the secondary plural -nt / -ntu (cf. TA ṣurm, pl. ṣurmant, TA sārm, pl. sārmäntu, TA ṣtām, pl. ṣtāmäntu).275 The plurale tantum TB särwāna ‘face, countenance’ has occasionally been compared with Ved. sŕ̥kvan-, later Skt. sr̥kvaṇī ‘corner of the mouth, lock-jaw’ (mw 1245; cf. also Skt. sr̥kva- ‘tooth, fang’, Schmidt 1980: 409; EWAia 2.783–784). There are two problems with this comparison, however: (1) the unexpected loss of *-k- (if original) and (2) the lack of cognates forms in other Indo-European languages. For these reasons, Hilmarsson (1989b) analysed TB särwāna as a *men-stem formed to PIE *streuH(d)- ‘to swell’.276 One might wonder whether TB särwāna ‘face, countenance’ was borrowed from a Middle Indian continuant of Skt. sr̥kvan-, but this is very unlikely because the cluster -kv- is expected to have yielded -kk- in Prākrits.277 Among nouns of doubtful etymology, we find TB kārak (pl. karākna) ‘branch (of a tree), foliage’ (cf. TA karak* ‘wooden part of a bow’, which is a hapax 272 See Adams (dtb 712) and Lubotsky (1988: 91) for etymological proposals. Peyrot (2008: 110) argues that the older plural must be TB ṣärmanma, as it is never attested in late and colloquial texts. He claims that ṣärmanma developed a plural in -na after dissimilation of the two labial nasals. This explanation is phonetically possible, but I think it is morphologically less probable. First, as pointed out by Peyrot himself, the plural -nma is much more productive than -na. Second, there are no other nouns with singular -m and plural -nma. As a consequence, I believe that the original plural is TB ṣarmna, which is attested in two archaic documents (B133 a3 ṣärmnā and THT1302 a3 ṣarmna) and represents the less attested plural variant. Later, two competing plurals were created: ṣärmanma (since the archaic stage) and ṣärmana (with epenthesis). The latter became the standard variant since it is attested only in classical and late documents. A similar analysis can also explain the plural of sārm ‘seed’, with old plural sārmna and late plural sarmana (attested in the late document AS14.1). 273 See Peyrot (2018a: 260–261), Adams (dtb 747), and Blažek & Schwarz (2017: 207). 274 The expected plural form would have been **stamana, **stāmna, or **stānma (cf. TA ṣtāmäntu). The lack of -m- in the plural led some scholars to reconstruct a PIE root enlarged by -d- (Hilmarsson 1986b) or -s- (Adams dtb 777), with the subsequent loss of the labial nasal in the cluster -Cmn-. 275 On the evolution of the PIE *men-stems in Tocharian, see Malzahn (2006) and Pinault (2008: 495). 276 See Adams (dtb 750–751) for yet another etymological proposal. 277 Cf. Pischel (1981: 240). See further Schmidt (1987: 296–297, 2018: 211) and Hackstein (1995: 121–122).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

212

Chapter 3

legomenon attested as a perl.sg. in A316 a1 [dta 102]). Adams (dtb 150) reports the nominative of this form as karāk /karák/, which is perhaps to be considered a separate word. Indeed, one can argue that TB karāk, with stressed final syllable, actually means ‘pot, vessel’. This noun is attested only three times, all in AS13D (Ikṣvākujātaka) at lines a4 (kauṃ-pirko kalymi war past ñārka-ñ karā(k) “water kept me away from the eastern direction, the vessel …”), b6 (/// ñiś karāk aimar war kewu “… I will take a vessel and I will pour water”), and b7 (karākmeṃ war kū(tär) “water from the vessel will be poured …”). This TB karāk can be interpreted as a loanword from Skt. karaka- ‘water vessel’ (mw 254; bhsd 168; swtf 2.21; cf. Georges-Jean Pinault apud CEToM s. AS13D). TB kārak ‘branch, foliage’ (with stressed initial syllable according to Hilmarsson 1996: 83) is considered to be attested once as the singular karak (B281 b5) and twice as the plural karakna (B554 a4) and karākna (B3 a8). The use of -a- (here /a/) instead of -ā- (/á/) is due to the archaic linguistic stage of the fragments B281 and B554. The long-spelled -ā- in the plural karākna /karákna/ (B3 a8) does not allow the reconstruction of a nom.sg. kārak /kárak/ with any certainty. This word has been traced back to Proto-Indo-European by Adams (dtb 150) and Hilmarsson (1996 83). On the other hand, one may wonder whether TB kārak* ‘branch’ and karāk ‘pot, vessel’ are actually just one word and that the ambiguous spelling TB karak in B281 b5 is to be interpreted as karāk. If so, this karāk would mean both ‘pot’ and ‘branch of a tree’ and could be a loanword from the same Skt. karaka-, which is also used as a proper name for several types of plants (mw 254). However, the analysis of this Tocharian B word cannot be separated from two further Tocharian A words: karak* ‘± limb of a bow’ (attested as a perl.sg. karkā in A316 a1) and karke ‘(small) branch’. Neither of the two words can be the formal equivalent of TB kārak* (Hilmarsson 1996: 88). An additional item that might be relevant is TB karāś (obl.sg.), ta kārāś ‘wilderness (?)’, which is usually taken as a loanword from Khot. karāśśi- ‘id.’.278 The origin of all these words is therefore problematic. 278 See Bailey (1967: 41–42, 1979: 54) and the comprehensive study by Dragoni (2022: 79–88). Cf. the Iranian root *karH- ‘to spread out, scatter, sow’ (Cheung 2007: 239–240). In Old Khotanese the word is attested in the nom.acc.pl. karāśśä (for its occurrences, see Dragoni 2022: 80–81). Theoretically, final -(i)ä may either belong to the i-declension or may be a later variant of nom.acc.pl. -e of the ā-declension. However, given the fact that the nom.acc.pl. karāśśä occurs twice in manuscript Or of the Khotanese Suvarṇabhāsottamasūtra (6.4.22 and 6.4.39a, see Skjærvø 2004: 1.136 and 140), where the variation -e ~ -ä is rare, I prefer to interpret karāsi-, karāśśi- as an i-stem. If so, I wonder whether the stage of borrowing of TB karāś (obl.sg.), ta kārāś ‘wilderness (?)’ could also predate the Late Khotanese period.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

213

We further find two pluralia tantum ending in TB -na with a clear singulative meaning: TB ersna ‘appearance’ and TB yasna ‘treasury’ (obl.pl. yasna in NS44 a5 and NS299 a4; loc.pl. prakrona yasnane “in a solid treasury” in THT1114 a4).279 Adams (dtb 103 and 526) argues that they are old derivatives of TB ere ‘form, appearance’ and TB yasa ‘gold’ respectively. The derivation of the first noun from a PIE *s-stem *h3er-os- > TB ere has long been accepted (cf. Gk. ὄρος ‘mountain’, Skt. r̥ ṣvá- ‘high’).280 The second noun is probably from *h2u̯ esh2 > PT *ẃəsa > TB yasa, ta wäs, originally a collective formation.281 If these derivations are correct, it can be argued that the plural ending PT *-na has been added in a Proto-Tocharian stage in order to recharacterise the plural form of some *s-stems. As far as TB ersna is concerned, there is another possibility: it can be argued that this noun goes back to the plural form of an original heteroclitic paradigm. Comparative evidence may support this reconstruction. In Hittite, we find the heteroclitic stem ḫaršar, ḫaršn- ‘head, person, beginning’. In past decades this noun was linked to PIE *ḱérsh2-s-r, *ḱérsh2-s-n- ‘head’, but this derivation bears formal difficulties (Kloekhorst 2008: 314–316). For this reason, Goetze (1937: 492) suggested a comparison with Gk. ὄρος ‘mountain’ and further reconstructed PIE *h3er-s-r, *h3r-s-n- (cf. also Hitt. ḫarši- / ḫaršai- ‘high, risen’ < PIE *h3ers-i-, *h3rs-ei̯-, Kloekhorst 2008: 315–316).282 Formally this reconstruction is unproblematic, and if we add TB ersna < *h3er-s-nh2 it acquires even more credit. If so, Tocharian could have continued both the s-stem *h3er-os- > TB ere and the derived heteroclitic stem *h3er-s-nh2 > TB ersna.283 Although TA aräṃ ‘appearance, form’ should belong here (dta 20), it is unclear how it is related to TB ersna, because the change *-rsn- > -rn- is without parallel in Tocharian A.284 Rather than deriving TA aräṃ from a different proto-form (Van Windekens 1976: 149; dtb 99), however, an original pre-TA 279 For the meaning of yasna, see Ogihara (2013c: 5). 280 See Adams (dtb 99). The fact that this noun is synchronically an e-stem (cf. the obl.pl. ereṃ in THT1223 b1 and perl.pl. ereṃtsa in B566 a6) may be secondary (cf. § 3.6.1). 281 See Pinault (2012a: 197) and Hackstein (2017: 1318–1319). Driessen (2003: 348–350) explains TB yasa, ta wäs as a loanword from Proto-Samoyedic *wesä. 282 See also Oettinger (2016: 322). 283 The singular TB śalna ‘quarrel’ may originally belong here as well, if it is an old plural form. For an etymological suggestion, see Malzahn (2011: 100). 284 Actually, I found only one certain word where the cluster -rsn- can be shown to predate Proto-Tocharian. It is TA |kärsnā-| (cf. TB |kǝrsǝ́ na-|), the present stem of kräsā- ‘to know’, where the cluster -rsn-, however, could have easily been restored (while TA |kärṣnā-| ‘to cut off’ is from *kärṣt-nā-).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

214

Chapter 3

*arsäṃ, the regular outcome of PT *ersna, may have been influenced by the noun TA ar* ‘form’, the unattested Tocharian A counterpart of TB ere. If so, TA *arsäṃ first lost internal *-s- and was then reinterpreted as a singular by aligning the singulative meaning with the singular number (cf. § 3.6.3.6 and § 3.6.4). A parallel case to that of TA aräṃ may be TA krāṃ ‘(outer) skin’ (dta 171). In a lecture held at the “Tocharian in Progress” conference (8th–10th December 2020), Georges-Jean Pinault linked the noun TA krāṃ to the plurale tantum TB krāna (attested once in THT1224 b1). Both nouns can be traced back to PIE *k(e)rH-n- ‘piece’ (cf. Lat. carō, carnis ‘flesh’, de Vaan 2008: 94). Thus we may have another example of a collective noun which was originally morphologically plural and has been reinterpreted as singular in Tocharian A only. 3.6.3.2 Alternating Nouns with the Plural Ending TB -una All other alternating nouns belonging to class ii.1 attest a slightly different plural formation ending in TB -euna > -auna or TB -una. The historical interpretation of these markers is debated. Before pursuing this diachronic matter, however, these Tocharian B nouns and the matching Tocharian A forms have to be scrutinised closely from a synchronic perspective. We find TB -euna > -auna in two separate groups. The first group contains lexical plurals with a singulative meaning: TB palauna ‘praise(s)’, TB tarśauna ‘deception(s)’, and TB krentauna ‘virtue(s)’. Matching nouns in Tocharian A are only found for the former two: TA paloṃ and TA tārśoṃ. Although they closely resemble their Tocharian B counterparts, these two nouns are grammatically singular. It can be argued that they were plurals in Proto-Tocharian and that Tocharian A later aligned the singulative value of the meaning with the singular morphology of the number (see § 3.6.3.6). The second group consists of nouns that have TB -i in the singular and TB -euna > -auna in the plural: TB reki ‘word’ : rekauna (TA rake : rakentu), TB ṣewi ‘pretext’ : ṣewauna, and TB yapoy ‘land’ : ypauna (TA ype : ypeyu). Finally, TB -una is the plural marker of only three nouns. Once again, their derivation is not clear. The first is TB akrūna ‘tears’, which is only attested in the plural and is matched by TA ākär (pl. ākrunt). The other two substantives are TB ṣotri : ṣotrūna (TA ṣotre : ṣotreyäntu) and TB lāṃs : laṃsūna (TA wles : wlesant). The comparison between Tocharian A and B does not allow us to reconstruct the Proto-Tocharian plural form of these nouns with confidence. In addition, the singular forms of some Tocharian B nouns do not match their respective plural forms, since they seem to be the outcome of different ProtoTocharian antecedents. This means that they cannot be reconstructed as mirroring the same PIE stem paradigm.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

215

In what follows, I will first focus on the previous diachronic explanations of TB -(a)una. I will then argue that this ending can be traced back to the n-form of heteroclitic stems in PIE *-ur/n-.285 Over the past decades, the origin of the plural morpheme TB -(a)una has been a major topic of debate. One of the most cohesive discussions is that of Hilmarsson (1988b). His basic claims are: (1) the ending *-una has been abstracted from the plural akruna ‘tears’, and (2) the ending *-auna is a conglomerate marker, formed by the collective formation in PT *-a and the new abstracted ending *-una. This proposal has to contend with some difficulties, however. First, some of the nouns with plural -auna attest a variant form -euna (sometimes spelled -ewna) in archaic texts. Examples are: krenteuna (B224 b1, B248 a2, B365 b4, krentewnaṣṣe B146 b8), paleuna (B248 b1), rekewna (THT1312 b6). This shows that the plural forms in -auna of classical Tocharian B — or at least a great deal of them  — are actually from older -euna (Peyrot 2008: 43). The second difficulty concerns the origin of the element *-una. Indeed, it is unlikely that the bulk of its spread lies in its abstraction from a single plural form, namely akruna ‘tears’, where, moreover, the na-element is taken as secondary too (see above). For these reasons, Hilmarsson’s proposal is to be rejected, all while acknowledging that he could not have known at the time that -euna and -auna were diachronic variants. Adams (1990) has dealt with the same topic. His main aim was to reconstruct hypothetical stems from which both the singular and the plural may have derived directly. However, his derivations are quite algebraic; he reconstructs chains of derivational morphemes containing the nasal suffix PIE *-h1en- as the last element. Furthermore, some of his explanations are phonetically dubious. As pointed out by Malzahn (2006: 400), the fact that the formations in -(a)una are somehow related to the Tocharian B singular forms in -(a)u seems obvious at first glance, but after a closer scrutiny this statement seems cryptic.286 Another explanation for TB -una ought to be found. I believe Hilmarsson was right in trying to find a way by which the element -una could have been abstracted and then generalised to other formations that are etymologically unrelated to this plural ending. On the other hand, the bulk of this spread cannot be sought in isolated words, but rather

285 Parts of this section appeared in Del Tomba (2020c). 286 Similarly, it is improbable that these nouns are the outcome of PIE *men-stems (as per teb § 106) and therefore need to be related with the Tocharian B nouns of the nāki-type (with singular ending -i and plural -nma, on which see Pinault 2008: 495–496).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

216

Chapter 3

in morphological formations where -una is an inherited morpheme. In what follows, I attempt to show that the marker PT *-una can be interpreted as the original plural ending of the heteroclitic paradigms in *-uer/n-. It has long been acknowledged that Tocharian inherited these PIE formations and that they were quite productive for a certain period. Pinault (2011a) has shown that the most productive type was derived with the suffix *-u̯ or > PT *-wer, a stem allomorph of the collectives in *-u̯ ōr (Pinault 2011a: 164, 2021: 117–118).287 This suffix became quite productive in Proto-Tocharian, where it was employed to form verbal abstracts (Malzahn 2014b: 265). Examples include (Pinault 2011a: 164): TB ārwer, ta ārwar ‘ready, willing’ < PT *arwer < PIE *h2er‘to fit’, TB malkwer ‘milk’ < *məlkwer < PIE *h2ml ̥ǵ- ‘to milk’, etc. In most of the cases, however, the outcome of PT *-wer has become synchronically opaque, as *-w- has been lost between vowels: TB yerter ‘wheelrim, fellow’ < PT *yertewer, TB rser ‘hate’ < PT *rəsewer, TB karyor, ta kuryar ‘commerce’ < *kwəryawer. In parallel to the formations in *-u̯ or, I believe there is evidence for claiming that Tocharian also inherited the regular paradigms in *-ur/n. Pinault (2011a: 164) claims that these formations were no longer productive in Tocharian, since they are limited to relics. From a comparative point of view, the best example is tb ṣñor ‘sinew’ (pl. TB ṣñaura), which has cognates in several Indo-European languages, like YAv. snāuuarə.bāzura- ‘having arms like sinews’, Ved. snā́van- ‘sinew’, a-snāvir-á- ‘having no sinews’, Gk. νεῦρον ‘string, sinew’, Lat. nervus ‘sinew, muscle, nerve’, Arm. neard-kʿ ‘sinew’. All these forms point to the reconstruction of a heteroclitic paradigm PIE *snéh1-ur/n-. The formal mismatch between the singular TB ṣñor and the plural TB ṣñaura gives cause for concern since they should be traced back to the same base PT *snewr-. It is generally assumed that the singular PIE *snéh1-ur > *snēu̯ r̥ developed differently, because the expected PT *ṣñewər (or the like) underwent some kind of contraction, yielding tb ṣñor (Þórhallsdóttir 1988: 199–200; Ringe 1996: 155–156).288 For instance, Hilmarsson (1985a, 1986d) argues that PT *ṣñewər first became *ṣñewur and then *ṣñowur (through u-umlaut) > tb ṣñor (either 287 The collective formation in *-u̯ ōr may have only continued in Tocharian in isolated forms (see § 3.6.3.4). 288 I could not find any convincing examples of a contraction of *-ewə- to *-o-. Ringe (1989) adduces the reduplicated preterite participle of root beginning with w-. For instance, he argues that TB ausu, ta wasu ‘having put of (clothing)’ (from TB wəs-, ta wäs- ‘to wear’) can ultimately be traced back to pre-PT *wewəs(ə)wu, which would have evolved according to the following path: *wewəs(ə)wu > PT *wos(ə)wə (> TA wasu) > pre-TB *wowsəw (reintroduction of -w-) > *owsəw > TB ausu. This reconstruction is quite cryptic and other solutions can be proposed. Indeed, TB ausu could reflect PT *we-wəs-u directly, through a development of PT *we to TB o, i.e. *we-wəs-u > *wewsu > *owsu > TB ausu (cf. 3sg.prt. TB otkasa < PT *wetksa from wotk- ‘to separate’; see Peyrot 2010b: 70–72, 2013: 530). On the

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

217

with contraction or with irregular reduction of *-owr to *-or). But this is an ad hoc reconstruction, which also requires a significant number of unattested intermediate stages. A different explanation ought to be found. Lubotsky (1994b) deals with the reconstruction of the PIE root *turḱ-, its outcome in the Indo-European languages (Av. ϑβōrəštar- ‘creator’, Ved. tváṣṭar-, the god-creator, Gk. σάρξ ‘meat’, OIrish torc ‘boar’, etc.), and related issues. One of these problems concerns the metathesis of PIE *CurC to *CruC (AiGr 1.206; Mayrhofer 1986: 161–163; Meier-Brügger 2003: 98; Byrd 2015: 142–143). After having scrutinised the data that may testify such a phonetic development, he concludes that “in PIE the metathesis *-ur- > *-ru- was phonetically regular in the final syllable only” (1994a: 191). I believe that Tocharian may bring new evidence in favour of this reconstruction. The plural TB ṣñaura ‘sinews’ can be traced back to *ṣñewra, which is from an older *ṣñewna with a generalisation of the r-stem, while the singular TB ṣñor ‘sinew’ is from *ṣñeru < *snēru < *snéh1-ur, through earlier metathesis of -ur# > -ru# and Tocharian u-umlaut of internal *-e-, which has been regularly modified to *-o-. In addition, there are a dozen nouns with plural ending TB -wa, ta -u (-wā, -unt), of which the majority can in my view be traced back to heteroclitic stems in PIE -ur/n-. These nouns have a singular in TAB -r and a plural in TB -rwa, ta -ru (-rwā, -ru-nt). Examples include: TB āmpär* ‘limb, member’ (pl. amparwa), TB kwarsär, ta kursär ‘mile, vehicle’ (pl. TB kwärsarwa ~ kursarwa, ta kursärwā ~ kurtsru), TB tarkär, ta tärkär ‘cloud’ (pl. TB tärkarwa, ta tärkrunt), TB yarpär ‘± enclosure’ (pl. yärparwa), TB tsaṅkär, ta tsäṅkär ‘top, summit’ (pl. TB tsäṅkarwa, ta tsäṅkrunt), etc. The morphological derivation of these nouns has not yet been clarified. Following Van Windekens (1944: 155–156, 1979: 15–16) and Isebaert (1980: 235, 2004: 395), Adams (1990, dtb s.v., 2015: 178) argues that they are old action nouns and verbal abstracts in *-r, which were extended with an u-suffix in the pre-history of Tocharian. The u-extension is assumed to explain the unexpected wa-plural.289 However, this explanation is difficult since it fails to identify a reason for the alleged spread of the neuter u-stems, which do not form a particularly productive category in Tocharian.290

other hand, TA wasu may be from *we-wes-u, as Michaël Peyrot (p.c.) suggested to me (cf. also Malzahn 2010: 248). See further Del Tomba (2020c: 55–57). 289 Cf. Adams (1990: 68): “These neuter r-stems were typically extended as neuter u-stems at some point in pre-Tocharian”. 290 Pinault (2008: 493) is of a slightly different opinion. He claims that the reanalysed plural PT *-wa of the old u-stems spread analogically to some stems and, in particular, to certain nomina actionis in *-l and *-r.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

218

Chapter 3

I believe that the derivational and inflectional issues related to these nouns can be solved by analysing them as old heteroclitic derivatives in *-ur/n-, which underwent the sound law *-ur > *-ru. That is to say, all original ur-forms of the paradigm underwent metathesis in the strong cases, becoming ru-stems.291 As far as the plural paradigm is concerned, all these nouns, including those derived with the suffix PT *-wer, have lost the archaic n-form in the plural and have generalised the r-stem, e.g. ṣñaura ‘sinews’, wmera ‘jewels’, tärkarwa ‘clouds’, amparwa ‘limbs’, pwāra ‘fires’, ysāra ‘blood (pl.)’, etc. The reason for this development is fairly easy to envision: the formal link between the r- and the n-stem became increasingly opaque in the pre-history of Tocharian. It follows that some of these nouns have been detached from the n-form of the plural, becoming either r-stems (pl. -ra) or ru-stems (pl. -rwa). Thus the n-plurals became easily abstracted and employed to mark the plural of other inherited formations. And these formations are in my view some of the nouns that synchronically attest the plural ending -(a)una. Let us now look at the diachronic evolution of these nouns within the framework established above, beginning with the nouns with the plural -una.292 The reconstruction of the PIE word for ‘tear’ is notoriously difficult, and the derivation of TB akrūna ‘tears’ is no exception.293 The most comprehensive study on this word is undoubtedly Pinault (1997a). Before his investigation, the 291 The loanwords assimilated to this class, i.e. TB kottär (pl. kottarwa), ta kotär ‘family, clan’ (from Skt. gotra-), TB cākkär (du. cakkarwi), ta cākkär ‘wheel, cakra’ (from Skt. cakra-), TB mittär* (du. mittarwi) ‘sun, mitra’ (from Skt. mitra-), TB yāntär (pl. yantarwa), ta yāntär ‘mechanism, tie’ (from Skt. yantra-), may be explained in the following terms: after the loss of final vowels, they became formally identical to indigenous nouns with singular -är /-ər/, plural -arwa /-ə́ rwa/. 292 Nouns with dubious etymology will not be considered. This is the case of TB lāṃs, ta wles ‘work’ and TB yapoy, ta ype ‘land’. The first noun is related to the homophonous verbal root TB lans-, ta wles- ‘to work on, perform’. Adams (dtb 594) takes the verb as a denominal formation. For an etymological suggestion, see Malzahn (2010: 834). The second noun has been the topic of controversial analyses which have been summarised and discussed by Hartmann (2013: 472–473). Although I am not convinced by the etymology given by Hilmarsson (1988b), I believe he was correct to link the evolution of TB yapoy, ta ype to that of TB soy ‘son’, ta se (see further Malzahn 2006: 402 and Blažek & Schwartz 2017: 49). As far as the plural form is concerned, it is possible that PT *yəpoy-wna evolved regularly in TB ypauna, after the loss of internal -y- (see the main text below). 293 Cf. Schulze (1927). In order to account for the initial *d- in some Indo-European forms (e.g. OIrish dér, Gk. δάκρυ, etc.), Kortlandt (1985a) claims that the archaic PIE *h2eḱruwas replaced by the compound *dr̥ḱ-h2eḱru- ‘eye-bitter’ in some languages (Greek, Latin, Celtic, Germanic, and Armenian). Following this reconstruction, de Vaan (2008: 322) tentatively reconstruct the plural of the second form as *dr̥ḱ-h2ekru-n-h2, continued in TB akrūna, OLat. dacruma, Gk. δάκρυμα. However, the classical theory that Lat. dacruma has been borrowed from Gk. δάκρυμα is probably to be preferred (Ernout & Meillet 1932: 336).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

219

stem akru° was considered to be the outcome of PIE *-u-h2 by Adams (1988a: 32) and Ringe (1996: 30).294 This explanation is contradicted by several examples of PIE neuter *u-stems, which have a plural ending TB -wa < PIE *-uh2 (e.g. TB ārwa ‘trees’ < PIE *d(o)ru-h2, TB ostūwa ~ ostwa ‘houses’ < PIE *u̯ eh2stu-h2).295 In Tocharian B, this noun is attested only in the plural; in Tocharian A, the singular TA ākär is also attested, alongside the late plural form ākrunt. In order to demonstrate that akru° does not represent the outcome of PIE *h2eḱruh2, Pinault (1997a: 224–225) notes that the singular TA ākär does not mean ‘tear’ but it has the collective meaning of ‘masse de larmes’. He therefore suggests that PT *akru- was the regular outcome of the collective PIE *h2eḱrōu̯ and that this form has been reinterpreted as the base of a new plural. This analysis has the advantage of not deriving PT *akrəw- from the plural PIE *h2eḱruh2, which one would rather expect to have yielded TB **akruwa. Pinault further argues that the plural endings TB -na and TA -nt appeared independently in the two Tocharian languages, i.e. when they had already split off from Proto-Tocharian. However, it is also possible that the ending *-na had already been added in a Proto-Tocharian stage: on the one hand, Tocharian B maintained the plural form *akruna unchanged, while, on the other hand, Tocharian A extended the apocopated form *ākrun to ākrunt (as for e.g. *wakmna > pre-TA *wākmän >> TA wākmant ‘separations’). Although this explanation is certainly possible, other Indo-European continuants of the word for ‘tear’ clearly point to the reconstruction of a heteroclitic *ur/n-stem.296 If we reconstruct this heteroclitic paradigm for pre-Tocharian, then the plural TB akruna, ta ākrunt may attest an important archaism: an original paradigm containing *akuna as a pre-Tocharian replacement of the inherited collective formation was levelled as an r-stem and the ending -una was blended in.297 On the other hand, the singular PIE *h2eḱ-ur underwent 294 See Kim (2018a: 98–100). 295 In order to solve these problems, Ringe (1996: 31) claims that the final -a of wa-plurals was introduced analogically after the alleged outcome of PIE *-eh2 > PT *-a. However, as we will see in the following sections, PIE *-eh2 yielded TB -o even in word-final position. Furthermore, since all other a-plurals continue PIE *-h2, it is preferable to say that PIE *-uh2 yielded PT *-wa > TB -wa, TA -u. 296 See the discussions in Hamp (1959, 1972), Eichner apud Mayrhofer (1986: 162), Matasović (2004: 87), Kloekhorst (2008: 391, 2011: 268), Kroonen (2013: 504–505), and Byrd (2015: 143). The fact that some other Indo-European languages point to the reconstruction of a u-stem may equally be interpreted as caused by the metathesis of *-ur-> *-ru (as if, in Ved. áśru-, Gk. δάκρυ, OIrish dér, etc.). 297 Judging from the Hittite data (with residues in Vedic and Old Avestan, cf. aiiārə̄ ‘days’), heteroclitic nouns formed the nom.acc.pl. on the r-stem in PIE (see Nussbaum 2014a: 300–301). However, several Indo-European languages have reshaped the nom.acc.pl. on

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

220

Chapter 3

metathesis *-ur > -ru, yielding pre-PT *akru > PT *akrə > TB ākär* /ákər/, ta ākär ‘tear’.298 Table 33

Evolution of the word for ‘tear’ in Tocharian

PIE strong stem weak stem

> *akru * h2éḱu̯ r̥ * h2eḱu̯ én- > *aku̯ én-

sg. pl.

pre-PT

PT

TB

> *akru >> *akuna

> *akrə > ākär* >> *akrəwna > akrūna

TA ākär ākrun-t

The etymology of TB ṣotri ‘sign, mark’ (pl. ṣotrūna, du. ṣotrūni) is unclear. The most recent attempt has been made by Adams (1990: 65). However, his reconstruction has some difficulties. He posits a vr̥ddhi formation in -r to PIE *su̯ edh- ‘to custom’, which, in the history of Tocharian, would have become a u-stem and would then have been recharacterised by a nasal suffix. The final proto-form would have been *su̯ ēdh-r-u-h1en-, which is extremely cryptic. The reconstruction of TB ṣotri is complicated by the derivative TB ṣotarye ‘signal, remarkable’ (PK DA M 507.32 a5 and a8) and the variant plural ṣotarnma (AS3B a1). These forms might point to the reconstruction of a parallel singular ṣotär*. If this singular form is original, then we can reconstruct a Proto-Tocharian paradigm with sg. *ṣotrə, pl. *ṣotrəwna, which morphologically matches sg. *akrə, pl. *akrəwna. At an older stage, this noun would have been inflected as *ṣotru in the singular and *ṣotuna in the plural. Later, the r-stem would have been generalised, resulting in the blended plural ṣotr-una. On the other hand, the singular PT *ṣotr-ey > TB ṣotri, ta ṣotre would have been analogically created on the model of TB reki (ta rake) ‘word’, pl. rekauna (on which see below). From a formal point of view, PT *ṣotər can be derived from PIE *seHdh- ‘to achieve a goal’ as follows: *seHdh-ur > *sēdh-ru > *ṣetru > *ṣotru (u-umlaut) > TAB ṣotr- (on the semantic side, ‘goal’ → ‘target’ → ‘mark’).299 the basis of the n-stem, cf. Ved. áhāni from áhar/n- ‘day’, OLat. femina from femur, feminis ‘thigh’, OAv. sāxvə̄nī ‘teachings’ (de Vaan 2003: 138), Gk. ἥπατα from ἧπαρ ‘liver’ (cf. also Cantera 2009: 21 fn. 9 on Middle Persian). Further evidence that the same replacement took place in Tocharian are dealt with in § 3.6.3.5. 298 I do not believe that PT *akər can come from *h2éḱ-ur directly, because the sequence *-ḱu(or *-ḱw-) is expected to give PT *-kw- (cf. PIE *h1éḱu̯ o- ‘horse’ > PT *yəkwe > TB yakwe; PIE *h2eḱuti̯o- > PT *akwǝtse > TB akwatse ‘sharp’, Kim 1999: 162). 299 Cf. Rix (1985) and de Vaan (2008: 562–563). Dedè (2013: 122–125) reconstructs *séHdh-ur as the antecedent of Gk. εἶθαρ ‘immediately’. For yet another suggestion, see Malzahn (2006: 402–403). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

221

All other nouns to be discussed attest a plural formation in -ewna / -auna. Among the pluralia tantum, TB palauna ‘praise’ and TB tarśauna300 ‘deception’ are action nouns derived from the subj. stem of TB pǝla- ‘to praise’ and the poorly attested verbal root TB tǝrk- ‘to wind’, ta träk- ‘to lose (consciousness)’ respectively.301 Although their exact derivation is not clear,302 the plural form -auna is of Proto-Tocharian origin, as demonstrated by the Tocharian A correspondents tārśoṃ ‘deception’ and TA paloṃ ‘praise’ (cf. the plural palonās and the adjective paloṃṣi), synchronically singular.303 In Tocharian A, the two terms have been reinterpreted as singular, due to the singulative meaning of the plural formation, which is still attested in Tocharian B. In Tocharian B, a parallel case is kerekauna ‘flood’ (= Skt. ogha- ‘torrent, flood’), which is also morphologically singular. According to Pinault (2001: 99) and Hilmarsson (1996: 132–133), TB kerekauna derives from a thematisation of the PIE root *gwerh3- ‘to devour’, enlarged with *-k-. In fact, in many Indo-European languages, this root appears in reduplicated nominal forms or in derivatives formed with a *k-suffix (cf. Skt. gárgara- ‘whirlpool’, MP galōg ⟨glwkˈ⟩ (Phl.) ‘throat’, Gk. γόργυρα ‘underground drain’, Lat. gurges ‘whirlpool’, Lat. vorāx, -cis, Lat. vorāgō, etc.). Hilmarsson (1996: 133) reconstructs a formation *gworh3o-ko- ‘devouring’ > PT *kereke-, to which the collective ending TB -una has been added. This formation would have regularly developed PT *kerekewna ‘violent stream’ > TB *kerekewna > kerekauna. Another Tocharian B plurale tantum that can be included in this class has no Tocharian A correspondent. It is TB krentauna ‘virtue(s)’, which evidently derives from the synchronically suppletive adjective TB kartse, obl.sg.m. krent ‘good’. Hilmarsson (1988b: 36–37) reconstructs a neuter plural *krenta enlarged with *-una. As pointed out above, however, the only problem with this reconstruction is that we find the spelling krentewna in archaic texts and this form 300 Adams (dtb 303) reconstructs a singular tārśi* on the basis of the dubious adjective TB tārśī(cce) in B133 b5. The form tarśītse in B255 a4 may be interpreted as either a nom.sg. of the same adjective or a genitive singular (for tarśītse). On the other hand, an obl.sg. tārśai seems to be attested in B496 a4, which makes the reconstruction of the singular paradigm difficult. As pointed out by Hannes A. Fellner (apud CEToM s. THT 496), tarśauna is expected to have a singular tārśi*, while the obl.sg. tārśai points to a nom.sg. tarśiye*. Following Pinault (2015c: 213), I assume that the development of the singular paradigm is a Tocharian B innovation, and that in Proto-Tocharian this noun was a plurale tantum. See also Malzahn (2006: 400–401). 301 Van Windekens (1979: 197) suggested that the ending -auna should be segmented as -au-na, where -au- is the marker of past participles. He therefore assumed that the ending -auna in tarśauna and palauna was original. For criticism, see Hilmarsson (1988b: 35). 302 See Malzahn (2006: 401–402) for recent proposals. 303 For the mismatching root vocalism between TB palauna and TA paloṃ, see Malzahn (2006: 401–402). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

222

Chapter 3

cannot be the regular outcome of PT *krentawna. However, the absence of any krente- among the case forms of kartse is striking. Furthermore, the derivatives of this adjective took their base from kartse (cf. the ṣṣe-adjective TB kärtseṣṣe ‘pertaining to the good’; the abstract kärtsauñe ‘goodness, virtue, service’). It follows that TB krentauna should be interpreted as an old derived form (perhaps from an abstract noun PT *krentey ‘goodness’, see below).304 We are thus left with two nouns with the deviant singular ending TB -i, i.e. TB reki ‘word’ (TA rake) and TB ṣewi ‘pretext’ (without equivalent in Tocharian A).305 In the first noun, the vocalism of the stem may derive from either PIE *-oor *-ē-, but the palatalisation in ṣewi points unambiguously to PIE *-ē-. On the other hand, the matching TB -i : TA -e must reflect PT *-ey, the outcome of an *oi̯-stem (Ringe 1996: 82–83). This reconstruction follows Klingenschmitt (1994: 400), who argues that TB reki, ta rake ‘word’ is from PIE *rok-oi̯ or *rēk-oi̯ (cf. OCS rěčь < *rēki-) > PT *rekey. The morpheme *-oi̯ represents either the generalised allomorph of the collectives in *-ōi̯ < *-oi̯-h2 of the other Indo-European languages (Pinault 2021: 117) or the outcome of a neuter counterpart of the type Ved. sákhā (< PIE *sékw-ō(i̯) m.), Gk. πειθώ (< PIE *bhei̯dh-ō(i̯) f.), Hitt. zaḫḫai‘fight’ (< PIE *ti̯eh2-oi̯-, commune), Hitt. ḫaštāi- (< *h2/3ést-h2-ōi nt.).306 According to Klingenschmitt, the plural ending of these Tocharian nouns should have been -ōi̯, but long diphthongs are thought to have lost the semivowel in absolute final position as early as the proto-language (Gk. πειθώ < PIE *bhei̯dh-ō(i̯), Ved. sákhā < PIE *sekwh2-ō(i̯)). In any case, neither pl. *-ōi̯ nor *-ō have been continued in Tocharian, as these nouns have a plural -euna > -auna, whose origin must therefore be sought in other formations. A possible solution is that the abstracted plural PT *-una has been added to the outcome of the form continuing *-oi̯ in order to recharacterise the plural form of these nouns. We can therefore outline the following development: *-ey-una (or -ey-wna) > *-ewna (loss of *-y-) > TB -euna > -auna.307 304 Malzahn (2006: 400) reconstructs an original derivative in *-ur/n for this noun, but she does not specify on what basis TB krenteuna was constructed. 305 TB ṣewi is the only member of this class that seems to be feminine (cf. B109 a6 [late] yalñeṣṣai ṣewisa). According to Sieg & Siegling (1953: 210), masculine agreement may be found in B325 a5 (alye)k ṣewisa, but it is conjectural. 306 See Ringe (1996: 83–84) and Del Tomba (2020b: 28–29). For the source of Hitt. ḫaštāi-, see Oettinger (2016: 322); for neuter i-stems in Hittite, see Melchert (2021). 307 See also Malzahn (2012c: 179). As far as Tocharian A is concerned, we can assume two different developments. If Tocharian A never had this ending, then the plural *-una originated in a pre-Tocharian B stage. However, if its spread took place in (pre-)Proto-Tocharian, then Tocharian A lost this ending and further rebuilt the plural with the productive ending -nt(u). The second hypothesis seems preferable, because the spread of *-una can be reconstructed for a pre-Proto-Tocharian stage.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

223

Gender in the Inflection of the Noun

3.6.3.3 A Note on TB śaumo ‘man, person’ So far, we have seen that the Proto-Tocharian ending *-na has various sources. What is quite uniform, however, is the meaning of these formations. Indeed, a relatively large group of Tocharian B pluralia tantum with this ending has a clear singulative meaning, mostly uncountable. This Proto-Tocharian value of *-na may also account for its attestation in the plural of TB śaumo ‘person, man’. The etymology of this word is clear: it is an original deverbal adjective in -mo from the ancestor of TB śaw- ‘to live’ < PIE *gwih3-u- (liv 2 215–216). The singular inflection (nom. śaumo, obl. śaumoṃ) is exactly the same as the adjective klyomo ‘noble’. On the other hand, the deviant plural śāmna (with reduction of pre-TB *-aw- before consonant clusters; cf. also TB śāmñe ‘human’, Lane 1938: 26) runs counter to the expected form nom.pl. **śaumoñ (cf. nom.pl. klyo­ moñ). Other substantivised adjectives in -mo also have a differentiated plural paradigm nom. -oñ, obl. -oṃ (e.g. TB wāṣmo ‘friend’, nom. pl. wāṣmoñ, obl. pl. wāṣmoṃ). However, one should note that the plural TB śāmna very rarely means ‘men (i.e. male people)’, since in the great majority of the attestations it must be translated with ‘people, mankind; relatives’ (e.g. B3 b3–4 [Udānālaṅkāra] śaul attsaik totka śāmnaṃts ñke wrīyeṣṣe pältakwä atyaṃts a[b4](k)entasa “the life of humans is now only short (as) a drop of dew on the tips of grasses”, cf. Peyrot 2016b: 204). Furthermore, as pointed out by Adams (dtb 698), TB śaumo is often used to designate humans as opposed to deities (e.g. the merism ‘men and gods’ in B30 b8 [Udānālaṅkāra] tu yparwe ñakti śāmna tsälpāre piś toṃ cmelameṃ “Thereupon gods and men were freed from the five rebirths”, cf. Zimmer 1976: 77). Thus in this noun the plural ending -na also conveys a collective meaning. As for its origin, it seems that before the loss of the neuter as a category of target gender in the adjectival inflection, the historical outcome of the neuter plural *-mna < *-mnh2 started to serve as the plural of śaumo ‘man’, conveying the collective meaning of ‘humankind’. This reanalysis may have occurred when the masculine and the neuter already merged morphophonologically in the singular, but the neuter was still differentiated from both the masculine and the feminine in the plural. 3.6.3.4 Distribution and Evolution of TA -äṃ The plural ending TA -äṃ, corresponding phonetically to TB -na, is not productive, since it is confined to only five substantives.308 As can be seen from the table below, the cognate nouns in Tocharian A and B belong to different inflectional classes. 308 Parts of this section appeared in Del Tomba (2019).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

224

Chapter 3

Table 34

Tocharian A nouns with plural -äṃ and their Tocharian B correspondents

tocharian a sg.

pl.

por ‘fire’ ysār ‘blood’ ytār ‘road’ wram ‘thing’

poräṃ ysāräṃ ytāräṃ wramäṃ

plāc ‘word’

plācäṃ

class

tocharian b sg.

pl.

ii.1 ii.1 ii.1 ii.1

puwar ‘id.’ yasar ‘id.’ ytārye ‘id’. wreme ‘object’

ii.1

plāce ‘id.’

pwāra ysāra ytariṃ (obl.) — plāci (nom.) plātäṃ (obl.)

class

i.1 i.1 vi.1 ? v.2

Of the five Tocharian A nouns, three are of alternating gender (TA por, TA ysār and TA wram), and two are of feminine gender (TA ytār and TA plāc). The core issue is which of the two languages preserves the older state of affairs. The present section aims to answer this question, analysing the synchronic distribution and the diachronic evolution of this ending in Tocharian. I intend to show that Tocharian A has generally preserved the original situation, while Tocharian B has mostly recharacterised the plural form of these nouns. If my analysis is correct, it also confirms that this inflectional class is relevant to the reconstruction and the further development of an archaic Proto-Indo-European class of nouns: the *r/n-heteroclites. Three of the five Tocharian substantives that belong to class ii.1 can be traced back to PIE heteroclites. They are: TA por, tb puwar ‘fire’, TA ysār, tb yasar ‘blood’, and TA ytār, tb ytārye ‘road’. That these nouns reflect PIE *r/n-stems was first noted decades ago, but the relevance of this fact to their plural formation has not, to my knowledge, been explicitly acknowledged.309 TA ytār, tb ytārye ‘road’ Let us begin our discussion with TA ytār, tb ytārye /y(ə)tárye/ ‘road, street, path’, both of feminine gender. These words must be compared with Lat. iter, gen. itineris, and the derivative YAv. pairiθna ‘the course of life’ (Yt 8.54, Panaino

309 The connection of these Tocharian nouns with the PIE *r/n-heteroclites had already been proposed in the past decades by leading scholars, like Petersen (1939: 75), Van Windekens (1944: 79–81), and Hilmarsson (1984b) but their treatments are in many points different from mine.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

225

1990: 141).310 The PIE form from which these nouns derive is usually reconstructed as *h1éi̯tr̥ , *h1itén- (from PIE *h1ei̯- ‘to go’, liv 2 232–233), although evidence for the full grade *h1éi̯tr̥ is meagre. A closer look at the Tocharian words reveals some issues to be discussed. To begin with, the a-vocalism does not represent the expected outcome of PIE *h1éit-r. This means that Tocharian continues a different formation, which can be traced back to the collective PIE *h1itōr (Hilmarsson 1986a: 44; Pinault 2011a: 163–164; dtb 559; Kim 2019b: 145). Kortlandt (1988a: 84–85) is against this derivation, preferring to postulate analogy after TA ysār, tb yasar ‘blood’.311 While this solution is certainly not unthinkable, analogy is in my view unnecessary here, because we can easily reconstruct a morphologically plausible ancestor from which the Tocharian words may derive.312 The unexpected feminine gender in both Tocharian A and B, and the element -ye /-(ə)ye/ in Tocharian B are problematic. Hartmann (2013: 470–472 and 519–520) has recently collected and commented on the previous interpretations of these problems. He posited PIE *h1itōr-ih2 or *h1itōr-ēn as the potential virtual ancestors of TB ytārye. The first reconstruction follows Klingenschmitt (1994: 396 fn. 140), who argued that both TB ytārye and TA ytār ́ evidence a recharacterised collective formation by means of the vr̥kī-suffix. The second reconstruction follows Hilmarsson (1987a: 47–48), who argued that a conflation of the r- and the n-stem took place in Proto-Tocharian, in such a way that a new form *itōr-en- was created from *itōr. The nominative singular of this preform should have been *itōr-ēn, which in turn became *yətarəye > TB ytariye ~ ytārye. Hartmann favours the first hypothesis, while Malzahn (2014c: 198) prefers the second. Both theories face difficulties, however. The first reconstruction is unsatisfactory from a phonological point of view, because PIE *-ih2 should have evolved into TB -(i)ya, ta -i, thus TB **ytār(i)ya, ta **ytāri. The fact that PIE *-h2 310 The oft-cited Hitt. †itar (alleged hapax legomenon in KUB 41.8 i 20, cf. Rieken 1999: 374–377; Kloekhorst 2008: 422) has been read by Miller (2008: 209 fn. 97) as DUMU-tar ‘offspring’. 311 In more recent times, Kortlandt (2020: 116) has claimed that TA ytār may represent *h1ei̯tr̥ “with pre-apocope lowering of *ä to *a before *-r”. I fail to understand what kind of apocope is involved here. Furthermore, Kortlandt does not provide further evidence for an alleged lowering of *ə before *-r. 312 One might object that, from the semantic point of view, the assumption of an original collective *h1itōr is difficult, as neither TA ytār nor TB ytārye denotes a multitude of streets, nor can it be proven that they did so at an earlier stage. Nussbaum (2014a: 251) points out this problem and convincingly suggests that this (morphological) collective formation has an “instantial” value, i.e. that it “denotes […] an individual instance of an action, event, or state” (p. 247), as in Gk. τέρμα ‘crossing’ < *tér(h2)-mn̥ vs. τέρμων ‘a boundary’ < *tér(h2)-mō(n).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

226

Chapter 3

yielded PT *-a > TB -a, and never PT *-e > TB -e (as per Hartmann 2013: 470) is corroborated by unambiguous examples (see further § 4.8.2.3). The second solution presents no difficulties from a phonological point of view (cf. TB yriye ‘lamb’ < PIE *u̯ erh1-ēn, Pinault 1997b: 185–187), but it faces chronological and morphological problems. Indeed, it implies a possibly circular development according to which an original *yətar, the regular outcome of PIE *h1itṓr, first became *yətarəye (continued without modifications in TB ytārye) and then *yətar > ytār in Tocharian A, according to the model of TA ysār ‘blood’. As per the other heteroclites, this noun should be reconstructed as neuter in Proto-Indo-European. It follows that the feminine gender of TA ytār, tb ytārye must be secondary, because PIE neuter nouns are usually continued as alternating in Tocharian. In my opinion, in the Proto-Tocharian phase, this substantive was influenced by the ancestor of the productive feminine nouns TB kälymiye, ta kälyme ‘direction, region’ because of its meaning, so that PT *yətar initially acquired feminine gender. Since the gender of TA kälyme also fluctuates between alternating and feminine (DTA 141; Peyrot 2012a: 212), one might assume a case of mutual influence. Subsequently, after the dissolution of Proto-Tocharian, it shifted inflectional class in Tocharian B, becoming a noun of the kälymiye-type.313 TA ysār, tb yasar ‘blood’ The second noun to be discussed is TA ysār, tb yasar /yə́ sar/ ‘blood’. It has cognate forms in several Indo-European languages, including Hitt. ēšḫar, gen. išḫanāš, Skt. ásr̥-k, gen. asnáḥ, Gk. ἔαρ ~ ἦαρ,314 Latv. asinis, OLat. as(s)yr (Paul. Fest. 12. 19; cf. also aser in CGL 2.23, 56 and the derivative OLat. assarātum, a kind of “bloody” drink, de Vaan 2008: 58), perhaps Lat. sanguen (Ennius, Ann. 108) ~ sanguīs, Arm. ariwn, etc. These forms may allow us to posit PIE *h1ésh2-r, *h1sh2-én-. The Tocharian words can easily be derived from this proto-form 313 A similar analysis has been proposed by Pinault (2015b). Malzahn (2014a: 200) tentatively tries to analyse the irregular feminine gender of these nouns as an archaism, by comparing it with Homeric Gk. ἐέλδωρ ‘desire, wish’, of unexpected feminine gender (see also Leukart 1987: 355). In parallel, Nussbaum (2014b: 253) also claims that there is no reason not to interpret the feminine gender of this noun as original, because the other continuants in *-ōr inherited by Tocharian are alternating. However, this statement can also be read the other way around: since the other continuants of *-ōr are alternating in Tocharian, *h1itṓr should originally have been neuter too and is thus expected to evolve as an alternating noun. 314 Gk. ἔαρ is unattested before the Hellenistic period. In the glosses by the fifth-century CE grammarian Hesychius we find both variants: ἦαρ · αἷµα . ψυχή (Hsch. sub ἤ-8) and ἔαρ · αἷμα . Κύπριοι (Hsch. sub ε-31).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

227

(Kortlandt 2010: 146). Otherwise, they may also be the outcome of the collective *h1ésh2ōr (Hilmarsson 1986a: 22; Pinault 2011a: 163; dtb 525). TA por, tb puwar ‘fire’ The last noun that can be traced back to a PIE heteroclitic stem is TA por, TB puwar ‘fire’.315 Cognates of these words are found in most Indo-European languages. Among these, Hitt. paḫḫur (gen. sg. paḫḫuenaš) continued the proterodynamic inflection almost intact and thus provides substantial evidence for reconstructing the heteroclitic paradigm as PIE *péh2-ur, *ph2-uén- (Kloekhorst 2013: 111). Other cognates include: Gk. πῦρ < *pūr, gen. πῠρός, Arm. howr < *pūr (Olsen 1999: 94), Umbr. pir < *pūr (cf. acc.sg. sim ‘pig’ < *suH-m), abl. pure < *pŭr-ed, Goth. fon, gen. funins, OHG fuir, ON fúr < *pūr (Simms 2009), Czech pyř ‘burning ash’ (Machek 1957: 502). There is no doubt that both TA por and TB puwar ‘fire’ are somehow linked to these formations. However, the exact apophonic grade and morphological formation from which they descend are notoriously problematic, since the phonological comparison between TA -o- and TB /-əwa-/ is awkward and complicates the Proto-Tocharian reconstruction. Winter (1965: 192–193) was the first to claim that Tocharian A and B point to different preforms: TA por would continue PIE *péh2-ur, while TB puwar would be from PIE *puh2-r. Other scholars propose that the word for ‘fire’ retained both simple and collective stems in Proto-Tocharian: Tocharian A would continue the former, Tocharian B the latter. This reconstruction is followed by Van Windekens (1976: 383) and Adams (dtb 421–422), and it has been recently advocated by Kim (2019b: 145). However, multiplying the number of proto-forms that cannot belong to the same morphological paradigm is questionable. Indeed, if Tocharian inherited both the regular and the collective formation of this noun, it is not clear what the distribution of these variants in the Proto-Tocharian paradigm might have been. It is probable that one of the two paradigms had been generalised before the breakup of Proto-Tocharian. 315 Hartmann (2013: 402–403) suggests that TB pūwar is in the process of changing grammatical gender (“Genuswechsel”). While masculine agreement in the singular is expected for both a masculine and an alternating noun, Hartmann points to two occurrences of masculine agreement in the plural, which may speak in favour of a gender shift (alternating → masculine). However, the texts where the relevant forms occur might be in verse, where masculine agreement in the plural is frequent (cf.  §  4.6.1): (1) IT201 a6 s(ru)kalñeṣṣeṃ pwārasa “through the fires of death” may correspond to a 4¦3 segment, just like the grammatically expected ᛫ tremeṣṣana pwārasa “through the fires of anger” (4¦3) in line a5; (2) THT1345.b b1 eṅkalṣeṃ pwāra “the fires of passion” may also be in a metrical passage (cf. the strophe number 10 in a1 and the form krento with o-mobile in a3). Therefore, evidence in favour the gender shift of TB pūwar is doubtful.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

228

Chapter 3

In an attempt to trace TA por and TB puwar back to a single preform, Hilmarsson (1985a: 42–43, 1989a: 135) argued that a collective *ph2u̯ ōr may have evolved in Proto-Tocharian as *pəwar and then TB puwar and TA por. A similar reconstruction has been supported by Ringe (1996: 17–18) and Hackstein (2017: 1314). In my view, there are two problems with this theory. The first is the outcome of the laryngeal. I expect PT *pawar > TB **pāwar /páwar/ as the regular outcome of PIE *ph2u̯ ōr. Ringe points to this problem and hesitantly argues that in a sequence *CHuV, the laryngeal evolved into *ə rather than *a. This sound change is difficult to evaluate, since it is not falsifiable. There is indeed no other clear parallel that can prove this evolution.316 However, PT *p(ə)war can be an expected outcome of the zero grade *puh2r, and it is therefore much more economical to start with this proto-form. Still, a more serious problem is the alleged contraction PT *-əwa- > TA -o-, because it again lacks any immediate parallel.317 In what follows, I base my reasoning on direct and indirect evidence in order to determine whether this sound change can be established or not. As we will see, however, the overall picture is still unclear. Let us look first at other potential outcomes of PIE *-uh2-. I have found the following clear examples: (1) PIE *suh2d-ro- ‘sweet’ (Gk. ἡδύς, Skt. svādú-) > PT *sware > TA swār, tb swāre; (2) PIE *uh2g- (liv 2 664–665) > PT *wak-a- > TB waka- ‘to split, flourish’, ta wākā- ‘to burst’. Other examples of the correspondence TB -wa- : TA -wā- are: (1) TA swāñceṃ, TB swāñco (obl.) ‘ray of light’ (to be linked in some way with the n-stem of PIE *séh2-ul / -u̯ én- ‘sun’) and (2) the dual TA pärwāṃ, TB pärwāne, from PIE *h3bhruH- ‘eyebrow’ (Gk. ὀφρῦς, Skt. bhrū́-). These examples evidently go against the proposed sound law, but they are still not conclusive. Indeed, TB pūwar may inform us about the original accentuation of this word, which should have been stressed on the shwa in Proto-Tocharian, thus /pə́war/. Some other indirect evidence may be adduced. Hilmarsson (1989a: 135, 1996: 187) saw a similar development in the oblique singular of the Tocharian A word for ‘dog’, which is TA koṃ (attested once in A360 a9), tb kweṃ. Both of these oblique forms are the outcome of PT *kwen(ə) < PIE *ḱu̯ on-m̥ . The development 316 The only parallel adduced by Ringe (1996: 18–19) is TB skiyo, which he traces back to PIE *sḱh2ieh2-. He attributed the lack of palatalisation in this word to an irregular development of the first laryngeal that yielded “some nonfront segment” in Proto-Tocharian (p. 19). However, the evolution of this term is even more complex than that seen in the word for ‘fire’. As a consequence, I do not think that it can be used as a solid argument in favour of the sound law *CHuV/*CHiV > *CəwV/*CəyV. See further § 3.5.2.1. 317 Hilmarsson (1989a: 135) hesitantly proposed that PT *pəwar became *powar in PreTocharian A, via umlaut. However, there is no evidence that u-umlaut operated in Tocharian A after the Proto-Tocharian period. See Burlak & Itkin (2003).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

229

shown by this inflected form is quite isolated, also because Proto-Tocharian labiovelars are expected to lose the labial element before PT *e < PIE *o (e.g. *kwólo- ‘± turning’ > PT *kele > TB kele ‘navel’; PIE *ǵhu̯ ono- ‘sound’ > PT *kene > TB kene, ta kaṃ ‘melody’). It is therefore possible that the labiovelar was reintroduced analogically after the nominative at some stage or that a Lindeman’s variant *ḱuu̯ - was generalised throughout the paradigm. Another parallel might be TA pl.ipv. plos for the expected *pälwäs (< *pələwasa), perhaps showing the same alleged contraction as TA por < *pəwar (Peyrot 2012a: 210, 2013a: 171 fn. 178). However, an analogical development after the singular TA plo* cannot be excluded, and it is even likely in view of the variant plamäs for the regular pl.ipv. pälmäs and the lack of root-final -ā in the Tocharian A pl.ipv. (Peyrot 2013a: 171 fn. 178). A last indirect parallel of the sound law PT *-əwa- > TA -omay be found in the evolution PT *-əye > TA -e-, which has quite a number of comparable items (see the previous section on TA ytār, tb ytāriye). All things considered, I believe that this sound law cannot be established with confidence, since other parallels (if any) still need to be found. However, in light of the data presented, one might say that the disyllabic sequence PT *ə́wa- became TA -o- if the first syllable was accented and the entire sequence came to occur in a closed syllable. If one is not inclined to accept this sound law, two last possibilities can be ventured. As hinted in § 3.6.3.2 and more thoroughly discussed in Del Tomba (2020c), I expect that if Tocharian inherited the proterodynamic paradigm of PIE *péh2-ur/n ‘fire’ then the strong cases underwent metathesis of *-ur > *-ru. Conversely, the weak stem regularly evolved into *ph2un-V́ - > *puh2n-V́ -. These proto-forms should have yielded PT *porə and *pwan-, which can account for both Tocharian forms. Accordingly, Tocharian A would have continued the strong stem *poru > PT *porə > TA por, while Tocharian B would have continued the weak stem PT *pwan- > *pəwan- (ə-epenthesis) >> pre-TB *pəwar > TB puwar (see also Schindler 1967: 242–243). Otherwise, if Tocharian inherited a double zero grade form *puh2r (from an older *ph2ur), the reverse development would have occurred. As a matter of fact, this case would not be isolated in the Tocharian nominal lexicon, as there are other — admittedly rare — cases where the two Tocharian languages have continued outcomes of different apophonic grades of one single paradigm. Clear examples in this sense are TA tsar, TB ṣar ‘hand’, which point to different inflected forms of PIE *ǵhesr- ‘hand’ (for explanations, see Schindler 1967: 244–245; Pinault 2006a: 80–81; Kim 2009b: 112 fn. 4; dtb 711).318 318 Other cases of formally different inflected forms due to either regular or contextual phonological developments can be found, mostly in Tocharian B: TB sg. āyo, pl. āsta ‘bone’ (cf.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

230

Chapter 3

One might think that the paradigm was already levelled as a r-stem in Proto-Tocharian. However, compelling evidence that Proto-Tocharian still preserved n-forms comes from Tocharian A, as I will show below (§ 3.6.3.5). TA wram (tb wreme) ‘thing, object’ The two last substantives that belong to class ii.1 are TA wram (tb wreme) ‘thing, object, matter’ and TA plāc, tb plāce ‘word’. They cannot derive from heteroclitic stems. From a synchronic point of view, TA wram is well attested, while TB wreme occurs twice in B197 (= Skt. viṣaya). This fragment is part of a Sanskrit Tocharian bilingual manuscript dealing with matters of Abhidharma. The Sanskrit parts are quotes from the Abhidharmāvatāra-prakaraṇa (Kudara 1974; Catt 2016). A recent study by Catt, Huard & Inaba (2020) has demonstrated that the apparent non-initial accent in the alleged compound käkse-wreme (B197 b4 and b5) is actually due to a mistake by the copyist, who incorrectly wrote the first akṣara of the adjective wakse*, a hapax legomenon that the authors convincingly translate as ‘cool’. Note that the usual Tocharian B noun for ‘thing, object’ is TB wäntare, which is not etymologically related to TA wram. However, since the gender of TB wreme is unknown and it is attested only in the nominative singular, we are not able to determine to which class it belongs. Indeed, TB -e is the nom.sg. of several Tocharian B inflectional classes, among which the most productive is class v.1 (continuing old thematic stems). For this reason, the authors of the Elementarbuch sorted this noun into this class. Following Van Windekens (1976: 580–581), TA wram is the exact cognate of Gk. ῥῆμα, -ατος ‘statement, word’, since both Greek and Tocharian A point to an action noun PIE *u̯ réh1-mn̥ . This etymology is supported by the plural form TA wramäṃ.319

TA āy, pl. āyäntu); TB or ‘wood’, pl. ārwa (due to different kinds of umlaut); TB ṣñor ‘sinew’, pl. ṣñaura. 319 I see no strong reason to reconstruct either Pinault’s *u̯ r̥h1-o-mo- (2008: 512) or Adams’ *u̯ rē-mēn- (dtb 672). Although these preforms have the advantage of deriving both Tocharian A and B words from a common ancestor, the former does not take into account the unproductive plural ending TA -äṃ (showing, say, the “morphologia difficilior”), while the latter requires the reconstruction of a lengthened grade in both the root and the suffix. Furthermore, on the basis of TB kälymiye, ta kälyme < PIE *ḱli-mēn, we expect reconstructed *u̯ rē-mēn to have evolved into TB **wremiye, ta **wrame.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

231

TA plāc, tb plāce ‘word’ The last noun to be discussed is TA plāc, tb plāce ‘word’. Among the five nouns with plural TA -äṃ, it is the only case where Tocharian B has the more archaic inflection, while Tocharian A has replaced the plural form. In the following, I will therefore refer more to Tocharian B than Tocharian A. An etymological connection with the verbal root TB pǝla-, ta pälā- ‘to praise’ is obvious. This verb is the outcome of either PIE *(s)pelH- ‘to proclaim, speak solemnly’ (cf. Gk. ἀπειλέω ‘to threaten’, Pinault 2008: 345; liv 2 576), or *bhelh1‘to yell, roar’ (cf. OHG bellan ‘to bark’, Klingenschmitt 1994: 127; dtb 403; liv 2 74), although the meaning of the Tocharian verb favours the first derivation. It is usually assumed that our noun is an old ti-derivative of this verbal root.320 From an inflectional point of view, TB plāce belongs to an unproductive class (class v.2, cf. teb § 183), whose few members display nom. sg. -e after a palatalised consonant, truncation of this vowel in the oblique singular, and a non-palatalised consonant in the oblique plural. The bulk of this class can be traced back to PIE *i-stems with original hysterodynamic inflection (Pinault 2013a: 345–346). This analysis is confirmed by TB maśce ‘fist’, which is to be equated with Proto-Indo-Iranian *musti- ‘fist’ (cf. Skt. muṣṭí-, Av. mušti-), although the Tocharian word continues a nom.sg. PIE *-tḗ(i̯), instead of the expected *-ti-s in Indo-Iranian (Pinault 2013a: 346–347; dtb 476; Malzahn 2014b: 259 fn. 2). All thing considered, the evolution of TB plāce is as follows: nom. sg. PIE *plH-tē(i̯) > PT *-ce > TB -ce, acc.sg. *-ti-m̥ > PT *-cə > TB -c, nom.pl. PIE *-tei̯-es > PT *-cəyə > TB -ci, acc.pl. *-ti-ns > *-cəns (?) >> PT *-təns > TB -täṃ.321 320 Klingenschmitt (1994: 401–402) reconstructed a hysterodynamic abstract derivative in *-tu (see recently Hackstein 2017: 1316). However, as correctly pointed out by Hartmann (2013: 486–487 with references), this derivation is implausible, because evidence for reconstructing hysterodynamic *u-stems is meagre (Neri 2003: 110–111) and the derivatives in PIE *-tu are usually either masculine or neuter (Adams 1988a: 125–126). Furthermore, we have no other clear continuants of hysterodynamic u-stems in Tocharian (as Klingenschmitt himself acknowledged). 321 The reconstructed paradigm of the PIE hysterodynamic i-stem follows Beekes (1973) and Lindner (UrIG § 4c. p. 29). Malzahn & Fellner (2015: 72 fn. 36) argue that the nom. sg. -e and the lack of palatalisation in the oblique plural are unexpected and that they are due to analogical development after the ubiquitous TB e-stems, on the one hand, and to the contrast between palatalised nom.pl. and non-palatalised obl.pl. in e.g. lāñc : lāntäṃ (from TB walo ‘king’), lyśi : lykäṃ (from TB lyak ‘thief’), on the other hand. I agree with them that the replacement of the non-palatalised obl.pl. TB plātäṃ for the expected TB *plācäṃ may be secondary. In Proto-Tocharian, the ending *-əns instead of *-’əns was ubiquitous, and an analogical change after the class of TB lyak (obl. pl. lykäṃ) is probable. On the other hand, I do not see any diachronic problem with the nom.sg. -e of TB plāce. Analogy after the TB e-stems is in my view unnecessary.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

232

Chapter 3

Now that we have clarified what type of PIE stems are continued in the Tocharian A class ii.1, we can proceed to the origin of the plural ending TA -äṃ. 3.6.3.5 Origin of the Plural Ending TA -äṃ There are two opposing ways to explain the plural forms of the nouns discussed above: (1) either Tocharian B has preserved the original situation and Tocharian A has introduced the morpheme -(ä)ṃ < PT *-na from other stems, or (2) Tocharian A has preserved the original situation and in Tocharian B the nasal plural *-na has been lost. At first sight, both hypotheses seem plausible. The former implies that Tocharian A inherited plural forms identical to those attested in Tocharian B. When final vowels were deleted in pre-Tocharian A, nominative and oblique would have become homophonous in both the singular and the plural. In order to reintroduce a distinction between singular and plural, the plural morpheme -äṃ would have been attached at a later stage (e.g. pl. PT *yəsara > pre-TA *ysār >> TA ysāräṃ). This hypothesis also faces some problems, however. As stated in the opening section, the fact that the marker TA -äṃ is the least productive among the plural endings of Tocharian A must be seriously considered if its origin is to be traced. As a consequence, analogical developments can hardly be involved: basically, there is no immediate source from which the plural *-äṃ could have been abstracted and then generalised. One might think that TA -äṃ was introduced from the neuter nasal stems. However, the only noun that could diachronically go back to a *men-stem and that synchronically shows this ending is wram ‘thing, object’, because other continuants of the PIE *men-stems have replaced their original plural forms, like TA ñom ‘name’, pl. ñomäntu (cf. TB ñem, pl. ñemna < PT *ñemna). This evidence implies that *-äṃ was not a convenient plural ending in pre-Tocharian A and that there is therefore no reason that words like TA por ‘fire’, ytār ‘road’, and ysār ‘blood’ should have selected this ending over other much more productive plural markers. I therefore believe that the latter scenario is the correct one, since it lends itself to a more elegant solution: the nasal element in TA -äṃ must be interpreted as an archaism not only in TA wram ‘thing, object’, which goes back to an old *men-stem, but also in those words that continue heteroclitic *r/n-stems, where the plural -äṃ historically coincides with the original n-form. It follows that Tocharian A, as opposed to Tocharian B, has continued the heteroclitic inflection by refunctionalising the n-form of the oblique cases in the plural. This is not an isolated trend of development, since it closely resembles similar cases in Latin and Iranian.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

233

In the history of Latin, the old heteroclites are normalised in two ways (Ernout 1914: 67–68; Leumann 1977: 359–360; Weiss 2020: 256–258). On the one hand, some nouns have analogically levelled the r-stem in all cases (e.g. Lat. ūber, -ris ‘udder; abundant’, cf. Skt. ū́dhar/n- ‘udder’), although in Old Latin a few of them were still heteroclitic. Compare, for instance, Lat. femur, gen. femoris ‘thigh’ (e.g. in femore, Cicero, Verr. Or. IV. 43, 93) with OLat. femur, gen. feminis ‘id.’ (e.g. femina in Plautus, Poen. 3.1, 68). On the other hand, nouns like iter, gen. itineris ‘street, way, journey’ or iecur, gen. iocineris ‘liver’ show spread of the r-stem from the strong cases to the n-stem of the weak cases. It follows that in the pre-history of Latin two paradigms of the word for ‘way, street’ can be virtually reconstructed: older *iter / *itinis and newer *iter / *iteris (Leumann 1977: 103). Latin speakers mixed up the two paradigms, forming a new inflection with a stem *itin-er-, from a pre-existing *itin-, in all weak cases and in the plural. Only the nominative and the accusative singular still attest the original distribution of the allomorphs. Let us now consider some examples from Iranian. In Khotanese, spellings with -rr-322 can be the outcome of original consonant clusters beginning with *r (e.g. Khot. ttarra- ‘grass’ < *tr̥ na-, cf. Skt. tr̥ ṇ́ a-; Khot. kārra- ‘deaf’ < *karna-, cf. YAv. karǝna- ‘ear [daēvic]; deaf’, Ved. kárṇa- ‘ear’, Emmerick 1969: 69). Based on this evidence, Emmerick (1980: 168) was able to trace OKhot. gyagarra- ‘liver’ back to a proto-form *i̯akr̥ na-. This reconstructed form ultimately derives from the heteroclitic stem IIr. *Hi̯akr̥ /n- (cf. Skt. yákr̥ -t, gen. yaknás, Av. yakarə), but it shows a conflation of the r- and the n-stem with subsequent thematisation. In parallel, the numeral OKhot. byūrru ‘10,000; myriad’ can be the outcome of *baiu̯ arnam (Emmerick 1980: 168 and 1993: 292; cf. Bailey 1979: 309). Although no clear Indo-European cognates of this word have been identified so far, OKhot. byūrru has cognates in several Iranian languages, from both the Western (e.g. MP bēwar ⟨bywl⟩ [Phl.], ⟨byw(r)⟩ [ManMP]) and the Eastern side (e.g. ManSogd. βrywr, Christian Sogd. brywr ‘myriad’, Oss. Iron biræ, Digor be(w)ræ), including YAv. baēuuarə/bāeuuan-, which points to the reconstruction of a heteroclitic *r/n-stem for Proto-Iranian (kewa 2.514). It is reasonable to assume that the same mixture of the two stems affected the words for ‘fire’, ‘blood’, and ‘road’ in the pre-Tocharian A stage. In Proto-Tocharian, these words must have continued the heteroclitic inflection, with r-stem in the singular and n-stem in the plural. Then, when Tocharian B and A split off from Proto-Tocharian, the former generalised the r-stem, and 322 The exact phonetic nature of r and rr is unknown. Emmerick (2009) suggests that r is a dental trill and rr a dental approximant. Note that rr is not a graphic combination of two ⟨r⟩. See further Hitch (2016: 37–39) and Skjærvø (2022: 122–123).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

234

Chapter 3

the latter refunctionalised the two stems, adding the reanalysed nom.obl.pl. PT *-na to the r-stem. Table 35

Heteroclitic inflection from Proto-Indo-European to Tocharian A

strong stem weak stem

pie

pre-PT

*it-ṓr *it-n-

> *yət-ar > *yət-ən-

sg. pl.

PT

pre-TA

TA

> *yətar > *yətə-na

> *yätār >> *yätār-än(ā)

> ytār > ytāräṃ

As Hock (1991: 189–190) has pointed out, in analogical changes old and innovative forms have to coexist as variants for some time before the effective realisation of the analogy. Occasionally they are affected by blending (sometimes also called contamination). The phenomenon of blending is usually treated as a sporadic lexical change by which a new word is created through the combination of two already existing lexemes. In some cases, however, blending also affects the morphological paradigm of words, especially when they develop competing stems. This is exactly what has happened to the three Tocharian A nouns. In Proto-Tocharian, the two stems were therefore maintained for some time, particularly because they had different grammatical functions: the r-stem was used to express the singular, and the n-stem the plural. But the entire paradigm was analogically levelled, and the r-stem became the basis on which the n-containing endings were added. Through this development, the functional correspondence between singular and plural has been formally maintained, and PT *-na has become a new plural marker. One may further wonder whether this phenomenon can be regarded as a process of “exaptation”, a term introduced in linguistics by Lass (1990: 81–82), according to which linguistic relics can be refuncionalised by being adapted according to existing regular templates (see recently Van de Valde & Norde 2016: 9–11). That is, in the original heteroclitic paradigm the function of the r-form was not that of marking the singular, nor was the function of the n-form to mark the plural. Yet, after the general weakening of case inflection and the syncretism of the case markers of Proto-Tocharian, the n-form was refunctionalised to specifically mark the plural. On the other hand, the competitive r- and n-forms developed differently in Tocharian B: the entire paradigm of these nouns was levelled in favour of the r-stem, while the n-form disappeared. This is a common trend of development that is also found in some other Indo-European languages. Examples include: Lat. ūber, gen. ūberis ‘udder’ (cf. Skt. ū́dhar, gen. ū́dhnas, Gk. οὖθαρ, gen. -ατος),

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

235

MP ǰagar ⟨ykl⟩ ‘liver’ (cf. Skt. yákr̥-t, gen. yaknás, YAv. yakarə), OHG wazzar ‘water’, OE wæter ‘id.’ vs. Goth. wato (n-stem) ‘water’, ON vatn ‘id.’ (cf. Hitt. u̯ ātar, gen. u̯ itenaš, Gk. ὕδωρ, gen. ὕδα-τ-ος), OHG fuir ‘fire; heart’, OD fuir ‘fire’, OE fȳr ‘id.’ vs. Goth. fon ‘fire’, ON funi ‘flame’ (cf. Hitt. paḫḫur, gen. paḫḫuenaš), and see further the doublet Goth. sauil ‘sun’ (dative sunnin) vs. Goth. sunno ‘id.’ (cf. OAv. huuarə̄,̆ gen. xvə̄ṇg). A similar analysis, mutatis mutandis, also accounts for TA wram ‘thing, object’. However, I was not able to find any clear explanation for the plural plācäṃ ‘words’ (cf. plācänyo ‘because of words’ in e.g. A75 b6). Indeed, among the words discussed above, this is the only case where Tocharian B attests remnants of the original inflection (cf. nom.pl. TB plāci < PT *pəlacəyə < PIE *(s)plH-tei̯-es). A tentative analysis suggests that TA plāc acquired the plural ending from TA wram. The reason this analogical development took place lies in the meaning of these nouns. Indeed, TA wram must originally have meant ‘speech, word’, as the etymology of the term seems to indicate. For a certain period, TA wram and TA plāc were consequently almost synonymous, and this favoured the transfer of the ending -äṃ to the paradigm of plāc. Only at a later time would TA wram have developed the meaning of ‘object’. Summing up the result of our findings, we have seen that, with the exception of TA plāc ‘word’, the Tocharian A nouns with plural ending -äṃ can be traced back to PIE *r/n-stems (TA ytār ‘road’, ysār ‘blood’, por ‘fire’) and to PIE *men-stems (TA wram ‘thing’). My final aim was to demonstrate that the plural ending TA -äṃ is an important archaism that in a way continued the Proto-Indo-European state of affairs. We have seen that the reconstruction of heteroclitic nouns requires strict comparisons between the older stages of the Indo-European languages, because in more recent times the same languages generalised one of the two stems. In Tocharian B we precisely find this development: the formal contention between r- and n-stems was resolved at the expense of the latter. The result of this process caused the collapse of the n-stem. On the other hand, we have seen that Tocharian A preserved the older state of affairs, since it has maintained both the r-form of the singular and the n-form of the plural. The final outcome of this development is a blended plural with the r-form as the stem and the n-form as the ending. This inflectional class therefore constitutes an important section of the Tocharian lexicon that offers a small but significant contribution to the diachronic evolution of Indo-European nominal morphology. 3.6.3.6 Conclusion From a synchronic point of view, it has become clear that TB -na and TA -äṃ display different distributions in the noun inflection. Tocharian A -äṃ is

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

236

Chapter 3

confined to archaisms, which mostly inherited this plural marker from the proto-language. The Tocharian B ending is characteristic of two groups of substantives: (1) a class of alternating nouns, where TB -na has to be interpreted as an inherited marker (both of Proto-Indo-European and Proto-Tocharian origin); (2) a flourishing class of feminine nouns (the so-called śana- and aśiya-types), where the origin of -na is debated. From a synchronic perspective, we have seen that the alternating nouns with the plural ending TB -na are a closed class; from a diachronic perspective, this class is quite heterogeneous since its members cannot derive from a common PIE nominal stem type. Nonetheless, the internal comparison between Tocharian A and B allows us to reconstruct *-na as quite a common marker of alternating nouns in Proto-Tocharian. Krause & Thomas (teb) divided class ii into two parallel subclasses: class ii.1 has a plural ending TB -na, while class ii.2 has a plural ending TB -nma. This bipartition is based on Tocharian B, since the metathesis of the cluster -mn- to -nm- entailed the formation of the second subclass. The corresponding Tocharian A nouns have different plural forms. On the one hand, a few inherited heteroclitic *r/n-stems and *men-stems continued to be members of class ii. Indeed, a small subclass continues neuter formations in PIE *-men-, where the ending -na derives from PIE *-nh2 (e.g. TB śāmna ‘mankind’, TB ñemna ‘names’, etc.). Another noun that may have inherited this plural marker from Proto-Indo-European is TB ersna ‘form, appearance’, which I have compared with Hitt. ḫaršar, ḫaršn- (Kloekhorst 2008: 314–315) as both reflecting the outcome of a heteroclitic paradigm. On the other hand, most nouns with the plural PT *-na were transferred to other classes with plural ending TA -nt /-ntu (class iii.1 and class iii.2). These Tocharian A nouns corresponds to Tocharian B nouns of both class ii.1 and class ii.2, as the examples below show: TB sārm, pl. sārmna : TA sārm, pl. sārmäntu; TB ñem, pl. ñemna : TA ñom, pl. ñomäntu; TB nāki, pl. nakanma : TA nākäm, pl. nākmant; TB wāki, pl. wakanma : TA wākäṃ, pl. wākmant, etc. Sometimes we can still see the old plural form -mnā in isolated Tocharian A derived forms, as in TA arkämnāṣi, an adjective derived from * arkänmā (cf. TB erkenma) or the gen.pl. TA wramnāśśi from wramäṃ. Furthermore, we have seen that several nouns with na-plural had a clear singulative meaning in Proto-Tocharian. This value has been maintained in both Tocharian languages, but it is morphologically expressed in different ways. Indeed, Tocharian A, as opposed to Tocharian B, has reanalysed most of the formations in PT *-na as singulars (cf. plural TB palauna ‘praise’ vs. singular TA paloṃ ‘id.’; plural TB tarśauna ‘deception’ vs. singular TA tārśoṃ ‘id.; perhaps plural ersna ‘form’ vs. singular TA aräṃ ‘id.’ and plural krāna ‘(outer) skin’ vs. singular TA krāṃ ‘id.’). The same development can also be observed in a few Tocharian B nouns, as in kerekauna ‘violent flood’ and possibly śalna Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

237

‘quarrel’. This peculiar value of PT *-na is understandable from a comparative perspective. Indeed, as recently argued by Pronk (2015b), the nasal suffix had a “singulative” meaning in Proto-Indo-European, where it was initially limited to neuters. Proto-Tocharian recharacterised this suffix with the original neuter collective *-h2 > PT *-a and this new ending *-na became a special marker of plural nouns with singulative and collective meaning. The origin of the plural ending TB -una has been the main topic of my discussion. I have argued that this marker was abstracted from the neuter plural of the PIE heteroclitic stems in *-u̯ er/n. In order to substantiate this claim, I have scrutinised the Tocharian lexicon with a view to finding continuants of these archaic stems. The results of my investigation are recounted below. Tocharian inherited both the regular heteroclites in *-ur/n and the derived collectives in *-u̯ ōr/n. In the latter type, the allomorph *-u̯ or > PT *-wer became a common suffix to form verbal abstracts (Pinault 2011a). In the former type, the PIE sequence *-u̯ r̥ underwent metathesis, yielding *-ru in all strong cases. These new *ru-stems converge in the Tocharian class i.2, where we find a conspicuous number of alternating nouns with sg. TB -är /-ər/, ta -är and pl. TB -arwa /-ə́ rwa/, ta -ru (-rwā, -runt). Additional evidence in support of the metathesis *-u̯ r̥ > *-ru comes from isolated words, where the o-vocalism in the root has always been a matter of debate (see Del Tomba 2020c). This vowel can be now explained through affection by final -u (e.g. tb ṣñor ‘sinew’ vs. pl. TB ṣñaura, TA kror ‘crescent of the moon’, TB kror-iya* ‘horn (?)’, TB plor-iya*, a wind instrument, etc.). From a diachronic perspective, the paradigmatic connection between metathesised *ru-forms (strong stem) and non-metathesised *n-forms (weak stem) became increasingly opaque in the prehistory of Tocharian and a new plural form based on the singular was created, thus pre-PT *-ru : *-wəna >> PT *-ru : *-rwa > TB -r : -rwa, ta -r : -ru. Indeed, while the singular *-ər could be from either pre-PT *-ru or *-ur, the plural *-rwa proves that the singular was pre-PT *-ru. The formal mismatch between r- and n-forms favoured the gradual abstraction of the plural ending *-wəna, which started to form pluralia tantum and to recharacterise the plural form of various inherited stems. Among these stems, *-wəna was attached to singular forms ending in PT *-e / -a and *-ey, forming a diphthongised plural *-ewna that regularly became -euna in archaic Tocharian B, -auna in classical Tocharian B, and -omna in Late Tocharian B (Peyrot 2008: 52). The original distribution of the heteroclitic forms has been partially retained in relics, like akrūna ‘tears’ and ṣotrūna ‘signs, markers’, where the r-containing stem has been generalised and the plural *-wəna blended in. A similar development occurred in Tocharian A, where the outcomes of the heteroclitic nouns for ‘road’, ‘blood’, and ‘fire’ show a blended plural with the r-form as the stem (ytār-, ysār-, por-) and the n-form as the ending (-äṃ). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

238

Chapter 3

3.6.4 On Some Tocharian B Pluralia Tantum Many languages contain nouns which are inflected either only in the plural (like Eng. clothes and Lat. dīvitiae ‘wealth’) or only in the singular (like Eng. dust and Lat. vulgus ‘folk’). These words can be descriptively labelled pluralia tantum and singularia tantum. In other words, they are nothing more than lexical plural or singular whose distinctive property is that they lack either a singular or a plural inflection (Acquaviva 2008: 15–16). In Tocharian there are many nouns that belong to these categories. Some rare cases of masculine pluralia tantum are TB kercci ~ kerccī (nom.pl.) ‘palace’ and TB meli, ta malañ (nom.pl.) ‘nose’. However, most Tocharian B pluralia tantum are the outcome of old neuter forms. These lexical plurals generally have collective semantics. Their paradigm can be exemplified with the noun TB mīsa ‘flesh’, whose inflection is as follows: Table 36

Inflection of TB mīsa

inflectional class

nom. pl. obl. pl.

gen. pl.

stem

mīsa-type

mīsa

mīsaṃts

misa-

mīsa

To this paradigm we can add the distributive plural misaiwenta ‘pieces of meat’. This noun is to be linked with PIE *mēms- / *mems- ‘meat’ (cf. Skt. māṃsá-, Goth. mimz, etc.). Although this etymological connection is evident, some details on the phonetic evolution of the word are still to be clarified (in particular, PIE *-ms- > *-ns- would be expected to yield **-nts- in Tocharian B). A recent discussion on this word and related issues can be found in Pinault (2013b: 350–353). As with TB mīsa, the historical analysis of the other nouns in this category present difficulties. For some, despite clear Indo-European cognates, the derivational process involved is unclear, while others lack a clear etymology. In previous sections I discussed a productive group of pluralia tantum which show a plural in -na or -(a)una. These are: TB särwāna ‘face’, TB krentauna ‘virtue(s)’, TB ersna ‘from, beauty’, TB yasna ‘treasury’, etc. For a diachronic analysis of these nouns, see sections  §  3.6.3.1 and  §  3.6.3.2. For a discussion of TB āsta ‘bones’, see § 3.5.1.2. Other alleged pluralia tantum are (dtb s.v.): TB stmānma ‘pipes, tubes’, TB proksa ‘grain (?)’, TB āka ‘millet’, TB tserekwa ‘deception’, TB mekwa, ta maku ‘nails’, and TB par(u)wa ‘feathers’.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

239

The first noun is a hapax legomenon attested in AS6C a5 wraṃtse stmānma ‘pipes of the water, gutters’, but we have no evidence to analyse it as a plurale tantum, since its singular form could simply be unattested in the surviving texts. If so, the singular of stmānma could be reconstructed as stanmau*, parallel to TB śanmau, pl. śanmānma (see Hilmarsson 1991a: 153). Two words, TB proksa ‘grain (?)’ and TB āka ‘millet’, refer to different types of grain. The former was identified by Schmidt (2002: 3–4) in the wooden document THT3998 a3. However, both its meaning and etymology are unknown (see Peyrot 2018a: 259–60 for critical remarks). On the other hand, the case of TB āka ‘millet’ is peculiar, because it is apparently inflected as both a singulare (cf. HWB74(1).3 āka laś ‘millet has been spent’, cf. Ching 2010: 309–310) and a plurale tantum (cf. SI P 136.b a3: āka lateṃ ‘millets went out’, cf. Ching 2010: 324–326; PK Bois C1 a2 maṅkarāñcana āka ‘old millets’, a7 ce maṅkarāñcana ce ñwemaṣṣana āka ‘old and fresh millets’, etc.).323 The noun is usually compared with Lat. acus, -eris ‘husk’ and Gk. ἀκοστή ‘barley’, both from *h2eḱ- ‘sharp’. If it belongs to this root, TB āka may be an original neuter plural from *h2eḱ-h2 (see Pinault 2008: 371 and Peyrot 2018a: 253–254 for alternative proposals). The other nouns to be discussed end in TB -wa. Pinault (2008: 25) interprets TB tserekwa ‘deception(s), deceit, illusion’ as a plurale tantum. Support for such an analysis lies in the fact that it occurs frequently with TB snai ‘without’, an adverb which usually combines with singular nouns. The noun is related to the verb TB tsere-ññ- ‘to deceive, trick’, whose base appears to have been borrowed from Khot. jsīr- ‘to deceive’ (cf. also jsīrgyā- ‘deception’, jsīrāka- ‘cheating pilferer’, see Bailey 1979: 115–116; cf. also Hilmarsson 1991b: 87–88).324 The two remaining nouns are also those with stronger Indo-European comparisons, even if their derivation and formal shape are not as expected. TB mekwa, ta maku ‘nails’ (both plural, pace Blažek 2001: 192, cf. A321 a2 tsres maku āṅkaräsyo “with hard.pl nails and fangs”) is connected to the familiar Indo-European noun for ‘nail’, PIE *h3noghu- (or *h3nogwh-?) > Lat. unguis ‘claw’, ungula ‘hoof’, Gk. ὄνυξ ‘talon’, Arm. ełungn ‘nail’, OHG nagal ‘nail’, etc. The unexpected m- is usually explained through labial assimilation *nekwa > mekwa (dtb 502 with references; cf. also Blažek 2001, who postulates a compound *sm̥ -h3nogh-u̯ o- or *sem-nogh-u̯ o-). Another problem is the lack of a-umlaut 323 Note however that the occurrences of āka showing singular agreement are few and may be regarded as grammatical mistakes in secular documents. 324 Michaël Peyrot (p.c.) pointed out to me that TB tser-ek* (pl. tserekwa) ‘deception(s)’ might be compared to TB tärr-ek (TA trak) ‘blind; blind person’, which is usually considered a compound with TB ek, ta ak ‘eye’. It may be claimed that the verb TB tsereññ- is from *tserek(w)ññ- through assimilation. However, the comparable case of TB weñ- ‘to say, speak’ < *wek-ññ- shows that degemination of *-ññ- > -ñ- is to be expected.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

240

Chapter 3

(cf. TB yākwa ‘body hairs’ < *yekwa). I see two possible ways to explain this irregularity. If the plural TB -wa is original, then one could invoke analogical levelling after an unattested singular of this noun. A second hypothesis is that TB mekwa was inserted into this inflectional type at a later stage (dtb 502), when a-umlaut ceased to operate. As a matter of fact, this noun is not expected to be alternating in Tocharian because all other Indo-European languages point either to a masculine or to a feminine. The last noun to be discussed is TB paruwa / parwa (?) ‘feathers’, usually considered to only be attested in the plural. This noun is attested four times with different spellings: (1) parwā in B282 b1; (2) loc.pl. parwāne in B282 a5 (cf. Peyrot 2013a: 815 fn. 819); (3) parwa in B89 a4; (4) paruwa in W32 b3. On the basis of these forms, it is unclear whether the root vowel was /ǝ/ or /a/. Indeed, B282 is an archaic text in which the spelling parwā seems to stand for /parwa/ (likewise parwāne /parwáne/, cf. Pronk 2009: 88 and Peyrot 2008: 33–39). The other occurrences are from classical texts: parwa in B89 a4 suggests /pǝ́ rwa/, while paruwa in W32 b3 suggests /parǝ́wa/. However, it should be noted that B89 has peculiarly spelled forms, like ksā (b6) for ksa ‘some, any’, tāmp (b6) for tamp ‘that’, träṅko (a1) for traṅko ‘sin’, käryaurtto (b6) for käryorttau ‘merchant’, so that parwa might stand for pārwa here. In addition, as pointed out by Hannes A. Fellner (apud CEToM s. THT1105), it is tempting to relate the hapax legomenon TB pār /pár/ ‘plumage (?)’ in THT1105 b3 to the plural TB parwa. In light of the above, I consider TB paruwa ‘feathers’ to be phonologically analysed as /parǝ́wa/. The historical analysis of TB paruwa is equally uncertain. Indeed, within a comparative framework, the reconstruction of the PIE word(s) for ‘feather, wing’ is notoriously difficult. As summarised by Pronk (2015b: 335), we can subdivide the Indo-European languages into two groups. Some languages point to *p(t)er-: Gk. πτερόν ‘wing, feather’, CS pero, Hitt. partāu̯ ar, -aun-, etc.; others attest an n-suffix: Skt. parṇá-, YAv. parəna-, Lith. spar̃nas, OE fearn ‘fern’. Latin penna < *pet-na falls in the middle. Beside these forms, Hittite has a heteroclitic paradigm pattar, pattan- (or pettar, pettan-; cf. also OW eterin ‘bird’ and atan ‘wing’). Sanskrit has a thematic noun páttra- ‘wing; leaf’, which is also attested in Germanic, cf. OHG fedara, OE feðer, etc. Kloekhorst (2008: 659) suggests that these words may preserve traces of an original heteroclitic stem PIE *péth2-r, *pth2-én- (Kroonen 2013: 138–139; Pronk 2015a) from the root *peth2‘to fly’ (LIV2 479). If the forms cited above (or at least a great number of them) are ultimately to be connected to this reconstructed noun, several analogical adjustments took place independently in the Indo-European languages. In this context the position of TB paruwa is still unclear since none of the aforementioned Indo-European cognate words points to the reconstruction of a u-stem. Pronk (2015b: 336) projects PIE *pth2-er-u- or *pt(h2)-or-u-h2 but these are ad Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

241

hoc reconstructions. It is further unclear to me what the fate of PIE *pt- would be in Tocharian, but I am not aware of any counterexamples for postulating an outcome PT *p-. In this case, TB pār ‘feather’ (?), pl. paruwa may be from PIE *péth2-r, *pth2-én- with levelling of the r-allomorph in the weak stem. The pl. in -wa would be influenced by nouns of class i.2 with sg. *-ǝr(u), pl. *-ǝrwa (of the type kwarsär ‘league, vehicle’, pl. kwärsarwa, see Del Tomba 2020c: 62–63 and further Isebaert 2004).325 A last substantive that probably merits inclusion in the group of pluralia tantum nouns is TB wīna /wə́yna/ ‘pleasure’. This noun occurs several times in the texts, especially in constructions with the verbs yam- ‘to do’ and kəlpa‘to obtain’ (Meunier 2013: 170–172). There is however some hesitation in the analysis of the grammatical number of TB wīna, as it is generally interpreted as either a singulare or a plurale tantum (dtb 654; Malzahn 2011: 85 fn. 7). Hilmarsson (1991c: 85–86) claimed that the nominative of TB wīna is not attested. Conversely, Adams (dtb 654) gives the nominative singular or plural as TB wīna, but he does not provide any attestation. A possible example of nom. wīna can indeed be found in SI B 75 a7 (= IT233 a4) (Udānavarga) taiknesa pälskontse wīna erepate “In the same way this image (is) a pleasure for the mind”. That wīna is an apposition of erepate ‘image, beauty’ is confirmed by the Sanskrit parallel Uv 1.29c–d tathā manoramaṃ bimbaṃ jarayā hy abhimarditam “in that manner an image gratifying the mind is destroyed by old age” (Bernhard 1965: 106).326 I therefore agree with Adams that this substantive 325 A last possibility to explain the change of the inflectional class of tb paruwa would involve a revision of the development of syllabic pie *r̥ in Tocharian. There may be evidence to claim that pie *r̥ passed through a stage *ur on the way to pt *ər. This reconstruction was first suggested by Adams (1984: 397–398, 1988: 16–17), who claims that all pie syllabic resonants first evolved into *uR, and then *u yielded *ə. Adams’ theory has been criticised (cf. e.g. Ringe 1991: 81–83), but it has also been supported by Burlak (2000: 123–124) and recently defended by Jasanoff (2018). As far I can see, Adams’ account can in fact be questioned for the outcome of syllabic nasals (cf. e.g. pie *ḱm̥ tom ‘100’ > tb kante /kə́ nte/, ta känt). However, some etymologies make the development *r̥ > *ur > *ər quite attractive: pie *ǵr̥h2- ‘to age, grow old’ (liv2 165) > *kura- > pt *kwəra- > tb kwəra-, ta kwärā-; pie *kr̥ s- ‘to run’ (liv2 165) > *kurs- > pt *kwərs- > tb kwars-är, ta kurs-är ‘league, mile; vehicle’; pie *udr̥ ‘water’ > *udur > *udru (metathesis) > pt *wər(ə) (regular loss of *d in the cluster *dr) > tb war /wə́ r/, ta wär. Note that the middle endings of Tocharian might also point to the reconstruction of a word-final development *r̥ > *ur > *ru > *rə (cf. Kim 2019b: 147). If this is correct, we may then account for the change of the inflectional class of *pth2-r̥ ‘feather’ without recurring to analogy: the singular would have evolved as *ptaru > *paru > pt *par > tb pār /pár/, and the plural would have been reshaped on the pre-Proto-Tocharian singular form as paruwa /parúwa/. Needless to say, the evidence should be tested further, but I believe that Adams’ account of the development of syllabic liquids is worth reconsidering. 326 See Wilkens, Pinault & Peyrot (2014: 12–13) for yet another possible attestation. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

242

Chapter 3

has an undifferentiated nom.obl. wīna. As for its grammatical number, the following attestation may be relevant: SI B 75 b7 /// no wīna tākoṃ “(how then) pleasure should arise?” (cf. Georges-Jean Pinault apud CEToM). This sentence should correspond to the Sanskrit passage Uv. 1.33b kā nu teṣāṃ ratir bhavet “how then should there be pleasure for them?”. As a consequence, TB wīna is the subject of the sentence in agreement with the 3pl.opt. tākoṃ and is to be considered as a plurale tantum.327 Tocharian B wīna is matched in Tocharian A by wañi, but the two words, though related, cannot go back to the same proto-form. From a derivational point of view, TA wañi might correspond to Lat. venia ‘favour, permission’ and possibly OIrish fine ‘kindred’ < PIE *u̯ enH-ih2. However, Tocharian A points to the reconstruction of either *-o- or *-ē- in the root. In the first case, TA wañi can be interpreted as a derivative in *-ih2 of a noun of the τόμος-type from *u̯ enh1‘to desire’.328 Otherwise, according to Adams (dtb 654) a vr̥ddhi-derivative *wēnH-iyo- might be reconstructed (ON vænn ‘fair’). As for TB wīna, mechanical reconstructions would be either *una, which is ad hoc and does not account for the internal -i- /ǝy/, or *u̯ ii̯na (Weiss 2018: 377). Indirect evidence that a similar word survived in Tocharian A is provided by the denominal verb TA wäynās-, tb wǝynask- ‘to venerate’ (Hackstein 1995: 101; Hilmarsson 1991b: 85–87; dtb 906; Weiss 2018: 377). Further etymological discussion is, for the time being, doubtful.329 There also a few nouns that have an ending TB -a in both the nominative and the oblique singular. The most attested of these are TB yasa, ta wäs ‘gold’ and TB śalna ‘quarrel’. If not loanwords, their formal structure invites an analysis as old collective plural forms in PIE *-h2. Pinault (2012a: 197) and Hackstein (2017: 1318–1319) trace TB yasa, ta wäs ‘gold’ back to PT *ẃəsa < *h2u̯ esh2, an original

327 The grammatical number of TB kerekauna ‘flood’ and TB särwāna ‘face’ is clear (contra Malzahn 2011: 84–85 fn. 7): the former is a singular (cf. ceu orocce kerekau(na) “this great flood” in Or.15009/296 b4, cf. Tamai 2009), and the latter is a plural (cf. kaklaiksauwa särwan(a) “the face is wrinkled” in B405 b3, cf. Hilmarsson 1989a; Saito 2006: 225). On these words, see Hartmann (2013: 330 and 369). 328 TA wañi is said to be masculine on the basis of the agreement in YQ II.13 a4: mäñcaṃ klyom wañi te napeṃsam “What is the noble pleasure among the mortals?” (cf. Ji et al. 1998: 131). If this passage has been well interpreted and translated, then the adjective klyom ‘noble’, inflected as a masculine singular, agrees with wañi (Hartmann 2013: 319; Poucha 1955: 285). However, Peyrot’s translation “Oh noble one, is there somehow pleasure among men?” (Peyrot 2018c: 85) is probably to be preferred, as it is perfectly compatible with the Old Uyghur parallel (cf. also Geng, Laut & Pinault 2004a: 364). 329 Etymological suggestions can be found in Weiss (2018: 376–378). For indirect evidence in support of the reconstruction of a noun *u̯ ēnH-, *unH- ‘desire’ (nt.?), see de Vaan (2008: 662). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

243

collective formation. Conversely, Driessen (2003: 348–350) has explained the noun as a loanword from Proto-Samoyedic *wesä.330 As we have already seen, the reinterpretation of collective formations as singulars is a well-attested phenomenon in Tocharian A. In this language, we find just a few pluralia tantum and, to my knowledge, they cannot be traced back to old neuter plurals. Collectives in PT *-a were mostly reinterpreted as singulars and transferred to other inflectional classes. Some examples include: TA paloṃ (sg.) vs. TB palauna (pl.), TA tārśoṃ (sg.) vs. TB tarśauna (pl.), TA aräṃ (sg.) vs. TB ersna (pl.), TA krāṃ vs. TB krāna. A more complex case is TA wmār ‘jewel’, a feminine noun with a count plural wmāri. This noun corresponds to Tocharian B wamer ‘jewel’ (pl. wmera), a masculine noun. It is clear that the two Tocharian words differ in both gender and inflection. Recently, Pinault (2011a: 160–164 and 171–173) commented on these forms and reconstructed an alternating noun with singular *wəmer, and plural *wəmera. After the fragmentation of Proto-Tocharian this word underwent independent developments in both Tocharian languages. TB wamer assumed masculine gender from the (quasi-)synonym TB yetwe ‘ornament’. A more significant development took place in Tocharian A. The plural form *wəmera first evolved into *wǝmara (through a-umlaut), and was then reanalysed as a feminine singular, thus *wǝmara > TA wmār. The expected singular PT *wəmer > TA **wmar vanished. The new singular wmār was then provided with a new countable plural wmāri. In my view, Pinault’s explanation is impeccable and allows us to insert TA wmār into the group of Tocharian A nouns coming from original collective formations. As a general tendency, the reanalysis of old plural forms as singulars was more extensive in Tocharian A. The reason is relatively easy to establish: after the general apocope of final vowels, these substantives would not have had any clear plural marker. Furthermore, given the fact that the great majority of these nouns had a collective meaning, the reanalysis of these plurals as singulars is easily understood. 3.7

Summary and Conclusions

The main questions addressed in the introduction to this chapter were related to the historical evolution of the feminine and the neuter genders in the inflection of the Tocharian noun. In each section issues related to these questions have been discussed and solutions have been proposed. In particular, I have identified and commented on those inflectional types that have been variously 330 Cf. also Peyrot (2019: 101). See Malzahn (2011: 99–100) for an etymological proposal of TB śalna ‘quarrel’. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

244

Chapter 3

connected to the feminine gender, in order to trace their evolution from Proto-Indo-European to Tocharian. The problematic status of the Tocharian genus alternans and its historical link to the PIE neuter has also been discussed in parallel. These two points will be synthetically recapitulated below. 3.7.1 Evolution of the Feminine in the Noun Inflection In this chapter I have tried to understand the evolution of the PIE inflection in *-eh2 > *-ā. To this end, I have identified the Tocharian inflectional classes in which we find synchronic continuants of this reconstructed type. Those classes are: the kantwo-type, the okso-type, the arṣāklo-type, and, in part, the palsko-type. I then discussed the etymological and derivational problems connected to the members of these classes. The results of my investigation show that Tocharian has probably inherited and generalised a hysterodynamic ablaut paradigm in *-(e)h2 throughout the inflection of the nouns. The outcome of this reconstructed paradigm was maintained in the Tocharian B kantwo-type, where the singular paradigm nom. -o, obl. -a mirrors the PIE opposition between strong stem *-eh2(-), and weak stem *-h2-. In Tocharian A, the formal difference between the Tocharian B okso-, arṣāklo-, and kantwo-types is only partially attested. Indeed, the majority of Tocharian A nouns matching these Tocharian B inflectional types are ā-stems (< PT *a-stems). I have therefore investigated whether this mismatch is to be interpreted as an archaism or an innovation. In other words: what was the Proto-Tocharian state of affairs? In order to answer to this question I discussed contradictory evidence revealed by a closer comparison between Tocharian A and B. I have attempted to reconstruct a partially syncretic inflectional type for Proto-Tocharian which underwent three splits in Tocharian B. There are several developments that caused these splits. In short, we can say that some endings and forms are the outcome of specific markers of the *(e)h2-inflection, others of the *ōn-inflection, and yet others originated after sound changes that are particular to Tocharian B. Some nouns belonging to these classes are masculine (like the archaic noun TB kantwo, ta käntu ‘tongue’, fauna terms, and entity-denoting nouns), but the great majority of them are feminine. In fact, feminine appears to be the default gender of these classes, as is shown by the fact that inanimate loanwords belonging to these classes are also feminine. Finally, we have seen that some *(e)h2-stems may have been continued in the so-called palsko-type, where they were reinterpreted as alternating in view of morphophonological confluences between the outcome of the thematic collectives in *-eh2 and the feminine singular of the *(e)h2-inflection. I also discussed the distribution, the origin, and the evolution of the two ́ *ih2-formations reconstructed for the proto-language, i.e. the devī-type and the

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Noun

245

́ vr̥kī-type. We have seen that the poorly represented śana-type can be traced ́ back to the devī-type, with the exception of TB śana, ta śäṃ ‘wife’ itself, whose singular paradigm nom. -a, obl. -o mirrors the PIE stem type *gwénh2-/-éh2-. On the other hand, the origin of the so-called aśiya-type can be traced to a more recent Proto-Tocharian stage, since the members of this class seem to have calqued their inflection from that of the adjectives (see Chapter 4). These two classes are exclusively made up of feminine nouns denoting female referents. In addition, we have seen that the formal and the functional distinctions ́ ́ between the devī-type and the vr̥kī-type ceased to exist in Tocharian: the final result was the merger of these formations, the outcome of which is partially continued in the wertsiya-type, where feminine abstract and concrete nouns can be found. 3.7.2 Evolution of the Neuter in the Noun Inflection As for the development of the PIE neuters, we have confirmed the common assumption that they are in principle continued as the Tocharian genus alternans. Our attention has been focused on the evolution of both the thematic and the athematic neuter paradigms. I have investigated the formal merger of the thematic neuter with the masculine inflection in the singular and with the feminine inflection in the plural. This development must have been nonuniform and gradual, since cases of fluctuation in the gender assignment of (pre-)Proto-Tocharian can be reconstructed. This led to sporadic cases of shift between inflectional classes and metaplasm of gender. I have also analysed in detail the outcome of various athematic neuters that played an important role in the creation of new endings (such as the alternating plural marker TB -na, ta -äṃ) and to the evolution of the Tocharian gender system in general (such as the heteroclitic stems in *-r/n, the s-stems, the neuter n-stems, and certain lexical plurals that preserved a collective meaning related to the original function of -a < PIE *-h2). Special attention has been devoted to the evolution of the heteroclitic stems in *-r/n and *-ur/n. Furthermore, I have laid the basis for the postulation of a new sound law PIE *-u̯ r̥ > *-ru in Tocharian and I have showed that, through this metathesis, we can historically account for (1) the source of r-stem nouns with plural in TB -wa, ta -u (-wā, -unt), (2) the unexpected o-vocalism in some isolated forms, and (3) the origin and the spread of the plural marker TB -una.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Chapter 4

Gender in the Inflection of the Pronoun and Adjective The present chapter investigates the evolution of the category of gender in the system of agreement targets in Tocharian. The goal is to understand what type of gender system Tocharian inherited from Proto-Indo-European and how it evolved in the inflection of pronouns and adjectives. Considering that the feminine has been a topic of debate within the diachronic investigation of Tocharian nominal morphology, particular attention is paid to the development of this gender value. The evolution of the neuter will also be investigated in order to test the theory that it merged with the masculine in the singular and with the feminine in the plural, and to understand how the Tocharian genus alternans developed as a result of these mergers. In order to achieve these aims, I consider endings and forms of the relevant declensions in both pronouns and adjectives. The ultimate aim of this study is to clarify whether Tocharian inherited a different gender system to the other non-Anatolian Indo-European languages and to what extent this reconstructed system differs from that attested by the Tocharian languages. 4.1

General Aim, Methodology and Structure of the Chapter

The general aim of the chapter is to understand how the system of gender developed in the Tocharian system of nominal modifiers. This leads to a range of issues which can be approached from the point of view of pronouns on the one hand, and adjectives on the other. The structure of the chapter mirrors this fact and is therefore divided into two parts. Following an overview of the pronominal system of Tocharian (§ 4.2), the first part deals with the development of the Tocharian demonstratives and other pronouns based on them (§ 4.3 and subsections), as well as the peculiar inflection of the pronominal adjective TB allek, ta ālak ‘other’ (§  4.4). The aim of this part is to discuss the problematic endings and forms within the Tocharian pronominal inflection. The feminine paradigm will be the core issue of my investigation. The final goal is to demonstrate that both masculine and feminine paradigms are to be interpreted as the regular outcome of their

© Alessandro Del Tomba, 2023 | doi:10.1163/9789004532892_005 Alessandro

Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

247

Proto-Indo-European ancestors, with some minor and motivated analogical changes. In the second part (§  4.6,  §  4.7,  §  4.8 and subsections), the development of gender in the adjectival system of Tocharian is investigated. A synchronic overview of the Tocharian adjectival system is offered. The main part is devoted to the diachronic evolution of both thematic and athematic adjectives from Proto-Indo-European to Proto-Tocharian and from Proto-Tocharian to Tocharian A and B. The two pivotal questions that this section addresses are: (1) what type of gender system Tocharian inherited from the proto-language, and (2) how it evolved in the adjectival system. These two questions lead to a number of sub-issues which revolve around the status of the feminine gender and its evolution in the thematic declension. In fact, this topic has become one of the most controversial sections of Tocharian historical morphology. Further pivotal issues concern the evolution of the neuter gender and its functional loss as a category of target gender. In order to solve these problems, I will first focus on the reconstruction of the Proto-Tocharian paradigms of those adjectival declensions that played a relevant role in the evolution of the gender system. The reconstruction is based on a systematic comparison between Tocharian A and B. I will then compare the Proto-Tocharian adjectival system with that reconstructed for Proto-Indo-European, in order to understand the relevant modifications that have occurred and to establish the morphophonological mergers that took place between the three inherited genders. Few methodological preliminaries are required. In general, I follow the commonly accepted principles of historical linguistics and comparative linguistic reconstruction. Furthermore, while every Indo-European language — Tocharian included  — may have preserved archaic elements that have the potential to refine (or even modify) elements of the reconstruction, any investigation into Tocharian historical morphology must take into consideration the relatively late attestation of Tocharian. This does not imply that studying the prehistory of Tocharian from a linguistic perspective may not prove relevant by reconstructing both backwards and forwards with respect to Proto-Indo-European. However, as compared with other ancient Indo-European languages, it is an undeniable fact that Tocharian underwent a much longer period of linguistic development before it reached the stage in which its gender system is attested. In this unattested period several structural modifications took place which frequently changed and remodelled the system of nominal inflection and derivation inherited from Proto-Indo-European. In the historical morphology of Tocharian regular changes caused irregularities

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

248

Chapter 4

in the nominal paradigms. These were mostly solved by analogical changes which restored regularity in the morphological patterns. These changes led to a widespread restructuring of the morphology. Therefore, an important issue within Tocharian historical linguistics is to carefully distinguish inherited material from inner-Tocharian innovations. As far as the diachrony of gender is concerned, any change in the number of gender values and gender agreement patterns must ultimately hinge on variation and change in the domain of gender distinction. A gender value is lost when the corresponding gender agreement patterns fall out of use. As pointed out by Di Garbo & Miestano (2019: 25–27), modification in the domain of gender can result in the reduction, loss, emergence or expansion of gender agreement. From a cross-linguistic perspective, the reduction and loss of gender agreement is a result of two distinct processes: (1) morphophonological erosion; (2) redistribution of agreement patterns. In the system of agreement targets, gender is usually cumulated with other category values, such as number and case. This means that when we consider the diachronic evolution of gender we also need to consider its co-evolution with those nominal categories that are regularly cumulated with gender. Morphophonological changes are caused by sound changes which can modify segmental morphology and ultimately result in its loss. This may, in turn, cause the loss of grammatical distinctions and the restructuring of inflectional paradigms. That is, the phonetic attrition of formal markers in word-final syllables may lead to reduction of gender and eventually in its complete elimination (Priestly 1983; Igartua 2006). As for the agreeing elements that first undergo modifications, attributive elements, such as adjectives or definite articles, generally lose gender distinctions at an early stage, while pronouns retain them longest amongst the agreement targets (Priestly 1983: 343–346; Corbett 1991: 143). This because gender agreement is more likely to be retained on those targets where it is most functional to reference tracking and reference identification, such as pronouns (Di Garbo & Miestano 2019: 26). Conversely, in view of a much stronger syntactic cohesion between adnominal agreement targets and their controllers, those agreement elements that are more adjacent to nouns — adjectives, numerals, etc. — are more prone to the loss of gender distinctions when morphophonological changes occur. As Corbett (1991: 143) clarifies, “[p]ronouns generally retain agreement, but attributive modifiers are losing gender agreement. […] Within the noun phrase, adjectives lose agreement in these languages before determiners, and adjectives lose agreement in number longer than agreement in gender”. As for the rise and further development of alternating genders, Igartua (2006) clarified that in the Indo-European languages where alternating Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

249

genders are attested phonetic changes and analogical pressures have both conspired to preserve a gender distinction within a system lacking robust formal oppositions. The outcome of these processes leads to the differentiation of a gender value that is alternating in its agreement patterns (Loporcaro 2018: 225–230). It is generally assumed that in the development of the gender systems in Indo-European languages there is a diachronic hierarchy of the relative order in which agreement markers are reduced (Priestly 1983: 340; Audring 2009: 178): nouns > attributives (i.e. agreement targets more adjacent to nouns) > relative pronouns > personal and demonstrative pronouns. According to this hierarchy, “retention of a formal gender-opposition in nouns presupposes the same in adjectives, and retention of a formal genderopposition in adjectives presupposes the same in pronouns; and, conversely, formal gender-loss in pronouns presupposes gender-loss in adjectives, and the latter presupposes gender-loss in nouns” (Preistlt 1983: 340). That gender is first displayed in nouns is intuitive, as it represents the conditio sine qua non for having agreement classes that display gender agreements. That is, grammatical gender must be considered, first and foremost, in terms of the morphosyntactic relations between a noun and a modifier. However, the emergence of alternating genders challenges this hierarchy because gender systems with an alternating gender usually involve an additional gender value in the controller nouns (controller gender) which is not represented in the agreements targets. Thus, the emergence and survival of the genus alternans presupposes a situation such that singular and plural inflection bear some unambiguous marking of gender in the paradigm but also such that the opposition between them in the agreement targets replicates a pattern characteristic of the distinction between the other gender values, such as the masculine and the feminine (Maiden 2016). Thus, the rise of the genus alternans is the result of a complex intersection of phonetic and morphological changes, which resulted from the morphophonological erosion of nominal paradigms and further conflation of cumulative morphemes bearing gender values. For this reason, investigating the development of gender markers in the system of agreement targets is of pivotal importance for understanding how the gender system evolved in a given language, particularly if this language also shows alternating agreement class(es). This is the goal of the present chapter. 4.2

Overview of the Tocharian Pronominal System

Like most ancient Indo-European languages Tocharian retains a large number of pronouns of varying functions and origins. They also display different inflections. We find: Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

250

Chapter 4

– Personal pronouns for the first and second persons, i.e. TB ñäś, ta näṣ ‘I’, TB tuwe, ta tu ‘you’, TB wes, ta was ‘we’, TB yes, ta yas ‘you (pl.)’ (and pronoun suffixes for the first, second, and third person); – Demonstrative pronouns, e.g. TB se, ta sa- ‘this’; – Indefinite pronouns, e.g. TB ksa ‘some, any’; – Interrogative and relative pronouns, e.g. TB kuse, ta kus ‘who, which’; – Pronominal adjectives, e.g. TB allek, ta ālak ‘other’, TB makte, ta mättak ‘self’. Some of these are inflected according to gender, number, and case; namely, the demonstratives, the relative and interrogative pronoun TB mäksu ‘which’, the interrogative pronoun TA äntsaṃ ‘which’,1 the pronominal adjectives TB makte, ta mättak ‘self’ and TB allek, ta ālak ‘other’, the personal pronoun TA näṣ ‘I’ [fem. ñuk]. Others inflect according to number and case: the personal pronouns TB ñäś ‘I’ and TB twe, ta tu ‘you’, the interrogative and relative pronoun TA kus ‘which’, etc. Yet others inflect according to case only: the indefinite TB ksa, the interrogative TB intsu ‘who? which?’ etc. A synthetic table of the Tocharian pronouns is the following: Table 37

Tocharian pronouns

pronouns and pronominal adjectives gender-number-case Anäṣ ‘I’; Bse Asäs ‘this’; Bsu Asäm ‘(s)he’; Bsamp Asaṃ ‘that’; Bseṃ ‘±this’; Aäntsaṃ ‘which’; Bmäksu ‘which’; Bmakte Amättak ‘self’; Ballek Aālak ‘other’ number-case Bñäś ‘I’; Btuwe, Atu ‘you’; Akus ‘which’ case Bksa ‘some’; Bintsu ‘which’; Bkuse ‘which’

Since the main focus of this work is on the gender system, it follows that only those pronouns that display gender distinctions will be the topic of my investigation. From a comparative perspective, it is notable that the first person pronoun distinguishes a feminine form in Tocharian A (TA ñuk, see tg § 266–270), since no other ancient Indo-European language displays gender differentiation in

1 On the evolution of the interrogative pronouns TB intsu, ta äntsaṃ, see recently Peyrot (2018c).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

251

the first person pronoun.2 Indeed, the overall development of the first person pronoun and the source of the gender-distinction are still a matter of debate. Scholars agree in attributing the origin of the feminine form to a Tocharian A innovation (Jasanoff 1989a; Pinault 2008: 534). It will therefore not feature in my analysis. Feminine inflected forms of the interrogative pronoun TA äntsaṃ ‘which’ are only attested in the oblique singular äntāṃ (cf. A4 a4 äntāṃ tkanā “in which land/where on earth” and A70 a2 äntāṃ kälymeyaṃ “in which direction”). This pronoun can be traced back to PT *ən-se-nə (m.), *ən-sa-nə (f.), where *-se-, *-sa- are the reconstructed outcomes of the PIE demonstrative *só, *séh2 (see Peyrot 2018c, with references). In fact, the great majority of the Tocharian gender-differentiated pronouns follow the inflection of the demonstratives; indeed, the demonstratives form the base from which these pronouns derive. For this reason, in the following paragraphs we mainly deal with the evolution of the demonstrative pronouns in Tocharian. 4.3

Evolution of the Tocharian Demonstrative Pronouns

Cross-linguistically, pronouns play a pivotal role in the emergence of gender markers and in their subsequent evolution.3 Demonstratives in particular have a special function in the rise, the further development, and the possible decline of gender values (Corbett 1991: 310–311; Claudi 1997; Luraghi 2011). However, despite their importance, demonstratives have not been a central matter of discussion in recent works on the diachronic evolution of the Tocharian gender system (e.g. in Hartmann 2013, where the pronouns are not discussed). Nevertheless, the history of the demonstratives constitutes a fascinating topic within the study of Tocharian nominal morphology, and several peculiarities in their inflection and historical evolution remain to be accounted for. A case in point is the plural paradigm of the feminine, where, as I will argue, an essential issue has been overlooked.

2 The Tocharian A paradigm is actually even more noteworthy from a typological perspective. Aikhenvald (2000: 252–253) argues that: “If gender oppositions are found in 2nd person, they will normally also be there in 3rd, and if they are found in 1st, which is rare, they will normally also be there in 2nd and 3rd”. In addition, the gender distinction in the Tocharian A first person pronoun violates Greenberg’s Universal 44: “If a language has gender distinctions in the first person, it always has gender distinctions in the second or third or in both”. See further Plank & Schellinger (1997). 3 Parts of this section appeared in Del Tomba (2018).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

252

Chapter 4

In the first section (§ 4.3.1), I briefly introduce the synchronic paradigms of the Tocharian demonstratives from a functional and derivational perspective. In the second, central section (§ 4.3.2, § 4.3.3), I outline the synchronic distribution and the diachronic evolution of the singular and the plural inflection. Some important issues concerning the distribution of the plural forms and the reconstruction of cases of homophony within the paradigms will come to light. Finally, in the third section (§ 4.3.4), I summarise the evolution of the demonstratives, identifying the most significant modifications and subdividing them into chronological stages. Further remarks and suggestions will conclude the discussion (§ 4.3.5). 4.3.1 Introduction to the Tocharian Demonstratives Tocharian shows a wide range of demonstrative pronouns which can be classified according to both functional — i.e. spatial deixis — and formal patterns. However, the form and function of the Tocharian A pronouns do not pair with the respective form and function of those in Tocharian B. For instance, we find four different paradigms in Tocharian B and only three in Tocharian A. In the table below, the demonstratives are presented according to their function (Stumpf 1971; Kümmel 2015: 109–110):4 Table 38

Tocharian B and Tocharian A demonstrative pronouns

function

tocharian b

Anaphoric Proximal Remote Medial (?)

su, sāu, tu se, sā, te samp, somp, tamp seṃ, sāṃ, teṃ

≈ ≈ ≈

tocharian a

meaning

säm, sām, täm säs, sās, tāṣ saṃ, sāṃ, taṃ

‘he, she, the’ ‘this’ ‘that’ ‘± this’

Formally, the Tocharian demonstratives differ chiefly in their derivation and in the suffixes employed in the two languages. The basic stem is the descendant of the PIE pronoun *só (masc.), *séh2 (fem.), *tód (nt.), which can unambiguously 4 Regarding the origin of the Tocharian system of demonstratives, Kümmel (2015: 114) observes that some Middle Iranian languages — like Sogdian, Khotanese, and Tumshuqese — and Gāndhārī show a similar ternary system, classified according to deixis as neutral, near, and remote (Sims-Williams 1994; Emmerick 1989: 387–388). Kümmel proposes that the new Tocharian system is the outcome of a contact-induced change with these Middle Iranian and Middle Indian languages.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

253

be compared with Ved. sá ~ sáḥ, sā́, tát, Av. hə̄ ~ hō, hā, tat̰, Gk. ὁ, ἡ, τό, etc. Taking the nominative singular masculine form as an example, we can identify five fusional elements and outline the following six derivations (Pinault 1989a: 115–116): (1) (2) (3) (4) (5) (6)

TB se < PT *se < PIE *só; TB su < pre-TB *sə-w (cf. Skt. asaú ‘that’, Gk. οὗτος ‘this’); TB samp < pre-TB *sə-mpə (cf. TB ompe ~ omp ‘there’);5 TB seṃ, TA saṃ < PT *se-nə (cf. perhaps TB -ṃ, 3sg.prs.act.);6 TA säs < pre-TA *sə-ṣə (cf. perhaps TA -ṣ, 3sg.prs.act.); TA säm < pre-TA *sə-mə (cf. perhaps Skt. ayám ‘this’).

The three Tocharian A demonstratives resulted from the addition of various particles to the original basis PT *se-, *sə-, which itself represents the descendant of the PIE demonstrative pronoun. Although these kinds of developments are generally well understood (see recently Pinault 2009a), some of the inflectional patterns of the Tocharian demonstratives remain a matter of debate. In the following paragraph I will focus on the singular paradigm, before moving on to the plural paradigm.7 4.3.2 Paradigm of the Singular Considering the Tocharian B pronoun of proximal deixis and the basic shape of the demonstratives in Tocharian A, we can outline the following paradigm of the singular: 5 According to Pinault (2009a), the Tocharian B graphic cluster mp corresponds phonetically to [βə]. 6 The status of TB seṃ and its Tocharian A functional correspondent is debated. Stumpf (1971: 100–133 and 1976) maintained that TB seṃ was functionally equivalent to TB se, while Winter (1975) argued that it had a second person deictic function. Similar proposals were advanced by Peyrot (2008: 122–124), who followed Winter (1975) in attributing an intermediate deictic function to it, but Pinault (2009a: 226–229) concluded that it had an endophoric function. Finally, Kümmel (2015) has now demonstrated that TB seṃ was used primarily in cases of medial deixis in the historical period, with dominant recognitional use. However, in Proto-Tocharian, *se-nə had distal function, as in Tocharian A, and it subsequently acquired a medial deictic function in Tocharian B, when its original value was taken over by the new TB samp (cf. TB omp ~ ompe ‘there’), which is more marked compared to TA saṃ. 7 In both Tocharian A and Tocharian B, the demonstrative pronouns sporadically show dual forms in the masculine inflection. Given the fact that these are not relevant to our discussion, I do not consider the pronominal dual here. See Hilmarsson (1989a: 36–39), Pinault (2008: 542), and Kim (2018a: 61–63, 69, 85–87).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

254 Table 39

Chapter 4 Singular paradigm of the demonstratives

masculine

feminine

“neuter”

TB

TA

TB

TA

TB TA

sa-, säca-, cä-

sā tā

sātā-

te ta-, täte ta-, tä-

nom. sg. se obl. sg. ce

As is clear from the above, the inflection is marked by endings and changes to the stem. Both masculine and feminine distinguish the nominative and the oblique by means of different stems, with s- for the former and c- (m.), t- (f.) for the latter. Furthermore, the masculine and the neuter are disambiguated through the palatalisation of *t- to c-. The origin of this palatalised allomorph c-, which is also found in the masculine plural, is debated. Cowgill (2006) and Pinault (2008: 541–543) argue that it represents the regular outcome of PIE *te-, through a conflation of the o-grade, characteristic of the strong cases, and the e-grade, characteristic of the weak cases.8 Another possibility is that the c-forms originated from a combination including the pronoun *h1e (Skt. ayám, Lat. is, etc.), but a precise explanation of how this development may have worked is lacking.9 Be that as it may, the palatalisation must first have arisen in the demonstratives before it spread as a morphological pattern in the adjectival inflection (see § 4.6, § 4.7.1). As we have already suggested (see  §  2.2.2), a further peculiarity of the demonstratives is the preservation of some frozen forms which are formal remnants of the PIE neuter gender: e.g. TB te, ta ta- < PIE *tód (Skt. tát, Gk. τό, etc.). They are limited to the singular inflection. As thoroughly demonstrated by Stumpf (1971: 47–48), these forms must be explained as archaisms. They can be used only with pronominal function and never attributively. Strictly speaking, in a noun phrase the neuter demonstrative cannot be used as a modifier of a noun, i.e. with adjectival function (see the examples in § 2.2.2). Moreover, the distribution of the genitive singular markers between the masculine and the neuter is significant: the former ends in TAB -i, while the latter ends in 8 In particular, Pinault (2008: 541) reconstructs the c-stem from the genitive singular TB cpi /cwi, which in turn derives from an archaic dative singular *te-smōy > *cəzβu > *cəβə > TB cp-i / cw-i with further addition of the ending -i. In the feminine paradigm, the gen.sg. tāy consists of a basis tā- and the same genitive singular marker -y that we descriptively find in the three substantives of the śana-type with nom.sg. -a, obl.sg. -o, gen.sg. -oy. 9 For yet another proposal, see Winter (1980: 551–552).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

255

TB -ntse, ta -is. Whereas TAB -i may go back directly to a PIE ending (most likely, the dative singular *-ei̯, Pinault 2014a: 275–277; contra Klingenschmitt 1994: 365–369), the endings TB -ntse, TA -is are a Tocharian innovation: they go back to PT *-nse, which was originally the genitive singular of the nasal stems and subsequently spread to other inflectional types.10 The spread of this ending to the pronominal neuter inflection must therefore be a late phenomenon. This inflectional evidence is further proof of the non-adjectival use of the neuter demonstratives, since the gen.sg. TB -ntse, ta -s can only be found as a marker of nouns in Tocharian. The fact that Tocharian maintained some neuter forms in the demonstratives is significant. From a typological and cross-linguistic perspective, the maintenance of the singular forms of a decayed gender value in the pronominal system is expected. Indeed, when gender distinctions are lost, traces of a former gender distinction are most likely to be preserved in the pronominal system, if anywhere in the language.11 We can compare the situation in the modern Ibero-Romance languages, where so-called “neuter” forms are found in the pronominal system (Hall 1965; Priestly 1983: 343–344; Loporcaro 2018: 79–81). For instance, Spanish has unique neutral agreement forms for non-nominal controllers, like e.g. eso/ello (nt.), ella (f.), el (m.) ‘that’. Similar triplets are found in Catalan and Portuguese. In these languages there is no set of nouns for which these neuter forms would be a normal antecedent. They are formerly neuter from a diachronic perspective, but, synchronically, they are neutral as they cannot refer to any nominal controller (Corbett 1991: 214–216): there are no neuter nouns in Ibero-Romance.12 Therefore, the situation we find in Tocharian parallels that of Ibero-Romance: there are no neuter nouns in Tocharian with which the neuter forms of the demonstratives could agree. A further peculiarity that Ibero-Romance and Tocharian have in common is that of having continued frozen singular form of the neuter only. Once again, this is typologically unsurprising. Indeed, the categories of gender and number are frequently cumulated in the languages of the word, but it is much common for a language to have the category of number than that of gender (Plank 10

For an in-depth analysis of this ending, see Pinault (2008: 489–490) and Jasanoff (2019). For the evolution of the cluster PT *-ns(-), see § 4.8.1 and Huard (2020: 219–223). 11 See Corbett (1991: 310–311), Priestly (1983), Di Garbo & Miestamo (2019: 626) and the remarks in § 2.2.2 and § 4.1. 12 A peculiarity that sets Spanish apart from the other Ibero-Romance varieties is that Spanish makes use of a dedicated neutral form of the article for nominalisations, such as lo bueno ‘the good’ vs. el bueno ‘the good (guy)’, el buen trabajo ‘the good work’ (Corbett 1991: 215; Loporcaro 2018: 80–81).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

256

Chapter 4

& Schellinger 1997: 55–57). Moreover, gender distinctions are not randomly distributed across the number category, as gender in non-singular numbers often implies gender distinction in the singular. This has been most clearly codified in Greenberg’s Universals 37 and 45. If we read this claim from a diachronic perspective, genders are more often neutralised in the plural than in the singular (with some exceptions, see Plank & Schellinger 1997). Thus, the maintenance of frozen forms of the neuter in the singular paradigm of the demonstratives is in line with the diachronic development of gender values across the languages of the world. Returning to the Tocharian demonstratives, a phonological problem that should be mentioned is the doublet forms in the masculine singular, cf. TB se, ta sa- vs. TB sə- (in samp < *səmp and su < *sə-u), ta sä- (in säs and säm) and in the neuter TB te, ta ta- vs. TB tə-, ta tä-. The development of PIE *o to *ə is unexpected, but it is not without parallel, cf. TB mäksu ‘which’ < PT *mə-kwə-sə-u from *mo-kwi-so-u (Peyrot 2018c), and TB ompe ~ omp ‘there’ (Pinault 2009a), TB kete ~ ket ‘whose’, TB ate ~ at ‘away’, TB pest ~ päst, etc. If all these forms must be regarded as attesting the same development, we must agree with Peyrot (2008: 164–165, 2018c: 69) that an irregular centralisation of *-e to *-ə caused by the weak accentuation of these words took place (cf. the non-accented forms of the definite article in Ancient Greek). It is possible that the doublet *sə ~ *se also resulted from two competing proto-forms: the former would have been the descendant of PIE *só, while the latter would have been the outcome of a recharacterised nom.sg. *só-s. A similar s-variant can be seen in e.g. Skt. sáḥ, OAv. hə̄, YAv. hō, alongside Skt. sá, Av. hā.13 As far as the feminine inflection is concerned, the nom.sg. TB sā /sá/, ta sā- has clear cognates in other Indo-European languages, like Skt. sā́, Gk. ἡ etc. However, such a straightforward origin is problematic, since the regular outcome of PIE *-eh2 > *-ā should have been TB -o (see § 4.8.2.2). To my knowledge, five different explanations have been outlined to account for the nominative singular TAB sā: (1) shortening of the original *ā in accented monosyllables, thus PIE *séh2 > *sā > *să > PT *sa > TAB sā (as per Ringe 1996: 94–96); (2) loss of the laryngeal in pausa (Kuiper’s law), thus PIE *séh2 > *să(h2) > PT *sa > TAB sā (as per Pinault 2008: 542; Fellner 2014a: 13); (3) replacement of final PT *-o by *-a through analogy with the athematic inflection (as per Fellner 2014a: 13, but with hesitation); (4) TAB sā is not from *séh2 but rather from *sih2 (Kortlandt 2017: 101, 2020: 120); 13

See Pinault (2009a: 232–233) for yet another hypothesis. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

257

(5) lowering of PT *-o > *-a in monosyllabic Auslaut position (as per Kümmel 2009: 172–173). Explanation (1) seems quite improbable to me, since a long vowel is expected to be maintained in accented position. The analogical replacement of *-o to *-a  — explanation (3)  — is problematic, since in the adjectival inflection we find nom.sg. -ya and not -a (cf. nom.sg.f. TB astarya, ta āṣtri ‘pure’; see further § 4.7.1, § 4.8.2.3). The reconstruction of the loss of the laryngeal in pausa  — explanation (2) — remains a credibly explanation of the Tocharian forms (Pinault 2009a: 231), although the exact syntactic context in which the reduction would have taken place is unclear. Another option recently put forward by Kortlandt (2017: 101, 2020: 120) — explanation (4) — would involve the reconstruction of PIE *si-h2 (cf. possibly Goth. si, OIrish sí, Skt. sī-m, OAv. hī and the Sophocles’ hapax ἵ ‘she’) as the antecedent of TAB sā. However, this hypothesis comes at the cost of taking non-palatal *s- as analogical after the masculine and recurring to a restructuring of the inherited paradigms.14 In addition, the nom.sg.f. would be the only case form based on an alleged stem ending with *-i-h2, while all other case forms more clearly continue *-e-h2. Finally, I found neither strong evidence in favour of, nor strong counterevidence against, explanation (5), i.e. lowering of PT *o to *a. It should be noted, however, that monosyllabic particles and conjunctions very often undergo irregular and unexpected modifications in Tocharian. Clear cases are those of PIE *ē and *o, which give PT *a rather than expected PT *(y)e in unstressed monosyllables in absolute Auslaut (Kümmel 2009: 173; dtb s.v.): PIE *meh1 ‘not (neg.)’ > *mē > *me > TB ma, PIE *kwiso > *kwse > *kwsa > TB ksa ‘some, any’, PIE *weh1 ‘or’ > *wē > *wa > TB wa ‘therefore, nevertheless’ (cf. possibly *nuh1 > *nū > TB no, ta nu ‘however, but’). Following Van Windekens (1941: 29), Hilmarsson (1996: 58) suggested that the interrogative adverb TB kā ‘why’ presumably reflects a PIE instr.sg.f. of the type Lat. quā ‘in what way, how’, Gk. πῇ ‘id.’ (see also dtb 158). If so, this would be another case of lowering of PT *o (< *-eh2) to *a in monosyllables. Note that nom.sg.f. TB sā /sá/ and obl.sg.f. TB tā /tá/ are frequently attested as unaccented sa /sa/ and ta /ta/ in the texts. 14

For the reconstruction of PIE *sih2, see Sihler (1995: 389), Kloekhorst (2008: 750–751), Kortlandt (2017: 100–101). According to Fellner (2014a: 14), the reconstruction of PIE *sih2 is phonologically (but not comparatively) possible, given the fact that he does not accept that the suffix *ih2 could have palatalised the preceding consonant. See also de Vaan (2019), who explains Goth. si and OIrish sí ‘she’ as new formations from the PIE anaphoric pronoun nom.sg. *(h1)ih2 (feminine of *h1e) plus *s- (from *so, tod). Note that the PIE anaphoric pronoun has left virtually no traces in Tocharian. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

258

Chapter 4

Moving now to the oblique singular, TB tā shows phonological problems closely related to those seen in the nominative singular. Indeed, an outcome TB **to from PT *to(m) < PIE *téh2-m should be expected, since in internal position *-eh2- should have yielded PT *-o- > TB -o-.15 Considering that a shortening of the original *ā in an accented monosyllable is quite improbable, TB tā may be the result of either a lowering of *-o- to *-a- or an analogical replacement of *to after the new nominative singular *sa (Pinault 2008: 542). The reason for this analogical replacement involves the diachronic development of the plural paradigm of the feminine and the neuter. We will focus on these and other problems in the following paragraph. 4.3.3 Paradigm of the Plural In the plural, Tocharian A shows a rigid system with clear formal markers (tg § 287): Table 40

Tocharian A plural paradigms

deixis Anaphoric säm Proximal säs Remote saṃ

nom. obl. nom. obl. nom. obl.

masculine

feminine

cem cesäm ceṣ cesäs ceṃ cesäṃ

tom tosäm toṣ tosäs toṃ* tosäṃ

From a synchronic point of view, these paradigms are easy to describe. All enclitic elements (-m, -ṣ/-s, -ṃ) are added directly to the basic shape of the pronoun, which attests the c-allomorph in the inflection of the masculine and the t-allomorph in the inflection of the feminine. In all the oblique plural forms, we note ä-epenthesis between the ending TA -s and the enclitic. In the pronoun of proximal deixis, Pinault (2008: 540) suggests that the final sibilant undergoes morphological palatalisation in the nominative plural. However, a different explanation is also possible: the original enclitic element was the 15

For the outcome TB -o from PIE *-eh2m compare TB ṣe ‘1’ with its obl.sg.f. somo, which is from PT *semo < PIE *someh2-m. The plural TB somona, ta ṣomaṃ ‘some’, evidently less frequent than the singular, goes back to the same Proto-Indo-European stem. See Pinault (2006a) for an in-depth discussion of the paradigm of both masculine and feminine inflections of the Tocharian numeral ‘1’. See also Adams (dtb 722) and Winter (1992: 98–103) for further suggestions. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

259

palatalised sibilant *-ṣ, which was depalatalised through assimilation in all the allomorphs with initial or internal (-)s- (as per Pedersen 1941: 116 and Kortlandt 1983: 320–321, cf. also the numeral TA sas ‘1’ < pre-TA *ṣas [tb ṣe]). Although the nom.pl.f. of the pronoun of remote deixis is not attested, it can easily be determined as TA toṃ* on the model of the other paradigms. In Tocharian B the situation is more complex, because three out of the four demonstratives that are differentiated in the singular have just one paradigm in the plural. Indeed, the only pronoun that features a formally distinct paradigm is TB sam(p): Table 41

Plural paradigm of TB sam(p)

deixis Remote samp

nom. obl.

masculine

feminine

caim(p) ~ ceym cemp*

toym* toym

Compared with the other demonstratives, the paradigm of TB sam(p) is the least frequently attested. This is true especially for the plural inflection. For the masculine, I have found only eight occurrences of the nominative plural: two in the London collection (caim in IT248 b4 [class.], IT899 a2 [class.]), one in the Paris collection (caim in AS17K a4 [class.]), and five in the Berlin collection (caimp in B83 a6 [class.], B85 a3 [class.-late], B88 a6 [class.], ceym in B107 b2 [late], caim in THT2381.e b2 [frgm.]).16 A form TB cem occurs once in AS16.7 b5, which, according to the above analysis, should be an oblique plural. However, the context requires a nominative plural instead: /// cem wa nraine tsäksenträ “nevertheless, those burn in hell”. Therefore, TB cem may be a late variant from caimp (cf. Georges-Jean Pinault and Melanie Malzahn apud CEToM). No oblique plural forms are attested: we have only one occurrence of the genitive plural form in B85 a2 (Araṇemijātaka) mā ñiś cempaṃts rakṣatsents aiṣṣäṃ “he must not give me to those rākṣasas!” (Schmidt 2001: 313). This form allows us to reconstruct with greater certainty the oblique plural of the masculine inflection: according to Krause & Thomas (teb § 269) it might have been ceympa* (?), but, looking at the genitive plural cempaṃts /cempə́ nts/, which must have been built on the oblique form, it was probably cemp*, from *cen-mp. The feminine plural paradigm is even more difficult to determine, since I have found only one plural form, the oblique toym in B19 a1 (Udānālaṅkāra) 16

As pointed out by Stumpf (1971: 133–134), the great majority of the attested forms of TB samp are from the Araṇemijātaka. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

260

Chapter 4

toym läklenta lkātsi “to see those sufferings”. No genitive plural forms or secondary cases are attested. Now, if we look at the plural forms of the other Tocharian B demonstratives, several difficulties come to light. In the following, I will summarise and compare two different hypotheses on this matter. I will then propose a new analysis in support of one of them. According to the classical view of Krause & Thomas (teb §§ 266–269), the three Tocharian B demonstrative pronouns of anaphoric, proximal, and remote deixis have three different sets of paradigms in the plural. See the table below: Table 42

Tocharian B plural paradigms (TEB §§ 266–268)

Deixis Anaphoric su Proximal se Remote seṃ

nom. obl. nom. obl. nom. obl.

masculine

feminine

cai, cey ceṃ cai, cey ceṃ cai, cey ceyna, cenäṃ

toṃ toṃ toy toy toyna toyna

A similar description of the paradigms can also be found in more recent literature and handbooks on Tocharian. As one can see, the paradigm of the masculine is the same in the three sets, with nominative and oblique differentiated. The only exception concerns the oblique plural of the pronoun of remote deixis seṃ, which is TB ceyna ~ cenäṃ.17 On the other hand, the paradigm of the feminine plural is quite peculiar: it never distinguishes the nominative from the oblique, but it shows different forms in the various pronominal inflections. However, a closer inspection of the language of the documents in which the various forms are attested allows for a different analysis. Stumpf (1971, 1974, 1976, 1990) has dealt with the Tocharian demonstratives, providing innovative insights both on their forms and functions. In a pioneering article (Stumpf 1974), he claimed that Tocharian B did not display any differences between se, seṃ, and su in the plural. As a consequence, Tocharian A and B would 17

The obl.pl. cenäṃ is only sporadically attested (in AS19.21 a5 [class.], THT2291 b2 [frgm.], and NS355 b4 [class.; but cf. ceṃ in the parallel text B85 b4]). This form can be interpreted as either a recharacterised obl.pl.m. or as a real occasional attestation of a ṃ-form plural of the regular TB ceṃ. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

261

differ significantly in the formation of the plural inflection of their respective demonstratives, since Tocharian B would not display any formal diversification in the plural paradigm of the pronominal sets. This analysis obviously stands against the traditional one of Krause and Thomas. Stumpf (1974) explains the different forms of the plural within the framework of a restructuring process from archaic to late Tocharian B. In recent years, this hypothesis has been closely evaluated and further confirmed by Peyrot (2008: 124–128). As for the masculine paradigm, Peyrot (2008: 58) claims that in word-final position “we find in archaic texts only ai, in classical texts both ai and ey, and in colloquial texts only ey”. One of the examples adduced by Peyrot is namely the nom.pl.m. of the demonstratives cai > cey, for which the distribution of the variants may be more intricate, however. Indeed, both nom.pl.m. cai and cey are found over all Tocharian B stages, from archaic to late and colloquial texts; however, cai is by far the most common variant, as it is attested seven times more frequently than cey. In particular, taking into consideration those texts for which a linguistic stage can be established, the following occurrences of TB cai and TB cey can be identified: cai: 21× in archaic texts (AS12c, B130, B132, B133, B138, B139, B140, B274, B275, B285, B295, B388, B389, B545, B587.a, THT1286, THT1309, THT1860, THT2247, SI B 119 (3), Or 8212.163), 5× in archaic-classical texts (AS7G, AS7H, NS39, B37, B212), ca. 110× in classical texts, 6× classical-late texts (AS14.2, AS8a [caiy], NS56, B231, B506, IT5), 7× late texts (B197, B330, THT1468, NS55, AS14.1.1, DA M 507.38 [caiy], SI Streklov D2 [caiy]), and 10× in colloquial texts (Taj 02, Dd 5.1, SI Strelkov-D84.a, Bois B57, Bois B58, Bois B70, Bois B114, THT2751.a, Si M-TD/31δ, SI BToch 12). cey: 6× in archaic texts (B255, B256, B362, B365, B587.a, IT6), one archaic-classical text (AS7G), 11× in classical texts (AS1B, AS7B, AS7E, AS7C, B363, B408, B431, B565, B585, B589, IT127), 4× in late texts (B331, B108, B347, B375), and 2× in colloquial texts (G-Qm 1, Bois B92). Thus, cai seems to be the standard variant in all Tocharian B stages. A general scheme of the distribution of the two variants with the proportion by which each is attested is offered in the graph below:18

18 The y-axis refers to the percentage of fragments attesting a given form. If, in the same text, more than one occurrence of the same variant is attested, it has not been reported in the graph. The x-axis refers to the linguistic stage of the fragments according to Peyrot (2008). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

262

Chapter 4

100 90 80 70 60 50 40 30 20 10 0

archaic

archaic-classical

classical

cai Graph 1

classical-late

late

colloquial

cey

Distribution of the masculine plural variants in the history of Tocharian B

As for the oblique plural, TB ceṃ is attested in archaic, classical, and late texts, while TB ceyna is only attested in classical, late, and colloquial texts (e.g. in B108 b3, b6, b9; B325 b1; B375 b4, b5; PK Bois B101 a1). From a diachronic perspective, TB cai, cey must be the regular outcome of PIE *tói̯ (Ringe 1996: 86, cf. Skt. té, Gk. οἱ), while TB ceṃ is from PIE *tóns (cf. Skr. tā́n, Gk. τούς). The palatalised allomorph c- is a Tocharian innovation. Going back to Tocharian A, the nominative plural ce- shows regular monophthongisation of the diphthong PIE *-oi̯ > TA -e, while the oblique plural continues PIE *tons >> PT *cens (with morphological palatalisation) > pre-TA *ces > TA ces-. In view of the larger number of variants, it is not surprising that the distribution of the feminine plural is more difficult to outline. Following Stumpf (1974, 1990), Peyrot (2008: 126–127) convincingly argues that TB toṃ is the old plural form (both in the nominative and in the oblique), since it mostly occurs in archaic and classical texts. I have found the following attestations of TB toṃ in archaic documents: toṃ (28×): AS7N b5; B117 a7, B117 b3; B123 b5; B127 a2; B128 a4; B133 a3; B133 a5; B137 a7; B274 a1; B275 a3; B284 a5; B338 b3; B338 b4; B341 b5; IT22 a7 (?); IT80 a2; IT157 a2; Or8212.163 b5bis; Or8212.163 b6; THT1254 a4; THT1450.a b5 (?); THT1535.a a6; THT2247 a3; THT2247 b4; THT2371.g b2; THT3597 a2; to(ṃ) (1×): B240 b1; toṃn (1×): B274 a3; ton (3×): SI B 120(1) a4; B291.a a1; B365 a2; toṃtsā (1×): B563 b6; tontsā (1×): B135 a4; tonmeṃ (1×): B274 a4; tonts (1×): B274 a1.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

263

Gender in the Inflection of the Pronoun and Adjective

On the other hand, the nom.pl. toy and the obl.pl. toyna are both new formations. The former might be attested only once in a fragmentary archaic document (IT853 a2, cf. the spelling träṅko at line a3), but it became the standard nominative form in classical-late texts. Finally, on the model of toy, a new nom.obl.pl. toyna was created, which is the common form in late texts. One can also compare the similar distribution of the feminine plural variants of the interrogative and relative pronoun TB mäksu, which is formed by TB su: mäktoṃ : Or8212.163 b5 [arch.]; NS54 a3 [class.], THT2386.j and .s a5 [class.]; IT174 a6 [class.] (mäkt(o)ṃ); IT733 a3 [class.], THT1603.a b2 [class.] ((mä)ktoṃ); NS76 b5 [class.] (mä(kt)oṃ); mäktoynas : B199 b1 [late]. Indeed, in the history of Tocharian B, the ending -na has become the ubiquitous marker of the feminine plural in the adjectival inflection, and in the śana- and aśiya-types of nouns referring to female entities (see § 3.4). A general scheme of the distribution of the variants is offered in the graph below:19 100 90 80 70 60 50 40 30 20 10 0

archaic

archaic-classical

classical

toṃ Graph 2

classical-late

toy

late

colloquial

toyna

Distribution of the feminine plural variants in the history of Tocharian B

19 The graph shows the number of attestations of the feminine plural variants in the Tocharian B texts. The y-axis refers to the percentage of fragments attesting a given form. If, in the same text, more than one occurrence of the same variant is attested, it has not been reported in the graph. The x-axis refers to the linguistic stage of the fragments according to Peyrot (2008).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

264

Chapter 4

4.3.4 Origin of the Feminine Plural At this point, a central question that needs to be answered is where the archaic form TB toṃ and the Tocharian A feminine plural paradigm (nom.pl. to-, obl.pl. tos-) come from. Their origin and historical evolution have never been precisely investigated, although they certainly constitute a problematic issue within the development of the demonstratives and the analysis of the system of gender in Tocharian. Let us start our discussion with Tocharian B. In view of the variant TB tonak, a first hypothesis which might explain the archaic form toṃ is that it is a phonetic development of an original *tona, where *to- would be the “regular” PIE outcome, recharacterised by the ending -na. This ending is indeed the most productive plural marker of the feminine in the adjectives. However, the sequence tona- is not attested elsewhere, and TB tonak can be analysed as /tonə́ kə/, which is from *ton + the emphatic particle *-kə (with ə-epenthesis), rather than /tónak/ (Thomas 1984: 224; Peyrot 2008: 126). Furthermore, if toṃ derived from *tona, we should postulate an ad hoc apocope, since the sequence -na in Tocharian B is always maintained in the nominal inflection, and final -a is not apocopated anywhere else. This hypothesis is thus to be rejected. As other inner-Tocharian explanations are doubtful, I believe it preferable to postulate an Indo-European source for these forms. In my opinion, the final nasal in TB to-ṃ is the regular outcome of the Indo-European accusative plural *-ns. Tocharian A confirms this hypothesis, since the obl.pl.f. TA tos- (cf. tos-äm, tos-äs, tos-äṃ) can go back to the same proto-form: both TA tos- and TB toṃ allow us to reconstruct an ancestor PIE *téh2-ns > *tāns (UrIG § 24., p. 138).20 The outcome of the PIE accusative plural *-ns (> TB -ṃ, ta -s) is clearly attested in the nominal inflection, where the historical interpretation is widely accepted. Alternatively, one might want to explain TB toṃ, ta tos- as the result of an analogical development on the basis of the masculine obl.pl. TB ceṃ, ta ces-. However, analogy is in my view unnecessary. Since in the masculine *-ns developed into TB -ṃ /-n/, ta -s, we would expect the same correspondence for the feminine (but see also § 4.8.2.2). The vowel match in TB toṃ : TA tos- may be problematic, since it is generally assumed that the outcome of PIE *eh2 yielded a in Tocharian A. However, the correspondence TB o : ta o is characteristic of a well-known group of words in which corresponding vowels in Tocharian B and Tocharian A partially violates the generally assumed evolution of PIE *-eh2- > PT *-o- > TB -o-, TA -a- (e.g. 20 I do not believe that the expected outcome of PT *tons is TA *tes, through intermediate *tois (see  §  4.8.1). In any case, the o-vocalism could have been adopted from the nominative.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

265

PIE *bhréh2tēr > TB procer, ta pracar ‘brother’). Even though they do not refer to the demonstratives, Burlak & Itkin (2003) have highlighted the fact that TB o matches TA o mostly when this vowel appears in initial syllables in Tocharian A. This is particularly evident in monosyllables, as in TB kos : ta kos ‘how much’; cf. also TB moko : ta mok ‘old’, TB pont- : ta pont- ‘all’ (Burlak & Itkin 2003: 28; Burlak 2000: 137–140). To this list, we can add without any difficulty the feminine plural of the demonstratives TB to- : ta to-. This further confirms that PT *o regularly yielded (or has been maintained as) TA o in monosyllables. One problem remains to be solved. The fact that Tocharian B, since the archaic stage, attests a nom.pl. toṃ formally identical to the oblique does not match the Tocharian A counterpart, where we find nom.pl. TA to- as the regular outcome of *téh2-es > *tās.21 The same outcome should also be expected in Tocharian B. In other words, we do not have any formal descendant of the reconstructed Indo-European nominative plural feminine in Tocharian B. The nom.pl.f. toṃ must therefore be a secondary Tocharian B innovation. In my view, the only plausible explanation is to reconstruct an analogical development, according to which the historical obl.pl.f. toṃ spread to the nominative plural in a pre-Tocharian B stage. Indeed, various reasons for this analogical development can be envisaged. To begin with, it is to be expected that certain forms of the feminine pronominal paradigm became homophonous in the prehistory of Tocharian. Most importantly, the oblique singular and the nominative plural feminine should have become identical after the loss of final *-m and *-s. In order to obviate to these coalescences, analogical replacements took place in unattested phases of Tocharian B, perhaps beginning as early as Proto-Tocharian, to both disambiguate the forms of the paradigm and to favour formal isomorphism of the stem. From a hypothetical PT *to (< PIE *téh2-m), parallel to PT *allo- (< PIE *ali̯eh2-m), a new oblique singular TB tā /tá/ was created, either by analogical levelling from the nominative singular TB sā /sá/ or by lowering.22 The expected neuter plural PT *to < PIE *téh2 was apparently lost, since we have only the singular of the neuter preserved. If the neuter plural survived into pre-Tocharian B, this additional homophony may further have favoured the creation of the new nominative plural toṃ. Be that as it may, this new feminine plural paradigm follows a general Tocharian B trend of development according 21 An example of nom.pl.f. TB toṃ in an archaic document is toṃ läklenta tne cmelants ṣärmtsa mäskenträ “these sufferings are here because of the rebirths” (B284 a5). 22 The comparison of TA tā- and TB tā suggests that the supposed evolution *to > *ta had taken place already in Proto-Tocharian. However, it cannot be excluded that the same development occurred independently in the two Tocharian languages.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

266

Chapter 4

to which the plural inflection of the feminine shows no difference between nominative and oblique in both adjectival and pronominal declensions. As we will see, a closer look at the feminine paradigm of TB allek confirms the evolution outlined above (see § 4.4). 4.3.5 Evolution of the Tocharian Demonstratives In this concluding section I summarise the diachronic evolution of the inflection of the Tocharian demonstratives, subdividing the analysis into four parts: (1) (2) (3) (4)

from Proto-Indo-European to pre-Proto-Tocharian; from pre-Proto-Tocharian to Proto-Tocharian; from Proto-Tocharian to Tocharian A; from Proto-Tocharian to Tocharian B.

I use a distinction between pre-Proto-Tocharian and Proto-Tocharian in order to distinguish evolutions that presumably took place in different non-attested chronological stages. I similarly refer to pre-TA and pre-TB to reconstruct transitional phases. Table 43

From Proto-Indo-European to pre-Proto-Tocharian

masculine PIE sg. nom. acc. pl. nom. acc.

*só *tóm *tói̯ *tóns

> > > >

feminine

pre-PT

PIE

*se *te(m) *tey *tens

*séh2 *téh2m *téh2s *téh2ns

> > > >

neuter

pre-PT

PIE

*so, *sa *to(m) *to(s) *tons

*tód *tód *téh2 *téh2

pre-PT > > > >

*teT *teT *to *to

Before the split of the two languages from Proto-Tocharian, most of the characteristic phonological developments of the vowel system had been completed. In this phase we can reconstruct: (1) the general loss of vowel length in the vowel system; (2) PIE *o > PT *e; (3) PIE *eh2 > PT *o; (4) the merger of the PIE series of stops into a single voiceless series (here, PIE *d > PT *T).23 23 The diachronic evolution of PIE *d in Tocharian is particularly difficult (see Winter 1962b). In a non-palatalising context, the regular outcome was PT *ts (e.g. PIE *der- ‘to

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

267

Gender in the Inflection of the Pronoun and Adjective

The different outcome of PIE *séh2 depends on the possible interpretations of TAB sā /sá/ pointed out in § 4.3.2: it is either the outcome of the loss of the laryngeal through Kuiper’s law, or it first became *so and then *sa by lowering in word-final position. Table 44

From pre-Proto-Tocharian to Proto-Tocharian

masculine pre-PT sg. pl.

nom. acc. nom. acc.

*se *te(m) *tey *tens

> → → →

feminine

PT

pre-PT

*se *ce *cey *cens

*so, *sa *to(m) *to(s) *tons

> > > >

neuter

PT

pre-PT

*sa *ta *to *tons

*teT *teT *to *to

PT > > > >

*te *te — —

In this phase, two important modifications took place: (1) the generalisation of the palatalised stem c° in all the t-cases of the masculine inflection (cf. § 4.3.2); (2) the gradual loss of the neuter plural, which started in a Proto-Tocharian phase. If I am correct in proposing that TB toṃ and TA tos- go back directly to PT *tons, it is impossible that the neuter plural became homophonous with the entire paradigm of the feminine plural. Instead, the neuter plural PT to became homophonous with the oblique singular and with the nominative plural of the feminine inflection and subsequently lost its function.24

24

split’ > PT *tsər- > TB tsǝr-, ta tsär- ‘to be separate’). Other outcomes may be: (1) PT *∅ in some consonant clusters (before *i̯, *u̯ , *r, e.g. PIE *du̯ oh1 ‘two’ > *dwū > PT *wu > TA wu); (2) PT *t in some other consonant clusters (e.g. PIE *neu̯ d- ‘to push’ > *nət- + -sk- > PT *nətk- > TB nǝtk- ‘to thrust away’). See also Ringe (1996: 64–66 and 146–147). With regard to the other demonstratives, it is possible that the Tocharian A demonstrative of remote deixis saṃ and the Tocharian B demonstrative of medial deixis seṃ were created before the split of the two languages from Proto-Tocharian: the original value of *se-nə was remote deixis, which was maintained in Tocharian A and further reinterpreted as medial deixis in Tocharian B. It is probable that a chain shift took place when the new demonstrative TB samp was created and caused the reanalysis of TB seṃ.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

268

Chapter 4

Table 45

From Proto-Tocharian to Tocharian A

masculine PT sg. pl.

nom. obl. nom. obl.

*se *ce *cey *cens

> > > →

feminine

TA

PT

sacaceces-

*sa *ta *to *tons

> > > >

neuter

TA

PT

TA

sātātotos-

*te(T) *te(T) — —

> > > >

tata— —

As we have already seen, Tocharian A recharacterised the basic outcome of the demonstrative by adding the enclitic suffixes *-mə (anaphoric), *-ṣə ~ *-sə (proximal) and, perhaps, -nə (remote). As far as the phonological evolution is concerned, we note regular monophthongisation of the Proto-Tocharian diphthong *ey > TA e in the nom.pl.m., and regular outcome of PT *e > TA a (e.g. PIE *ǵombho- ‘tooth’ > PT *keme > TA kam). Both masculine and feminine oblique plural forms continue the ending PIE *-ns by sound-law *-ns > -s. Table 46

From Proto-Tocharian to Archaic Tocharian B

masculine PT sg. pl.

nom. obl. nom. obl.

*se *ce *cey *cens

> > > >

feminine

TB

PT

se ce cey, cai ceṃ

*sa *ta *to *tons

> > → >

neuter

TB

PT

sā tā toṃ toṃ

*te *te — —

TB > > > >

te te — —

In Tocharian B, the situation is less straightforward than in Tocharian A. Several analogical replacements took place which worked to diversify the paradigm and create formal isomorphism. In the table above I outline the evolution from Proto-Tocharian to archaic Tocharian B. The singular paradigm does not show any substantial modifications over the course of the evolution of the language. We now turn to the evolution of the plural paradigm from archaic to late Tocharian B, as schematised in Table 47 (adapted from Peyrot 2008: 127).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

269

Gender in the Inflection of the Pronoun and Adjective Table 47

Evolution of Tocharian B plural paradigm

stage

nom. pl. m.

obl. pl. m. nom. pl. f.

obl. pl. f. ling. phase

i ii

*cey *cey

*cens *cen

*to *to

*tons *ton

PT pre-TB

iii iv v vi vii

cai, cey cai, cey cai, cey cai, cey cai, cey

ceṃ ceṃ ceṃ ceṃ cey-na

toṃ toṃ to-y toy toy

toṃ toṃ toṃ toy-na toyna

Archaic TB

Late TB

In the masculine inflection, we see the preservation of the diphthong *cey > cai, cey in the nominative plural (stages i–iii). The oblique plural of the feminine TB toṃ is the regular outcome of PT *tons < PIE *téh2ns (stages i–v). However, the homophonous nominative plural form TB toṃ cannot go back directly to PIE *téh2-es: the regular outcome should have been TB **to (stage iii). The original oblique plural was generalised to the nominative plural when the nominative was not well characterised and possibly homophonous with both the oblique singular and the neuter plural (stage iv). In the historical development of Tocharian B, new inflected forms were created. Firstly, the nominative plural masculine TB cey was subject to reanalysis: -y was reanalysed as a nominative plural marker and spread to the feminine plural. As a result, a new nominative plural feminine toy was created (stage v).25 Finally, in classical and late Tocharian B, a new oblique toyna was formed by the addition of -na, the plural marker of the adjectival feminine inflection, to a basis toy- (stages vi-vii). This element -na was further reanalysed as the oblique plural marker in the pronominal inflection, and it also spread to the oblique plural of the masculine (stage vii). I believe that TB sam(p) also points to the same evolution. Indeed, next to the nominative plural caim(p), one occurrence of the variant ceym is attested in B107 b2, a well-preserved document drafted in late Tocharian B. No oblique plural is directly attested, but we can reconstruct cemp* on the basis of the 25

According to Peyrot (2008: 126), of a sample of 33 attestations, 19 are nominative, and only one (in B504 a4) is an oblique, probably a mistake (the other 13 attestations come from fragmentary documents where the case is unclear).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

270

Chapter 4

genitive plural cempaṃts (§  4.3.3). With respect to the feminine, we would expect a form tomp* as the mp-variant of toṃ through assimilation of the dental nasal before the labial nasal. This reconstruction is confirmed by the oblique of the masculine cemp* < cen- + -mp. After checking all the attestations of tom(p) and toym(p) in order to evaluate whether they may be plural variants I found only three attestations: tom in B42 a4, which is undoubtedly an oblique singular, because it agrees with abl.sg.f. arṣāklaimeṃ; tomp in AS17K b5, which is used with pronominal function in a context that seems to require a singular; and, finally, the aforementioned toym in B19 a1.26 It seems to me that the plural paradigm of samp was thus affected by the same modifications that we have seen for the other demonstratives: an original nom.pl. tom(p), reconstructed for phonological reasons, evolved into toym(p). This analysis highlights the fact that the plural paradigm of TB sam(p) differs from the others solely by the presence of the enclitic particle -m(p): the inflection and the evolution of the various endings are the same as those of the other demonstratives. 4.4

Evolution of the Pronominal Adjective TB allek, TA ālak ‘other’

The paradigm of TB allek, ta ālak ‘other’ contains certain peculiarities, displaying as it does an inflection halfway between that of the demonstratives and that of the so-called “thematic” adjectives. The aim of this section is to clarify how the inflection of this pronominal adjective evolved from Proto-Indo-European to Tocharian. As will become clear, the historical evolution of TB allek, ta ālak ‘other’ has much in common with that of the demonstratives, especially their feminine inflection. The etymological connection of TB allek, ta ālak ‘other’ with Gk. ἄλλος, Lat. alius, OIrish aile, Arm. ayl, Goth. aljis, etc. was among the very first insights into Tocharian (cf. the equation “ālyeḵ = alius” in Sieg & Siegling 1908: 927). All these cognate formations can be traced back to PIE *ali̯os (cf. perhaps also Ved. anyá- ‘other, different, alien’, Av. aniia-, if n instead of l can be a 26

The obl.pl.f. toym (< *toymp) is somewhat peculiar, since TB toy usually serves as a nominative plural. In my view, this problematic form can be interpreted in two ways: (1) the expected obl.pl. **tomp < *tonmp was replaced by toym(p) at an early stage, because it would have been homophonous with the attested obl.sg. tomp; or, (2) if the obl.pl. toy in B504 a4 is to be taken seriously, then the obl.pl. form toym(p) could be interpreted as the “regular” pre-form of a later toynamp*. These two proposals are not mutually exclusive. Admittedly, the analysis of this pronominal set is challenging because we have only one attestation of the feminine plural paradigm. More generally, too few forms are attested to establish the evolution of the paradigm from archaic to late Tocharian B.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

271

secondary replacement, as per Mayrhofer EWAia 1.80).27 The singular and plural paradigms run as shown in the following table (Winter 1992: 151–152; Peyrot 2008: 127): Table 48

Paradigm of TB allek, ta ālak

masculine

sg. pl.

nom. obl. nom. obl.

feminine

TB

TA

TB

TA

allek alyek alyaik alyeṅkäṃ

ālak ālyakäṃ ālyek ālykes

alyāk allok alloṅk(na) alloṅkna

ālyāk ālyäkyāṃ ālkont ālkont

The Proto-Tocharian paradigm is quite difficult to reconstruct, since Tocharian A and B do not match in more than one case form, particularly in the feminine inflection. There exist a number of variant and misspelled forms in Tocharian B. Let us start with the paradigm of the masculine. According to Krause & Thomas (teb § 282), Pinault (2008: 548), and Fellner (2017: 156 fn. 33), the nom.sg.m. allek has a variant alyek, but I was not able to find clear evidence for this form. Even if some occurrences of a nom.sg.m. alyek really existed, they would not have been sufficient in number to claim that alyek was a real variant of the regular allek. As far as the obl.sg.m. is concerned, Peyrot (2008: 127–128) points out that alongside the regular alyek we find one example of alyeṅk, attested in B346 a6 (late). He argues that the nasal may have been taken from the obl.pl.m. Otherwise, it may have been analogically introduced after certain thematic adjectives in obl.sg.m. TB -eṃ /-en/.28 Also, in the plural, an isolated nom.pl. alyaiṅk is found in B580 b1 (late frgm.), which may have acquired the nasal from the rest of the plural paradigm (cf. obl.pl.m alyeṅkäṃ and pl.f. alyoṅk-). 27 Pace Adams (dtb 31), there is no need to reconstruct PIE *h2el-no- ‘that, yonder’ as the ancestor of the Tocharian forms. He further compares Tocharian with Lat. ollus ‘that’, OIrish ol ‘beyond’, OCS lani ‘in the past year’, but this connection is difficult for both semantic and comparative reasons (all these forms clearly point to an o-grade *(h2)ol-no-). 28 The forms alleksa in B42 b4 wnolm=alleksa and IT24 b1 nanw alleksa are not to be interpreted as perl.sg. but as sandhi variants of allek ksa (cf. IT137 a2: /// (a)llek ksa käryorttau lyakā-ne istak /// “… a certain merchant saw her. Suddenly …”, cf. Ogihara 2009: 403).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

272

Chapter 4

The singular paradigm of the feminine does not display any relevant variant. An obl.sg.f. alyok is sometimes mentioned (cf. e.g. teb  §  282; Pinault 2008: 516). Winter (1992: 151) hesitantly gives this variant as attested in B244 a1 (Buddhastotra, class.), /// (a)lyok weś(e)ñ(ai)sa brahmasvar “with another brahmasvara-sounding voice”, but the initial part of the lacuna is probably to be restored as (uppā)l-yok weś(e)ñ(ai)sa brahmasvar “with [his] brahmasvara-sounding lotus-voice” (as suggested by Georges-Jean Pinault apud CEToM). As a consequence, variants may only be found in the feminine plural paradigm. Peyrot (2008: 127–128) claims that alloṅk (with the graphic variant alloṃṅk in B173 a5) is only attested as a nominative, alloṅkna (frequently written alloṅna; cf. also ālloṅkna in B45 a2) only as an oblique, while the morphological hapax alloykna in B200 a1, though unclear in the case, can successfully be compared with the late oblique plural toyna ~ mäktoyna. In essence I agree with Peyrot’s paradigm. However, we must also recognise that the attestations of the nom.pl.f. are few, and are by no means conclusive (three certain attestations in total, ālloṅ[kna] in B133 a5, alloṅk in B379 b2, and alloṃṅk in B173 a5). There is one further form that Broomhead (1962: 1. 24) reads alloṅna, in a context that clearly requires a nominative: IT195 a6 (literary) /// yerkwantalañ mewīyañ alloṅna lwāsa leopard(?).nom.pl tiger(m).nom.pl other.f.nom.pl animal(a).nom.pl św(ātsi) food.inf “Leopards (?), tigers, and other animals (crave for?) food”. Conversely, Peyrot (l.c.) reads alloṅtä in this line, which he interprets as a mistake either for alloṅna or for alloṅkä. However, there is not sufficient evidence for arguing that a nom.pl.f. alloṅkna did not exist. Turning back to the historical evolution of the paradigm, a peculiarity of this pronominal adjective is that in a Proto-Tocharian phase the emphatic particle *-kə was added to the base *alle-. This particle is often suffixed to pronominal and other deictic words (cf. TB ykāk ‘still, TB ṣek ‘always’, TA okāk ‘until’). As Winter (1992: 151–152) observes, it was initially inflected before the particle and, subsequently, after it as well (cf. gen.sg.m. alyekepi). What we see before -k should therefore be the expected outcome. This is particularly evident in the case of Tocharian B but less so in the case of Tocharian A. Indeed, Tocharian A has largely reshaped the inherited paradigm of ālak, following a general tendency to eliminate the ending before -k and move it after the enclitic, with subsequent generalisation of the nominative form as the basic stem (Winter

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

273

1992: 153). All Tocharian A variant forms can be explained in light of this development. Thus, we have: obl.sg.m. ālakäṃ (cf. nom.sg.m. ālak) vs. ālyakäṃ (for expected **ālyak); obl.sg.f. ālyäkyāṃ (< *ālyākyāṃ, cf. nom.sg.f. ālyāk); obl.pl.m. ālyekäs (cf. nom.pl.m. ālyek) vs. ālykes (for expected *ālyesk), etc. The plural paradigm of the feminine was totally remade by the addition of -ont to the stem ālk- (< *ālak-). A similar recharacterisation also affected the f.pl. TA mätkont vs. tb mäktona* from TA mättak, tb makte ‘himself (= Lat. ipse)’ and probably originated after the f.pl. TA pont (tb ponta) from TA puk ‘whole, all’ (Pinault 2008: 549). In light of all these replacements, Tocharian B is the best candidate for reconstructing the Proto-Tocharian paradigm of this pronominal adjective. It is important to note that TB allek, ta ālak displays allomorphy TB all- ~ aly-, ta āl- ~ āly- throughout the paradigm. In Tocharian B, the allomorph allis found in the nom.sg.m. and in the entire inflection of the feminine, with the exception of the nom.sg. The contrast -ll- vs. -ly- has been the topic of controversial interpretations. Pinault (2008: 419–420 and 548–549) suggests that they are graphic variants of the palatalised lateral /ĺ/. A different analysis is provided by Malzahn (2010: 5). She argues that PIE *li̯ became *ĺĺ and that it could have two different outcomes: (1) it became palatalised -ly- /ĺ/ or (2) it was depalatalised to -ll-. This assumption faces some theoretical problems, given the fact that the twofold outcome of PT *ĺĺ would not have been conditioned by any phonetic context. Following Winter (1992: 152), Peyrot (2013b: 223–225) proposes another explanation for the alternation between all- and aly-. He suggests that PIE *li̯ became PT *ll, which underwent regular degemination in Tocharian A. It follows that the stem-allomorph TB all- preserves the archaic state of affairs, while the stem-allomorph aly- is a secondary innovation. A similar proposal has recently been provided by Fellner (2017: 156 fn. 156); however, he questions the evolution PIE *li̯ > PT *ll. Indeed, he claims that the expected outcome of PIE *li̯ must have been PT *-l- (continued as such in Tocharian A) and that the other forms of the paradigm showing the allomorph TAB -ly- and the geminated TB -ll- could be motivated through analogy after the gerundives in TB -lle, ta -l (see § 4.7.1). Analogy from the gerundives is, in my view, not very convincing, and Fellner’s sound law PIE *li̯ > PT *-l- is difficult to test. I found the following attestations of the degeminated stem TB al-: archaic: aleksa (B284 b7); archaic-classical: alekk (B207 b2), ālek (B221 a3); classical: ale(k) (THT1109 b1), aloṅkna (THT1115 a1); classical-late: alekä (B289 b3); late: alekä (B197 b1), alekak ‘in addition’ (Ot 13.1 a5), aloṅkna (B189 b5).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

274

Chapter 4

As is clear, the distribution of the variants raises difficulties. Indeed, the stem al- can be found since the archaic period but is only rarely attested. However, rather than considering the stem al- an archaism that occasionally surfaces in sporadic forms, we can attribute it either to scribal mistakes or to prior examples of the reduction ll > l that characterises late and colloquial texts in particular (Peyrot 2008: 66; Fellner 2017a: 151).29 Turning back to the historical evolution of allek, it is now possible to assume that the regular outcome of this paradigm would have displayed *all- as the basic stem. In a second stage, a new stem *aĺĺ- was introduced analogically, through morphological palatalisation, after the paradigm of the demonstratives which displays the same distribution between palatalised and non-palatalised stems (Winter 1992). After analogy took place, the paradigm would have displayed non-palatalised nom.sg. vs. palatalised stem in the rest of the paradigm of the masculine. On the other hand, if we look at the inflection of the feminine, we notice that the distribution is the other way around: a form aly- is only attested in the nominative singular, while a stem al- is attested in all other cases. As a consequence, we must accept that the nom.sg.f. alyāk was created at a later stage. This reconstruction is rejected by Malzahn (2011: 97), who suggests that PIE *-eh2 yielded TB -a and assumes that TB alyāk is the regular outcome of PIE *ali̯eh2. If so, however, a non-palatalised form *allāk would be expected. As a consequence, I believe that the form TB alyāk was secondarily recharacterised after the feminine inflection of the adjectives that always displays the pattern -[+pal.]a (see § 4.8.2.3). The reason for this replacement is that the 29

There are problems with some derived forms, like TB aletstse ‘foreign, unrelated’ (= Skt. ajñāti-), TB alokälymi ‘leaving all other things aside’ (= Skt. ekānta, ekoti), and TB āläṃ ‘elsewhere’, since they all show single -l-. It is generally assumed that some kind of degemination in preaccentual position took place (so Hilmarsson 1996: 16; Winter 1992: 154–155, who also dismiss a direct relation of āläṃ with PIE *ali̯o-). In the case of TB alokälymi, we also find the variant āllokälymi in the archaic fragment B125 a1 and all(o)kä kälymīsa in THT1520 a1 (arch.), which might be used as an argument that an original phrase allok kälymi (obl.sg.f.) was univerbated in *allokəkəĺmi > *allokəĺmi > *alokəĺmi in the archaic phase of Tocharian B. Otherwise, we must assume that the original form was alo-, which is indeed the lectio difficilior, and that the variants with allo- were influenced by allok (see Hackstein, Habata & Bross 2019: 181–182). As far as TB aletstse is concerned, the obl.sg.f. alletst(s)ai in THT1544.b a2 and the derived abstract alletsñe ‘± foreignness’ in B327 a4 are of no value, since they are from fragments written in late Tocharian B. Even though it is tempting to explain all these forms as directly derived from allek, it is also possible to trace TB ale- in some of these forms back to a different morphological formation (Pinault 2008: 549), namely PIE *h2el-o-, which is found, for example, in Germanic, where a form *ala° still occurs in compounds (Goth. ala-, OHG ala-, see Kroonen 2013: 23 and lipp 2. 19).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

275

Gender in the Inflection of the Pronoun and Adjective

expected outcome of PIE *ali̯eh2 would have merged morphophonologically with that of PIE *ali̯eh2-m, both resulting in *allo-. In the feminine plural, the stem TB alloṅk- is common to both nominative and oblique. Winter (1992: 153) and Hilmarsson (1996: 18) analyse it as a reduced form of a pre-existing *allonakə, which lost the *-a- before the enclitic. This reconstruction is totally ad hoc.30 In my view, the form alloṅk must be explained just like the feminine plural paradigm of the demonstratives: TB allon- is the regular outcome of the obl.pl. PT *allons, which continues the acc.pl. PIE *ali̯eh2-ns. The nom.pl. alloṅk obviously cannot go back to PIE *ali̯eh2-es > *ali̯ās, given that it would then have evolved into TB **allo-. Since this nom.pl. has an obvious parallel in the demonstratives, we can assume that in the paradigm of allek the inherited oblique was also extended to the nominative. A distinction between nominative and oblique was then reintroduced by the addition of -na to the oblique form, which resulted in the attested alloṅk-na (cf. also the late obl.pl.f. TB toy-na). To sum up, it is clear that the historical evolutions of both the pronominal adjective TB allek and the demonstratives have much in common, especially as regards their feminine inflection. The general development of TB allek is recounted in the table below.31 Table 49

Evolution of the paradigm of TB allek

masc. nom.sg. acc.sg. nom.pl. acc.pl. fem. nom.sg. acc.sg. nom.pl. acc.pl.

PIE

pre-PT

PT

pre-TB

TB

*ali̯os *ali̯om *ali̯oi̯ *ali̯ons *ali̯eh2 * ali̯eh2m *ali̯eh2es * ali̯eh2ns

> *alle > *alle > *alley > *allens > *allo > *allo > *allo > *allons

>*alle-kə >> *aĺ(ĺ)e-kə >> *aĺ(ĺ)ey-kə >> *aĺ(ĺ)ens-kə >> aĺ(ĺ)a-kə > allo-kə > allo-kə > allons-kə

> *alle-kə > *aĺ(ĺ)e-kə > *aĺ(ĺ)ay-kə > *aĺ(ĺ)en-kə > *aĺ(ĺ)a-kə > *allo-kə >> allon-kə > allon-kə

> allek > alyek > alyaik >> alyeṅk-äṃ > alyāk > allok > alloṅk >> alloṅk-na

PIE Reconstructed PIE paradigm of *ali̯o-. pre-PT Regular outcome of the paradigm, where the sequence PIE *li̯ evolved into *ll /ll/.

30 31

Even more improbable in my view is Van Windekens (1979: 267 and 273) and Adams’ (dtb 31) nom.pl. *allo-ñ-kə. For a different reconstruction of the Proto-Indo-European paradigm of this pronominal adjective, see Kortlandt (2020: 120–121).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

276

Chapter 4

PT Two important modifications took place, which reshaped the whole paradigm: (1) morphological palatalisation of the masculine paradigm, analogically extended after the demonstratives (*-ll- > *-l(l)y- /ĺ(ĺ)/); and (2) palatalisation of the nom.sg.f. after the pattern of the thematic adjectives (*allo(-kə) >> *aĺ(ĺ)a-kə > alyāk /aĺák(ə)/). These developments solved cases of homophony in the paradigm, especially between nominative and oblique in the singular inflection of both the masculine and the feminine. pre-TB Extension of the regular outcome of the obl.pl. to the nom.pl. (*allokə >> alloṅk), which must have become homophonous with the obl.sg.f. TB Finally, a new distinction between nominative and oblique was reintroduced in the feminine since the latter took the ubiquitous marker of the feminine plural TB -na. On the model of this new case form, the obl.pl.m. was probably remarked with -äṃ /-ən/, according to the following diachronic proportion: obl.pl.f. alloṅk-na : obl.pl.m. alyeṅk >> obl.pl.f. alloṅk-na : obl.pl.m. alyeṅk-äṃ.

4.5

Conclusion

This section has focused on the diachronic evolution of the pronominal inflection in Tocharian. Although a large number of endings and forms have been discussed, my results are easily summarised. The main goal was to demonstrate that the great majority of the endings of both Tocharian demonstratives and the pronominal adjective TB allek, A ālak ‘other’ can be directly traced back to Proto-Indo-European. Furthermore, I have adduced new evidence in support of the scenario provided by Stumpf (1974 and 1990) for the evolution of the plural inflection in Tocharian B. The main part of my analysis has involved the paradigm of the feminine. In particular, I have argued that the nominative plural PIE *téh2-es > *tās evolved regularly into Tocharian A to-, while the accusative plural PIE *téh2-ns > *tāns yielded Tocharian A tos-. The archaic Tocharian B plural form toṃ was explained as the regular outcome of the accusative plural PIE *téh2-ns > *tāns, while its secondary spread to the nominative plural was motivated on paradigmatic grounds. The feminine paradigm of TB allek can be analysed under the same light. 4.6

Overview of the Tocharian Adjectival System

Tocharian adjectives agree in number, gender, and, with certain limitations, in case with their target. Indeed, when the head noun is inflected in the nominative or oblique, the adjective consistently agrees with it; when the head noun is inflected in one of the secondary cases, the adjective is in the oblique. When the head noun is in the genitive, the adjective often agrees in case

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

277

but is occasionally inflected as oblique.32 See the following examples from Tocharian B: B350 a4 (literary) ipreräntse ānte snai tärkarwa sky(?).gen.sg surface(m).nom.sg without cloud(a).obl.pl astare klautka ᛫ pure.m.nom.sg become.3sg.prt.act “The surface of the sky became pure without clouds”. (cf. Thomas 1957: 93) AS5A a1–2 (Udānastotra) (pe)[a2]laiknetse kreñcepi stamalñeṣṣe Law(a).gen.sg good.m.gen.sg prtng.to.establishment.m.obl.sg akālksa wish(a).perl.sg “[…] through the wish for the establishment of the good Law”. AS7J b1 (Karmavibhaṅga) lwāke tatākau mä(sketär) (s)u pot.nom.sg be.prt.part.m.nom.sg be.3sg.prs.mid this(m).nom.sg cm(e)lane kreṃt (pe)laiknetse ᛬ birth(a).loc.pl good.m.obl.sg Law(a).gen.sg “He becomes a pot of the good Law in the rebirths”. (cf. Georges-Jean Pinault, Melanie Malzahn & Michaël Peyrot apud CEToM) In the Udānastotra fragment both the controller and the target are inflected as gen.sg. in the noun phrase pelaiknetse kreñcepi ‘of the good Law’, while in the Karmavibhaṅga fragment the target is inflected as obl.sg.m. kreṃt ‘good’ although it agrees with the gen.sg. pelaiknetse ‘of the Law’. From a derivational point of view, Tocharian adjectives can be derived from nominal bases (e.g. TB käṣṣīññe ‘related to a teacher’ from käṣṣī ‘teacher, master’) and verbal bases (e.g. gerundives and preterite participles), occasionally from adverbs (e.g. TB späntaitstse ‘having faith’ from spantai ‘trustingly’), and from pre- and postpositions (e.g. TB eṃṣketstse ‘lasting, permanent’ from eṃṣke ‘up to’). From an inflectional point of view, they are traditionally grouped into four classes. This classification was established by the authors of the Elementarbuch (teb §§ 213–247), who selected the masculine plural paradigm of Tocharian B 32

See Pinault (2008: 467), Adams (2015: 73–75), Meunier (2015a: 6–7), Carling (2017: 1354).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

278

Chapter 4

as the standard criterion, as shown in the following table (corresponding forms in Tocharian A are put in square brackets): Table 50

teb adjectival classes

class

nom.pl.

obl.pl.

example

i.

TB -i [ta -e] TB -ñ [ta -ṣ] TB -ñc [ta -ṃś, -ṣ] TB -ṣ [ta -ṣ]

TB -eṃ [ta -es] TB -(nä)ṃ [ta -ñcäs] TB -ntäṃ [ta -ñcäs] TB -ṣäṃ [ta -ñcäs]

TB astari | -eṃ ‘pure’ [ta āṣtre | -es ‘id.’] TB klyomoñ | -oṃ ‘noble’ [ta klyomäṣ | -äñcäs ‘id.’] TB perneñc | -entäṃ ‘worthy’ [ta parnoṣ | -oñcäs ‘id.’] TB yāmoṣ | -oṣäṃ ‘having done’ [ta yāmuṣ | -uñcäs ‘id.’]

ii. iii. iv.

Each class can in turn be divided into subclasses on the basis of minor differences in their inflection. Historically, class i continues the PIE thematic inflection, while the other classes go back to the PIE athematic inflection. Class ii is divided into five subclasses which are usually traced back to different types of PIE nasal stems. Tocharian A and B often diverge in the respective inflection of this class. A good example of this divergence is provided by the adjectives in TB -mo, ta -m of class ii.5, where, in the paradigm of the masculine, Tocharian A has taken over some of the endings from the nt-declension (class iii) in the oblique singular and plural, and from the declension of the preterite participle (class iv) in the nominative plural (cf. TB obl.sg.m. klyomoṃ vs. TA klyomänt; obl.pl.m. TB klyomoṃ vs. TA klyomäñcäs;33 nom.pl.m. TB klyomoñ vs. TA klyomäṣ, all from TB klyomo, ta klyom ‘noble’). Another salient mismatch between Tocharian A and B can be found in class ii.4, where, in the paradigm of the masculine, Tocharian B n-forms correspond to the regular continuants of the thematic inflection in Tocharian A (cf. nom.pl.m. TB täpreñ /təpréñ(ə)/ < *-eñə < *-on-es vs. ta täpre < *-ey < *-oi̯ from TB tapre /tə́ pre/, ta tpär < *dhub-ró- ‘deep’, see below § 4.7.1). As far as class iii is concerned, in some (isolated) cases, the comparison between Tocharian A and B is straightforward, e.g. obl.sg.m. TB krent, ta krant ‘beautiful’ < PT *krent (but cf. also the Tocharian A variant krañcäṃ, which has taken over palatalisation from the nom.pl.), or nom.pl.m. TB poñc, ta poñś ‘all’ 33

See Peyrot (2010b).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

279

< PT *poñcə. A productive set within this class can be traced back to the PIE possessive formations in *-u̯ ent-, where in Tocharian A the expected nom.pl.m. *-ñś < *-ñc has been remade in -ṣ after the preterite participle. Class iv corresponds to the Tocharian preterite participles, which continue the PIE perfect participles in *-u̯ os-. This brief overview makes clear that a grammatical sketch of the Tocharian A and B adjectival systems taken together faces difficulties, since a number of changes have occurred independently in both languages. These changes have sometimes obscured the derivations of some adjectival types from the common antecedent. As a consequence, the classification given in teb has given rise to criticism. Among its problematic aspects is the fact that it is entirely based on Tocharian B, and the endings of Tocharian A rarely match those of Tocharian B, both synchronically and diachronically. However, since teb’s classification is the only standard so far, and given that the aim of this chapter is to discuss the inflections diachronically, teb’s classification will be taken as a starting point. 4.6.1 Metrical Constrains and Adjectival Agreement A peculiarity of the Tocharian B adjective is that of allowing a masculine plural form to be substituted for a feminine plural in specific metrical contexts (teb § 66 Anm 1). This phenomenon was examined in detail by Claus-Peter Schmidt (1972). He shows that agreements of this type are more common with alternating nouns, though occurrences with feminine nouns can also be found.34 In the adjectival declension of Tocharian B, masculine plural forms generally differentiate from feminine plural forms in having one fewer syllable (cf. e.g. TB nom.pl.m. astari, obl.pl.m. astareṃ vs. nom.obl.pl.f. astarona, where the count is 3× syllables of the masculine vs. 4× syllables of the feminine). The use of the masculine form is therefore an artificial strategy attested since the archaic stage to fit the metre of difficult passages.35 34 Cf. e.g. NS39 a3 orocci arṣaklañ (for expected orotstsana arṣāklañ) ‘great snakes’ or B553 b4 piṃś indriṣi mai(yyāñä) (or mai(yyana)) (for expected piṃś indriṣṣana mai(yyāñä)) ‘the five indriya-powers’. 35 Based on the data collected by P.-C. Schmidt (1972) and Hartmann (2013), this expedient is attested in: 5 (6×) archaic texts (B226 a2 kleśänmaṣṣeṃ śänmanmasa; B284 b4–5 kätkauu ekmätte ¦ yneśäññi kuse läklenta ᛬; B554 a3 kleśanmaṣṣi sta(na no) ¦; THT1191 a3 ᛬ yukāntaṣṣeṃ pwāra, a5 ///ṣṣeṃ kwärsärwā; THT1254 b3 /// pwār= empelyi ᛬), 1 (2×) archaic-classical text (B212 b1–2 kleśanmaṣṣeṃ ¦ tekänmasa, b4–5 kleśanmaṣṣeṃ ¦ tekänma po), 29 (33×) classical texts (M500.1 a4 ¦ kleśanmaṣṣeṃ tekanma ¦ [= B312 b5–6 kleśanmaṣṣeṃ tekanma; AS4B b1 kl(e)ś(an)m(a)ṣṣ(eṃ) (te)kanma], a2 ¦ naumyenta arthanmaṣṣ(eṃ) ¦; B3 a2 ¦ aścemaṣṣeṃ bhājanta; B14 a8 (śaiṣṣe)ṣṣeṃ skwanma ¦; B23 b2 ¦

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

280

Chapter 4

See the following example: B73 b4–536 (Lakṣana-Buddhastotra) pälskoṣṣi śpālmeṃ ¦ ckenta piś reskeṃ ¦ räddhinmaṣṣi späntaitsñeṣṣi ¦ nāgi yākṣi [b5] gandhārvi ᛬ ompalskoñeṣṣi ¦ snai keś wärttonta ¦ pelaikneṣṣi preṅki aurcci ¦ krentaunaṣṣi naumyenta ᛬ “The five great rivers of consciousness flow. Powerful and trustful (are) the Nāgas, the Yakṣas, the Gandharvas. Countless forests of meditation, wide islands of the Law, the jewels of virtues”. The passage above contains the first two pādas of strophe 15 of a Buddhastotra describing the lakṣanas of the Buddha. In such a short passage we find three examples of the “masculine-for-feminine” expedient: pälskoṣṣi … ckenta ‘rivers of thought’, ompalskoñeṣṣi … wärttonta ‘forests of meditation’, and krentaunaṣṣi naumyenta ‘jewels of virtues’. In all noun phrases we find plurals of three alternating nouns (cake ‘river’, wartto ‘forest’, and naumye ‘jewel’) for which a feminine agreement would be expected on the target. However, in all three cases the adjectival feminine form, namely pälskoṣṣana, ompalskoñeṣṣana and krentaunāṣṣana, would have added extra syllables to each of the pādas, yielding an unacceptable metrical sequence. As a remedy, the composer of the text used masculine plural adjectives in agreement.37

piś-cmelaṣṣeṃ ñya(t)s(e)ntane ᛬; B73 b4 pälskoṣṣi śpālmeṃ ¦ ckenta piś reskeṃ [= IT89 b1 pälskoṣṣi śpālme(ṃ) cke(nta) ///; IT466 a3 (pälsko)ṣṣ(i) śpālmeṃ ckenta]; B45 b5 tary= āsaṃ(kh)y(ai)ṣṣi yärponta ¦; B75 b2 ¦ krentaunaṣṣi naumyenta ᛬ [= IT89 b2 krentaunaṣṣi naumye(nta)]; B204 b2–3 ¦ tsärśalñenta läkleṣṣeṃ, b3 ¦ eṅkalñeṣṣeṃ ¦ ñemna po ᛬; B271 a1 arañcäṣṣi uppālta ¦; B523 b4 kleśanmaṣṣi tsnamñenta; AS4A b1 ¦ aiśamñeṣṣeṃ mīsa po ¦ [= NS27 a2 aiśamñeṣ(ṣ)eṃ mī(sa po)]; AS5B b5 ñiś ñiññeṣṣeṃ wäntarwa ¦; NS25 and 26 a2 eneṅkaññeṃ indrinta ¦; NS37 and 38 a1 (up)p(ā)lṣi stāna ¦; NS54 a1 yāmornta ¦ bodhisatve wakīceṃ, a6 ¦ (po saṃ)sārṣṣeṃ tserekwa ¦ karsal(ñ)esa [vs. b3 saṃsārṣṣana tserekwa]; IT5 b3 ᛬ kwīpe-onmiṣṣeṃ pwārasa [vs. a4 onmiṣṣana pwārasa]); IT97 a5 (artha)nmaṣṣeṃ skwanmasa (?); IT166 a2 uppālntasa eśneṣṣeṃ ¦, b1 /// – ṣṣi naumy(e)nta ¦; IT214 a6 ᛬ mlyu(weṣṣ)eṃ mīsa; IT219 a3 /// ¦ träṅkontasa ¦ pilkontaṣṣeṃ ᛬ (?); IT272 a4 śäktālyenta ¦ kätnālyi krentaunaṣṣi ᛬; S1 b1 bodhyāṅkṣi naumyenta ¦; SI P2 b3 kärsanälyeṃ wäntarwane ¦), 1 (1×) classical-late text (THT2237 a2 pañiktiññi naumyeṃ(ta) ¦), and 1 (1×) late text (B11 b4–5 (kleśa)nmaṣṣi lwāsa cwi ¦). 36 Verse: metre 4 × 5 ¦ 5 ¦ 8 ¦ 7 [5 ¦ 5 ¦ 4+4 ¦ 4+3]. 37 The same metrical artifice also occurs in other manuscripts of the same text, cf. IT89 b1 pälskoṣṣi śpālme(ṃ) cke(nta) ///, b2 /// (ompalsko)ññ(e)ṣṣi snai keś wärttonta pelaikneṣṣi preṅki aurcci krentaunaṣṣi naumye(nta) ///, and IT466 a3 (pälsko)ṣṣ(i) śpālmeṃ ckenta piś reskeṃ ///.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

281

The “masculine-for-feminine” plural agreement appears more frequently with fixed expressions, such as kleśanmaṣṣeṃ tekanma ‘kleśa-diseases’, and with the plural of pūwar ‘fire’. As a confirmation that this phenomenon is artificial, we observe that it is more common for an adjective to take the masculine plural form when the following word begins with a consonant (including the controller noun). This shows that sandhi could not have taken place to control the number of syllables. Indeed, from the data collected by Schmidt (1972) and Hartmann (2013), “masculine-for-feminine” plural agreement with words beginning with a vowel is attested only three times: (1) NS25 and 26 a2 eneṅkaññeṃ indrinta “internal senses”; (2) B271 a1 arañcäṣṣi uppālta “lotuses of the heart”; (3) uppālntasa eśneṣṣeṃ “with lotus eyes (?)”.38 While noting the irregular agreement in NS25 and 26 a2, Georges-Jean Pinault & Melanie Malzahn (apud CEToM s. PK NS 24, PK NS 25 and 26) observe that in the parallel passage in NS24 a3 we find a slightly different rephrasing eneṅkāmeṃ indrinta, where eneṅkāmeṃ ‘from within’ is a derived adverb from eneṅka ‘inside, within’ (dtb 88). In order to account for the two variants, they suggest that the original text ought to have had a sandhi form eneṅkāññana= ndrinta and that, apparently, the two copyists reintroduce the i-vowel with the aforementioned strategies. Furthermore, as Hartmann (2013: 286) suggests, we also find cases where both masculine and feminine agreement are found on agreement targets of the same alternating plural noun, as in B3 a2 aścemaṣṣeṃ bhājanta kautalñ= āke po toṃ ᛬ “earthenware.m bowls, all these.f [have] breaking [as their] end” (cf. Peyrot 2013: 709). Here aścemaṣṣeṃ ‘earthenware’ and toṃ ‘these’ are both in agreement with bhājanta ‘bowl’, but the former takes masculine agreement while the latter takes feminine agreement. It is generally assumed that this phenomenon is not attested in Tocharian A, because in this language masculine and feminine plural forms usually have the same number of syllables in most of the adjectival paradigms (teb  § 66 Anm 1.; Hartmann 2013: 100).39 However, other strategies are used in Tocharian A to fit the metre of difficult passages. Indeed, the highly productive adjectives 38 Broomhead’s (1962: 1.108) translation of IT166 a2 is tentative. 39 Tocharian A examples of an artifice similar to that found in Tocharian B remain to be discovered. A possible, though problematic, example might be A298 b4–5, containing the beginning of act 22 of the Maitreyasamiti-Nāṭaka. The line runs as follows: (o)[b5]ki slamasyo wäntoñcäs arkañcäs tarkru(nt) “(li)ke black clouds shrouded in flam(es)”, where the adjectives obl.pl.m. wäntoñcäs ‘surrounded’ and arkañcäs ‘black, dark’ unexpectedly agree with the alternating noun tarkrunt ‘clouds’ (Hartmann 2013: 372). Both adjectives would have shown shorter forms if they were inflected as feminine plural (wäntont* and arkant*). However, it is unclear whether this passage is in verse or in prose.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

282

Chapter 4

in TA -i and -ṣi show synchronic variants for the feminine plural (see § 4.7.1.3). These variants differentiate a shorter form from a longer one, such as obl.pl.f. saṃsārṣṣās, napeṃṣās (3× syllables) vs. saṃsārṣinās, napeṃṣinās (4× syllables). Compare the following examples: A69 a240 (verse, avadāna) aviś ṣolār saṃsārṣṣās ¦ puk klopäntw(ä)ṣ sn(e) praski ¦ śaśyu pältskum « ᛬ » “From all sufferings of the saṃsāra right down to the Avīci-hell, without fear, … content”.41 A69 b3 (prose, avadāna) päñ cmolwāṣi(nä)s wrasaśśi saṃsārṣinās klopäntwaṃ kāruṃ stäk “Pity spread to the sufferings of the saṃsāra of beings of the five circles of births”. (cf. Seržant 2014: 202) A254 a642 (verse, Maitreyasamiti-Nāṭaka) (yśe) kraṃś sne kärye ¦ purpāc skuntu napeṃṣās ᛬ “(O) you good ones, enjoy without concern the human delights”. (cf. Geng, Laut & Pinault 2004b: 54) A14 b243 (verse, Puṇyavantajātaka) (pñi) (p)uk cmolwaṃ särki yṣäṃ ¦ eṣṣäṃ sukuntu ñäkciyās ¦ napeṃṣinās ᛬ “(Merit) follows it in (a)ll incarnations, gives him divine [and] human pleasures”.44 As is clear, in a metrical passage in A69 we find the obl.pl.f. saṃsārṣṣās ‘of the saṃsāra’ (3× syllable) at the end of a colon and in agreement with all.pl.alt. klopäntw(ä)ṣ, while in the prose passage of the same fragment we find the longer form of the adjective in the noun phrase saṃsārṣinās klopäntwaṃ. Conversely, both A254 a6 and A14 b2 are in verse, but in the former we find the shorter form obl.pl.f. napeṃṣās (3× syllables) ‘of a human being’ in agreement with skuntu ‘delight, pleasure’, while in the latter we find the longer form napeṃṣinās to fit the final colon of 4 syllables. Similar variants are frequently found in verse passages in Tocharian A. 40 Verse: metre 4 × 7 ¦ 7 ¦ 4. 41 Cf. dta 17. The prt.ptc. TA śaśyu is a hapax legomenon of unknown meaning (Malzahn 2010: 920; Peyrot 2013: 826 fn. 889). 42 Verse: metre 4 × 5 ¦ 7. 43 Verse: metre 4 × 7 ¦ 7 ¦ 4. 44 Cf. Gerd Carling (apud CEToM s. A14).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

4.7

283

Reconstruction of the Proto-Tocharian Adjectival Paradigms

In what follows, I discuss the outcome of thematic and athematic types in the Tocharian adjectival system. The aim of this paragraph is twofold: (1) to provide a more detailed overview of the synchronic inflectional patterns that define the classes, and (2) to explain how these classes must be reconstructed for Proto-Tocharian. In the first section (§  4.7.1), I deal with the thematic type; in the second section (§ 4.7.2), I discuss some athematic types; in the third section (§ 4.7.3), I summarise the results, providing a general overview of the reconstructed adjectival system of Proto-Tocharian. The reconstruction of the Proto-Tocharian paradigms will be then used to investigate the development of the gender system from Proto-Indo-European to Tocharian (§ 4.8). 4.7.1 The Thematic Type (class i) Class i is the most productive class by far. It consists of both primary and secondary adjectives, which are derived by means of a relatively large number of suffixes. The fact that these formations can ultimately be traced back to the PIE thematic type is made evident by the masculine inflection: nom.obl.sg. TB -e, ta -∅ < PT *-e < nom.sg. PIE *-o-s and acc.sg. PIE *-o-m; nom.pl. TB -i, ta -e < PT *-ey < nom.pl. PIE *-oi̯; obl.pl. TB -eṃ, ta -es < PT *-ens < acc.pl. PIE *-o-ns (see § 4.8.1 for further remarks). The suffixes employed and the adjectives derived are the following:45 Table 51

Adjectival suffixes of the thematic type

adjectival suffix

example

TB -re, ta -r TB -lle, ta -l (the gerundives) TB -tte, ta -t (the privatives) TB -te, ta -t (the ordinals)

e.g. TB astare, ta āṣtär ‘pure’ e.g. TB pralle, ta präl ‘to be carried’ e.g. TB etaṅkätte, ta atäṅkät ‘not obstructed’ e.g. TB trite, ta trit ‘third’

45

In addition, Krause & Thomas (teb § 220 and § 232) list a handful of adjectives in TB -ke, ta -k, whose inflection is shifting between class i and class ii.5 (nom.pl.m. TB -añ). These formations are almost exclusively found in loanwords, and they are mostly used as substantives e.g. TB aṣanīke, ta āṣānik ‘worthy, arhat (epithet of the Buddha])’ from TB aṣāṃ /aṣán/, ta āṣāṃ ‘worthy’ from Bactr. αζανο (cf. Khot. āṣaṇa- ‘id.’ [ˈaːʐaɳa-], see Weber 1985 and Dragoni 2022: 39–40), TB eynāke ~ aināke, ta enāk ‘vulgar, bad’ derived from an Old Iranian source related to Christian Sogd. ʾynʾqwc ‘blasphemer, abusive’ (see Isebaert 1980: 115).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

284 Table 51

Chapter 4 Adjectival suffixes of the thematic type (cont.)

adjectival suffix

example

TB -iye, ta -i TB -ṣṣe, ta -ṣi TB -ññe, ta -iṃ TB -tstse, ta -ts TA -ñci

e.g. TB ñakc(i)ye, ta ñäkci ‘divine’ e.g. TB oraṣṣe, ta orṣi ‘wooden’ e.g. TB lwāññe, ta lweṃ ‘pertaining to an animal’ e.g. TB kramartstse, ta krāmärts ‘heavy’ e.g. TA kuleñci ‘female’

Krause & Thomas (teb  §§  213–229) grouped these thematic suffixes under various subclasses on the basis of two parameters that pertain to Tocharian B. These parameters are: (1) the feminine plural form TB -ona vs. -ana; (2) the paradigmatic alternation between palatalised and non-palatalised stem-final consonant in the masculine inflection. The intersection of these criteria leads to the creation of four different subclasses: (1) adjectives with no palatalisation alternation and f.pl. -ana (adjectives in -ṣṣe, -ññe, and -(i)ye); (2) adjectives with a palatalisation alternation and f.pl. -ana (adjectives in -tstse-); (3) adjectives with no palatalisation alternation and f.pl. -ona (adjectives in -re and -lle); (4) adjectives with a palatalisation alternation and f.pl. -ona (tte-adjectives and te-adjectives). Some criticism can be levelled against this classification, which requires that Tocharian A be adapted to it. Before proceeding further, however, we must explore the role of palatalisation in Tocharian in more detail. Indeed, it is important to distinguish between “phonological/etymological” and “morphological/analogical” palatalisation. The first type is the “regular” palatalisation, i.e. the assimilation of a consonant in front of etymological high (semi-)vowels, which results in a palatal (or palatalised) consonant. On the other hand, palatalisation is also a morphological phenomenon in Tocharian: “it is not a palatal feature added to a consonant, but it is a system of morphological alternations of non-palatal and palatal consonants” (Peyrot 2013b: 223). It is “morphological” because (1) it is no longer caused by sound change, but has an analogical mechanism behind it, and (2) it has morphological functions, since the contrast palatalised vs. non-palatalised consonant marks different grammatical forms (Peyrot 2013a: 69–70). We can now turn to teb’s subgrouping. If we consider only phonological/etymological palatalisation and not the analogical type, adjectives from class i can be grouped into two subclasses, which account synchronically for several mismatching forms in the plural of both Tocharian A and B. Indeed,

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

285

based on this parameter the paradigm can be predicted: (1) those adjectives without a palatalised suffix throughout the paradigm (i.e. without phonological/etymological palatalisation) take the f.pl. TB -ona, ta -aṃ, while (2) those adjectives with a palatalised suffix throughout the entire paradigm (i.e. with phonological/etymological palatalisation) take f.pl. TB -ana, TA nom.pl.f. -āñ, obl.pl.f. -ās. Morphological/analogical palatalisation is found in the first type only. In this regard, a special problem is posed by the derivatives in TB -tstse, ta -ts since they belong to different subgroups in the two Tocharian languages. Indeed, in Tocharian B they have morphological palatalisation and nom.obl.pl.f. -ana (subclass i.2), while in Tocharian A they have no palatalisation and nom.obl.pl.f. -aṃ (subclass i.1). This mismatch is certainly problematic, but I will argue that the incoherent synchronic distribution of the derivatives in TB -tstse, ta -ts can be explained diachronically: in Proto-Tocharian, these adjectives inflected just like adjectives in TB -re, ta -r (subclass i.1) and Tocharian A has preserved the archaic state of affairs (see § 4.7.1.1). A general scheme of the two subclasses is given below: Table 52

Thematic type (class i)

adjectives class

TB

TA

i.1

-re – -lle -tte -te -ññe -ṣṣe -iye -tstse -ñci

-r -ts -l -t -t -Vṃ -ṣi -i – –

i.2

palatalisation phonol. morphol. no no no no no yes yes yes yes yes

no no yes yes yes no no no no no

plural paradigm masculine

feminine

nom. TB -i, ta -e nom. TB -ona, ta -aṃ obl. TB -eṃ, ta -es obl. TB -ona, ta -aṃ

nom. TB -i, ta -ñi nom. TB -ana, ta -āñ obl. TB -eṃ, ta -näs obl. TB -ana, ta -ās

The subgrouping outlined above does not only predict the plural paradigm of the feminine, but that of the masculine too. Since these two subclasses pose independent diachronic difficulties, they are treated separately in the following paragraphs.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

286

Chapter 4

4.7.1.1 Subclass i.1 Adjectives in TB -re, ta -r are, for the most part, primary adjectives in Tocharian. They are built with the PIE thematic suffix *-ro-, which is well attested in adjectives describing “property concepts” and has a prominent role in the Caland system. See the following examples: TB ratre, ta rtär ‘red’ < PIE *h1rudh-ró- (cf. Gk. ἐρυθρός, Lat. ruber), TB swāre, ta swār ‘sweet’ < PIE *suh2d-ró- (cf. Gk. ἡδύς, Skt. svādú-), TB pärkare, ta pärkär ‘long’ < PIE *bhr̥ǵh-ró- (cf. Arm barjr ‘high’, Hitt. parkuš ‘id.’, Skt. br̥hánt- ‘id.’). Isolated adjectives in TB -re, ta -r derived from verbal bases are found rarely (e.g. TB kätkare ‘deep, far’, possibly from kətk- ‘to put down (?)’;46 TB cäñcare ~ ciñcare, ta ciñcär ‘charming, pleasant’ from TB cənk- ‘to please’, dtb 272). As demonstrated by Hilmarsson (1991a: 14–17), the so-called privatives in TB -tte, ta -t continue a common Indo-European derivational construction of the type Ved. amr̥ t́ a- ‘immortal’, Gk. ἄμβροτος ‘id.’; Ved. ákṣita- ‘imperishable’, Gk. ἄφθιτος ‘id.’; Lat. invictus ‘invincible’ (Pinault 2015d: 162). The general construction is as follows: *n̥ -[verbal base]-to- ‘one who is not x’. In Tocharian, these formations are synchronically based on the subjunctive stem. See the following examples: TB ekamätte ‘future’ (cf. kəm- ‘to come’) < *e(n)-kwəmə-te ‘not (yet) come’ < *n̥ -gwm̥ -tó-; TB etaṅkätte, ta atäṅkät ‘unhindered’ (cf. TB tənk-, ta tänk- ‘to stop’) < *e(n)-tənkə-te ‘not stopped’ TB -ll- is explained by Fellner as a secondary development on the model of -ly-, which he interprets as a geminate /-ĺĺ-/. The fact that these two different PIE formations coalesced in Proto-Tocharian in a single paradigm is attributed to the fact that the masculine and the neuter singular merge morphophonologically in other thematic formations. Fellner’s reconstruction implies that the Proto-Tocharian paradigm of the gerundives had a substantial number of variant forms, because the alleged merger between the formations in *-lom and those in *-lii̯o- would have been a very scattered development, which started in Proto-Tocharian but ended independently in the two Tocharian languages, i.e. after the breakup of Proto-Tocharian. Furthermore, I do not see any place where the Proto-Tocharian outcome of *-lom and *-lii̯o- could have coalesced, because the former would allegedly have formed abstract substantives, and the latter verbal adjectives. In addition, Fellner explains the gemination of TB -ll- analogically after the 49

The different origins of TB -ll-, ta -l-, on the one hand, and TAB -ly-, on the other hand, has already been proposed by other scholars, like Couvreur (1947a), Krause (1952: 203), and Van Windekens (1979: 81–82). This analysis cannot be further supported. See the remarks by Thomas (1985: 59) and the main text above.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

290

Chapter 4

geminated -ly-. Although I agree with him that -ly- may stand for /-ĺĺ-/, the claim that an original sequence pre-TB *-le < PT *-le < PIE *-lom would first have been levelled in -lle and then become -le in late and colloquial texts appears circular to me. Moreover, I believe that the distribution between non-palatalised nom.sg.m. -lle vs. palatalised obl.sg.m. -lye is well established in archaic Tocharian B. Indeed, I found that the nom.sg.m. is consistently spelled as -lle in archaic texts, while a nom.sg.m. -lye is only sporadically attested (e.g. IT7 a2 [Vinayavibhaṅga] ma wär tärkalye īkene “not at the place where the water is to be sprinkled”, Ogihara 2009: 93 and 333–334; but cf. also Adams’ 2015: 132 translation “in a place not accessible to water”).50 As a consequence, the variant -lye for the nominative singular only starts to appear in classical texts and does not become the standard variant even in late texts, where the original sequence -lle has regularly been reduced to -le. This is consistent with Thomas’ findings (1967), who concluded that the distinction between nom.sg.m. -lle and obl.sg.m. -lye only started disappearing in classical and late Tocharian B (and it never fully disappeared) (Peyrot 2008: 118–119). I therefore agree with Winter (1962c, 1992: 152) and Pinault (1989a: 102–103, 2008: 458) that the gerundives in TB -lle, ta -l are to be derived from a single PIE ancestor, which can indeed be reconstructed as *-li̯o- (cf. Arm. -(e)li, Olsen 1999: 395–398). As already outlined above (§  4.4), Peyrot (2013b; cf. Winter 1992) has recently argued that the expected development of the PIE sequence *-li̯- was PT *-ll-, which evolved regularly into TB -ll- and TA -l- (cf. Gk. ἄλλος ‘other’ < PIE *ali̯os). If that is correct it follows that, in the paradigm of the gerundives, all forms with palatalised -ly- must be explained as secondary and that the non-palatalised forms should be the forms on which the reconstructed paradigms are based. In both Tocharian A and B, we have seen that forms in -ly- are found throughout the paradigm except in the nominative singular masculine and the feminine plural. This type of paradigm strongly resembles that of the privatives, the ordinals, and the pronominal adjective TB allek, ta ālak. This analysis was first proposed by Winter (1962c: 1068–1069 fn. 2), and it is further supported by

50

For instance, I found the spelling -(l)le in the following archaic texts: IT47 b2 aille; IT80 a2 smille; IT106 a4 yamäṣäle; IT122 a3 (yama)ṣṣälle; IT157 b2 yamäṣälle; IT234 b3 pralle; IT268 a2 tsrelle; AS7N a4 sarkäṣṣälle; AS9A b8 sonopälle; AS12C a2 yänmālle; AS12D b4 yātalle; B123 b2 källālle, b7 ///ṣṣälle; B134 a5 prekṣälle; B135 b7 aiṣälle; B139 a5 srukālle, b6 tsäṅkāll(e); B140 b3 kly(eu)ṣṣälle; B118 b4 srukalle; B127 a4 yātalle; B132 a4 weṣṣälle, etc.; B240 b1 släṅkäll(e); B251 a3 klyelle; B279 b4 śmälle; B291.a b6 (kata)lle; B336 a5 (tr)īwaṣälle; THT1193 b5 yatalle (?); THT1536.a a3 källalle; THT1540.i a2 källālle; THT1184 a2 paṣṣälle; THT1535.d a1 yamäṣälle.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

291

both the distributions of the variants in Tocharian B texts and a closer examination of the Tocharian A paradigm.51 The status of ⟨ly⟩ is ambiguous, but Fellner is in my opinion correct that it could stand for /ĺĺ/. Indeed, evidence for a palatalised geminate /ĺĺ/ can be found in the occasional attestations of ⟨lly⟩ in archaic, classical, and even in late Tocharian B texts, as in tärkänallya (IT7 a6 [arch.]), paṣṣallyi (B67 b5 [class.]), lkaṣṣällye (THT3599.a b2 [arch.-class. (?)]), trīwäṣällya (W39 b1 [class.]), naṣṣallyanasa (B324 a5 [late]), – – ṣṣäll(y)i (B133 b8 [arch.]), ///-llyi (IT289 a2 [class.]) and in the paradigm of TB allek ‘other’, e.g. allyek (B133 b3 [arch.]), allye(ṅkaṃtso) (B137 a7 [arch.]), (a)lly(e)kämpa (B144 b3 [arch.]), āllyaik (B273 a1 [arch.]), and allyeṅkä (THT1860 a4 [arch.]). The spellings with geminate -lly/ĺĺ/ occur specifically in the inflection of gerundives and TB allek ‘other’; this indicates that -lly- /-ĺĺ-/ is a secondary palatalisation of geminate *-ll- /-ll-/. Conversely, in the inflection of e.g. the ekṣalye-type (nom.sg. -lye /-ĺe/, obl.sg. -ly /-ĺ(ǝ)/, nom.pl. -lyi /ĺǝy/, obl.pl. -lyäṃ /ĺǝn/) we only find spellings with -ly-, but never -lly- (Peyrot 2021). In Tocharian A, the obl.sg. -l-äṃ instead of the expected **-ly-äṃ can easily be explained diachronically: PT *-lye > pre-TA *-ly (apocope) > *-l (depalatalisation in word-final position, cf. PT *-ññe > TA -ṃ /-n/) >> TA -läṃ (regular recharacterisation of the inherited oblique, cf. obl.sg.m. -rä-ṃ PT *ts) might in turn undergo palatalisation, resulting in *ś, as the following isolated example seems to confirm: PIE *déḱm̥ ‘ten’ > *dz’əkə > PT *śəkə > TB śak, ta śäk (Pinault 1989a: 49–50; Ringe 1996: 146–148). The contrast ts vs. ś was probably extended in the Tocharian A verbal system. Examples include: the present

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

293

TB -ana : TA -aṃ can be explained in the same light. Indeed, if I am correct that palatalisation *-ts- > -c- is secondary, then the original feminine plural was *-ona in Proto-Tocharian, which regularly yielded TA -aṃ. Then, in the prehistory of Tocharian B, the sequence *-tsona (and not the expected **-ccana) was analogically adapted to -ññana and -ṣṣana with subsequent generalisation of the a-vocalism. To recapitulate, three arguments suggest that adjectives in TB -tstse, ta -ts originally belonged to subclass i.1 in Proto-Tocharian: (1) *ts had no reconstructable palatalised counterpart in Proto-Tocharian; (2) Tocharian A has pl.f. -tsaṃ, which clearly points to -tsona; (3) Tocharian B does not have a pl.f. *-ccana, with analogical palatalisation (cf. the paradigm of the singular, which has -cc-, while in Tocharian A we find -ts- throughout). In light of the above, I think that the original paradigm of subclass i.1 was mutatis mutandis that of the adjectives in TB -re, ta -r and that of the Tocharian A adjectives in -ts. However, before the breakup of Proto-Tocharian, analogical palatalisation affected those derivatives whose formant suffix could undergo palatalisation. Through this process, a new differentiation between the nominative and the oblique was reintroduced in the singular paradigm of the masculine (-lle vs. -lye; -tte vs. -cce; -te vs. -ce; -tstse vs. -cce). On the other hand, the re-adjectives, which did not have any palatalised counterpart, took the obl.sg. marker -ṃ, which was not a mandatory ending in Proto-Tocharian (§ 4.8.1). The evolution of the masculine paradigm can be summarised as follows: Table 57

nom. sg. obl. sg. nom. pl. obl. pl.

Evolution of the masculine paradigm

TB

TA

-e -e(ṃ) -i -eṃ

-∅ -ä-ṃ -e -es

PT < TB -ai is phonologically problematic. As a consequence, I follow the reconstruction recently defended by Peyrot (2012a), according to which the obl.sg.f. TB -yai is to be compared with the gen.(-dat.) sg. TA -ye.56 As a matter of fact, Tocharian B does not synchronically display any gen.sg.f. form, which may be an indication of the functional reanalysis of this ending as an oblique marker (see further § 3.4.2, § 3.5.2.5). Furthermore, the generalisation of the oblique marker TA *-n to the paradigm of the feminine can be easily explained as an innovation: on the one hand, if Proto-Tocharian had obl.sg.f. *-yan, there is no reason why it should not have been maintained in Tocharian B; on the other hand, TA *-n is the ubiquitous oblique ending in Tocharian A. It follows that the Proto-Tocharian obl.sg.f. cannot be reconstructed as either *-yai or *-yan, but as an unmarked ending *-ya (see Peyrot 2012a: 203–204 and the evidence from the TA (ṣ)i-adjectives below). The Proto-Tocharian paradigm of the feminine would have been as follows: Table 58

nom. sg. obl. sg. gen. sg. nom. pl. obl. pl.

Proto-Tocharian feminine paradigm of subclass i.1

TB

TA

-ya -yai — -ona -ona

-i -yāṃ -ye -aṃ -aṃ

PT < Ved. mádhya-, Gk. μέ(σ)σος, Lat. medius, etc., cf. Meillet 1937: 261–262). A good example is TB patarye ‘paternal’, Skt. pítr(i)ya- ~ pítriya-, Gk. πάτριος, Lat. patrius, etc. Among the suffixes from subclass i.2, it is not very productive, and it is only employed to derive adjectives from nominal bases. On the other hand, TB -ṣṣe, ta -ṣi is by far the most productive adjectival suffix in both Tocharian languages. It has genitival semantics and denotes appurtenance in a broad sense (i.e. material, origin, designation, etc.). In addition, derived adjectives in TB -ṣṣe, ta -ṣi are frequently used instead of a noun inflected in the genitive (Zimmer 1982; Meunier 2015a), and they translate the determiner (i.e. the first term) of Sanskrit karmadhāraya-compounds (Meunier 2015b). A derivational peculiarity of this suffix is that it can form denominal adjectives from singular, dual, and plural stems only when these stems form their plurals by means of suffixation (e.g. sg. TB läkleṣṣe ‘sorrowful’ |lǝklé-ṣṣe|, pl. TB läklentaṣṣe |lǝklénta-ṣṣe| ‘pertaining the pains, painful’; sg. TB paiyyeṣṣe |payyé-ṣṣe| ‘pertaining to the foot’, du. paineṣṣe* |payné-ṣṣe| ‘pertaining to the feet’).58 Adjectives in TB -ṣṣe, ta -ṣi can be derived from nouns, pronouns, and adverbs. The origin of the suffix has always been in question. Some scholars have traced it back to *-s(i)i̯o- (cf. Lat. -ārius and the Anatolian adjectives in *-ssa/i-),59 while others have derived it from *-sk(i)i̯o- (cf. Arm. -cci).60 The development of the adjectives in TB -ññe is problematic, since it is generally assumed they have two formal equivalents in Tocharian A: adjectives in -ñi and adjectives in -(e)ṃ. In Tocharian B, this suffix is quite productive and forms adjectives of appurtenance with genitival semantics. Adjectives in TB -ññe are 57 58 59 60

For an overview of the meanings of the suffixes, see Adams (2009a), Fellner (2013), and Meunier (2015a: 199–217). See Hajnal (2004). See Ringe (1996: 117), Pinault (2008: 515), and Adams (2009a: 30). See Pedersen (1941: 95), Couvreur (1947b: 141), and Fellner (2013: 63–67).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

296

Chapter 4

mostly derived from substantives referring to living beings (animals, humans, demons, deities, etc.) or from personal pronouns (TB ñiññe ‘my, pertaining to me’ from the genitive of ñäś ‘I’; TB taññe ‘your’ from the genitive of tuwe ‘you’; TB ṣañäññe ‘own; nature, essence’ from ṣañ ‘id.’). Additionally, they can occasionally be derived from terms for body parts (cf. TB paiyyeññe ‘related to the foot’ from paiyye ‘foot’) and concrete nouns (TB pyapyaiññe ‘related to flowers’ from pyāpyo ‘flower’).61 Furthermore, the feminine -ñña has been grammaticalised as a suffix of feminine oppositional nouns (e.g. ñakte ‘god’ : ñäkteñña ‘goddess’, see Malzahn 2013a: 115–116 and § 2.3.2, § 3.4.2). The reasons for this grammaticalisation are easy to envision: (1) on a comparative level, oppositional feminine nouns are typologically very often formed through denominal adjectives denoting appurtenance; (2) among the Tocharian suffixes denoting appurtenance, only TB -ññe displays such a clear derivational animacy-based feature, which makes it the best candidate to express gender-marking, i.e. a motion suffix. Returning to the origin of the suffix and its Tocharian A counterparts, scholars have long debated the obscure correspondence between TA -(e)ṃ, -ññi and TB -ññe. These suffixes have traditionally been traced back to PIE *-n(i)i̯o-. Hilmarsson (1987b), followed by Pinault (2011c), deals with the history and the distribution of the suffixal alternations *-ii̯-/-i̯-. He argues that Tocharian developed two variants of this suffix, i.e. *-nii̯o- and *-ni̯o-, which were originally conditioned by Sievers’ Law. According to him, PIE *-nii̯o- and *-ni̯o- yielded PT *-ñəye and *-ñe respectively. Later, they merged in Tocharian B -ññe, while

61

The Khotanese suffix -īña- is suspiciously similar to TB -ññe, ta -(e)ṃ. Konow (1932: 62) argues that Khot. -īña forms denominal adjectives from substantives. Degener (1989: 129–130) clarifies that it is only used with nouns denoting living beings. According to Degener (1989: 130–131), this suffix is attested with 23 adjectives, of which the majority are derived from nouns denoting animals (LKhot. kavīña- ‘pertaining to a fish’ from kavā‘fish’; LKhot. pasīña- ‘pertaining to a sheep’ from pasa- ‘sheep’; LKhot. makalīña- ‘pertaining to a monkey’ from makala- ‘monkey’). It is also used very frequently with loanwords (ca. 10 certain derivatives), although important examples with inherited nouns are also attested (cf. OKhot. dahīña- ‘belonging to a man’ from daha- ‘man’). See Degener (1989: 130) for an attempt at explaining the etymology. It goes without saying that the Khotanese and the Tocharian suffix share a core semantic feature. However, Khot. -īña is not very productive, while TB -ññe, ta -(e)ṃ is very productive. Particularly curious is the fact that -īña is mostly attested in Late Khotanese, and, as in Old Khotanese, it is predominantly confined to loanwords from Indo-Aryan. As a consequence, Khotanese might have borrowed this suffix from Tocharian and inserted it to a quite productive class of adjectival derivatives that have -ī- before the nasal. See also the following correspondences: TA nāgeṃ ‘snakelike, related to the nāga’ : OKhot. nāgīña- ‘id’; TA kinnarñā* ‘(female) Kinnara-’ (probably from TB kinnaräñña*) : OKhot. kindarīña- ‘pertaining to a Kinnara-’; TB yakṣaññe ‘pertaining to a Yakṣa’ : Khot. *yakṣīña- ‘id.’. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

297

Tocharian A maintained them distinguished: *-ñəye evolved into TA -ñi and *-ñe yielded TA -(e)ṃ. Fellner (2013: 45–60) questions this reconstruction. He claims that TA -(e)ṃ cannot correspond to TB -ññe, because the inherited PIE sequence *-ni̯- never palatalised the nasal in Tocharian. Accordingly, Tocharian would have inherited only *-nii̯o-, which evolved TB -ññe, ta -ñi. He bases this reconstruction on the non-palatalised nom.sg.f. TB sana, ta säṃ ‘1’, which he traces back to PIE *smih2 (cf. Gk. μία, Arm. mi) > *smya > PT *sənya- (see also Fellner 2017a: 154 fn. 17). However, there exist counterexamples to Fellner’s hypothesis. See the following clear correspondences, where, in the same context, a palatalised nasal of Tocharian B is matched by a non-palatalised nasal in Tocharian A: (1) the isolated adjective TB arkwañña : TA ārkiṃ ‘white’; (2) the adjectival type TB klyomña : TA klyomiṃ ‘noble’ (class ii.5); (3) the noun TB śamñā-ṃ-śka: TA śomiṃ ‘girl’; (4) the noun type TB weśeñña : TA waśeṃ ‘voice’; (5) the adjectives TB pokaiññe ‘related to the arm’ : TA pokeṃ ‘bracelet’, etc. Fellner comments on (some of) these counterexamples and he consistently resorts to either analogical changes in order to explain the palatalisation of the nasal in Tocharian B or to accidental attestation of the suffix -eṃ in the matching forms of Tocharian A. However, in light of the examples outlined above, it is simpler to only reconstruct analogy for the nom.sg.f. TB sana, ta säṃ ‘1’, where, in fact, the dental nasal cannot be the regular outcome of the sequence *-my-.62 Nonetheless, if one formally compares TB -ññe and TA -eṃ, another problem immediately comes to light: how are we to explain the vowel -e- in Tocharian A? Winter (1977), Hilmarsson (1987b), and Pinault (2008: 458–489) deal with this problem and convincingly suggest the following set of diachronically ordered changes: 1. 2. 3.

Raising of anaptyctic *i  PT *-VññV > pre-TA *-ViññV Apocope and degemination  pre-TA *-ViññV > pre-TA *-Viñ(ñ) Depalatalisation in absolute word-final position and monophthongisation  pre-TA *-Viñ > TA -Vn

62 Despite the fact that an evolution PIE *-m- > *-n- in front of the semivowel *-i̯- is sometimes attested in other Indo-European languages (cf. Gk. βαίνω, Lat. veniō < PIE *gwm̥ -i̯e/o-), Fellner’s path PIE *smih2 > *smya > PT *sənya- is, to my knowledge, without parallels in Tocharian. In any case, I do not think this is a good example of the proposed sound change, because the possibility cannot be ruled out that the palatalising effect of *-y- was lost before the development of *-my - > *-ny- and that the bilabial nasal had already absorbed the palatalisation. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

298

Chapter 4

This set of phonetic developments explains several (apparently) irregular mismatches between Tocharian A and B: (1) TB -əñ- : TA -in-, e.g. TB ostaññe /ostə́ ññe/ ‘related to the house’ vs. TA waṣtiṃ < pre-TA *wastəiñ < PT *wostəññe;63 (2) TB -añ- : TA -en, e.g. TB lwāññe /lwáññe/ ‘related to an animal’ vs. TA lweṃ < pre-TA *lwāiñ < PT *lwaññe; (3) TB -eñ- : TA -en, e.g. TB weśeñña ‘voice’ vs. TA waśeṃ < pre-TA *waśaiñ < PT *weśeññe; TB weñ- ‘to say’ vs. TA weñ< pre-TA *waiñ < PT *weññ- (Winter 1977; Peyrot 2013a: 469–470).64 Fellner (2013) dismisses this development, claiming that one would expect to find vowel raising also before the nom.pl. -ñ (< *-ñə). I do not think that this is a relevant parallel, because this phonetic change is not expected to have occurred in word-final position or before *-ñə, and, even if it did occur, it could have been easily removed through analogy (cf. the similar development in the outcome of the PIE cluster *-ns-, which developed anaptyctic *i only word-internally, see  §  4.8.1). Furthermore, as already pointed out by Winter (1977: 149–150), only Proto-Tocharian geminated sequences of the type *-VññV are affected by this Tocharian A sound law. Lastly, the claim by Fellner that the suffix TA -eṃ is either inherited from PIE *-no- (as per Couvreur 1947a) or borrowed from Skt. -na- seems problematic, and it does not explain how TA -ecame about. As argued above, TB -ññe and TA -(e)ṃ can be found in several comparable pairs of words which also share the same animacy distribution of the base from which they derived (cf. e.g. TB aśiyaññe : TA aśśeṃ [< *āśyāiñ] ‘pertaining to a nun’ from TB aśiya, ta aśi ‘nun’). It is therefore evident that TB -ññe and TA -(e)ṃ must be traced back to the same suffix, which can be reconstructed as PIE *-ni(i̯)o-.

63

Perhaps we may also add TB warñe*, ta wriṃ* ‘aquatic’ < PT *wərəññe, which is used in both Tocharian languages as a modifier of the word for ‘animal’, thus ‘aquatic animal(s)’, cf. B588 a4 wärñi lwasā; A154 a4 wrināñ lwā; A394 a2 wrinās lwā. As one can see, in B588 the adjective wärñi is inflected as a nom.pl.m., but it agrees with the alternating noun lwasā ‘animals’. As already pointed out by C.-P. Schmidt (1972; cf. also Hartmann 2013: 109, 534–535), in Tocharian B metrical passages alternating nouns sometimes agree with a masculine modifier in the plural, replacing the usual feminine agreement. This is one of the poetic devices used to adjust the syllable count in poetry (see § 4.6.1). Similarly Peyrot (2008: 116) explains the plural variant palskalñi of palskalñe ‘thought’ as an artificial masculine by-form that was created to fit the metre (vs. alternating palskalñenta). Agreement shows that the nom.pl. palskalñi is indeed a masculine form (cf. B8 b3 trai palskalñi ‘three.m thoughts’), while the pl. palskalñenta is alternating, as it takes feminine agreement (B8 b8 (yo)laina palskalñe(nta) ‘evil.f thoughts’). 64 Cf. further TB oṅkolmaññe : ta oṅkalmeṃ ‘of the elephant’ and TB rṣākaññe : ta riṣakeṃ ‘sage’.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

299

As far as TA -ñi is concerned, it is very sporadically attested, being limited to just three adjectives: TA oñi ‘human’, TA yokañi ‘thirsty’, and TA praskañi ‘fearful’. TA praskañi is a hapax legomenon attested in A111 b4, while yokañi is attested twice in construction with kaśśi (kaśśi yokañi “hungry and thirsty”, in A13 a1 and A105 b5; cf. also ///ime kälpo yokañ(i)/// in THT1143 a3). The only adjective that displays the expected semantics of the base from which it is derived is TA oñi ‘human’. It is attested only once as a free word (A51 a2 oñi pātrukaṃ “in a human skull”); otherwise it normally figures in a compound with cmol ‘birth’ (cf. also the derived adjective TA oñi-cmolṣi ‘pertaining to the human birth’). It is generally assumed that this adjective is the counterpart of TB eṅkwaññe ‘human’ (Van Windekens 1979: 119; Hilmarsson 1987b: 85; Pinault 2011c: 454). Winter (1961: 277) questions this equation, claiming that the paucity of the attestations of the suffix TA -ñi (vs. the productivity of its supposed counterpart TB -ññe) may be an indication of its late creation. The stems from which praskañi (vs. praski ‘fear’) and yokañi (vs. yoke ‘thirst’) derive is not clear. If the adjectives were derived from the nouns, a different form might have expected, i.e. **praskiñi and **yokeñi (cf. TA ypeṣi ‘pertaining to the land’ from ype ‘land’; TA pekeṣi ‘pertaining to the drawing’ from peke ‘drawing’).65 Nonetheless, the very fact that these adjectives are derivationally and semantically obscure may be an indication of their early creation. Furthermore, the relation between TB eṅkwaññe and TA oñi < *onk-ñi cannot be questioned (the loss of *k is parallel to TB epiṅkte : ta opänt, TB piṅkte : ta pänt, see Peyrot 2013a: 538–539; cf. also TA āñcäṃ vs. āñm-, with ñcm > ñm). Moreover, these adjectives appear to be uninflected, and they mostly occurred in fixed expressions and derivatives. This piece of evidence may indicate that they continue frozen forms of the adjectival paradigm, without renewed case endings. However, a precise explanation of the origin of the suffix TA -ñi is wanting. One possibility is that in TA oñi a different development of PT -ññe took place, due to the fact that *-ññe was before a consonant cluster, i.e. *enkññe > oñi. However, this explanation is very tentative. Finally we turn to the suffix -ñci, which is a peculiarity of Tocharian A. It is limited to a handful of adjectives. The most prominent members are kuleñci ‘womanly, female’ from kuli ‘woman’ (obl. sg. kule) and ātläñci ‘manly, 65

Similar considerations have been put forward by Sieg, Siegling & Schulze (tg § 29), who claim that praskañi and yokañi are derived from the respective verbal roots and not from the nouns, since “[d]ie alleinnachweisbaren Substantivformen […] lassen sich lautlich mit den Adjektiven nicht gut vermitteln”. As far as TA oñi is concerned, Winter proposes a formation in TA -i, thus *oṅk-i > *oñśi > oñi, although the reduction *-ñś- > -ñ- is, to my knowledge, unattested (cf. also 3sg.opt. nśitär from TA näk- ‘to perish, disappear’).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

300

Chapter 4

masculine’ from ātäl ‘man’. These formations sometimes correspond to the Tocharian B ññe-adjectives, as in TA atroñci ‘of a hero’ : TB etreuññe* ‘id.’. In fact, TA -ñci and TB -ññe share the same semantic distribution. Furthermore, ordinals based on decades are also formed with TA -ñci, like taryākiñci ‘thirtieth’ from taryāk ‘30’. Pinault (2017a: 1343) traces it back to a palatalised doublet of *-nte < PIE *-nto- (of the type TB ṣuktante, ta ṣäptänt ‘seventh’, TB oktante, ta oktänt* ‘eighth’, etc.; see also Van Windekens 1979: 123–124). More specifically, I see in this suffix a conglomerate of *-nt- + *-ii̯o-. 4.7.1.4 Inflectional Patterns and Related Problems of subclass i.2 In Tocharian B, the derivatives in -(i)ye, -ṣṣe, and -ññe inflected according to the following paradigm: Table 59

nom. sg. obl. sg. nom. pl. obl. pl.

Inflection of class i.2 adjectives in Tocharian B

masculine

feminine

-(ć)Će -(ć)Će -(ć)Ći -(ć)Ćeṃ

-(ć)Ća -(ć)Ćai -(ć)Ćana -(ć)Ćana

When these adjectives are compared with those of subclass i.1, it is apparent that the two most relevant differences are precisely those which define the distinction between the two subclasses: (1) phonological palatalisation throughout the paradigm; (2) feminine plural -ana. A related question is therefore the nature of the relation between the plurals -ana and -ona. We will return to this issue in the following paragraphs. In Tocharian A we find a different situation. Indeed, a heavy restructuring process affected the paradigm of these derivatives. This process resulted in a significant number of synchronic variants, especially in the case of the adjectives in TA -i and -ṣi (tg §§ 110–112). In the following, I will first outline the synchronic paradigms of these derivatives and I will then discuss them diachronically. The paradigm of the masculine is as follows (teb § 215):

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective Table 60

301

Masculine paradigm of the i- and ṣi-adjectives in Tocharian A

masculine singular

plural

nom.

-(ṣ)i

obl.

-(ṣ)i -(ṣ)iṃ -(ṣ)inäṃ

-(ṣ)iñi [-(ṣ)iñ] -(ṣ)inäs [-(ṣ)is]

The obl.sg. -(ṣ)i is common and coexists with the nasal variant (tg §§ 111–112). Examples of the i-adjectives include: ñäkci ‘divine’ (A13 b3) ~ lāñciṃ ‘royal’ (A71 b1, b5), ñäkciṃ (A145 b6; A257 b3) ~ lāñcinäṃ (A56 a2; A57 a1). The case of obl.sg. lāñci ‘royal’ (A1 b4; A16 a4, b1; A276 a7; A394 a2; A403 a1) is less certain, since it consistently occurs before waṣt ‘palace’, so it cannot be excluded that it is in a compound with the noun (cf. also lāñci waṣtantu “royal palaces” in A319 b5). In the plural, the variants nom.pl. -(ṣ)iñ, obl.pl. -(ṣ)is are uncommon. I found the following examples: nom.pl. -ṣiñ (tsraṣiñ A1 a3, b6; A447 b5, from tsraṣi ‘energetic’), -iñ (kaśśiñ A341 a4; A340 a4 (?), from kaśśi ‘hungry’, cf. TB keściye), obl.pl. -ṣis (all.pl. tsraṣis-ac A1 a3; perl.pl. tsraṣis-ā A354 b3), -is (instr.pl. kärpis-yo SHT4438 b2 [= instr.pl. Skt. anāryaiḥ ‘vulgar, inferior’], from kärpi ‘common, vulgar’, cf. TB kärpiye*). Note that they are mostly used as substantivised adjectives. Compare the following examples: A1 b6 (Puṇyavantajātaka) māski kätkāläṃ ktäṅkeñc tsraṣiñ difficult(ty) cross.ger.m.obl.sg cross.3pl.prs energetic.m.nom.pl sāmuddrä ᛬ ocean(a).obl.sg “The energetic ones cross the ocean that is hard to traverse”. (cf. Thomas 1952: 34) A447 b5 (literary) (ṣñi)kek nu cem tsraṣiñ ṣeñc however but this.m.nom.pl energetic.m.nom.pl be.3pl.ipf “However, they were strong”. (cf. Knoll 1996: 17)

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

302

Chapter 4

In the feminine paradigm we find even more variants:66 Table 61

Feminine paradigm of the i- and ṣi-adjectives in Tocharian A

feminine

nom. obl.

singular

plural

-(ṣ)i -(ṣ)iṃ -(ṣ)i -(ṣ)iṃ -(ṣ)ināṃ -(ṣ)yām, -ṣṣāṃ

-yāñ, -(ṣ)ṣāñ -(ṣ)ināñ -yās, -(ṣ)ṣās -(ṣ)inās

In the nominative singular, -(ṣ)i frequently alternates with -(ṣ)iṃ, which is, however, less frequently attested. On the other hand, TA -(ṣ)i used as an obl.sg. has very limited productivity. See the following attestations (tg §§ 110–101; Peyrot 2012a: 201–203): (1) lāñci kuleyac “to the royal woman” in A6 b5; (2) ñ(ä)kci naweṃsi(n)e “of divine and human …” in A410 b4; (3) kn(āṃ)muneṣi kapśiññis “of the body of wisdom” in A244 b2 (from knānmuneṣi ‘related to knowledge’); (4) opp{a}lṣi pārenā “on the lotus throne” in A316 b5.67 The obl.sg.f. -(ṣ)iṃ is more frequently attested but it is not the standard variant (tg § 112 counted 9 attestations in total), because -(ṣ)ināṃ represents the most productive obl.sg.f. For the last variant (TA -(ṣ)yām, -ṣṣāṃ), I found the following attestations:

66 Sieg, Siegling & Schulze (tg §§ 110a and 111) give two attestations of forms ending in -eṃ and -i used as feminine plurals. The former is attested in A378 a1 wsāṣy-ople« ṃ » tsākkiñ “tsākkis of golden lotuses” (see Peyrot 2014 fn. 46 for the correct reading and translation), where the anusvāra has to be restored and we cannot exclude that wsāṣy-opleṃ was an uninflected adjectival compound. The second is lāñci waṣtantu “the royal palaces”, which is better explained as a compound (Bernhard 1958: 158). 67 I am not convinced by Peyrot’s (2012a: 202) interpretation of TA waṣti ‘related to the house’ in A102 a2 // (wa)ṣti ñäkteññānac as an obl.sg. of an i-adjective in agreement with ñäkte­ ññānac ‘to the goddess’. Indeed, if it were an i-adjective, I would expect palatalisation of the cluster *waṣt-i > *waśśi (cf. lāñci ‘royal’ from the obl.sg. lānt ‘king’). Furthermore, TA waṣti is a hapax legomenon that appears to be at the beginning of a broken line so that the reading is effectively only ///ṣti.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

303

ñäkcyāṃ A35 b1, A63 a6, A208 a3, THT3020 a2; //-ṣāṃ A5 b1 (?); putti­ śparṣṣāṃ A257 a3, A313 a2, A338 a2, THT2399 a6, YQII.12 a8 (from putti­ śparṣi ‘relating to Buddhahood’); añcwāṣṣāṃ A340 a7 (from añcwāṣi ‘related to iron’); wsāṣṣāṃ A378 5 (from wsāṣi ‘golden’); oñi-cmolṣāṃ A379 a3 (from oñi-cmolṣi ‘related to the human birth’), ñemiṣyāṃ A227–228 a1 (from ñemiṣi ‘pertaining to joy’). The distribution of the variants in the plural paradigm is more intricate. As far as I know, among the i-derivatives only two adjectives attest a feminine plural inflection: TA ñäkci ‘divine’ and TA lāñci ‘royal’. The former consistently has a nom.pl. ñäkcyāñ (e.g. in A25 b2, A59 a1, A187 a6, A189 a2, A249 a1, A257 b4, A268 a1, A269 and 290 b1, A272 b4, etc.), and an obl.pl. ñäkcyās (e.g. in A73 b1, A77 a2, A144 b2, YQII.14 a6, etc.), while the latter always has a nom.pl. lāñcināñ (A64 b1, A76 and 83 a4), and an obl.pl. lāñcinās (A76 and 83 a3). In the ṣi-adjectives, the plural set -ṣināñ | -ṣinās constitutes the standard variant, but the second set is equally attested: nom.pl. waśirṣṣāñ A264 a2 (from waśirṣi ‘pertaining to a diamond’); añcwāṣṣāñ A295 a3, YQ N.3 a7. obl.pl. saṃsārṣṣās A69 a2 (from saṃsārṣi ‘related to the saṃsāra’); cmo­ lwāṣṣās A152 a6 (from cmolwāṣi ‘related to the birth); puttiśparṣṣās ~ puttiśparṣās A25 b4, YQ II.12 a6; parnoreṣṣās YQ II.12 a6 (from parnoreṣi ‘of splendor’); arkämnāṣṣās A375 b5 (from arkämnāṣi ‘of the place of the dead’); ñemiṣṣās YQ N.4 a6; kapśiṃñāṣās A7 b5–6 (from kapśiññāṣi ‘related to the body’); napeṃṣās YQ I.2 a4, YQ III.6 a3 (from napeṃṣi ‘of a human being’); wlaluneṣṣās A454 b3 (from wlaluneṣi ‘belonging to death’). From both a synchronic and diachronic point of view, all these variants can be divided into two parallel paradigms: one is based on the historically regular form of the suffix -(ṣ)i-, and the other on an extended nasal variant -(ṣ)in-. The problems involved are various. They relate to both the diachrony of Tocharian A and the comparison with the corresponding Tocharian B paradigms. The first issue concerns the origin of the nasal stem and how the variant forms are to be interpreted diachronically. Then, if we look at the Tocharian B counterparts, two further questions arise: (1) what is the relation between nom.pl.f. TA -āñ, obl.pl.f. -ās vs. pl.f. TB -ana (nom. = obl.)? (2) what was the Proto-Tocharian paradigm of these adjectives? Let us start with the first problem. If we compare the two-layer system of Tocharian A with the much simpler Tocharian B system, the n-paradigm of

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

304

Chapter 4

Tocharian A appears to be an innovation. It follows that the shorter forms are to be interpreted as the archaic ones (Peyrot 2012a: 201). The precise origin of the n-paradigm is not entirely clear, since it may have had multiple sources. The influence of the nasal inflection in the Tocharian adjectival system has been notably profound, and it affected both Tocharian A and B even after the dissolution of Proto-Tocharian. A good point of comparison may be the case of the re-adjectives in Tocharian B. Indeed, we find two types of re-adjectives in this language (Pinault 2008: 513–514): (1) the first is the regular outcome of the PIE thematic formations, which are ranged under subclass i.1 (the so-called astare-type, cf. TB astare ‘pure’, nom.pl.m. astari, obl.pl.m. astareṃ); (2) the second differs from the first in having developed a nasal inflection that is limited to the paradigm of the masculine (class ii.4, the so-called tapre-type, cf. TB tapre ‘deep’, nom.pl.m. täpreñ, obl.pl.m. täprenäṃ).68 In addition, these two types of re-adjectives are differentiated by the number of syllables (disyllabic for the nasal type, polysyllabic for the thematic type), the subsequent position of the stress (synchronically on the ending in the nasal type, but on the root in the thematic type), and the formation of the verbal abstracts (the suffix is -auñe for the nasal type, but -(əñ)ñe for the thematic type). Tocharian A does not display this division in the thematic adjectives and there is no evidence that it ever had such a binary system. Therefore, one may wonder whether a similar recharacterisation of certain “thematic” adjectives took place in the Tocharian A derivatives in -(ṣ)i. Another possibility is that Tocharian A has generalised the singular form as the basic stem of the plural in all adjectival paradigms of class i.2. A clear example is provided by the TA (e)ṃ-adjectives, whose paradigm is as follows (tg § 253): Table 62

nom. sg. obl. sg. nom. pl. obl. pl.

68

Inflection of the adjectives in -(e)ṃ in Tocharian A

masculine

feminine

-(e)ṃ -(e)ṃ -(e)näṃ -(e)ñi -(e)näs

— -(e)nāṃ -(e)nāñ -(e)nās

A similar contrast is also found in the dual inflection (cf. i-duals TB āstry ‘pure’, kätkri ‘deep’ vs. ne-duals TB tparyane ‘high’, prakaryane ‘firm’). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

305

As can be seen, in an unattested phase of Tocharian A, the singular stem -(e)ṃ (the regular outcome of PT *-(V)ññe) was generalised and the endings were reattached to this new stem. Indeed, if we look, for instance, at the paradigm of the masculine, we notice that the nom.pl. -eñi, obl.pl. -enäs cannot be the expected outcomes of nom.pl. PT *-ññey, obl.pl. PT *-ññens, since the diphthong -ey was expected to yield TA -e and we have no continuant of either the thematic vowel PIE *-o- > PT *-e-, or the cluster PT *-ññ-, which is expected to yield TA -ñ- in non-final position. I believe that the same kind of recharacterisation should be reconstructed for the derivatives in TA -i, -ṣi, where a new stem *-(ṣ)in- was created, probably based on a recharacterised oblique singular *-(ṣ)in. The masculine paradigm nom.pl. -(ṣ)iñi, obl.pl. -(ṣ)inäs can indeed be descriptively interpreted as the oblique singular -(ṣ)in- plus the palatalising nom.pl. -i on the one hand (< PT *-’əyǝ), and plus the “athematic” obl.pl. -äs on the other hand (< PT *-əns < PIE -n̥ s). The generalisation of the oblique singular *-n may have been favoured by the productivity of the nasal stems in Tocharian. This restructuring development produced the contrast between nasal and nasalless stems. The latter is to be interpreted as the regular outcome (Peyrot 2012a: 201): nom.obl.sg. TA -ṣi, -i : tb -ṣṣe, -iye (< *-ṣṣye, *-(ə)ye) nom.obl.sg. TA -(e)ṃ : tb -ññe (< *-(V)ññe) The fact that the nasal recharacterisation is a secondary development is also confirmed by the paradigm of the feminine, which shows a clear contrast between nasal and nasalless stems in the plural. As a matter of fact, it is in the feminine that we find more variants. If we isolate the n-forms, we are left with the following paradigm: nom.sg.f. -i; obl.sg.f. -i, -yāṃ, -ṣyāṃ (> -ṣṣāṃ);69 nom.pl.f. -yāñ, -ṣṣāñ, obl.pl.f. -yās, -ṣṣās. We now turn to the relation between the plural TA nom. -āñ, obl. -ās vs. TB nom.obl. -ana. The matter at hand here is which of the two languages preserves the older state of affairs. Some scholars, such as Kim (2009a: 74, 2018: 84) and Fellner (2013, 2014: 19 fn. 35), claim that neither Tocharian A nor Tocharian B have continued the Proto-Tocharian ending. That is to say, in the Proto-Tocharian continuant of the PIE thematic type, there existed a single feminine plural ending, which is reconstructed as *-ona (= subclass i.1). I cannot agree with this reconstruction. Indeed, the precise synchronic subdivision of class i as exemplified above speaks in favour of the split of the two subclasses at a Proto-Tocharian stage (cf. TB -ona, TA -aṃ vs. TB -ana, TA -āñ | -ās). 69

The evolution TA -ṣyā- > -ṣṣā- is an inner-Tocharian A gemination, cf. perl. sg. poṣṣā from poṣi ‘wall, side’, nom.pl. āśyañ ~ aśśāñ from aśi ‘nun’. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

306

Chapter 4

Again, I believe that Tocharian B has preserved the original situation. Indeed, I cannot envision any reason why a plural paradigm with nom. PT *-añə (cf. TA -āñ), obl. *-ans (cf. TA -ās) should not have been maintained in Tocharian B, nor why these endings would have come about in Proto-Tocharian in the first place. On the other hand, if we reconstruct pl.f. *-ana (nom.=obl.) for Proto-Tocharian, we can envisage a plausible diachronic development thanks to which this ending was eliminated in Tocharian A. Let us start with the reconstructed Proto-Tocharian paradigm of the feminine as Tocharian B allows us to reconstruct: Table 63

Proto-Tocharian feminine paradigm of subclass i.2

nom. obl. gen.(-dat.)

singular

plural

*-(ć)Ćya *-(ć)Ćya *-(ć)Ćyay

*-(ć)Ćyana *-(ć)Ćyana —

This paradigm was continued without relevant modifications in Tocharian B (for the replacement of obl.sg. PT *-a with the gen.sg. *-ay, see § 3.5.2.5 and § 4.7.1.2). Before vowel apocope in Tocharian A, a distinction between nominative and oblique was reintroduced in the singular: as is regular in Tocharian A, a nasal ending *-n was added to the inherited oblique singular, which led to a contrast between nom.sg. *-ā, obl.sg. *-ān. Then, vowel apocope took place and the new obl.sg.f. became homophonous with the apocopated plural *-ān < PT *-ana. Such homophony between the obl.sg., nom.pl., and obl.pl. in the paradigm could not have been maintained for long. As a consequence, a new distinction between nominative and oblique plural was introduced: the nom.pl. *-ān was palatalised into *-āñ, and the obl.pl. *-ān was levelled with the ubiquitous oblique plural marker -s, thus *-ās. A similar development can be inferred from the paradigm of some of the athematic declensions, as I will discuss in the following paragraph. The diachronic evolution of the Tocharian A paradigm can be exemplified as follows:

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective Table 64

nom.sg. obl.sg. gen.sg. nom.pl. obl.pl.

307

Evolution of the feminine paradigm from Proto-Tocharian to Tocharian A

PT

pre-TA

*-Ćya *-Ćya *-Ćyay *-Ćyana *-Ćyana

> *-Ćyā >> *-Ćyān > *-Ćyay > *-Ćyāna > *-Ćyāna

TA >*-Ći >*-Ćyān > *-Ćye > *-Ćyān > *-Ćyān

> -Ći > -Ćyāṃ > -Ćye >> -Ćyāñ >> -Ćyās

4.7.2 The Athematic Type (classes ii, iii, iv) In this section, I deal with the remaining adjectival classes of Tocharian in order to clarify which adjectival types are relevant to the development of the gender system and to reconstruct their Proto-Tocharian paradigms. Since the inflection of the Tocharian preterite participle (class iv) was heavily remodelled in both Tocharian languages, it does not constitute a central topic of my discussion.70 According to Krause & Thomas (teb §§ 230–239), class ii is very heterogeneous. It is divided into five subclasses on the basis of the inflected form of the Tocharian B nominative plural masculine: (ii.1) -iñ; (ii.2) -aiñ; (ii.3) -añ; (ii.4) -eñ; (ii.5) -oñ. Given the fact that each of these subclasses presents individual problems and different degrees of productivity, I introduce them separately to establish which subclasses can be used to reconstruct the Proto-Tocharian state of affairs. Class ii.2 is practically non-existent, since the plural -aiñ is limited to the paradigm of TB yolo ‘bad, evil’, which has a unique and unusual paradigm (cf. also the alternating stem yolo- ~ yolai- ~ yoloy- ~ yoly-). Peyrot (2016a) deals with the inflectional problems and the etymology of this adjective, supporting its foreign origin (from Khot. yaulā̆- ‘falsehood, evil’,71 cf. OUy. yavlak) and proposing that this nominal was first borrowed as a noun which subsequently developed adjectival use (Hilmarsson 1987a: 36).

70 71

See Saito (2006), Pinault (2008: 527–533), and Peyrot (2010b) for a recent discussion on the evolution of the paradigm of the preterite participle. For a discussion of the inflectional class of the Khotanese word, see Del Tomba (2022: 171) and further Dragoni (2022: 173–175).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

308

Chapter 4

The derivatives with plural -iñ (class ii.1) and -añ (class ii.3) have been the topic of controversial interpretations. The latter plural is characteristic of a number of agentive formations that are both morphologically and semantically connected. They are built on different verbal stems by means of the following suffixes: (1) TB -tsa, ta -ts (TB aknātsa, ta āknats ‘foolish’); (2) TB -ntsa (TB wapāntsa ‘weaver’); (3) TB -nta, ta -nt (TB kauṣenta, ta koṣant ‘killer, killing’); (4) TB -uca (TB kärstauca ‘cutting’); (5) TB -a in verbal governing compounds (TB yolo-rita ‘seeking evil’). In recent years, these formations have become one of the most debated topics within Tocharian nominal morphology. The problems involved are various, but they revolve around (1) the part of speech to which they belong and (2) the origin of their inflection.72 I basically agree with Peyrot (2013b, 2017) and Friis (2021) in arguing that they are to be analysed as nomina agentis, i.e. as substantives, including the so-called nt-participles (Malzahn 2010: 480–481). Indeed, they lack some of the characteristics that allow us to distinguish Tocharian adjectives as an independent part of speech. We can say that a prototypical adjective shares the following characteristics in Tocharian:73 (1) inflectional characteristics, i.e. special case markers, like the gen.sg.m. TB -(e)pi, ta -āp; (2) syntactic characteristics, i.e. semi-rigid position with respect to the head noun (inversion is sometimes attested in metrical texts or even in prose as a stylistic devise); (3) paradigmatic characteristics, i.e. different forms with respect to number, gender, and case; (4) morphosyntactic characteristics, i.e. agreement with the head noun in number, gender, and case. In fact, these formations lack any differentiation according to gender, some of their endings are characteristic of the noun inflection (cf. gen.sg. TB -ntse, ta -es), they are used to translate Sanskrit agent nouns in -in- (Peyrot 2017), and they are only sporadically employed to modify a noun (where they may be interpreted as being in apposition, rather than as attributive adjectives). There exist, however, counterarguments, as participial features include the fact that they are sometimes modified by adverbs and that they govern an object in the oblique (Fellner 2017b: 73–84). 72 73

See Malzahn (2010: 481–491), Pinault (2012a), Hackstein (2012), Fellner (2014c and 2017a), and Friis (2021). Tocharian has also a number of uninflected adjectives which often blur the boundary between adjectives and adverbs (Carling 2017: 1352). Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

309

The characteristics of class ii.3 are, in part, also shared by the derivatives of class ii.1. This subclass is mostly represented by verbal governing compounds that in the singular end in TB -i, ta -e (see Malzahn 2012b and Fellner 2018). Examples include: TB °ākṣi ‘announcing, proclaiming’ (from aks- ‘to announce’); TB °aiśi ‘knowing’ (from ayk- ‘to know’); TB °yāmi ‘doer, doing’ (from yam- ‘to do’); TB °plaṅṣi ‘seller, selling’ (from plǝnk- ‘to sell’); TB °nakṣi ‘destroyer, destroying’ (from nǝk- ‘to destroy); TB °pilṣi ‘listening’ (from pǝyl‘to listen’); TA °käṃṣe ‘occurring’ (from kän- ‘to occur’); TA °pāṣe ‘protecting’ (from pās- ‘to protect’). These formations are mostly used as nouns rather than as adjectives. Even when they are used to modify a noun, they can be interpreted as appositions without any difficulty (e.g. B229 b4 [arch.] läkle-näkṣi säkw-aiṣṣeñcai käṣṣi “Oh master, destroyer of suffering, giver of fortune”). From an inflectional point of view, they are inflected as nouns since they have the characteristic gen.sg. TB -ntse (e.g. IT159 a5 /// (wā)ki po-aiśintse snay allaiknesa “the superiority of the all-knowing in no other way”, cf. Broomhead 1962: 1.229; cf. also °yamintse in B251 a4 and B304 b3). Furthermore, some of them develop a different plural marker, like TB po-aiśi ~ poyśi ‘all-knowing, the omniscient one’ (calque from Skt. sarvajña-, Pinault 2008: 561), which has a plural poyśinta, taken after käṣṣinta ‘masters’ (plural of TB käṣṣi), both frequently used as epithets of the Buddha (Pinault 2003b: 338). An argument against the above interpretation comes from the fact that these formations are supposed to have paradigmatic gender-differentiation (teb § 230). Indeed, formations ending in TB -iñña are usually interpreted as the paradigmatic feminine counterparts of these nomina agentis. I was only able to find the following examples of the formations in question: (1) poysiñña ‘all-knowing’; (2) pkänte-yamiñña ‘hindering’; (3) käryor-pläṅṣiñña ‘selling (?); woman seller (?)’. The latter is a hapax legomenon attested without context in IT129 b5. It is therefore impossible to determine whether it is used to modify a noun or not. Ogihara (2009: 351) and Malzahn (2013a: 111) favour the second hypothesis. Malzahn interprets the suffix -ñña as the Tocharian marker of feminine motion (see § 4.7.1.4 above), thus ‘female seller’. The other two formations are consistently attested as modifiers of a head noun: TB pkänte-yamiñña is only found in agreement with wäntarwa ‘things’, thus pkänte-yamiññana wäntarwa “hindering things” (in IT27 b4; AS19.8 b1; THT1111 a4; THT1113 b5); TB poyśiñña is found several times without context (nom.sg. poyśiṃña AS17B a5; obl.sg. poyśiññai THT1247 b5, THT1260 b4; pl. poyśiññana IT272 a2), but in all other attestations it modifies a head noun (poyśiññai ekṣalympa “with the feast of the all-knowing” IT2 a2; poyśiññana rekauna “the words of the all-knowing” IT144 b5; poyśiññana ekṣalyänmeṃ “from the feasts of the all-knowing” IT271 b2; poyśiññana krentauna “the virtue(s) of the all-knowing” B205 a1). This fact is clearly at odds with that of the respective masculine forms Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

310

Chapter 4

and it may invalidate our analysis. Still, I believe that these formations in -ñña are not to be interpreted as the paradigmatic feminine counterpart of the verbal governing compounds in TB -i, but rather as feminine inflected forms of derived ññe-adjectives. Clear evidence in support of this analysis is that the adjective TB poyśiññe ‘pertaining to the all-knowing’ (from poyśi ‘all-knowing’, cf. also poyśiññeṣṣe ‘id.’) is attested in the same morphosyntactic context as the feminine poyśiñña. Thus, agent nouns in TB -i do not have a paradigmatic gender-differentiated feminine form. To sum up, I believe that the Tocharian formations of subclass ii.1 (TB -i, ta -e) and subclass ii.3 (TB -a, ta -∅) are to be interpreted as (agent) nouns. They may sporadically modify a head noun in apposition since there is no strong morphosyntactic (inflected like nouns; no rigid position; seldom agreement with a head nouns) or paradigmatic evidence (no feminine paradigm) to claim that they can be labelled as “adjectives” (but see Fellner 2017b). Therefore, their inflection will not be considered in the present chapter.74 On the other hand, there exists an isolated nominal that is formally ranged under class ii.1, although it is not derived from any verbal root and its adjectival use is beyond dispute. This is the adjective for ‘white’, which seems to belong to an original nasal inflection in Tocharian B. Its paradigm is as follows (Hilmarsson 1996: 40): Table 65

Paradigm of TB ārkwi ‘white’

masculine

nom. obl.

feminine

sg.

pl.

sg.

pl.

ārkwi arkwiṃa (?)

arkwiñ* arkwinäṃ

arkwañña arkwaññai

arkwīna arkwīna

a The obl.sg.m. is allegedly attested in IT170 a2 saiwaisa arkwiṃ tseñceṃ “on the left, white and blue (?)”. The context is however difficult, because no head noun is attested with which arkwiṃ may be in agreement, and tseñceṃ ‘blue’ is a hapax legomenon based on the stem of tseṃ ‘id.’. As a matter of fact, this arkwiṃ may also be a late variant of nom.pl.m. arkwiñ (see Hilmarsson 1996: 40).

74

Friis (2021) recently argued that the agent nouns in TB -ntsa, ta -nts continue the PIE feminine active participle in *-nt-ih2. She assumes that gendered inflection was lost when, in pre-Proto-Tocharian, these formations were lexicalised as agent nouns.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

311

In Tocharian A, this adjective shifts to the nt-inflection in the plural (class iii), cf. nom.pl.m. ārkyaṃś, nom.obl.pl.f. ārkyant, possibly adopted after TA arkant-* ‘black’, tb erkent- < PIE *h1r̥gw-ont- (DTA 15–16: 15–16; dtb 101). The identification of TB ārkwi, ta ārki with Gk. ἀργός, Skt. árjuna-, etc. goes back to the early years of Tocharian studies (Meillet 1911: 149). All these cognate forms are descendants of PIE *h2erǵ- ‘shining, white’. However, the exact derivational mechanism involved is still a matter of debate. Indeed, the Tocharian adjective seems to have been variously suffixed. Hilmarsson (1996: 41) argues that a reconstructed PIE *h2erǵu- ‘white’ (a Caland adjective) was extended with an individualising n-suffix *-ion-/-ien- in Tocharian. Indeed, the PIE root *h2erǵwas variously resuffixed in the Indo-European languages, sometimes with *-ior *-i-n- (cf. Hitt. ḫarki-; Gk. ἀργι- in compounds and further ἄργιλλος ~ ἄργιλλα ‘herbe à chèvres’, ἀργινόεις ‘whitish, shining’ [Hom.; Plut.], ἀργαίνω ‘to be white’, Chantraine 1933: 249), sometimes with *-u-n- or *-u-r- (Skt. árjuna- ‘white’, Gk. ἄργυρος ‘white, silver’, ἄργυφος ‘silver-shining’, cf. also Lat. argentum ‘silver’). Be that as it may, the fact that TB ārkwi, ta ārki goes back to an n-stem adjective is assured by its inflection, cf. obl.pl.m. arkwinäṃ (acc.pl. < *-n-n̥ s) and nom.sg.f. TB arkwañña /arkwə́ ñña/, ta ārkiṃ, which can be interpreted as the direct cognate of Ved. árjunī-, outcome of PIE *h2erǵu-n-ih2. The lack of palatalisation in the nom.obl.pl.f. TB arkwīna is unexpected. This evidence is at odds with the paradigm of TB tseṃ ‘blue’, a loanword from MChin. *tsheng > cāng 蒼 (Lubotsky & Starostin 2003: 265), which shows palatalisation of the nasal throughout the paradigm (f.nom.sg. tseñña, obl.sg. tseññai and the nom.obl.pl. tseññana). It goes without saying that the plural arkwīna cannot therefore be historically analysed as an original feminine inflected form; that is, it is not the outcome of a reconstructed form containing the athematic feminine suffix *ih2. More specifically, we can say that it does not attest palatalisation because it is the regular outcome of the old neuter plural form. We will return to the paradigm of TB ārkwi in the following section. Adjectives with nom.pl.m. TB -eñ (subclass ii.4) are mostly those thematic re-formations that developed a nasal inflection (of the tapre-type). It seems that this pattern was also extended to other original thematic adjectives, which are all disyllabic, like TB tute ‘yellow’, obl.sg. tuceṃ, obl. pl. tucenäṃ (dtb 318), and some we-adjectives, like maiwe ‘young’ and raiwe ‘slow’, etc. Since this subclass is agreed to be a Tocharian B innovation, it will not be used for the reconstruction of Proto-Tocharian (Pinault 2008: 513–515). The last group to be considered is subclass ii.5. It is the only inherited adjectival class of the nasal type that is quite productive in both Tocharian B and A. It consists of adjectives in TB -mo, ta -m. The most prominent member

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

312

Chapter 4

is TB klyomo, ta klyom (< PT *kĺəwmo < PIE *ḱleumōn, cf. Av. sraoman- ‘hearing’, Skt. śrómata- ‘reputation’, OHG (h)liumunt ‘id.’), which was inflected as follows (teb § 238): Table 66

Paradigm of the klyomo-type

masculine

nom. sg. obl. sg. nom. pl. obl. pl.

feminine

TB

TA

TB

TA

klyomo klyomoṃ klyomoñ klyomoṃ

klyom klyomänt klyomäṣ klyomäñcäs

klyomña klyomñai klyomñana klyomñana

klyomiṃ klyomināṃ klyomināñ klyominās

These formations go back to the PIE type in *-mon-/-mn-. As pointed out by Hilmarsson (1996: 156) and Pinault (2008: 520), the nom.sg.m. *-mōn regularly yielded TB -mo, ta -m; the rest of the masculine paradigm was remodeled after this case-form in both Tocharian languages. Thus, we have nom.pl.m. TB -moñ for expected **-mañ > *-mōn-es or **-meñ > *-mon-es. We have already observed that in Tocharian A the masculine paradigm was heavily influenced by the nt-stems (cf. also the late variant obl.sg.m. TB klyomont, on which see Peyrot 2008: 119). As far as the feminine is concerned, we can see that both the singular and the plural paradigm of the klyomo-type closely mirror those of the thematic type of subclass i.2. The basic stem can be traced back to the zero grade *-mn̥ -ih2- > PT *-məññya-. Subsequently, Tocharian B degeminated the palatal nasal *klyoməñña- > *klyomñña- > klyomña-, while Tocharian A underwent the following development: *klyoməñña > *klyoməiñña (raising) > *klyoməiñ > klyomiṃ (depalatalisation).75 This form was generalised to the rest of the 75 The reduction of PT *-məññ- to TB -mñ- is testified by several other formations, like the abstract nouns TB cämpamñe /cəmpə́ mñe/ ‘ability, power’ < PT *cəmpəməññe, TB aiśamñe /ayśə́ mñe/ ‘wisdom’ < *ayśəməññe (Pinault 2011b: 454) vs. TB arkwañña /arkwə́ ñña/ ‘white’, TB eṅkwaññe /enkwə́ ññe/, TB täṅkwaññe /tənkwə́ ññe/ ‘pleasing, lovely’. The same reduction can be seen in the type TB cäñcarñe ‘love’ from cäñcare ‘lovely, agreeable’ and in the ññe-adjectives, cf. TB gautamñe ‘pertaining to Gautama’ from gautame ‘Gautama’, TB eṣerñe ‘related as a sister’ from ṣer ‘sister’ vs. TB ostaññe ‘domestic’ from ost ‘house’, TB yäkweññe ‘related to a horse’ from yakwe ‘horse’ (Kim 2007b).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

313

feminine paradigm through paradigmatic levelling. The contrast in the plural TB klyomñana : ta klyomināñ | -ās is to be interpreted as that of subclass i.2 (see § 4.7.1.3). We can therefore reconstruct the following Proto-Tocharian feminine paradigm: Table 67

Proto-Tocharian feminine paradigm of the klyomo-type

nom. obl. gen.(-dat.)

singular

plural

* -məñña * -məñña * -məññay

*- məññana *- məññana —

We now turn to class iii, which can be divided into two groups. The first group comprises two isolated adjectives, which share some peculiarities in their inflection and are synchronically characterised by suppletion in their paradigm: TB po, ta puk ‘all, each’ and TB kartse, ta kāsu ‘great, good’. The former adjective has pont- as the basic stem in both Tocharian languages and it has been connected with Gk. πᾶς, πᾶσα, πάν, as if from PIE *peh2-nt(Lévi 1933: 38). Pinault (2008: 522–524) and Kim (2019c) discuss the origin and forms of this adjective. A relevant issue is that in Tocharian B it does not show gender and case distinction between nominative and oblique in the singular.76 In Tocharian A, puk marks the nom.sg. of both masculine and feminine, but the oblique is usually differentiated, i.e. obl.sg.m. poñcäṃ, obl.sg.f. pontsāṃ. One can assume that this adjective was inflected for gender and case at an older stage of Proto-Tocharian, and that the gradual loss of this distinction in the singular started only at a later stage.77 Another peculiarity is that the feminine plural TB ponta, ta pont does not show assibilation of the stem final consonant in Tocharian B or A (cf. the obl.sg.f. pontsāṃ and the singular feminine paradigm of the nt-adjectives, nom. TB -ntsa, ta -ṃts, obl. TB -ntsai, ta -ntsāṃ). This also applies to the feminine plural of the second adjective, TB kartse (fem. kartsa), TA kāsu (fem. kräts), which builds the majority of its forms from 76

The uninflected form TB po, ta puk also occasionally occurs in agreement with plural forms, as well as when it is used as a pronoun. Thomas (1997) recognises that uninflected forms are more common in poetic texts, probably for metrical reasons. 77 According to Pinault (2008: 523), this development was triggered by the uninflected TB māka, ta māk ‘much, many’.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

314

Chapter 4

the stem TB krent(-), ta krant(-). Though synchronically suppletive, there is general agreement that these stems are diachronically related (with the exception of nom.sg.m. TA kāsu; see Pinault 2008: 521–522 and Kim 2019c). In the feminine we find a clear contrast between the singular and the plural: indeed, the singular is built on an assibilated stem, TB kartsa, ta kräts, while in the plural we have no assibilation, TB krenta, ta krant.78 The same pattern can be found in the second subclass of class iii, which comprises a productive group of derived adjectives which go back to the PIE suffix *-u̯ ent-. This suffix underwent various modifications depending on the stem final vowel to which it was attached (cf. TB perneu, ta parno ‘worthy’ from the ancestor of TB perne, ta paräṃ ‘glory’; TB tallāw, ta tālo ‘miserable’ from the Proto-Tocharian present stem of TB təll- ‘to bear’). Again, a feminine singular TB -ntsa, ta -nts (with assibilation) corresponds to the non-assibilated -nta, ta -nt in the plural. At this point, it is clear that the singular and the plural feminine paradigms cannot go back to the same Proto-Tocharian stem. As for the case of TB ārkwi ‘white’, the singular continues the feminine singular *-ntya- < *-ntih2-, while the plural goes back to the neuter plural *-nta < *-nth2. All things considered, the Proto-Tocharian paradigm of the feminine can be reconstructed as follows: Table 68

Proto-Tocharian feminine paradigm of class iii

nom. obl. gen.(-dat.)

singular

plural

* -ntsa * -ntsa * -ntsay

*- nta *- nta —

4.7.3 The Proto-Tocharian Adjectival System: A Summary Before commenting on the ultimate evolution of the adjectival system from Proto-Indo-European to Tocharian, let us summarise the Proto-Tocharian paradigms as they have been outlined in the previous sections.

78

For a discussion of TA kräntso ~ kräṃtso ‘beautiful, pretty’, see Kim (2019c).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

315

We have seen that class i, which continues the thematic type, can be synchronically divided into two subclasses in both Tocharian A and B. We have also seen that there are good reasons for claiming that such a binary system must be traced back to Proto-Tocharian as well. Their respective paradigms are reconstructed as follows: Table 69

Proto-Tocharian class i.1

masculine sg. nom. obl.

pl.

*-e *-ey *-e(ṃ) *-ens

Table 70

sg.

pl.

*-ya *-ya

*-ona *-ona

Proto-Tocharian class i.2

masculine

nom. obl.

feminine

feminine

sg.

pl.

sg.

pl.

*-e *-e

*-ey *-ens

*-a *-a

*-ana *-ana

The remaining classes continue the athematic inflection. We have seen that Tocharian A mostly remade the inherited paradigms, since they mutually influenced each other and sometimes merged. For this reason, Tocharian B constitutes our main source for reconstructing the Proto-Tocharian state of affairs. As far as the masculine inflection is concerned, a contrast between nominative and oblique singular can be reconstructed: the oblique was marked by the pure stem in Proto-Tocharian, which in the case of the n-stems was *-n, and in the case of the nt-stems was *-nt. In the plural we have the residue of the original stem in the nominative, which undergoes palatalisation in front of the PIE athematic ending nom.pl. *-es. As far as the feminine is concerned, the paradigm of the singular matched that of class i, while the nominative and oblique plural ended in *-a. The general paradigm is as follows:

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

316 Table 71

Chapter 4 “Athematic” adjectival paradigm of Proto-Tocharian

masculine

nom. obl.

feminine

sg.

pl.

sg.

pl.

*-∅ *-Cə

*Ćə *-Cəns

*-Cya *-Cya

*-Ca *Ca

The klyomo-type (class ii.5) deviates from the paradigm outlined in the feminine plural, where we can reconstruct an ending *-a-na preceded by palatalisation of the stem-final consonant, thus PT *-məññana. Now, if we look at these reconstructed paradigms from a comparative Indo-European perspective, a number of diachronic issues come to light. These problems are addressed in the following paragraph, where I deal with the ultimate evolution of the gender system from Proto-Indo-European to Tocharian. 4.8

Evolution of the Gender System: From Proto-Indo-European to Tocharian

The goal of this section is to trace the Proto-Indo-European origin of the Tocharian gender system in the inflection of agreement targets. The problems revolve around the evolution of the feminine, its merger with the neuter, and the functional loss of the neuter as a gender value. In order to understand how these gender values evolved in Tocharian, I recount the most important theories on their evolution, discussing the morphophonological convergences that led to the attested situation. I first deal with the masculine inflection, before moving on to the feminine, which constitutes the core of my discussion. Particular attention is devoted to the thematic inflection as the locus of most of the mergers between the three inherited genders. 4.8.1 Evolution of the Masculine Paradigm and the Neuter Singular From a formal point of view, the singular inflection of the masculine evolved without relevant modifications from Proto-Indo-European to the two Tocharian languages. The inherited distinction between nom.sg. *-o-s, acc.sg. *-o-m has been blurred due to the process of consonant erosion that affected Proto-Tocharian in word-final position. Apocope also affected the neuter inflection, which became homophonous with the masculine in the singular:

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective Table 72

Formal merger of the thematic masculine and neuter in the singular

PIE

nom.sg. acc.sg.

317

PT

masc.

nt.

*-o-s *-o-m

*-o-m *-o-m

> >

*-e *-e

Before the dissolution of Proto-Tocharian, a new distinction between nominative and oblique started to be reintroduced through the addition of the oblique marker *-n, taken from the nasal stems (Pinault 2008: 476–477). This ending only became mandatory in Tocharian A, while in Tocharian B it had a limited distribution (teb § 142), since it only appeared in those paradigms where analogical palatalisation did not differentiate the nominative from the oblique.79 The origin of the oblique marker *-n must certainly be sought in a Proto-Tocharian stage, at which time, however, it may not yet have been grammaticalised as a fixed inflectional marker. The regular obl.sg. PT *-e may therefore have had *-en as a variant form, which originally only marked a direct object characterised as [+ human] (as in the substantives, cf. obl.sg.m. ṣamāne-ṃ /ṣamánen/ ‘monk’ vs. obl.sg.m. yakwe /yə́ kwe/ ‘horse’).80 This reconstruction also has the advantage of explaining the lack of any continuant of *-en > **-aṃ in Tocharian A adjectives, where we instead find -äṃ (e.g. obl.sg. āsträṃ ‘pure’ vs. -aṃ in the noun, e.g. oṅkaṃ ‘man’). In other words, since the obl.sg. *-n was not a mandatory adjectival ending in Proto-Tocharian, it could not have protected the original obl.sg. *-e from the regular apocope of final vowels in Tocharian A.

79 Rarely, a nasal oblique singular seems to alternate with the nasalless form, cf. (a)s(t)are śaul ś(a)ye(ñc)ai “one who lives a pure life” (IT579 b4) and se laiko yetse as(tar)e yamaṣäṃ “this lotion makes the skin pure” (W11 b1) vs. śīlne stmoṣo astareṃ “remaining in the pure moral behaviour” (NS55 b4) and (śīla)ṣṣ=āstreṃ weresa “with the pure smell of the moral behaviour” (B313 a3=AS5b a2–3). However, forms without -ṃ can also be interpreted as scribal errors, i.e. the copyist forgot to write the anusvāra. 80 According to Sims-Williams (1990) and Pinault (2002), the marking of a direct object characterised as [+ human] and [+ definite] with specific forms is a peculiarity that Tocharian shares with some Eastern Middle Iranian languages. Cf. the similar use of the Bactrian accusative preposition αβο.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

318

Chapter 4

As far as the plural inflection is concerned, the inherited nominative plural PIE *-ōs < (virtually) *-o-es (preserved in Ved. -āḥ, Goth. -os, Osc. -ús, etc.) was replaced by the pronominal PIE *-oi̯, as happened in many Indo-European languages including Gk. -οι, Lat. -ī, OCS -i, etc. This regularly yields PT *-ey > TB -i, ta -e.81 Conversely, Ringe (1996: 81–82), Hackstein (2017: 1315), and Kim (2018a: 64–65) believe that PIE *-oi̯ monophthongised very early in the pre-history of Tocharian, resulting in a front vowel PT *-ẹ (in their notation) before palatalisation ceased to operate. According to their proposal, proof of this early monophthongisation of PIE *-oi̯ is seen in relic nouns, whose nom.pl. form has palatalisation before the ending TB -i. Indeed, in all other nouns that regularly continue PIE *-oi̯, the palatalised nom.pl. would have been eliminated through levelling after the rest of the paradigm. Kim (2018a: 64) adduces the following three relics (cf. also teb § 181):82 (1) nom.pl. TB kokalyi /kokǝ́ ĺǝy/ ‘chariots’ vs. obl.pl. kokaleṃ* /kokǝ́ len/ ~ kokleṃ; (2) nom.pl. TB kerc(c)i /kérc(c)ǝy/ ‘swords’ vs. obl.pl. kert(t)eṃ /kért(t)en/; (3) nom.pl. TB trici /trǝ́ycǝy/, ta trice ‘third (pl.)’ vs. nom.sg. TB trite /trǝ́yte/, ta trit. The palatalisation in the plural paradigm of TB trite, ta trit is of no value, because ordinals in -te show morphological (i.e. analogical) palatalisation in all case forms of the masculine (with the exception of the nom.sg.). There is therefore no contrast between e.g. palatalised nom.pl. vs. non-palatalised obl.pl (cf. nom.pl. trici*, obl.pl. triceṃ or nom.pl. waci, obl.pl. waceṃ, from TB wate ‘second’). TB kercci ~ kerci is usually considered to be the nom.pl. of kertte ~ kerte ‘sword’. This case form is attested twice: IT89 b1 (= B73 b4) sūryakāṃtṣi kercci ram no läktsecci “like bright sūryakānta-swords” (Thomas 1968: 211; Couvreur 1954: 103; Adams 2012: 28); AS17D a2 ylaiñäkti ñī kerci ra aiskeṃ traike lkālñesa “The Indra gods provide confusion to me through their [mere] appearance like swords [do]” (edition and translation follow Georges-Jean Pinault & Melanie Malzahn apud CEToM). Since TB kercci is homonymous with TB kercci ‘palace’

81 82

An investigation of the evolution of word-final PIE *-oi̯ in Tocharian can be found in Del Tomba (2020a). I have omitted TB recci (attested once in B423 b6), obl. recceṃ (cf. reccenmpa B307 b7), probably the plural forms of a derived tstse-adjective. Indeed, Chams Bernard (p.c.) pointed out to me that these forms may belong to the paradigm of another word, and may not be inflected forms of TB retke ‘army’ (cf. already dtb 585). In any case, TB retke, ta ratäk is certainly not an inherited word: it is a loanword from Iranian *rataka- ‘line (of an army)’ (MP radag ⟨ltkˈ⟩ (Phl.), ⟨rdg⟩ (ManMP), ManParth. rdg, NP rada, and further Oss. Iron rad, Digor radæ). See Winter (1971: 150–151) and Isebaert (1980: 189).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

319

(< *kerc(c)əyi, cf. obl.pl. kerc(c)iyeṃ), a plurale tantum, these kercci-forms may actually belong to the paradigm of ‘palace’ rather than to that of ‘sword’. We are now left with kokalyi. Here the contrast between palatalised nom.pl. kokalyi and non-palatalised obl.pl. kokleṃ is clearly attested.83 However, in this case the palatalisation of the nom.pl. may also have been analogical after the inflection of the gerundives in -lle (Hilmarsson 1996: 163–164), or — more likely in my view — it may reflect a secondary palatalisation of TB -li /-ləy/ > -lyi /ĺəy/.84 Additional evidence against the sound law PIE *-oi̯ > PT *-ẹ is that palatalisation never occurs in those nouns that continue PIE *oi̯-stems, like TB reki, ta rake ‘word’, TB leki, ta lake ‘bed, couch’, TB telki, ta talke ‘sacrifice’, etc. In these cases, one cannot invoke paradigmatic levelling to eliminate the palatalised allomorph, because palatalising *-ẹ should have been maintained throughout the entire paradigm. A further argument against the early monophthongisation of *-oi̯ > PT *-ẹ is that one would not expect to find any trace of this vowel in Tocharian A. Indeed, Proto-Tocharian vowels *-a, *-o, *-e were all lost in pre-Tocharian A. As a consequence, when Tocharian A shows a final vowel, this vowel can only be the outcome of a contraction over two syllables or it can be traced to final diphthongs that were maintained down to Proto-Tocharian. In light of the above, I am therefore sceptical of an early monophthongisation of *-oi̯ > *-ẹ in general, and of reconstructing a palatalising value for this alleged monophthongised new vowel in particular.85

83 If derived from the nominative plural kokalyi ‘wagons’, the noun TB kokalyiśke* ‘little wagon’, attested once in B352 a2, would be very irregular, because Tocharian derivatives based on plural stems select nouns with suffixed plurals (with nom. = obl., like säswaśkañ ~ säsuśkañ ‘dear sons’ from säsuwa, pl. of soy, ‘son’). Furthermore, the nominative form is not used as the stem of a noun. One may therefore wonder whether this kokalyi is actually the dual of kokale ‘wagon’, with regular nom. = obl. 84 Oscillations between -li- /lǝy/ and -lyi- /ĺǝy/ are frequently attested: TB lyipär ‘remainder, residue’ (e.g. B119 b3 lyīprä; B99 b2 lyīpär; IT187 a5 lyipär) vs. lipär (AS15C a1 l(i)pär; B44 b6 līpä(r); THT1579 a3 liprä); añcali ‘gesture of palms together (← Skt. añjali-)’ (e.g. IT97 a3) vs. añcalyi (AS13J b1; B602.b b4); meli ‘nose, nostrils’ (B527 a5; IT491 a2) vs. melyi (IT306 a2); loc.sg. āline ‘in the palm of the hand’ (IT803 b2; AS19.6 b4; THT1107 b4) vs. ālyine (AS16.2 b4; B567 a1 and a2); loc.sg. śoline ‘in the hearth’ (e.g. IT4 b4; B153 a2; AS19.3 b3) vs. śolyine (IT4 b3). 85 Cf. also the nom.pl.m. TB alyaik ‘others’, where, according to Ringe, the addition of the emphatic particle PT *-kǝ must have been added after the supposed sound change *-oi̯ > *-ẹ. It is more convenient to say that PIE *ali̯-oi̯ regularly evolved into pre-PT *alley (or *aĺĺey, with analogical palatalisation) and then the diphthong PT *ey yielded TB ai because it was protected by the newly added *-kǝ in Proto-Tocharian.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

320

Chapter 4

The history of the accusative plural is slightly more complicated, especially from the point of view of Tocharian A. Indeed, while the obl.pl. TB -eṃ unambiguously continues PT *-ens < PIE *-ons, the obl.pl. TA -es is historically less clear. If we consider the equation TB -eṃ : ta -es in the adjectives, one would be tempted to include the obl.pl. TA -es among the list of environments where vowel raising before the inherited cluster *ns occurred. This view is shared by e.g. van Brock (1971), Adams (1988a: 116), Hilmarsson (1987c: 69–70; cf. also 1986: 342), Kim (2012), but there exists direct and indirect evidence that calls the hypothesis into question. First of all, among the phonological developments of Tocharian, the evolution of the inherited cluster *ns is quite a peculiar one (Winter 1961). Indeed, the unconditioned outcome is TB -nts-, ta -is- as corroborated by unambiguous examples:86 TB āntse, ta es (< pre-TA *aise) ‘shoulder’ < PT *anse < PIE *ōmso(?) (cf. Gk. ὦμος, gew 2.1148); gen.sg. e.g. TB -entse, ta -es (< pre-TA *-eise) < PT *-ense (Jasanoff 2019: 48–52); TB klǝntsa-, ta kläysā- (< pre-TA *kləisa-) ‘to sleep’ < PT *klənsa- < PIE *ḱlei̯- ‘to rely on’ (Malzahn 2010: 625); cf. also TA wlāys-, tb lans- ‘carry out’ (cf. also the noun TA wles, tb lāṃs ‘work, service’),87 TA eṣäk, tb eṃṣke ‘while’, and the perl.pl. TB -ntsa < pre-TB *-n-sa. This outcome is more clearly attested in word-internal position, since there is no evidence that PT *-ns yielded TB -nts, ta -is word-finally. Indeed, the correspondence obl.pl. TB -eṃ : ta -es is never found in the inflection of the noun, where TB -eṃ is consistently matched by TA -as.88 Another important piece of paradigmatic evidence is that we find the obl.pl. TA -es only in those (adjectival) paradigms that have nom.pl. -e < PT *-ey < PIE *-oi̯, while we find obl.pl. TA -as only in those (nominal) paradigms that have nom.pl. -añ (old PIE *o-stems, e.g. nom.pl. yukañ, obl.pl. yukas from TA yuk ‘horse’ < PIE *h1éḱu̯ o-). It goes without saying that analogical levellings have taken place in one of the two plural sets. If vowel raising of PT *ns > pre-TA *is only occurred in internal position, we should assume that the unconditioned development of PT *-ens (< PIE *-ons) was TA -as (Pinault 2008: 458), and that the vocalism of TA -es was taken over from the nominative plural. A further piece of evidence in favour of this reconstruction is that the continuants of the PIE athematic type have an obl.pl. TA -äs < pre-TA *-əns (cf. TA mañäs, tb meñäṃ ‘moons’ < PT *meñǝns; TA konäs, tb kaunäṃ ‘suns’ < PT *kawnǝns; TA lāñcäs, tb lāntäṃ ‘kings’ < PT *lantǝns; TA 86 87 88

See Huard (2020: 219–223). For the spelling variants of TB lans- and similar verbs, see Mazahn (2010: 749 and 833). The only exception is the obl.pl. pracres of TA pracar ‘brother’, where the “thematic” plural paradigm nom. -e, obl. -es cannot be original (Peyrot 2008: 114).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

321

poñcäs, tb pontäṃ ‘all’ < PT *pontǝns; cf. also TA tos, tb toṃ ‘these.f’ < PT *tons, etc.) and not the **-is < pre-TA *-əins we would expect if raising took place (cf. TA waṣtiṃ ‘related to the house’ : TB ostaññe; gen.sg. TA -is : TB -äntse /-əntse/, -antse /-ə́ ntse/; TA kläysā- ‘to sleep’ : TB kləntsa-). Possible counterexamples could be the gen.pl. TB -ṃts, ta -is and TB weṃts, ta wes ‘excrement, urine’. However, the former had a final shwa in ProtoTocharian, as the spelling -ṃtsä and -ṃtso (with o-mobile) in poetic and/or archaic passages of Tocharian B clearly show (cf. e.g. krentäṃtsä in B17 b6 and krentaṃtso in B416 a3; onolmeṃntsä Or 8212.163 b6 and onolmentso in IT183 b1, see Malzahn 2012a). As far as TB weṃts and TA wes are concerned, both words are only rarely attested: in Tocharian A, we find nom.sg. wes in A124b4 and gen.sg. wesis (< pre-TA *waisəise?) in A150 b6; in Tocharian B, the nom.obl.sg. weṃts is always found together with its derivative weṃṣiye ‘excrement, urine’ (B42 b6; B522 a4; B524 a8; THT4122 b4), while the perl.sg. weṃtsa is attested three times (AS3A b4; B497 b4; W2 a5). Its etymology is unknown, but Adams (dtb 662) traces it back to PT *wen(ə)sə.89 Be that as it may, I do not think that TB weṃts, ta wes is a good example for claiming that PT *-ns also yields TB -nts, ta -is word-finally. Therefore, in the adjectival paradigm of Tocharian A the following developments can be outlined: PT nom.pl. *-ey, obl.pl. *-ens > pre-TA nom.pl. *-e, obl.pl. *-as >> TA nom.pl. -e, obl.pl. -es. A related problem concerns why Tocharian A does not show any continuant of the nom.pl. *-ey in the noun inflection (apart from TA nom.pl. pracre, obl.pl. pracres, where the nom.pl. -e is unexpected). I see two possibilities which may explain this state of affairs. The first is that Tocharian A replaced the nom.pl. *-e with the productive nasal plural -a-ñ because TA *-e came to be homophonous with a relatively large and heterogeneous group of nouns (tg § 82; teb §§ 88, 102, 105), which has TA -e as a singular marker (nom. = obl.). Alternatively, Tocharian A may have maintained a more archaic state of affairs, and the spread of the nom.pl. *-oi̯ may have developed as follows: pronouns → adjectival pronouns → thematic adjectives → thematic nouns. If so, in Proto-Tocharian, this development had not yet reached the nouns, but only the adjectives, and Tocharian A attests the older distribution. After the breakup of Proto-Tocharian, the Tocharian B continuant of the PIE thematic nouns replaced the inherited nominative plural with -i < PT *-ey, while Tocharian A developed -añ, adding the productive nom.pl. -ñ to the stem final vowel -a < PT *-e. Unfortunately, there is no proof in support of either one of these theories. From a comparative point of view, the former is probably to be preferred, 89

See Huard (2020: 222).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

322

Chapter 4

because several Indo-European languages replaced the original nom.pl. *-o-es > *-ōs with the pronominal *-oi̯ from an early stage, and, to my knowledge, we have no continuant of a nom.pl. *-ōs in Tocharian. To sum up, the evolution of the masculine plural paradigm in the adjectival thematic inflection can be schematised as follows: Table 73

Evolution of the adjectival masculine plural from PIE to Tocharian

PIE nom. sg. acc. sg.

*-ōs *-ons

>>*-oi̯ >*-ons

PT

TB

TA

> *-ey > *-ens

> -i > eṃ

> -e > *-as >> -es

4.8.2 Evolution of the Feminine Paradigm and the Neuter Plural The historical analysis of the Tocharian feminine poses several problems. Some of these problems may be relevant to the reconstruction of the PIE gender system, since they revolve around the status of Tocharian with respect to the branching of the Indo-European tree and the evolution of the gender markers within Proto-Indo-European. As outlined above, the Tocharian singular paradigm of the feminine is peculiar, since it shows palatalisation or assibilation of the stem-final consonant in the outcomes of both thematic and athematic adjectival types. This is unexpected from a comparative perspective. Indeed, the ancient Indo-European languages, especially Greek and Indo-Iranian, indicate that the potentially ́ palatalising suffix *-ih2/-i̯éh2 of the devī-type was originally specialised in athematic adjectives, like nt-stems, s-stems, u-stems, etc. On the other hand, the feminine-marking suffix *-eh2 > *-ā was confined to the thematic type. Thus we find feminine forms such as Skt. návā-, Gk. νέᾱ, Lat. nova, OCS nova ‘new’ < *néu̯ eh2- for the respective thematic masculine adjectives Skt. náva-, Gk. νέος, Lat. novus, OCS novъ < *néu̯ o- ‘new, young’, but feminine forms such as Skt. svādvī-,́ Gk. ἡδεῖα < *su̯ eh2d(e)u-ih2 for athematic masculine adjectives Skt. svādú-, Gk. ἡδύς < *su̯ eh2dú- ‘sweet’ (Kim 2014: 121).90 90 See Ringe (2006: 313–315) on Germanic relics of *-ih2 in athematic adjectives. We further recognise, however, that we do find some instances of feminine in *-ih2 derived ́ from thematic bases, such as *dei̯u̯o- (Skt. devá-) ‘celestial, god’ → *dei̯u̯ih2 (Skt. devī-) ‘celestial (f.), goddess’ or Goth. þius ‘servant’ → Goth. þiwi ‘bondwoman, maid’. However, ́ devī‑inflection in the feminine of vr̥ddhi and secondary adjectives has been assumed to be an Indo-Iranian innovation (Rau 2007: 62–63). The use of *-ih2 to form feminine

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

323

The following table shows the contrast between (Proto-)Tocharian and some other Indo-European languages in the outcomes of the nom.sg. of PIE adjectives in *-ro- (Fellner 2014b: 65): Table 74

Evolution of thematic adjectives in some Indo-European languages

nom.sg.

PIE

post-PIE

Gk.

Skt.

Lat.

OCS

PT

masc. fem.

*-ros > *-reh2 >

*-ros > *-rā >

-ρος -ρᾱ

-raḥ -rā

-rus -ra

-rъ -ra

*-re *-rya

While Greek, Latin, Sanskrit, and Old Church Slavonic have the regular outcome of *-reh2, no continuant of the same ending can be reconstructed for Proto-Tocharian, since this would be expected to have yielded PT *-ro > TB **-ro, ta **-r, without -y- (see § 4.7.1.1). This mismatch between Tocharian and the other Indo-European languages has given rise to debate. As summarised by Fellner (2014b: 67), two mutually exclusive theories have been proposed, both aiming to explain the evolution of the feminine:91 ́ (1) Tocharian inherited the devī-suffix as the only standardised feminine marker in the adjectival inflection; ́ (2) Tocharian analogically extended the outcome of the devī-suffix from the athematic to the thematic type. The first theory indirectly revisits the development of the feminine gender within Proto-Indo-European. It implies that Tocharian preserves a more archaic status than the other Indo-European languages (with the exception of the Anatolian branch), according to which *-eh2 was not completely grammaticalised as a feminine marker when Tocharian was separated from the proto-language. It follows that the gender system might provide new evidence for the phylogenetic position of Tocharian as the second branch that split counterparts of thematic masculine nouns only shows that *-ih2 was a feminine motion suffix, i.e. a derivational morpheme (Kim 2014: 118–119). 91 Hartmann (2013) does not deal with this central problem of the Tocharian gender system. According to him, the peculiar distribution of the outcomes of *-ih2 and *-eh2 deserves an explanation (pp. 35–38), but “[o]b die angenommene Zweitausgliederung des Tocharischen von ihren Vertretern nun ausreichend begründet ist oder nicht, sei dahingestellt” (p. 530). See further Pinault (2015b: 189–192).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

324

Chapter 4

off from Proto-Indo-European, after the earlier departure of Anatolian. Kim (2009a, 2014) was the first to propose this theory, which received some scholarly consensus (cf. Hackstein 2012, 2013: 94–95; Kortlandt 2017, 2020).92 On the other hand, the second theory implies that, like the other non-Anatolian Indo-European languages, Tocharian inherited *-ih2/-i̯eh2- (of ́ the devī-type) as a feminine athematic suffix and its spread to the thematic type must be regarded as a secondary development (Pinault 2008, 2012a; Fellner 2014a, 2014b). In what follows, I will discuss a number of arguments supporting the second view. 4.8.2.1 Theories on the Origin of the Feminine in Tocharian Let us introduce Kim’s theory in more detail, highlighting the results of his investigation and outlining the consequences from a comparative perspective. Kim developed his idea in two separate articles, published five years apart (Kim 2009a and 2014).93 Considering that the first article presents the theory in an embryonic way, while the second article covers the matter more extensively and reviews a few shortcomings, they will be presented jointly. Kim’s central idea is based on the assumption that the element *-ya in the feminine inflection of the thematic adjectives is to be taken as an archaism in Tocharian. In support of this claim, he offers a brief revision of the gender system of Anatolian, concluding that the *eh2-stems were continued only as an inflectional class and that the PIE suffixes *ih2 and *(e)h2 had no feminine value in Anatolian (Kim 2009a: 70–72). It follows that, at an older stage of Proto-Indo-European, they did not serve as gender-marking suffixes, but they had other functional values. According to Kim, the former had an original “possessive-instantive” function (i.e. referring to an instance of an action or state), while the latter was mostly employed to mark collective formations, 92 Cf. Hackstein (2012: 167): “In contrast to other branches of Indo-European, […] Tocharian is peculiar in preserving a second stage, which precedes the functional extension of the collective-abstract to denote natural and grammatical feminine gender. At this intermediate stage, we observe the incipient association with male and female referents of those collective-abstract formants that are firmly associated with feminine grammatical gender in most other Indo-European branches, namely *-ih2 and *-eh2”. According to Hackstein (2012: 167–168), the use of collectives in PIE *-h2 to form agent nouns is an indication that the collective suffix had only begun its development into a feminine when Tocharian split off from Proto-Indo-European. As pointed out by Friis (2021: 10), this development relies on a supposed development of collectives and abstracts to agentive nouns in Tocharian. This claim, however, is debatable. See the criticism of Fellner (2014c: 60–62) and Friis (2021: 10–11). 93 An overview can also be found in Kim (2018a: 83–85).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

325

individual and abstract nouns, and had an endocentric function. It is generally assumed that the feminine value of these suffixes must have been a secondary development that took place in the proto-language only after the departure of the Anatolian branch (Rieken 2005; Melchert 2014). Kim’s proposal is that the relative chronology of this development implies that *ih2 was grammaticalised earlier than *(e)h2 as a feminine motion suffix and that the strongest evidence for this reconstruction comes precisely from Tocharian. Accordingly, the fact that the continuants of the thematic adjectives are marked in the feminine by *-ya < *-ih2 and that “the reflex of PIE eh2-stems had no particular association with feminine referents, but were simply another [Tocharian] inflectional class” (Kim 2009a: 81) is a strong indicator of this internal development. As a consequence, the common ancestor of both Tocharian and the so-called “Brugmannian languages” grammaticalised *ih2 as the feminine marker of both nouns and adjectives. However, this suffix could not be attached to the demonstratives or to primary adjectives because they were not derived from nouns and “made use of the suffix *-h2 in its endocentric sense” (Kim 2014: 127). Therefore, an important difference between the “Brugmannian languages” and Tocharian was a differentiation in the marking of the feminine gender between primary and secondary adjectives: the former took *(e)h2 and the latter took *ih2. Only after the split of Tocharian, so-called “Core Indo-European” grammaticalised the opposition between *eh2 and *ih2 as one between thematic and athematic types. This new contrast was favoured by the demonstrative pronouns, which regularly took *(e)h2 (thus *séh2).94 Though this theory is fascinating and innovative, I believe that it contains flaws, particularly from the perspective of Tocharian phonology and morphology. First of all, it is not fully falsifiable. On the one hand, there is no evidence in favour of any previous grammaticalisation of *ih2 in Anatolian as an adjectival feminine marking suffix, nor is there any against it. On the other hand, the other Indo-European languages attest a well-established opposition between thematic *-eh2- vs. athematic *-ih2-. Only Tocharian serves as evidence for this reconstruction, which cannot be supported comparatively. To be more precise, I agree with Kim (2014: 118–119) that, in the Indo-European languages, “the derivation of nouns specifically denoting females generally involves *-ih2- rather than *-(e)h2-”. Indeed, traces of the early grammaticalisation of *-ih2- as a 94

According to Kortlandt (2017: 101), a feminine *sih2 was created before the rise of *séh2. On the centrality of the demonstrative pronoun in the rise of the feminine gender, see Meillet (1931) and Martinet (1956). See also Luraghi (2011) and Pinault (2011b) for a recent overview of the deictic origin of the feminine.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

326

Chapter 4

feminine motion suffix can already be identified in Anatolian. As Kloekhorst (2019: 223–230) recently showed, the Kanišite Hittite suffix -e occurring in a few women’s names ending in -ašue of the type šakriaḫšu (m.) : šakriašue ~ šikriašue (f.), šupiašu (m.) : šupiašue (f.), and watniaḫšu (m.) : watniašue (f.) can be traced back to *-ih2.95 As Kloekhorst demonstrates, the element -e in these formations was not used as an agreement marker but rather as a derivational morpheme, on a par with -šar in f. -ḫšušar- vs. m. -ḫšu- ‘birth, offspring’. This implies that Anatolian possessed two derivational suffixes specifically used to build derivatives denoting female entities, namely -šar and -e: the former was used in both nouns and names, while *-ih2 occurred in names only. Therefore, the evidence from Kanišite Hittite confirms that the outcome of *-ih2 was already used to build feminine names from masculine names that were morphologically athematic: there is no trace of the suffix *-ih2 as an agreement marker, either in the thematic or in the athematic inflection of adjectives. Kim’s theory also presents difficulties from the perspective of Tocharian phonology. If the feminine continuants of the PIE *ro-adjectives formally go back to *-rih2 > PT *-rya (in the singular), the reconstruction of a feminine suffix *-ih2 could not account for the feminine form of certain other adjectival derivatives. Let us consider, for instance, the case of the ordinals in *-to-, whose nominative singular feminine ends in TB -ca, ta -ci. This form cannot be historically analysed as the outcome of *-tih2 > *-tya, since this would be expected to yield TB **-tsa, i.e. with assibilation of the dental stop rather than with palatalisation. Similar considerations can be put forward for adjectives in TB -tte, ta -t, nom.sg.f. TB -cca, ta -ci < PT *-cca (not *-tsa), and gerundives in TB -lle, ta -l, nom.sg.f. TB -lya, ta -lyi (not *-lla).96 This evidence strongly speaks in favour of a secondary generalisation of the pattern *-[+pal.]a, which was abstracted from the outcome of the athematic feminine, rather than representing a direct preservation of *-ih2 as an inherited suffix in the thematic inflection (see § 4.8.2.2, § 4.8.2.3).97 95 This derivation was already suggested by Goetze (1960). For the development PIE *-ih2 > Hitt. -ē,̆ Kloekhorst (2008: 91, 2019: 224) points to the nom.acc.pl.nt. kē < PIE *ḱih2, the nom.acc.pl.nt. 2-e ‘two’ < *du̯ ih2, and the nom.acc.pl.nt. 3-e ‘three’ < *trih2. 96 See Peyrot (2013b: 223–224) for the outcomes of the PIE sequences *li̯, *li, and *le. 97 The status of the adjectives in TB -tstse, ta -ts is slightly more complicated, since no palatalisation can be reconstructed in the paradigm of the feminine. As a matter of fact, no clear paradigmatic alternation between palatalised and non-palatalised -ts- is synchronically attested, especially not in Tocharian B. If such a contrast really existed, it was therefore already levelled out in Proto-Tocharian. Another possibility is that the feminine of the adjectives in TB -tstse, ta -ts was created on the model of the assibilated feminine PT *-ntsa < PIE *-nt-ih2 (class iii).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

327

From a morphological point of view, we must question the claim that the primary adjectives took *-eh2, while the secondary adjectives took *-ih2. Indeed, some scholars claim that adjectives did not constitute an independent derivational category in Proto-Indo-European. For instance, in Vedic only a handful of non-derived adjectives can be identified, and it cannot be excluded that these synchronically primary adjectives are derived from unattested verbal roots (Alfieri 2009, 2016, 2018). In any case, whenever we reconstruct adjectival roots for Proto-Indo-European, they are simply too limited in number to favour the generalisation of *eh2 in the thematic type. Furthermore, there exist some inherited adjectival forms in Tocharian that unambiguously show the expected outcome of the PIE *eh2-inflection. Out of the demonstratives, we find relics in the obl.sg. allok ‘other’, pl. alloṅk (see § 4.4), in the obl.sg. somo ‘one’, pl. somo-, and perhaps in the adverb TB wato ‘again’, which may be a frozen feminine form of wate ‘second’ (cf. Skt. dvitā ‘twofold; certainly, indeed’, dtb 626; Fellner 2014b: 68 fn. 9). As far as the Tocharian continuants of PIE *ali̯o- ‘other’ and *du̯ ito- ‘second’ are concerned, Kim’s opinion is not altogether clear. On the one hand, he advocates the reconstruction of a feminine paradigm with *-ih2/-i̯eh2- for *ali̯o- ‘other’, which, he suggests, would have produced TB allok in the oblique and TB alyāk in the nominative (Kim 2009a: 78–79, 2014: 122 fn. 18).98 On the other hand, he states that the aforementioned PIE *ali̯o- ‘other’ and *du̯ itó- ‘second’ could have maintained *-eh2 in the feminine inflection of primary adjectives as “possible relics” (Kim 2014: 127). Of these two analyses, only the latter can be accepted, because the stem allomorph alyā- is clearly secondary (see  §  4.4), and a reconstructed acc.sg. *ali-i̯eh2-m (with the full grade of the suffix taken from the weak cases, Kim 2009a: 79) would probably not have yielded obl.sg. TB allo-. Another weakness of Kim’s theory concerns, in my view, the evolution of the feminine plural paradigm and the morphophonological mergers between the feminine and the neuter in Tocharian. In his recent article, he modifies his previous view according to which “[…] in all clear cases without exception, feminine thematic adjectives also exhibit a suffix which can only continue PIE *-ih2!” (Kim 2009a: 76, emphasis by the author). This view is criticised by Pinault (2012a: 190–191). Indeed, in subclass i.1 we find the plural TB -o-na, ta -a-ṃ (without palatalisation of the preceding consonant), where the correspondence TB -o- : ta -a- can only be the outcome of a reconstructed form that must have contained PIE *-eh2- > PT *-o-. In order to account for this problem, Kim (2014: 122) traces the vowel *-o- back to the PIE neuter plural *-e-h2 in his later article (cf. also Winter 1962a: 126–127; Marggraf 1975: 200–201; 98

See also Fellner (2014a: 13 fn. 20) and § 4.4 above.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

328

Chapter 4

Hackstein 2017).99 Although this reconstruction poses no problems from a formal point of view, there are some issues related to the diachrony of the merger between the feminine and the neuter. Indeed, if the neuter plural was *-eh2 in the thematic inflection and *-h2 in the athematic inflection, while the feminine was only marked by the suffix *-ih2 in both inflectional types, there would not have been any formal context where the feminine and the neuter could have merged morphophonologically, either in the singular, or in the plural.100 Therefore, the reanalysis of the neuter *-o-na as a feminine marker would have had no basis.101 Kortlandt (2017, 2020) proposes a similar reconstruction, though differing from Kim’s account in several details and notably in its theoretical approach. Indeed, while Kim’s and my reconstruction of the diachronic evolution of Tocharian is much closer to the traditional reconstruction of Proto-Indo-European, Kortlandt’s adheres to the Proto-Indo-Uralic hypothesis. Kortlandt (2017: 100, 2020: 123) suggests that the split between Tocharian and the other non-Anatolian Indo-European languages preceded the creation of the feminine paradigm of thematic adjectives and that Tocharian reflects an intermediate stage at which *-ih2 was generalised to the singular inflection of the adjectives. According to Kortlandt (2020: 121) the formations in *-ih2 that provided the feminine of some Tocharian adjectives were singularia tantum, and their plural was supplied by the *-eh2- of the h2-stems. Kortlandt (2017: 99 Kim’s opinion on the Tocharian outcome of PIE *-eh2 in word-final position is not clear (cf. also Kim 2018a: 105–107). If TB -o-, ta -a- in TB -ona, ta -aṃ is from the thematic neuter plural, then PIE *-eh2 must have yielded PT *-o, because the spread of the ending *-na must have occurred after most of the Proto-Tocharian vowel modifications took place. Nonetheless, in the same article (2014: 122 fn. 16; cf. also 2009: 80), he seems to suggest an outcome PT *-a, since “[…] the evidence for the Tocharian treatment of PIE *-eh2 is effectively reduced to *seh2 (> PT *sa) and the neuter plural”. By “neuter plural”, he is not referring to TB -ona, ta -aṃ, but to those plural markers ending in -a, which are characteristic of athematic adjectival classes and alternating nouns, where, according to Kim, the final vowel reflects either *-h2 or *-eh2 (as per Ringe 1996: 94–97; contra Pinault 2008: 491–497). On the outcome of word-final *-eh2, see the next paragraph below. 100 Kim (2009a: 74) maintains that masculine and neuter fell together in the singular of the thematic inflection, but he gives no explanation for the generalisation of “the originally neuter plural nom./acc. ending *-(e)h2 […] to the feminine”. 101 One might wonder whether the merger of the feminine with the neuter originated in the athematic inflection, where the distinction between feminine (*-ih2(-) > *-ya(-)) and neuter plural (*-h2 > *-a) consisted only in the palatalisation/assibilation of the stem in the feminine. However, I believe that this reconstructed quasi-homophony is too meagre to justify the merger. Kim (2018a: 83–84) reconstructs a mixed paradigm for pre-Proto-Tocharian: the singular and the dual would continue PIE *-ih2/-i̯eh2- (of the ́ devī-type), while the plural would continue PIE *-eh2-. I cannot agree with this reconstruction, which is ad hoc.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

329

100) argues that the generalisation of *-ih2 as a distinct feminine marker would have been a more logical solution than the introduction of the predicative ending *h2, which was also found as a neuter plural ending and would have rendered the agreement rules more complex. I consider the reconstruction of a mixture of the ih2-inflection in the singular contrasted with the (e)h2-inflection in the plural a contrived projection of problematic Tocharian forms onto the reconstructum. I see no way to reconstruct this paradigm either for the Indo-European proto-language or for any older stages of Tocharian.102 In addition, Kortlandt’s explanations (2020: 121–123) of the various adjectival formations of Tocharian may also entail formal problems in their derivations. In particular, he only deals with the feminine formations when he lists the examples of the alleged semantic difference between *-h2 and *-ih2, but the argumentation is sometimes confusing as one must explain the palatalisation in both the masculine and the feminine in Tocharian adjectival suffixes. Contrary to Kortlandt (2020: 121), the synchronic inflection of the gerundives is not the same as TB allek, TA ālak, particularly in the feminine inflection. Another shortcoming of Kortlandt’s account is that he only deals with the development of the feminine inflection and does not consider the evolution of the gender marking as a whole. As a consequence, I fail to understand how the alternating gender would have come about if the feminine was formally differentiated from the neuter in the first place. Indeed, the rise of the genus alternans has not been considered by Kortlandt. For the reasons given above, both Kim’s and Kortlandt’s distribution of *-ih2 in the Tocharian thematic type as an inherited feature are, in my view, problematic. We should rather follow the second view, according to which Tocharian inherited a more classical Indo-European three-gender system, where the feminine was marked by *-eh2 in the thematic adjectives. In accordance with previous theories on this topic, I will show that the drastic modifications in the feminine adjectival inflection of Tocharian are innovations that took place gradually in the long period of linguistic development that separates Proto-Indo-European from Proto-Tocharian and then Proto-Tocharian from Tocharian A and B. This makes no claim about the alleged early split of Tocharian: the evolution of the feminine gender in the adjectival system

102 Similarly, Hackstein (2012: 151–152) suggests that Tocharian adjectives show suppletion of the feminine markers with *-ih2 in the singular but *-eh2 in the plural. According to him, a possible explanation for this semi-thematic inflection is that the emergence of *-ih2 and *-eh2 as feminine-marking suffixes represents an incipient innovation in Tocharian. See further below.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

330

Chapter 4

cannot serve as proof of the so-called “Indo-Tocharian”103 hypothesis, because the spread of *ih2 in Tocharian is an innovation. Nonetheless, the second hypothesis is not without problems either. Each of these problems can be framed as independent working questions, which have led me through my investigation of the evolution of the Tocharian feminine. They can be summarised as follows: (1) How did the non-ablauting *eh2-type evolve in Tocharian? (2) How and why was the outcome of the *ih2-type generalised in the thematic inflection? (3) Why did the feminine plural continue the neuter plural in the athematic inflection? (4) Why is there a contrast between palatalised singular vs. nonpalatalised plural in subclass i.1, and how did class i.1 and class i.2 become differentiated in Proto-Tocharian? Below, we will deal with these problems in the order outlined above. 4.8.2.2 Evolution of the Non-Ablauting *eh2-Inflection in Adjectives In the previous sections and chapters, we have variously dealt with phonological and morphological problems related to the outcome of the PIE *eh2-inflection in Tocharian, acknowledging that its evolution has given rise to major disagreement. Having considered evidence from the nominal and pronominal inflection, it is now time to discuss more extensively how the non-ablauting *eh2-inflection evolved in Tocharian. Van Windekens (1976: 24–25) and Adams (1988a: 20–21, 1998: 615–616) maintain that the unconditioned outcome of PIE *eh2 was PT *a. However, the majority of scholars currently agree that this phonological development should be modified to TB o, ta o, a.104 Nonetheless, the development of *-eh2 in word-final position is still a matter of debate: (1) some scholars suggest

103 See Peyrot (2019a) and Olander (2019). 104 Adams is virtually alone in still adhering to a sound change *eh2 > PT *a. On the other hand, Winter (1981: 935–941) was the first to suggest a development PIE *eh2 > PT *o. A counterexample that is sometimes adduced is TB mācer /mácer/ ‘mother’ > PIE *meh2tēr, instead of the expected **mocer (cf. Skt. mātár-, Av. mātar-, Gk. μήτηρ, Lat. māter), but an analogical a from TB pācer /pácer/ ‘father’ can be assumed in order to explain the unexpected vowel in mācer (Marggraf 1975). On the twofold outcomes of Tocharian A, see Burlak & Itkin (2003).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

331

PIE *-eh2 > PT *-a > TB -a;105 (2) other scholars maintain PIE *-eh2 > PT *-o > TB -o.106 With regard to the *eh2-inflection, it goes without saying that the main point of debate is the outcome of the nominative singular, which is the only case-form where we can reconstruct word-final *-eh2. I side with those scholars who claim that the regular development of *eh2 > *ā was PT *o in all positions. Indeed, the adduced forms where *-eh2 allegedly yielded PT *-a by sound law are not probative, since most of them require alternative explanations. The relevant forms are: (1) feminine thematic adjectives with nom.sg. ending TB -a, like -ñña, -ṣṣa, etc. (Ringe 1996: 94; Hajnal 2005; Malzahn 2011: 89); (2) the Motionsfemininum TB -a in e.g. oṅkolma ‘she-elephant’ or mañiya ‘maid-servant’ (Ringe 1996: 94); (3) the productive alternating plural TB -a (Adams 1988a: 32; Ringe 1996: 31; Kim 2014: 122 fn.16); (4) the pronominal nom.sg.f. TB sā, ta sā- < PIE *séh2 (Ringe 1996: 94; Jay Jasanoff apud Ringe 1996: 96–97 n. 1); (5) substantives with nom.sg. in TB -a of the wertsiya-type (Adams dtb s.v.; Malzahn 2011: 89); (6) the nom.sg.f. alyā-k from allek ‘other’ (Malzahn 2011: 97); (7) the nom.sg.f. TB ñuwa ‘new’ (Hackstein 2012; Fellner 2014a: 14; Kim 2014). Starting with the data from the noun, we have already explained the bulk of the substantives of the wertsiya-type (evidence 5) as reflecting formations of either ́ ́ the devī-type or the vr̥kī-type (§ 3.5.3). In these nouns, the final sequence -ya is to be interpreted as reflecting *-ih2. On the other hand, the regular outcome of a nom.sg. *-eh2 > TB -o in the noun inflection is found in several other types, such as the kantwo-, okso-, arṣāklo-, and palsko-types (see the relevant sections in § 3.5 and § 3.6.2.1). There is no need to reconstruct a sigmatic nom.sg. *-ās 105 See e.g. Peters (1990), Ringe (94–95), Kim (2009a, 2014), Malzahn (2011), and Hackstein (2012). Cf. Ringe (1996: 96): “If post-PIE word-final *ā developed into PT a by regular sound changes alone, the crucial change was probably a shortening of *-ā to *-a, since inherited short *a underwent no changes before the PT period”. Kim (2018a: 84) gives pre-PT *-a as the outcome of PIE *-eh2 but he does not further discuss why this ending would have merged with pre-PT *-o arising in the “thematic” feminine plural. Also, in the chapter on the Auslautgesetze Kim does not deal with the outcome of PIE *-Vh2 at length, since he claims that *-eh2 “has left relatively few unambiguous traces in Tocharian” (Kim 2018a: 105). 106 See e.g. Hilmarsson (1986), Pinault (2008), and Fellner (2014a, 2014b).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

332

Chapter 4

to explain TB -o:107 in my view, both *-ās and *-ā would have evolved into PT *-o in any case. There is also no evidence for claiming that the feminine suffix TB -a (evidence 2) of the oṅkolma-type (cf. oṅkolma ‘she-elephant’ vs. oṅkolmo ‘elephant’) is the outcome of PIE *-eh2, since TB -a could have been abstracted from the adjectival inflection at any stage of Tocharian B. Indeed, these feminine nouns follow the inflection of the aśiya-type, which follow the paradigm of the adjectives (see § 3.4.2, cf. the plural mañ(i)yana from mañiya ‘maid-servant’ ← mañiye ‘male servant’). As far as the alternating plural ending TB -a is concerned (evidence 3), there is no comparative evidence to trace it back to the thematic nt.pl. *-eh2. Indeed, in the noun inflection it is consistently found as the outcome of athematic neuter formations, whose nt.pl. is reconstructed as PIE *-h2 > *-a (see Pinault 2008: 491–497). As for the pronominal nom.sg.f. TAB sā < PIE *séh2 (evidence 4), we have already seen that the isolated outcome TAB -a of PIE *-eh2 may have had multiple sources (see § 4.3.2). Turning now to the adjectival inflection, Malzahn (2011: 89) hints at “a large number of feminines to thematic adjectives […] that one would want to derive from non-ablauting PIE *eh2-stems, which show a nom.sg. ending in TB -a and not in TB -o”. Though she does not mention what these formations are, she is in all likelihood referring to those adjectival derivatives from class i.2 that show phonological palatalisation as a structural characteristic of the suffix, i.e. m. -ññe, f. -ñña, m. -ṣṣe, f. -ṣṣa, etc. (evidence 1). In my view, this explanation is invalidated by other outcomes of thematic derivatives that display palatalisation only in the feminine (e.g. m. -re, f. -rya, m. -lle, f. -lya, m. -tte, f. -cca etc.). That is to say, the feminine singular forms of these thematic formations are all formed through a secondary addition of the pattern *-[+ pal.]a, which applied variously to the adjectival derivatives, depending on the basic structure of the suffix. Those adjectival suffixes that were not already palatalised took “explicit”, i.e. visible, palatalisation in the feminine, while those adjectival suffixes that were already palatalised took “implicit”, i.e. invisible, palatalisation (because the suffix could not be further palatalised). Similar considerations can be made to account for the mismatching stem in nom.sg.f. alyāk vs. obl.sg.f. allok (evidence 6), where the contrast -ly- vs. -ll- speaks in favour of a secondary palatalisation of the former form (§ 4.4). On the other hand, the pattern *-[+pal.]a surfaced as *-ya when the consonant preceding the suffix did not have a palatalised counterpart (cf. nom.sg.f. TB -rya, ta -ri of the adjectives in TB -re, ta -r). 107 Pace Kim (2009a: 80); similarly, Peters (1990: 243) and Malzahn (2011). See § 3.5.1.2.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

333

Hackstein (2012: 153) adduces one further instance where PIE final *-eh2 allegedly yielded PT *-a, i.e. TB ñuwa*, ta ñwi* ‘new’ (evidence 7).108 The problem here is the lack of palatalisation, because, according to Fellner, an analogical nom.sg.f. TB **ñuwya or **ñuyya would have been expected. However, I do not think this is a problem, since TA w cannot be palatalised and in Tocharian B synchronic alternations between w and y are limited to the causatives. In all other cases, alternations between y and w were levelled, and y was no longer felt to be the palatalised counterpart of the w-allomorph (cf. with levelling of the y-allomorph e.g. TB śay- ‘to live’ < *śay- ~ *śaw; cf. also Kim 2018a: 66). Finally, there exists another cogent grammatical argument that may indirectly prove the evolution PIE *-eh2 > PT *-o. As per Fellner (2014a), this evolution must be postulated for the prehistory of Tocharian. Indeed, the source of the Tocharian alternating gender and the neuter origin of some Tocharian feminine plural endings and forms can only be attributed to the morphophonological mergers of the feminine with the neuter plural (see below). If PIE *-eh2 yielded PT *-a, no cases of homophony between feminine and neuter could be reconstructed, since the thematic neuter plural would only have merged phonologically with the nominative singular of the feminine. This conflation would not have been sufficient to account for the formal merger of the two genders. I therefore agree with Pinault (2008) and Fellner (2014a) that the evolution of the feminine singular and the neuter plural in the thematic inflection was as follows: Table 75

Evolution of the feminine singular and the neuter plural in the thematic inflection

*eh2-declension

thematic neuter

nom.sg. acc.sg.

pie *-eh2 *-eh2-m

PT > *-o > *-o

nom.pl. acc.pl.

pie *-eh2 *-eh2

PT > *-o > *-o

108 Cf. Kim (2014: 32). Fellner (2014a: 14) also points to this form, albeit with some hesitation. Note that Fellner’s nom.sg.f. TA †ñwa (2014: 13) is not attested and is morphophonologically impossible since final -a does not occur in Tocharian A. We would not even expect the more regular TA †ñwā as the morphological correspondent form of TB ñuwa*; a form TA *ñwi would be expected instead (cf. Michaël Peyrot apud Kortlandt 2017: 100 fn. 4).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

334

Chapter 4

As can be seen, a merger of the neuter plural with (at least) the feminine singular can be reconstructed.109 This situation strongly resembles the historical evolution of the gender system from Latin to Romance. In fact, a typological comparison between Tocharian and Romance languages (particularly Romanian) has often been made (see, for instance, Ringe 1996: 97; Igartua 2006; Kim 2009a: 73–74; Fellner 2014a: 15–16). As a matter of fact, systems with a third gender value that combines alternating agreement traits of the masculine and the feminine between the singular and the plural are cross-linguistically uncommon, especially in the Indo-European domain. Within this typological comparison, however, an important diachronic fact has thus far been overlooked. Although it is true that the masculine and the neuter must have merged in the singular, the rise of the Romanian genus alternans is not due to a merger of the neuter and the feminine in the plural! Such a merger cannot have occurred, because the nt.pl. ended in *-a (< Lat. -a), while the f.pl. ended in *-e (< Lat. nom. -ae and/or acc. -ās, if it developed through *-ay as per Faraoni 2016: 392). In fact, the Romanian genus alternans originated more gradually. See the following schema from Loporcaro (2018: 223; see further pp. 219–239 and Loporcaro 2016): Table 76

m nt f

Transition of the gender system from Latin to Romanian

i. classical latin

ii. transition

sg

sg

 -us -um  -a

pl I III II

-i -a -ae

>

m f

-u -a

iii. romanian

pl  I III II

-i -a -e

sg

>

m f

-∅ -ă

pl  I III II

-i -e

Loporcaro claims that in a transitional phase between Classical Latin and Romanian, the third gender value (old neuter) experienced a double optional agreement set in the plural (f.pl. and nt.pl.). In this stage, the neuter displayed full syncretism with the masculine in the singular, “[…] with optional preservation of the contrast in the plural, where dedicated agreement targets persisted alongside the innovative option, that is, feminine plural agreement […]” (Loporcaro, loc. cit.). Comparative evidence from Old Italian, and other (Western) Romance languages and dialects confirms this reconstruction 109 See further § 3.6.2.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

335

Gender in the Inflection of the Pronoun and Adjective

(cf. Old Italian ill-a brachia ‘those arms’ vs. ill-e brachia ‘id.’ in the Codice Diplo­ matico Longobardo; see Loporcaro, Faraoni & Gardani 2014 and Loporcaro & Paciaroni 2011).110 On the strength of this diachronic comparison, we must consider whether the rise of the Tocharian genus alternans began with the merger between the masculine singular and the neuter singular and between the neuter plural and the feminine singular. A possible scheme of this development is given below: Table 77

Morphophonological mergers between the masculine, feminine, and neuter

pie

nom. acc.

PT

masc.sg.

nt.sg.

*-o-s *-o-m

*-o-m *-o-m

>*-e >*-e

pie

nom. acc.

PT

fem.sg.

nt.pl.

*-eh2 *-eh2-m

*-eh2 *-eh2

> *-o > *-o

Nonetheless, a problem is posed by the evolution of the feminine plural paradigm. While, on the one hand, the nom.pl. *-eh2-es is expected to have evolved into *-ās > PT *-o, the evolution of the acc.pl. *-eh2-ns is more complex from both an Indo-European and a Tocharian comparative perspective. Indeed, the reconstruction of this case form for Proto-Indo-European is not clear. A summary of the various reconstructions can be found in Olander (2015: 246– 247). In the following, I briefly review the data from the main Indo-European branches: (1) Ved. -āḥ and OAv. -ā ̊ point to IIr. *-ās (contra e.g. Kuryłowicz 1927: 222–223); (2) Attic-Ionic Gk. -ᾱς is ambiguous (cf. also Lesbian -αις), but Cretan Gk. -ανς clearly suggests *-āns (with Osthoff’s Law);

110 See also Paciaroni, Nolè & Loporcaro (2013), and Maiden (2011: 172–173; 2016: 12–13). As Faraoni (2016: 383–384) states: “[I]l toscano antico, e con esso le tante varietà centrome­ ridionali antiche e moderne analogamente analizzabili, possedeva un sistema a tre generi. Certo, […] tale sistema non era in tutto e per tutto simile a quello del latino, dove anche i sostantivi neutri, al pari di quelli maschili e femminili, disponevano di un paradigma di accordo specifico, con marche dedicate e non sincretiche come accade per il neutro alternante rumeno e italo-romanzo”.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

336

Chapter 4

(3) Lat. -ās is ambiguous, since it may go back either to *-āns (with loss of the nasal, cf. -ōs < *-ons, see Ernout 1945: 25) or *-ās (Weiss 2020: 253); (4) Umbr. -ass, Osc. -af may directly result from *-āns, with the change of word-final *-ns > Umbr. -ss, Osc. -f (Pisani 1964: 12); (5) Goth. -os suggests PGerm. -ōz < *-ās, but, according to Boutkan (1995: 141–142), it may also reflect PGerm. -ōns < *-āns (cf. the doublets nom.acc.pl. OE -e ~ -a and cf. further Guus Kroonen apud Olander 2015: 248); (6) the evidence from Balto-Slavic is notoriously difficult: in Baltic, Litv. def. adj. -ą́ sias and Old Prussian -ans point to *-āns, while Latv. -as, and Lith. -as point to -ās; in Slavic, OCS -y, -je̢ is from *-(i̯)āns (see Vaillant 1958: 83–4, Olander 2015: 248, Kortlandt 2016, and Kim 2019a with references therein). As one can see, the Indo-European comparative evidence is quite tricky, as some languages point to *-ās, while others point to *-āns. That is to say, was *-eh2ns/*-eh2m̥ s reduced to *-ās in the proto-language (i.e. IE languages pointing to *-n- restored the nasal) or was *-eh2ns maintained (i.e. IE languages without *-n- have independently lost the nasal)? The reconstruction is further complicated by the effect of the so-called “extended” Stang’s Law, i.e. a PIE sequence of a vowel followed by a semivowel (or a laryngeal) and a nasal is simplified word-finally with loss of the semivowel (or the laryngeal) and compensatory lengthening of the preceding vowel, thus *-VHN > *-V̄ N (Stang 1965). Stang’s Law has given rise to debate, especially with regard to the *eh2-inflection.111 The Tocharian data are equally ambiguous. In the adjectival inflection we do not find any clear continuant of a nasal variant *-eh2ns, as the plural inflection of the feminine paradigm was completely remodelled. However, we have seen that in the pronominal inflection the match obl.pl.f. TB toṃ : ta tos < PT *tons clearly suggests a reconstruction of *-eh2ns (cf. also TB alloṅk < *allons(-), see Hackstein 2017: 1313). Various explanations for these inconsistencies are conceivable. These largely depend on which reconstruction of the accusative plural of the *eh2-stem one favours. The first hypothesis is the least probable: the reconstruction of different accusative plural forms of the thematic *eh2-stems in adjectives and pronouns. On the one hand, pronouns should have taken *-eh2-ns, while, on the 111 The bibliography on Stang’s Law is abundant. See e.g. Vaux (2002), De Decker (2011), Pronk (2016), and Kortlandt (2017), with references.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

337

other hand, adjectives should have taken *-eh2-s. This hypothesis is linked to the late development of the feminine gender within the proto-language: when the new feminine agreement environment started to develop in late Proto-Indo-European, the feminine adjectival inflection was marked in the plural (nom. = acc.) by *-eh2- (originally the neuter plural) + the plural marker *-s. This hypothesis is totally ad hoc. The second hypothesis requires the more likely reconstruction of a uniform plural paradigm for both pronouns and adjectives. The paradigm was nom.pl. *-eh2-es and acc.pl. *-eh2-n̥ s (< **-eh2-m̥ s)112 in the older stages of PIE. The accusative plural then underwent Stang’s Law, yielding *-āns and then *-ās in the proto-language (as per AiGr, but cf. also Rix 1986; Weiss 2020: 253; De Decker 2011). As a consequence, those Indo-European languages that point to the nasal would have reintroduced it analogically after other stems in which the nasal was retained (as per e.g. Kim 2019a). As far as Tocharian is concerned, this implies that the pronominal obl.pl.f. forms TB toṃ, ta tos and TB alloṅk(-) reintroduced the nasal (perhaps after the masculine) at a later stage. I am personally reluctant to support this hypothesis, since I believe that the pronominal form of the obl.pl.f. is better explained as an inherited archaism (see § 4.3.4). Indeed, these forms represent a morphological irregularity and they are likely the result of regular sound change. The third hypothesis does not require Stang’s Law in the *eh2-inflection: the acc.pl. *-eh2-n̥ s may or may not have resulted in *-āns in the proto-language, but it retained the nasal in both cases (as per Beekes 2011: 200). It follows that those Indo-European languages that do not point to the nasal lost it independently.113 There are then two different working hypotheses for Tocharian: the outcome of *-āns was continued in pre-Proto-Tocharian or developed into *-ās at an older stage. If the former holds, then the expected Proto-Tocharian outcome would have been *-ons (just as PIE *-ons > PT *-ens). The reason this ending disappeared in favour of PT *-o-na is not immediately clear, but one can certainly envision a morphological replacement. Indeed, at a pre-Proto-Tocharian stage the feminine paradigm of the thematic inflection should have been marked by *-o, with the possible exception of the accusative plural. This of course caused the merger between the feminine and the neuter (nt.pl. PIE *-eh2- > PT *-o). After the formal merger of the two genders, the new remarked neuter ending 112 Hittite suggests the reconstruction of an older acc.pl. *-ms, cf. Hitt. -uš < *-ms and *-oms (Meier-Brügger 2003: 163; Kloekhorst 2008: 928–929; Beekes 2011: 188; Kim 2012). 113 Cf. also Martínez & de Vaan (2014: 58): “One thinks of different dialectal (or already IE?) treatments of *-eh2-ns: in one group, the nasal was lost in this sequence, while in the other group, it was maintained (or restored?)”.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

338

Chapter 4

*-o-na was generalised to the feminine. The redeployment of the neuter plural endings as feminine took place when gender marking was restructured in Proto-Tocharian leaving only a contrast between masculine and feminine in the inflection of the targets. Though in a different framework, this hypothesis has been supported by Kim (2014) and Hackstein (2017), who both take TB *-ona as *-o- (collective) with additional plural marker *-na. Similar considerations have been put forward by Winter (1962b: 26–27) and Marggraf (1975: 200). On the other hand, if *-āns yielded *-ās before Proto-Tocharian, the nasal may have been lost phonologically. In particular, it may be tentatively suggested that the inherited sequence *-V̄ ns underwent two different changes depending on the prosodic environment: in non-accented position *-V̄ ns > *-V̄ ns > *-V̄ s; in accented position *-V̄ ns > *-V̄ ́ ns > *-V̄ ́ns.114 This would explain why in the pronominal inflection the cluster -ns was maintained in the acc.pl. *tā́ns > PT *tóns > TB toṃ, ta tos, while it was lost in the adjectives. Although this explanation poses no relevant problems from a phonetic perspective, it is equally difficult to test. Indeed, it is hard to find other inherited sequences of *-V̄ ́ns in word-final position that may prove the genuineness of this sound law.115 Since I take the reconstructable obl.pl.f. PT *-ons in the pronominal inflection as an archaism, I believe that Tocharian inherited the acc.pl. of the *eh2-stems as *-āns, and that this ending was lost in the adjectival inflection 114 If so, one may wonder whether the nasal was retained as nasalisation of the preceding long vowel in a transitional stage. See Hilmarsson (1991a: 197–199) for this possibility. 115 Hilmarsson (1984a) claims that the nom.sg.m. of the numeral for ‘1’, TB ṣe, ta sas, continues PIE *sḗms > *sḗns. However, I agree with Pinault (2006a) that Gk. εἷς ‘1’ does not point to such a proto-form: the long vowel of the Greek form is best explained starting with an original nom.sg.m. *sem-s > *sens, which lost the nasal in Greek, with compensatory lengthening of the vowel (cf. Gort. εν[δ] δ- from ἕνς δ-, see gew 1.471; Beekes 2010: 394). The vocalism of the Tocharian forms cannot therefore mirror *-ē-, but rather originated by analogical levelling with the rest of the paradigm, which is built on the thematic stem *seme- < *somo- (Ved. samá-, OP hama-, Gk. ὁμός, Goth. sama, etc.). The feminine form TB sana, ta säṃ testifies that the nasal in the masculine survived for a while. Indeed, it can hardly mirror PIE *smih2 directly (cf. nom.sg.f. Gk. μία), and the internal n was likely introduced from the nom.sg.m (Pinault 2006a: 89–90). The expected palatalisation caused by *-ih2 may have been lost when the palatalised *ḿ was replaced by the non-palatalised *n. To my knowledge, there is no evidence for Fellner’s evolution *smih2 > *sm̥ ya > *sənya (2014b: 66 fn. 6). On the other hand, an example from Tocharian historical morphology that may support the reconstruction of *-V̄ ns > *-V̄ s is the development of the sequences acc.pl. *-ōn-n̥ s vs. *-on-n̥ s in the nasal inflection. Indeed, the former sequence evolved *-ōns > TB -aṃ, and the latter *-on-n̥ s > TB -enäṃ (e.g. in the nouns of the saswe-type, if not of recent origin [see Pinault 2008: 477–478], and in the adjectives of the tapre-type). Cf. also obl.pl. śrānäṃ ‘elders’ as if from PT *śəranəns < *keră-n-n̥ s < PIE *ǵerh2-n-n̥ s (Georges-Jean Pinault apud Carling 2003: 93 fn. 47).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

339

either morphologically (replaced by the neuter *-o-na) or phonologically (reduction of *-āns > *-ās in non-accented syllable), but it survived in the pronominal inflection. Note that Lindner (UrIG  § 1b. p. 2; § 23f. p. 138) supports a reconstruction of an early reduction of PIE *-eh2ms > -ās for the *eh2-stems contrasted with the maintenance of PIE *téh2ms > tā́ns in the demonstratives. As a consequence, for a pre-Proto-Tocharian stage, the neuter can be reconstructed as having no distinct singular marker, since it merged with the masculine singular, while the feminine did not have either a transparent singular or a transparent plural: on the one hand, the singular merged with the neuter plural; on the other hand, the plural (partially?) merged with its own singular and with the neuter plural. As a consequence, neither feminine nor neuter had unambiguous paradigms in either the singular or the plural. However, the fact that the feminine plural was remarked with neuter endings may indicate that the feminine was the grammatical gender less clearly differentiated morphologically. At this stage, function could have played a role in the reassignment of both case and gender markers. The development which led to the reassignment of the gender values in Tocharian must have begun with formal mergers, but, after the merger of the gender markers, function may have favoured the spread of the endings and forms of the historical neuter to the feminine plural. This led to a new paradigmatic differentiation between the singular and the plural within the paradigm of the feminine. Of the research questions listed at the beginning of this section, I have discussed the phonological evolution of the eh2-inflection (1). We can now move on to the secondary spread of *-ih2 in the Proto-Tocharian continuant of the feminine thematic paradigm (2). 4.8.2.3

Evolution of the Ablauting *ih2-inflection in the Adjectives and its Spread to the Feminine Thematic Type ́ Now that it has become clear that the generalisation of the devī-type in the (singular) thematic inflection must be regarded as a Tocharian innovation, we must clarify how it evolved in Tocharian and what type of internal change caused its spread. Fellner (2014a, 2014b) has recently dealt with the latter topic. He recurred to non-proportional analogy in order to explain the spread of *ih2. He suggests that this analogical development was favoured by a derivational mechanism that is quite common in Indo-Aryan, where this suffix was often used to form the feminine of secondary thematic adjectives, including vr̥ddhi formations. The starting point of this evolution would have been the opposition between PIE *dei̯u̯-o- ‘god’ (Lat. deus ‘god, deity’, dīvus ‘godlike’, Ved. devá-, Av. daēuua-, etc.) and *dei̯u̯-ih2 ‘goddess’ (Ved. devī-,́ Gk. δῖα-), both derived

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

340

Chapter 4

from PIE *di̯eu̯ -/ *diu̯ - ‘sky, heaven’. According to Fellner, “Pre-Proto-Tocharian speakers” reworked the relation between these two isolated words and generalised the pattern of *dei̯u̯-o- : *dei̯u̯-ih2 to the whole adjectival system, abstracting the element *-ih2. This analogical change would first have affected other vr̥ddhi formations before spreading throughout the entire thematic inflection, such that “the extension of the pattern to thematic adjectives in Pre-Proto-Tocharian finally eliminated almost all traces of old *-eh2 feminine adjectives, thus giving rise to the attested situation” (Fellner 2014a: 11). Though I agree with Fellner in the basic assumption that Tocharian did not inherit a different gender-marking system to that of the other Indo-European languages, I do not find his explanation totally convincing. Despite the fact ́ that a similar phenomenon took place in Indo-Iranian, where the devī-type with vr̥ddhi became the model of several derivatives that often formed the feminine with the outcome of *-ih2, I do not see any evidence for claiming that a similar development took place in Tocharian. The core of this analogical development would be based on the hypothetical opposition between *dei̯u̯-oand *dei̯u̯-ih2, but this reconstruction is doubtful because, in my opinion, it is too meagre a basis to explain the spread of *ih2. Furthermore, the continuants of these two Indo-European words are not attested in Tocharian (as Fellner acknowledges), where vr̥ddhi formations are, moreover, not productive. One must therefore agree with Kim (2014: 123) that “they would not […] amount to a sufficient basis for generalization of *-ih2- as the feminine suffix”. An alternative way to account for the spread of *ih2 must therefore be sought. I essentially agree with Pinault (2008: 516–519) that the generalisation ́ of the devī-type to the thematic declension was a very scattered development that was caused by the interplay of both phonological and morphological factors.116 Parallels from Romance languages suggest that this development may well have proceeded in a gradual manner. The basic principle is that sound changes caused irregularities, i.e. mergers and intransparencies, and that analogical developments took place to solve them. Therefore, I believe that the generalisation of *ih2 was caused by two types of analogical development: (1) analogical levelling favouring the isomorphism of endings; (2) non-proportional analogy resolving opaque morphological markers. 116 Cf. also Kim (2009a: 74): “[T]he morphological merger of genders was a relatively recent development”. Although I agree with this claim, the Tocharian A feminine endings nom. -āñ, acc. -ās of the adjectives in -(ṣ)i are, in my view, a Tocharian A innovation (see § 4.7.1.4).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

341

Let us first discuss how the athematic type in *-ih2 evolved in Proto-Tocharian. ́ Comparative evidence allows us to reconstruct the devī-type as characterised by paradigmatic ablaut: the allomorph *-ih2- was characteristic of the strong stem, and *-i̯eh2- of the weak stem. Nonetheless, no direct continuant of the allomorph *-i̯eh2- > *-yo- can be reconstructed on the basis of the Tocharian data. It may have been continued in the plural, where, however, it was mostly replaced by neuter forms (see e.g. class iii pl.f. TB ponta, ta pont and TB krenta, ta krant < *-nt-h2, and class ii.1 f.pl. TB arkwina < *-n-h2).117 A different replacement occurred in the klyomo-type (class ii.5), where the f.pl. TB klyomñana (cf. TA klyominā-) consisted of the singular stem (PT *klyoməñña- < *ḱleu̯ mn-ih2-), which was recharacterised by the nasal neuter plural *-na. The generalisation of historical neuter plural forms was caused by the morphophonological merger of the neuter and the feminine in the thematic inflection (on which see the previous paragraph). Therefore, old plural markers and plural forms of the neuter were exapted to convey the morphological meaning of the feminine. The reassignment of a new functional value to the neuter forms took place in parallel with the disappearance of the neuter as a category value. That is to say, with the disappearance of the neuter as a grammatical category of controller gender, neuter forms moved towards a more central, but still adjacent, area of the morphology.118 The exact relative chronology of these replacements is difficult to determinate, but indirect evidence that an allomorph *-yo- (< *-i̯eh2-) might have survived for a certain period in the plural could be adduced (see below). We first turn to the spread of *ih2 in the thematic inflection. Kim (2009a: 77) is essentially right in saying that the athematic adjectives are less productive than the thematic ones, so that analogical developments from the athematic type would have been implausible. However, among the thematic adjectives, the so-called “secondary derivatives” are more common and productive in Tocharian, i.e. thematic adjectives with etymological palatalisation of the 117 Historical forms of the neuter plural are mostly preserved when the feminine is assibilated (i.e. in old *nt-stems). Peyrot (2010b: 76–78) proposes that the feminine of the nt-stem *-ntsa may have been reanalysed as *-nt-sa in late Proto-Tocharian. If so, one may assume that, in the plural, this *-nt-sa was homophonous with the f.sg., and that the isolated plural marker *-sa was replaced by *-a, giving the attested *-nt-a as a result. Otherwise, if we reconstruct a recharacterised f.pl. *-ntsa-nta (parallel to *-ñña-na of the n-stems), it may have been reduced to *-nta by haplology. A third possibility is that the singular paradigm of the feminine became homophonous with its own plural, both resulting in *-ntsa, and that the plural was marked by the original nt.pl. *-nta in order to resolve these mergers. 118 For the relation between grammaticalisation and exaptation, and the reanalysis of Latin neuter plural ending -a as a collective when gender marking was restructured in the Romance languages, see Giacalone Ramat (1998).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

342

Chapter 4

suffix (formed with PIE *-ii̯o-/-i̯o-). These adjectives synchronically correspond to class i.2. In my opinion, the generalisation of the athematic feminine *-ih2 was favoured by a progressive convergence of the feminine inflection of these thematic derivatives with that of the athematic type, thanks to the common palatalisation of the stem-final consonant. Similar considerations were proposed by Pinault: “Il est vraisemblable aussi que l’extension du féminin de type devī-́ fut favorisée par le fait que la plupart des suffixes d’adjectifs thématiques comportaient déjà l’élément yod au masculin, d’où résultait ensuite la palata­ lisation” (Pinault 2008: 516–517).119 This development took place when, in the athematic inflection, a contrast between *-Ća- (< *-C-ih2-) in the singular, and *-Ćo- (< *-C-i̯eh2-) in the plural, still existed. As a matter of fact, the formal difference between thematic derivatives of class i.2 and athematic adjectives was only found in the singular paradigm, which was marked by *-Ćo- (< *-Ci̯-eh2-) in the thematic type, and *-Ća- (< *-C-ih2) in the athematic type. As a consequence, the inherited opposition between thematic and athematic feminines was gradually blurred, such that the thematic derivatives of class i.2 started to replace thematic *-Ćowith athematic *-Ća- in the singular. The feminine therefore evolved according to the following analogical proportion: athematic thematic Ć Ć sg. *- a- : pl. *- o- = sg. *-Ćx- : pl. *-Ćox = *-a *-mni̯ă > PT *-məñña :: nom.sg. PIE *-n(i)i̯eh2 > *-ni̯ā > *-ñño >> PT *-ñña. This development had an important morphological advantage since it disambiguated the feminine singular from the plural inflection of the feminine and the neuter. Once the result of this analogical process had been fixed, the pattern *-[+pal.]a was reanalysed, abstracted, and then generalised to the remaining thematic adjectives, which synchronically belong to class i.1 (e.g. nom.sg.f. rtar-ya, but nom.sg.m. ratre ‘red’ < PIE *h1rudhro-). Then, the plural paradigm was replaced by the neuter plural of nasal stems PT *-na. This recharacterisation affected the plural paradigm of the adjectives of the entire class i and the adjectives of class ii (old n-stems, cf. TB klyomñana). 119 See also Pinault (2012a: 190–191).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

343

The overall development that led to the secondary generalisation of the athematic declension to the feminine singular can be explained as a case of what Lazzeroni (1992) calls synergetic drift (“deriva sinergica”). Lazzeroni explains the concept of synergetic drift as a set of changes oriented towards a reorganisation of linguistic traits based on new parameters that govern them. He exemplifies this development through the reorganisation of grammatical gender and inflectional classes in Sanskrit. Comparing Vedic Sanskrit with Classical Sanskrit data, he notes that the feminine i- and u-stems gradually adhered to the ī- and ū-declension respectively, while the masculine īand ū-stems underwent a symmetrical development, becoming i- and u-stems. As a consequence, in the history of Old Indo-Aryan, vowel quantity became a morphological marker of gender opposition: the masculine took short vowels, and the feminine long vowels. This development started in a pre-documentary stage at the stage during which the Indo-European opposition between masculine *o-stems and feminine *ā-stems vanished in Indo-Iranian and was replaced by a quantity-conditioned distinction between masculine a-stems (< *o-stems) and feminine ā-stems. When the opposition between short -a(-) and long -ā(-) was reanalysed as an opposition between masculine and feminine gender, the morphologisation of vowel quantity of the stem was extended to the other declensions. As far as Tocharian is concerned, the drift was triggered by regular phonological developments. In pre-Proto-Tocharian, the Proto-Indo-European opposition between the feminine *eh2-inflection of the derivatives in *-(i)i̯o- and the feminine *ih2-inflection was conveyed by the difference of the quality of the vowels in *-[+pal.]o- vs. *-[+pal.]a- in the singular. At this stage, the thematic feminine was poorly marked in both the singular and the plural, because some endings and forms were morphophonologically merging with the neuter plural of the thematic declension. In an attempt to solve these confluences, the vowel *-a- was generalised in place of the inherited *-o- (< *-eh2-) of the thematic inflection. The pattern *-[+pal.]a- was then gradually grammaticalised as a morphological marker of the feminine singular and spread to the rest of the thematic type, reaching those thematic derivatives where palatalisation was not etymological. 4.8.2.4 Origin of the Split of class i The last point that needs to be discussed is how the differentiation within class i originated in Proto-Tocharian. After all the phonological and morphological modifications outlined above, the feminine paradigm of the thematic adjectives should have had the following endings:

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

344 Table 78

nom. obl.

Chapter 4 Feminine paradigm in Proto-Tocharian class i.1

singular

plural

*-[+pal.]a *-[+pal.]a

*-ona *-ona

This reconstructed paradigm evolved without relevant modifications in subclass i.1, which retained a contrast between palatalised singular with vowel TB -a-, ta -ā- < PT *-a- vs. non-palatalised plural with vowel TB -o-, ta -a- < PT *-o-. However, those adjectives with etymological palatalisation of the suffix, which had a palatalised stem even before the plural ending, also generalised the singular pattern *-[+pal.]a- of the singular to the plural. This generalisation led to the creation of a different subclass. Table 79

nom. obl.

Evolution of the feminine paradigm in Proto-Tocharian class i.2

singular

plural

*-Ća *-Ća

*-Ćona >> *-Ćana *-Ćona >> *-Ćana

Thus, we can divide the Proto-Tocharian continuants of the Proto-IndoEuropean thematic adjectives into two groups: (1) PT adjectives with no etymological palatalisation of the suffix; (2) PT adjectives with etymological palatalisation of the suffix. These two groups differed in the paradigm of the feminine plural: both had pl. *-na (nom. = obl.), but in the former this ending was preceded by *-o- and no palatalisation of the suffix (thus *-ona), while in the latter it was preceded by *-a- with palatalisation of the suffix (thus *-yana). Palatalisation must therefore have played a central role in the split of the two classes. In essence, I propose that the original plural ending was *-ona. This marker was already accompanied by etymological palatalisation in the second group of derivatives (continuing the PIE type in *-(i)i̯o-). When the ending *-ya was generalised in the feminine paradigm of the singular, the vowel *-a- was levelled to the plural paradigm of the adjectives from the second group. In this way, *-yana replaced *-yona (class i.2), while, in the first group, *-ona was retained.120 120 An indirect confirmation of this change may come from the gerundives in TB -lle. We have seen that the feminine plural attests a transitional stage: the original non-palatalised Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Gender in the Inflection of the Pronoun and Adjective

345

4.8.3 Summary of the Evolution of the Gender System in the Adjectives After having recounted the most important theories on the origin of the Tocharian gender system and their importance from a comparative perspective, I have discussed the relevant modifications that the gender system underwent. The comparison between Tocharian and Romance languages suggests that the evolution of the gender system may have been a gradual development during the course of which the masculine, the feminine, and the neuter mutually influenced each other morphologically, before being fixed in the attested agreement system. While the masculine evolved without relevant modifications from Proto-Indo-European to Tocharian, the feminine underwent a nuḿ ber of characteristic changes, since it generalised the outcome of the devī-type in the singular, and it developed endings and inflectional forms from the neuter in the plural. The principle of this heterogeneous set of developments is recounted below. Regular phonological change caused cases of homophony within the paradigm of the feminine and formal mergers with the neuter plural. Indeed, in the continuant of the PIE non-ablauting *eh2-inflection, the feminine was not marked either in the singular or in the plural: on the one hand, the singular merged with the neuter plural; on the other hand, the plural partially merged with its own singular and with the neuter plural. In order to re-mark a distinction between the singular and the plural, the feminine assumed plural endings from the neuter inflection, while the singular was influenced by the athematic ́ inflection of the devī-type. The loss of the neuter was a gradual development as is shown by the fact that the morphology of the neuter gender was refunctionalised as feminine in the plural. The reassignment of a new gender value to neuter forms took place in parallel with the gradual disappearance of the neuter as a grammatical category and with the rise of the alternating gender. In a non-attested stage, variants and fluctuations in the paradigms can be reconstructed. Parallels from the oldest phases of the Romance languages show that even when the alternating agreement pattern becomes more established, the neuter as a gender value does not automatically cease to exist but competes for a while with the conservative agreement. During this stage, the functional reanalysis and

plural -llona was replaced by the palatalised TB -lyana in late texts (Pinault 2008: 519; cf. Peyrot 2008: 118: “it is striking that the new pl.f. -ana was introduced together with palatalisation”). We have also seen that the morphological contrast between palatalised vs. non-palatalised case endings was being lost in the historical development of the gerundives in Tocharian B, since they started to shift from class i.1 to class i.2. Within this diachronic drift, the hypothetical plural **-[+pal.]ona must have been felt to be ungrammatical, because the plural -ona always occurs with non-palatalised stems. Thus, a new plural -[+pal.]ana (not **-[+pal.]ona) was introduced analogically. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

346

Chapter 4

exaptation of the endings and forms of a moribund category are likely to have taken place. Among the continuants of the feminine thematic type, those adjectives with etymological palatalisation of the suffix substituted *-yo (< PIE *-(i)i̯eh2-) with *-ya- (< PIE *-ih2-). This process was caused by two complementary developments: levellings of case and gender markers, and non-proportional analogy to resolve opaque morphological markers. Once this process was completed, the pattern *-[+pal.]a- was abstracted as a morphological marker of the feminine singular and it spread to the rest of the thematic type. It mostly surfaced as *-ya when the consonant preceding the suffix did not have a palatalised counterpart. This new opposition between singular stem *-[+pal.]a- and old plural stem *-[-pal]o- was retained in those derived adjectives whose suffix was not etymologically palatalising; on the other hand, those derivatives with etymological palatalisation of the suffix also generalised the vowel *-a- in the plural. The late Proto-Tocharian paradigm of the feminine in class i can be schematised as follows: class i.1 PT f.sg. *-[+pal.]a- vs. f.pl. *-[-pal.]o-na class i.2 PT f.sg. *-[+pal.]a- vs. f.pl. *-[+pal.]a-na. After the breakup of Proto-Tocharian, the two Tocharian languages independently re-marked the oblique singular. The Proto-Tocharian gen.sg. *-ay was reanalysed as the new oblique in Tocharian B, while, in Tocharian A, it continued to serve as a genitive. As a general tendency of Tocharian A, the obl.sg. marker *-n was generalised in the feminine before the Tocharian A apocope of final vowels took place, and the obl.sg.f. became pre-TA *-ān. Then, vowel apocope took place and in class i.2 some markers became homophonous again: indeed, the f.pl. *-ana was apocopated to *-ān and it coalesced with the new obl.sg. In an attempt to resolve these mergers, a new distinction between nominative and oblique plural was introduced, and the ubiquitous endings nom.pl. -ñ, obl.pl. -s were added. To conclude, the peculiarities of the feminine in the adjectival inflection of Tocharian are best explained as the outcome of internal developments that took place within the evolution of the language.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Chapter 5

Retrospective and Conclusion This study has dealt with the grammatical gender system of Tocharian, its synchronic description, and its diachronic evolution. The main findings are recapitulated below. 5.1

Synchronic Analysis

The main questions posed in Chapter 2 were whether the genus alternans is a real gender value and, consequently, how many genders Tocharian has. After an introduction to the linguistic typology of grammatical gender and the Canonical Typology elaborated by G. G. Corbett (§  2.1), the synchronic analysis of Tocharian gender was discussed (§ 2.2). In § 2.3, the mechanism of synchronic gender assignment was examined by individuating inflectional, derivational, and semantic strategies for predicting the gender of Tocharian nouns. Finally, a short description of the reconstructed Proto-Indo-European category of grammatical gender was offered (§ 2.4) On the basis of typological and cross-linguistic comparisons with Romance languages in general, and Standard Italian and Romanian in particular, it has been demonstrated that Tocharian has three gender values, including the genus alternans (§ 2.2.1). The reasons behind this analysis are recounted below: (1) alternating nouns have specific agreement patterns which are different from those of the masculine and the feminine; (2) alternating nouns belong to individual inflectional classes whose peculiarity is that they have no formal distinction between nominative and oblique in either the singular or the plural; (3) alternating nouns form a productive group: loanwords and inherited words are inserted into this class of nouns; (4) alternating nouns are exclusively inanimate; (5) the agreement of an alternating noun provides for a nominal concord where a controller agrees with a target inflected as masculine in the singular but as feminine in the plural. This agreement set is respected even when coordinated noun phrases headed by alternating nouns agree with nominal modifiers or pronouns. Indeed, they regularly select feminine plural agreement.

© Alessandro Del Tomba, 2023 | doi:10.1163/9789004532892_006 Alessandro

Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

348

Chapter 5

Thus the genus alternans is a systematic and productive category of Tocharian grammar. Applying the principles and criteria of Canonical Typology, we are therefore forced to analyse the Tocharian controller and target differently: just as in the masculine and the feminine, so too do the alternating nouns tend to belong to individual inflectional classes, which predict the gender of a noun. Conversely, it is the target that has no dedicated and unique forms for the genus alternans and this leads to a mismatch between controller and target genders. The genus alternans pertains to the domain of the controller gender and can therefore be regarded as a non-canonical gender value. Nevertheless, it is fully embedded in the grammar of the language. Finally,  §  2.2.2 discussed terminological problems associated with the name of the third Tocharian gender. It was shown that the label “alternating”, as opposed to “neuter”, is to be preferred. Indeed, although alternating nouns historically derived from PIE neuters, the Tocharian genus alternans is the outcome of formal and functional mergers between the three inherited genders. In Tocharian, the term “neuter” is more appropriately used for a relic class of forms limited to pronouns and ordinal numerals. Although these relics derive from the PIE neuter, they do not comprise a distinct gender from a functional-synchronic perspective, since they do not share any exponent in any agreeing word class: they are used for non-gender reference and should therefore be regarded as “neutral” forms. 5.2

Diachrony of the Masculine

The evolution of the PIE masculine in Tocharian is not complex. Nouns reconstructed as masculine for the proto-language are generally continued as masculine in Tocharian, in both the thematic and athematic types. In the singular inflection, the opposition between nom.sg. *-s, acc.sg. *-m was lost, caused by the Proto-Tocharian apocope of final consonants. This produced the formal merger of the nominative and the oblique (< PIE accusative) in the singular. In the thematic type, they coalesced in PT *-e < PIE *-o- (§ 3.6.1.1). In the inflection of the noun, a distinction between nominative and oblique was reintroduced through the addition of PT *-n to the oblique singular. This new obl.sg. *-en (thematic) ~ *-ǝn (athematic) was confined to nouns denoting male human beings (cf. obl.sg. tb eṅkweṃ, ta oṅkaṃ ‘man’; obl.sg. TB lykaṃ /ĺkǝ́ n/, ta lykäṃ ‘thief’). In Tocharian A, the obl.sg. -ṃ /-n/ became more productive and was used for feminine nouns as well. On the other hand, some other types continuing the athematic inflection generalised the original weak

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Retrospective and Conclusion

349

steam to the oblique in order to remark the opposition (cf. TB pācer ‘father’ < PT *pacer < PIE *ph2tēr vs. obl. pātär < PT *patǝr < PIE *ph2tr-; TB maśce ‘fist’ < PT *mǝśce < PIE *mustē(i̯) vs. obl. maśc < PT *mǝścǝ < PIE *musti-; TB yriye /yrǝ́ye/ ‘lamb’ < PT *ẃǝrǝye < PIE *u̯ erh1ēn vs. obl. yari /yǝ́ rǝy/ < PT *ẃǝrǝy < PIE *u̯ erh1en-). Usually, Tocharian A levelled the paradigm with the generalisation of one of the two stems (cf. nom.obl.sg TA pācar ‘father’). In the inflection of the adjective, different strategies were employed. Analogical palatalisation was introduced after the inflection of the demonstratives (§  4.3,  §  4.6). This palatalisation affected the masculine paradigm with the exception of the nominative singular. This development reintroduced the formal opposition between nominative and oblique (cf. e.g. TB allek [ta ālak] ‘other’ vs. obl. alyek [ta ālyak-äṃ]; TB trite [ta trit] ‘third’ vs. obl. trice [ta tricäṃ], TB ayāmätte ‘not done, not able to do’ vs. obl. ayāmäcce, etc.). On the other hand, those adjectival suffixes which could not have any palatalised counterpart took the obl.sg. marker *-n, which, in the adjectival inflection, became a mandatory ending after the Proto-Tocharian period (§ 4.7.1, § 4.8.1; cf. TB -eṃ vs. TA -äṃ [not **-aṃ] in TB astareṃ, ta āṣträṃ ‘pure’). Indeed, in Tocharian A the obl.sg.m. -ṃ was only generalised after the loss of final vowels in Tocharian A (cf. TA obl.sg. tricäṃ ‘third’ [cf. TB trice] < pre-TA *tric < PT *trice; TA gerundival obl.sg. -läṃ [cf. TB -lye] < pre-TA *-l [apocope and depalatalisation] < PT *-(ĺ)ĺ; cf. also TA obl.sg. -i vs. -iṃ in the i-adjectives, TA obl.sg. ñäkci ‘divine’ [cf. TB ñäkc(i)ye] vs. ñäkciṃ). In the plural inflection, the thematic nom.pl. *-oi̯ was clearly continued in adjectives and pronouns. In Proto-Tocharian, it evolved into *-ey, and was then monophthongised in TB -i, ta -e (non-palatalising; see § 4.8.1). In Tocharian B, word-final PT *-ey was maintained only in monosyllables (cf. nom.pl. PIE *toi̯ ‘these’ >> PT *cey > TB cai, cey; du. PIE *toih1 ‘these’ > PT *tey > TB tai). There is no strong evidence for an alleged early monophthongisation of PIE *-oi̯ > (palatalising) PT *-ẹ. In the noun inflection, the nom.pl. -i is the marker of a productive class of Tocharian B e-stems, which can be traced back to the thematic type (§ 3.6.1.1). On the other hand, nom.pl. TA -e is not found in the noun inflection (with the exception of pracre ‘brothers’, where it is unexpected). Indeed, the Tocharian A counterparts of Tocharian B e-stems probably remade the expected nom.pl. *-e with TA -añ, levelling the a-vocalism of the stem (< *-e- < PIE *-o-; cf. obl.pl. -as). In the athematic type, the PIE nom.pl. *-es caused palatalisation of the preceding consonant. When apocope took place in Proto-Tocharian, this palatalised consonant was reanalysed as the nominative plural marker (cf. nom.pl. TB -ñc, ta -ṃś < PT *-ñcǝ < PIE *-nt-es; nom.pl. TAB -ñ < PT *-ñǝ < PIE *-n-es).

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

350

Chapter 5

As far as the development of the thematic oblique plural is concerned, I side with those scholars who claimed PIE *-ons evolved into PT *-ens > TB -eṃ, ta -as without vowel raising in Tocharian A (cf. § 4.8.1). In Tocharian A, the obl.pl. -as is regularly found in the continuants of the PIE o-stems. On the other hand, in the adjectival and pronominal inflection we find obl.pl. TA -es. It has been attempted to explain the e-vocalism in this marker as the result of analogical levelling from the nom.pl. -e. In the athematic type, PIE *-n̥ s regularly evolved into PT *-ǝns > TB -äṃ, ta -äs without vowel raising in Tocharian A. 5.3

Diachrony of the Feminine

Among the issues discussed in the book, the evolution of the feminine gender has been a central subject of study. Indeed, it has recently been claimed that Tocharian inherited a gender system different from that reconstructed for the other Indo-European languages: in this system, the feminine had either not yet risen as a grammatical category or was marked by the suffix *-ih2/-i̯eh2- in both thematic and athematic declension. As a matter of fact, in the thematic inflection of Tocharian several endings seem to continue a declension in PIE *-ih2, which is conversely attested in the athematic inflection in the other Indo-European languages. These theories were scrutinised in § 4.8.2.1, where it was concluded that they cannot account for the Tocharian evidence; therefore, another solution was needed. In  §  4.8.2.3, the generalisation of the *ih2-inflection in place of the non-ablauting *eh2-inflection in the adjectival paradigm was explained as a secondary innovation, internal to Tocharian. This was caused by a gradual and heterogeneous set of developments caused by formal and functional mergers of the feminine within the inherited *eh2-inflection and with the neuter plural in the thematic paradigm (§ 4.8.2.4). At this stage, endings and forms that originally belonged to the neuter paradigm started to shift to the plural paradigm of the feminine, which, synchronically, does not attest any differentiation between nominative and oblique plural (e.g. f.pl. TB -na, ta -ṃ < PT *-na < PIE *-nh2 or f.pl. TB -nta, ta -nt < PT *-nta < PIE *-nth2). In order to solve the merger of some of the case markers, the thematic type took over endings from the athematic type, reintroducing a distinction between singular and plural in the feminine paradigm. The process involved can be interpreted as a “synergetic drift”, i.e. a set of changes aimed at reorganising linguistic traits by means of new parameters (§ 4.8.2.3). The drift began when the Proto-Indo-European opposition between the feminine *eh2-inflection of the thematic derivatives in *-(i)i̯o-, and the

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Retrospective and Conclusion

351

feminine *ih2-inflection of the athematic stems started to be conveyed only by the difference between *-[+pal.]o- vs. *-[+pal.]a- in the singular. In an attempt to solve the cases of homophony mentioned above, the vowel *-a- (< *-h2-) was generalised in place of the inherited *-o- (< *-eh2-). This ending must have become too ambiguous, being used to mark e.g. the nominative singular (< *-eh2), the oblique singular (< *-eh2-m), the nominative plural (< *-eh2-es), as well as the plural of the thematic neuter (< *-eh2). The generalisation of *-afirst affected those thematic formations that continued Proto-Indo-European suffixes derived with *-(i)i̯o-. The pattern *-[+pal.]a was then abstracted as a morphological marker of the feminine singular, and it spread to the rest of the thematic type (§ 4.8.2.3). The opposition between the new singular stem *-[+pal.]a- and old plural stem *-[-pal]o- was retained in those derived adjectives whose suffix was not etymologically palatalising (cf. sg. TB -rya, ta -ri vs. pl. TB -rona, ta -raṃ; sg. TB -lya, ta -lyi vs. pl. TB -llona, ta -laṃ; sg. TB -cca, ta -cci vs. pl. TB -(t)tona, ta -(t)taṃ), but it was lost in the other types formed with *-(i)i̯o-, which also levelled the vowel *-a- in the plural (cf. sg. TB -ṣṣa, ta -ṣi vs. pl. TB -ṣṣana, ta -ṣāñ | -ṣās; sg. TB -ñña, ta -ināṃ (obl.) vs. pl. TB -ññana, ta -ināñ | -inās). The final result of this process was the attested bipartition of the feminine inflection of class i, which continues the PIE thematic type (§  4.8.2.4). The mismatching plural paradigm TB -ana vs. TA -āñ | -ās was explained as a secondary innovation of Tocharian A (§ 4.7.1). The diachronic evolution of the pronominal inflection was investigated in § 4.3 and § 4.4, where it was demonstrated that the majority of the endings of Tocharian demonstratives and pronominal adjectives can be directly traced back to Proto-Indo-European. A main aim of Chapter 3 was to identify the reflexes of the PIE *eh2-type and the PIE *ih2-type in Tocharian nominal morphology. It was argued that Tocharian inherited and generalised a hysterodynamic ablaut paradigm in *-(e)h2 throughout the inflection of the noun. The outcome of this reconstructed paradigm was maintained in the Tocharian B kantwo-type (§  3.5.1), where the PIE opposition between strong and weak stem *-eh2/-h2- produced the contrast between nom.sg. -o (< *-eh2) and obl.sg. -a (< *-h2-). A similar paradigm was also reconstructed for the Proto-Tocharian ancestor of some nouns of the okso-type (ai-stems with nom.sg. -o, obl.sg. -ai), and the arṣāklo-type (a-stems with nom.sg. -o, obl.sg. -ai) (cf. § 3.5.2). The original situation was possibly preserved in Tocharian A, where the three inflectional types mentioned above more clearly correspond to just one type with an ā-stem (< PT *a-stem). Finally, some *(e)h2-stems may have been continued in the palsko-type (cf. § 3.6.2.1), where they were reinterpreted as alternating as a result of the morphophonological merger of some endings of the *(e)h2-stems (cf. nom.sg.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

352

Chapter 5

PT *-o < nom.sg. PIE *-eh2 and acc.sg. PIE *-eh2-m) and the neuter plural (cf. nom.obl.pl. PT *-o < nom.acc.pl. PIE *-eh2). The members of the śana-type were traced back to two different PIE stem types which both inflected proterodynamically in Proto-Indo-European (§ 3.4.1): TB śana, ta śäṃ ‘wife’ is from a stem in PIE *-h2/-eh2-, while TB lāntsa, ta lānts ‘queen’ and TB ṣarya ‘(beloved) lady’ are from a stem in PIE *-ih2/-ieh2́ (of the devī-type). So too in this class does the contrast between nom.sg. -a, obl.sg, -o mirror the ablauting alternation between strong and weak stem of the suffix *-(i)h2/-(i̯)eh2-. The generalisation of the PIE weak stem in some Tocharian obliques was caused by the formal merger of the inherited nominative and accusative in many inflectional types. To solve this homophony, a trend of development in the evolution of Tocharian nominal morphology was to reanalyse the inherited genitive or dative singular as the new oblique (cf. obl.sg. TB -a < gen.sg. PIE *-h2és vs. nom.sg. TB -o < nom.sg. PIE *-éh2; obl.sg. -yo < gen.sg. PIE *-i̯éh2s vs. nom.sg. -ya < nom.sg. PIE *-ih2; obl.sg. -ai < dat.sg. PIE *-h2éi̯). On the other hand, it has been shown that, apart from few exceptions, the nouns of the aśiya-type are all of late origin and they calqued their paradigm from the adjectives of class i.2 (§ 3.4.2, § 4.7.1). From the adjectives, the śana-type and the aśiya-type also took the plural marker PT *-na. In this respect, it is probable that Tocharian inherited and further generalised PIE ́ *-ih2/-ieh2- (of the devī-type) in the derivation of nouns specifically denoting females, as suggested by Kim (2014). However, it is hard to establish whether ́ the apparent outcome of the devī-suffix in the noun inflection is a preservation of the original state of affairs or a secondary generalisation that had its epicentre in the adjectives. Furthermore, it has been argued that Tocharian inherited the feminine suf́ ́ fix *-ih2 of both the devī-type and the vr̥kī-type, and that these two formations merged in Proto-Tocharian. The outcome of this merger formed the inflection of the wertsiya-type (§ 3.5.3). 5.4

Diachrony of the Genus Alternans

The Tocharian genus alternans generally reflects the PIE neuter, but it originated after several morphophonological mergers, of which its peculiar alternating agreement is a direct outcome. As suggested by a typological comparison with languages with similar gender systems and further confirmed by a close reconstruction of the ProtoTocharian adjectival paradigms, the thematic neuter must have become

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Retrospective and Conclusion

353

homophonous with the masculine singular in the singular and with the feminine singular in the plural. This coalescence first took place in the adjectival system, and it was a gradual process. Indeed, in some pronouns and adjectives, we still find relics of frozen forms of the historical neuter. Synchronically, they are either used adverbially or with pronominal function (i.e. non-attributively). Since these relics are found in paradigms where, in the masculine, the contrast between nominative and oblique singular was secondarily remarked by analogical palatalisation, it is likely that the neuter also survived for a while in the inflection of the modifiers, where it was differentiated from the masculine in the oblique singular. When the new alternating agreement was grammaticalised, these old neuter forms could no longer correlate with any neuter nouns and were reanalysed: they became “neutral” with respect to gender. On the other hand, in the plural inflection, historical neuter forms spread to the feminine in certain adjectival paradigms. The neuter plural endings in -a were refunctionalised as marker of the feminine. In the noun inflection, the development of both the thematic and athematic neuter was analysed. It was attempted to identify the reflexes of PIE thematic neuters in Tocharian: some of these were continued as alternating and thus converged in the āke-type, while others were reassigned to the masculine gender, as they synchronically belong to the yakwe-type (§ 3.6.1). This fluctuation was caused by the formal merger of the PIE masculine and neuter in the thematic paradigm of the singular. A similar coalescence also characterised the PIE neuter plural and the feminine. Evidence of this merger can be found in the palsko-type, where old thematic plural forms may have been reanalysed as singular (§ 3.6.2.1). Thus, in Tocharian neuter nouns were sometimes absorbed into the masculine class, while some neuter plural forms were reinterpreted as singular or as feminine, with the reassignment operating on the basis of formal resemblance. The reassignment of neuter nouns took place in parallel with the disappearance of the neuter as a grammatical category and with the gradual grammaticalisation of the alternating behaviour of the agreement class. Variations and fluctuations in this reconstructed stage are likely to have occurred. The majority of the neuter nouns, however, were retained in a separate category, which evolved into the genus alternans. I also investigated in detail the outcome of those athematic neuters that have played an important role in the creation of new endings and in the evolution of the Tocharian gender system in general. In particular, isolated pluralia tantum and lexical plurals were discussed in  §  3.6.4; s-stem formations and neuter root nouns were treated in § 3.5.1 and § 3.6.1.2. Particular attention has been paid to the evolution of an archaic class of nominals, the heteroclites in PIE *r/n. We have seen that some heteroclitic

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

354

Chapter 5

stems have been continued in class ii.1 (pl. TB -na, A -äṃ), where Tocharian A maintained both the r-stem of the singular and the n-stem of the plural. The final outcome of this development was a blended plural with the r-form as the stem and the n-form as the ending (§ 3.6.3.5). The evolution of the PIE *ur/n-stems was carried out in § 3.6.3.2, where the basis for the postulation of a sound law PIE *-u̯ r̥ > *-ru was presented. This metathesis accounts for the origin of r-stem nouns with plural in TB -wa, A -u (-wā, -unt), for the unexpected o-vocalism in some isolated forms, and, probably, for the spread of the plural marker TB -una. 5.5

Outlook

In this study, it has been argued that, despite the many peculiarities of its gender system, Tocharian does not necessarily preserve a different, more archaic, gender marking than the other non-Anatolian Indo-European languages (cf.  §  2.4); rather, it inherited a regular three-gender contrast. In an unattested stage, formal mergers took place, and the masculine, the feminine and the neuter influenced each other morphologically, before being fixed in the attested agreement classes. These mergers caused the functional loss of the neuter as a target gender, the rise of the new alternating agreement class, and other morphological developments which remarked the feminine. The final result of this process is the attested suppletion of the thematic feminine markers with the outcome of *-ih2 in the singular and the outcome of *-eh2 in the plural. Therefore, the results of this study indicate that the gender system cannot primarily be used to support the early separation of Tocharian from Proto-Indo-European: the peculiarities of the Tocharian gender system were caused by internal developments that often blurred the boundary between inherited archaisms and internal innovations. Indeed, Tocharian shows a patchwork of innovations and retentions in its gender system just like the other Indo-European languages. We do find formal traces of, for instance, the older bipartite system with relics of sex-indifferent thematic inflection in such nouns as TB yakwe (m./f.) ‘horse; mare’ < PIE *h1éḱu̯ os or TB santse (f.) ‘daughter-in-law’ < PIE *snusós (§ 2.3.3, § 3.6.1.1), and ́ also possible evidence that the suffix *-ih2 of the devī-type was used for deriving feminine nouns denoting females from the respective masculine ones. However, these conservative elements are not distinctive features of Tocharian, as other Indo-European languages have variously retained them too. Thus, the thematic o-stem paradigm of the (feminine) word for ‘daughter-in-law’ is also preserved in Greek νυός and Armenian now, gen.sg. nowoy, while it was rebuilt

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Retrospective and Conclusion

355

in e.g. Sanskrit snuṣā́-, OE snoru, OHG snura (Peyrot & Xiaoqiang 2021); traces of the “common gender” of *h1eḱu̯ os can be found in Latin equus, Greek ἵππος (Laconian hίππος), and Vedic áśva- together with the remodelled nouns Latin equa, Vedic áśvā- (Hackstein 2013); the feminine-marking noun morpheme *-ih2 is widespread in Indo-European languages (Kim 2014), with traces as early as in (Kanišite) Hittite (Kloekhorst 2019: 223–230). On the other hand, it is generally assumed that Tocharian nominal morphology is less conservative and archaic with respect to what is reconstructed for the Indo-European proto-language. Accordingly, Tocharian offers relatively little of Indo-European interest in the domain of nominal morphology. This would be hardly surprising in a language where the inherited system of declension was drastically simplified and then overlaid by external influences. Although the nominal system has certainly innovated and eroded, the statement that Tocharian is of only minor importance for the reconstruction of the PIE nominal system has been proven to be wrong. We have seen that extensive reductions and phonological changes caused several irregularities in the nominal paradigms, which were mostly resolved by analogical changes to restore regularity in morphological patterns. This led to a heavy restructuring of the morphology. However, in the regularities and the irregularities of its nominal system, Tocharian has mostly used and refunctionalised inherited material, which is also relevant from an Indo-European comparative perspective.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography Acquaviva, Paolo 2008 Lexical Plurals. Oxford: Oxford University Press. Adams, Douglas Q. 1984 The position of Tocharian among the other Indo-European languages. Journal of the American Oriental Society 106: 395–402. 1987 Marginalia to the Tocharian Lexicon. Tocharian and Indo-European Studies 1: 1–9. 1988a Tocharian Historical Phonology and Morphology. New Haven: American Oriental Society. 1988b The expansion of the PIE n-stems in Tocharian: The systematic development of a paradigm. Tocharian and Indo-European Studies 2: 7–30. 1990 Some reflexes of the PIE neuter n-stems in Tocharian. Tocharian and Indo-European Studies 4: 65–78. 1991 “Head” and “horn” in Tocharian and Albanian. In: Lambert Isebaert (ed.), Studia Etymologica Indoeuropaea. Memoriae A. J. Van Windekens (1915–1989) dicata. Leuven: Peeters, 3–11. 1998 Review of Don Ringe (1996). Language 74: 615–617. 1999 A Dictionary of Tocharian B. Amsterdam–Atlanta: Rodopi. 2004 The lessons of two words for materials, ‘clay’ and ‘iron’, in Tocharian. In: Adam Hyllested, Anders Richard Jørgensen, Jenny Helena Larsson & Thomas Olander (eds.), Per aspera ad asteriscos. Studia Indogermanica in honorem Jens Elmegård Rasmussen sexagenarii Idibus Martiis anno MMIV. Innsbruck: Institut für Sprachwissenschaft, 27–31. 2009a Genitive and Adjective in Tocharian. Journal of Indo-European Studies 37: 299–320. 2009b How to tell a sheep’s age in Tocharian B. Tocharian and Indo-European Studies 11: 1–12. 2012 Shedding light on *leuk- in Tocharian and Hittite and the wider implications of reconstructing its Indo-European morphology. Tocharian and Indo-European Studies 13: 21–56. 2015 Tocharian B: A grammar of Syntax and Word-formation. Innsbruck: Institut für Sprachwissenschaft. 2016a Review of Markus Hartmann (2013). Tocharian and Indo-European Studies 17: 249–254. 2016b Tocharian B arkañ yet again. Tocharian and Indo-European Studies 17: 1–10. 2017 The lexicon of Tocharian. In: Jared Klein, Brian Joseph, Matthias Fritz & Mark Wenthe (eds.), Handbook of Comparative and Historical Indo-European Linguistics. Vol. II. Berlin–Boston: de Gruyter, 1365–1388. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

358

Bibliography

2021

A note on two Tocharian B nominal case markers. Journal of Indo-European Studies 49.3: 265–278. Adams, Douglas Q. & Václav Blažek 2021 Tocharian B yerkwantala*. In: Matteo Tarsi (ed.), Studies in general and historical linguistics offered to Jón Axel Harðarson on the occasion of his 65th birthday. Innsbruck: Institut für Sprachwissenschaft, 81–88. Adams, James N. 2011 Late Latin. In: James Clackson (ed.), A Companion to the Latin Language. Chichester–Malden: Wiley-Blackwell, 257–283. AiGr 1905–45 Jacob Wackernagel & Albert Debrunner, Altindische Grammatik. 3 vols. Göttingen: Vandenhoeck & Ruprecht. AirW 1904 Bartholomae, Christian. Altiranisches Wörterbuch. Strassburg: Trübner. Aikhenvald, Alexandra Y. 2000 Classifiers. A Typology of Noun Categorization Devices. Oxford: Oxford University Press. Alfieri, Luca 2009 La categoria dell’aggettivo in vedico. Archivio Glottologico Italiano 94: 3–47. 2016 The typological definition of the (apparently historical) notion of root. Archivio Glottologico Italiano 101: 129–169. 2018 La definizione tipologica della radice e la teoria del segno lessicale. In: Luca Alfieri, Maria Carmela Benvenuto, Claudia A. Ciancaglini, Alessandro De Angelis, Paolo Milizia & Flavia Pompeo (eds.), Linguistica, filologia e storia culturale. Studi in memoria di Palmira Cipriano. Roma: Il Calamo, 25–44. Anreiter, Peter 1982 Gedanken zur Abbildung id./vortoch. *d → toch. t/c. In: Wolfgang Meid, Hermann Ölberg & Hans Schmeja (eds.), Sprachwissenschaft in Innsbruck. Innsbruck: Institut für Sprachwissenschaft, 19–31. Aronoff, Mark 1994 Morphology by Itself. Cambridge, MA: MIT Press. Āṭhavale, Ananta Dāmodara (ed.) 1980 Śrīmadvr̥ddhavāgbhaṭaviracitaḥ Aṣṭāṅgasaṅgrahaḥ Induvyākhyāsahitaḥ [The Aṣṭāṅgasaṅgraha by the illustrious Vr̥ddha Vāgbhaṭa, accompanied by Indu’s commentary]. Pune: Śrīmad Ātreya Prakāśam (1st ed. 1902). Audring, Jenny 2008 Gender assignment and gender agreement: Evidence from pronominal gender languages. Morphology 18: 93–116. 2009 Reinventing Pronoun Gender. Diss. Vrije Universiteit, Amsterdam. Utrecht: LOT.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

359

2014

Gender as a complex feature. In: Tania Paciaroni, Michele Loporcaro & Anna M. Thornton (eds.), Exploring Grammatical Gender, special issue of Language Sciences 43, 5–17. 2016 Gender. Oxford Bibliographies Online: Linguistics. 2019 Canonical, complex, complicated? In: Francesca Di Garbo, Bruno Olsson & Bernhard Wälchli (eds.), Grammatical Gender and Linguistic Complexity. Volume I: General Issues and Specific Studies. Berlin: Language Science Press, 15–52. Babou, Cheickh Anta & Michele Loporcaro 2016 Noun classes and grammatical gender in Wolof. Journal of African Languages and Linguistics 37: 1–57. Bailey, Harold W. 1937 Ttaugara. Bulletin of the School of Oriental Studies 8: 883–922. 1959 Ambages Indoiranicae. Annali dell’Istituto Universitario Orientale di Napoli. Sezione Linguistica 1: 113–146. 1967 Indo-Scythian Studies: Being Khotanese Texts. Vol. VI: Prolexis to the Book of Zambasta. Cambridge: Cambridge University Press. 1979 Dictionary of Khotan Saka. Cambridge: Cambridge University Press. Barbati, Chiara 2016 The Christian Sogdian Gospel Lectionary E5 in Context. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Barbera, Gerardo 2000 L’espressione dello spazio in tocario. Studio morfosintattico e lessicale. Tesi di laurea, Istituto Universitario Orientale di Napoli. Barber, Peter J. 2014 Sievers’ Law and the History of Semivowel Syllabicity in Indo-European and Ancient Greek. Oxford: Oxford University Press. Bateman, Nicoleta 2011 On the typology of palatalization. Language and Linguistics Compass 5: 588–602. Bateman, Nicoleta & Maria Polinsky 2010 Romanian as a two-gender language. In: Donna Gerdts, John Moore & Maria Polinsky (eds.), Hypothesis A — Hypothesis B. Linguistic Explorations in Honor of David M. Perlmutter. Cambridge, MA: MIT Press, 41–77. van Beek, Lucien 2013 The development of the Proto-Indo-European syllabic liquids in Greek. Diss. Universiteit Leiden. Beekes, Robert S. P. 1973 The Greek i- and u-stems and πόλις, -ηος. Glotta 51: 228–245.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

360

Bibliography

1985

The Origin of the Indo-European Nominal Inflection. Innsbruck: Institut für Sprachwissenschaft. 1994 The neuter plural of thematic nouns: Derivatives from a stem in -e- from thematic nouns. In: George E. Dunkel, Gisela Meyer, Salvatore Scarlata & Christian Seidl (eds.), Früh-, Mittel-, Spätindogermanisch. Akten der IX. Fachtagung der Indogermanischen Gesellschaft vom 5. Bis 9. Oktober 1992 in Zürich. Wiesbaden: Reichert, 1–15. 2010 Etymological Dictionary of Greek. With the assistance of Lucien van Beek. 2 vols. Leiden–Boston: Brill. 2011 Comparative Indo-European Linguistics. An Introduction. Revised and corrected by Michiel de Vaan. Philadelphia: Benjamins. Belardi, Walter 1985 Filosofia, grammatica e retorica nel pensiero antico. Roma–Firenze: Lessico intellettuale europeo. Benveniste, Émile 1936 Tokharien et Indo-Européen. In: Helmuth Arntz (ed.), Germanen und Indo­ germanen. Volkstum, Sprache, Heimat, Kultur. Festschrift für Herman Hirt. Vol. 2. Heidelberg: Winter, 229–240. 1942–45 Review of A. J. Van Windekens (1944). Bulletin de la Société de Linguistique de Paris 42: 48–50. 1954 Notes avestiques. In: Johannes Schubert & Ulrich Schneider (eds.), Asiatica. Festschrift Friedrich Weller. Zum 65. Geburtstag gewidmet von seinen Freunden, Kollegen und Schülern. Leipzig: Harrassowitz, 30–34. Bernard, Chams B. forth. Like Dust on the Silk Road. An investigation of the earliest Iranian loanwords and of possible BMAC borrowings in Tocharian. Diss. Universiteit Leiden. Bernard, Chams B. & Ruixuan Chen 2022 A Fall into the Pit. Remarks on Tocharian B koṣko, koṣkīye. Indo-Iranian Journal 65.1: 1–31. Bernhard, Franz 1958 Die Nominalkomposition im Tocharischen. Diss. Göttingen. 1965 Udānavarga. Band I, Einleitung, Beschreibung der Handschriften, Textausgabe, Bibliographie. Göttingen: Vandenhoeck und Ruprecht. Bianconi, Michele 2021 “There and back again”: A hundred years of Graeco-Anatolian comparative studies. In: Michele Bianconi (ed.), Linguistic and Cultural Interactions between Greece and Anatolian. In Search of the Golden Fleece. Leiden–Boston: Brill, 8–39. bhsd 1953 Franklin Edgerton, Buddhist Hybrid Sanskrit Grammar and Dictionary. Vol. II: Dictionary. New Haven, CT: Yale University. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

361

Blažek, Václav 2001 Tocharian A muk ‘yoke’ and A maku, B mekwa pl. ‘(finger)nails’ — Why m-? Historische Sprachforschung 114: 191–195. 2005 Tocharian A kuli, B klyiye “woman” < *ĝ/gleH2u̯ i-H1en-? Historische Sprach­ forschung 118: 92–100. 2012 Indo-European laryngeals in the light of Afroasiatic. In: Benedicte Nielsen Whitehead, Thomas Olander, Birgit Annette Olsen & Jens Elmegård Rasmussen (eds.), The sound of Indo-European. Phonetics, Phonemics, and Morphophonemics. Copenhagen: Museum Tusculanum Press, 1–26. 2016 Review of Douglas Q. Adams (2013). Journal of Indo-European Studies 44: 218–243. Blažek, Václav & Michal Schwarz 2017 Early Indo-Europeans in Central Asia and China. Cultural relations as Reflected in Language. Innsbruck: Institut für Sprachwissenschaft. Bolelli, Tristano 1948 Il valore semasiologico delle voci ἧτορ, κῆρ e κραδίη nell’épos omerico. Annali della Scuola Normale Superiore di Pisa 17: 65–76. Bond, Olivier 2019 Canonical Typology. In: Jenny Audring & Francesca Masini (eds.), The Oxford Handbook of Morphological Theory. Oxford: Oxford University Press, 409–431. Booij, Geert 1994 Against split morphology. In: Geert Booij & Jaap van Marle (eds.), Yearbook of Morphology 1993. Dordrecht: Kluwer Academie Publishers, 27–49. 1996 Inherent versus contextual inflection and the split morphology hypothesis. In: Geert Booij & Jaap van Marle (eds.), Yearbook of Morphology 1995. Dordrecht: Kluwer, 1–16. 2005 The Grammar of Words. An Introduction to Linguistic Morphology. Oxford: Oxford University Press. Bouckaert, Remco, Ph. Lemey, M. Dunn, S. J. Greenhill, A. V. Alekseyenko, A. J. Drummond, R. D. Gray, M. A. Suchard & Q. D. Aktinson 2012 Mapping the origins and expansion of the Indo-European language family. Science 337: 957–960, 989. Boutkan, Dirk 1995 The Germanic Auslautgesetze. A new interpretation. Amsterdam: Rodopi. Broomhead, J. W. 1962 A textual edition of the British Hoernle, Stein and Weber Kuchean manuscripts with transliteration, translation, grammatical commentary and vocabulary. Doctoral dissertation, Trinity College Cambridge. 2 vols. Brugmann, Karl F. C. 1891 Zur Frage der Entstehung des grammatischen Geschlechts. Beiträge zur Geschichte der deutschen Sprache und Literatur 15: 523–31. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

362

Bibliography

1911

Grundriss der vergleichenden Grammatik der indogermanischen Sprachen. Band 2, Teil 2. Strassburg: Trübner. Brugmann, Karl F. C. & Berhold Delbrück 1893 Grundriss der vergleichenden Grammatik der indogermanischen Sprachen. Band 3 Vergleichende Syntax. Strassburg: Trübner. Burlak, Svetlana 2000 Istoričeskaja fonetika toxarskix jazykov. Moskva: Institut Vostokovedenija, Rossijskaja Akademija Nauk. Burlak, Svetlana & Ilja B. Itkin 2003 A sound change that never happened. The fate of Proto-Tocharian *o (B o) in Tocharian A. Tocharian and Indo-European Studies 10: 17–35. Bussmann, Hadumod 2008 Lexikon der Sprachwissenschaft. Alfred Kröner Verlag: Stuttgart. Byrd, Andrew Miles 2015 The Indo-European Syllable. Leiden: Brill. Cantera, Alberto 2009 On the History of Middle Persian Nominal Inflection. In: Werner Sundermann, Almut Hintze & François de Blois (eds.), Exegisti monumenta. Festschrift in Honour of Nicholas Sims-Williams. Wiesbaden, 17–30. Cardona, George 2003 Sanskrit. In: George Cardona & Dhanesh Jain (eds.), The Indo-Aryan Languages. London–New York: Routledge, 115–178. Carling, Gerd 2000 Die Funktionen der lokalen Kasus im Tocharischen. Berlin–New York: de Gruyter. 2003 New look at the Tocharian B medical manuscript IOL Toch 306 (Stein Ch.00316.a2) of the British Library — Oriental and India Office Collections. Historische Sprachforschung 116: 75–95. 2004 Tocharian B erkatse [A *arkäts] and related phenomena. In: Adam Hyllested, Anders Richard Jørgensen, Jenny Helena Larsson & Thomas Olander (eds.), Per aspera ad asteriscos. Studia Indogermanica in honorem Jens Elmegård Rasmussen sexagenarii Idibus Martiis anno MMIV. Innsbruck: Institut für Sprachwissenschaft, 95–102. 2005 Proto-Tocharian, Common Tocharian, and Tocharian. On the value of linguistic connections in a reconstructed language. In: Karlene Jones-Bley, Martin E. Huld, Angela Della Volpe & Miriam Robbins Dexter (eds.), Proceeding of the Sixteenth Annual UCLA Indo-European Conference. Los Angeles November 5–6, 2004. Washington: Institute for the Study of Man, 47–71.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

363

2008

Phonological erosion and semantic generalization: Notes on the grammaticalization of the Tocharian case paradigm. Working Papers 53. Lund University, 5–14. 2009 Dictionary and thesaurus of Tocharian A. Part 1: A-J. In collaboration with Georges-Jean Pinault and Werner Winter. Wiesbaden: Harrassowitz. 2012 Development of form and function in a case system with layers: Tocharian and Romani compared. Tocharian and Indo-European Studies 13: 55–74. 2017 The Tocharian syntax. In: Jared Klein, Brian Joseph, Matthias Fritz & Mark Wenthe (eds.), Handbook of Comparative and Historical Indo-European Linguistics. Vol. II. Berlin–Boston: de Gruyter, 1352–1364. Carling, Gert & Chundra Cathcart 2021 Reconstructing the evolution of Indo-European grammar. Language 97.3: 561–598. Catt, Adam A. ̆ 2016 Tocharian B ly(ī)ptsentar: A new class VIII present. Tocharian and IndoEuropean Studies 17: 11–28. Catt, Adam, Athanaric Huard & Yuima Inaba 2020 Tocharian Abhidharma Texts I: A philological study of B 197. Journal Asiatique 308.2: 183–204. CEToM 2011– A Comprehensive Edition of Tocharian Manuscripts. www.univie.ac.at/tochar ian/ (accessed in November 2022). Chang, Will, David Hall, Chundra Cathcart & Andrew Garrett 2015 Ancestry-constrained phylogenetic analysis supports the Indo-European steppe hypothesis. Language 91: 194–244. Chantraine, Pierre 1933 La formation des noms en grec ancien. Paris: Champion. 1999 Dictionnaire étymologique de la langue grecque. Histoire des mots. Avec un Supplément sous la direction de: Alain Blanc, Charles de Lamberterie, Jean-Louis Perpillou. Paris: Klincksieck. Charpentier, Jarl 1917 Die ethnographische Stellung der Tocharer. Zeitschrift der Deutschen Morgenländischen Gesellschaft 71.4: 347–388. Cheung, Johnny 2002 Studies in the Historical Development of the Ossetic vocalism. Wiesbaden: Reichert. 2007 Etymological Dictionary of the Iranian Verb. Leiden: Brill. Ching, Chao-jung 2010 Secular Documents in Tocharian: Buddhist economy and society in the Kucha region. Thèse de doctorat, École Pratique des Hautes Études, Paris.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

364

Bibliography

2012 Lüèlùn Huáng Wénbì suǒ fāxiàn zhī sì jiàn Qiūcíyǔ shìsú wénshū [A brief discussion on the four Kuchean secular documents discovered by Huang Wenbi]. In: Huáng Jiànmíng, Niè Hóngyīn & Mǎ Lán (eds.), Shǒujiè Zhōngguó shǎoshù mínzú gǔjí wénxiàn guójì xuéshù yántǎo huì lùnwénjí [Proceedings of the 1st International Colloquium on Ancient Manuscripts and Literatures of the Minorities in China]. Běijīng: Mínzú Chūbǎnshè, 303–324. 2014 Perfumes in Ancient Kucha: On the word tuñe attested in Kuchean monastic accounts. Tocharian and Indo-European Studies 15: 39–52. 2015 Animal husbandry in Ancient Kucha: a historical perspective. In: Melanie Malzahn, Michaël Peyrot, Hannes A. Fellner & Theresa S. Illés (eds.), Tocharian Text in Context. International Conference on Tocharian Manuscripts and Silk Road Culture, June 25–29th, 2013. Bremen: Hempen, 37–56. 2016 On the names of cereals in Tocharian B. Tocharian and Indo-European Studies 17: 29–65. Ching, Chao-jung & Hirotoshi Ogihara 2013 A Tocharian B sale contract on a wooden tablet. Journal of Inner Asian Art and Archaeology 5 [2010]: 101–128. 2020 Qiūcí shíkū tíjì [Cave Inscriptions in Ancient Kucha]. Shànghǎi: Zhōngxī Shūjú. Ciancaglini, Claudia A. 2001 Sciti, iranici, nomadi: problemi di etnonimia in Strabone. In: Giusto Traina (ed.), Studi sull’XI libro dei Geographika di Strabone. Galatina: Congedo, 11–83. 2009 How to prove genetic relationship among languages. Rivista di Studi Orientali 81: 289–320. Clackson, James 2022 Methodology in Linguistic Subgrouping. In: Thomas Olander (ed.), The IndoEuropean Language Family. A Phylogenetic Perspective. Cambridge: Cambridge University Press, 18–32. Claudi, Ulrike 1997 Some thoughts on the origin of gender marking. In: Robert K. Herbert (ed.), African Linguistics at the Crossroads. Papers from Kwaluseni. Köln: Rüdiger Köppe, 63–74. Čop, Bojan 1975 Studien im tocharischen Auslaut I. Ljubljana: Oddelek za primerjalno jezikoslovje in orientalistiko, Filozofska Fakulteta, Univerza v Ljubljani. Corbett, Greville G. 1991 Gender. Cambridge: Cambridge University Press. 2000 Number. Cambridge: Cambridge University Press. 2006 Agreement. Cambridge: Cambridge University Press. 2007 Gender and noun classes. In: Timothy Shopen (ed.), Language Typology and Syntactic Description. Volume III: Grammatical categories and the lexicon, Cambridge: Cambridge University Press, 241–279. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

365

2010

Implicational hierarchies. In: Jae Jung Song (ed.), The Oxford Handbook of Linguistic Typology. Oxford: Oxford University Press, 190–205. 2011 The penumbra of morphosyntactic feature systems. Morphology 21: 445–480. 2012 Features. Cambridge: Cambridge University Press. 2013 Gender Typology. In: Greville G. Corbett (ed.), The Expression of Gender. Berlin–New York: de Gruyter, 87–130. Corbett, Greville G. & Sebastian Fedden 2016 Canonical gender. Journal of Linguistics 52: 495–531. 2018 New approaches to the typology of gender. In: Sebastian Fedden, Jenny Audring & Greville G. Corbett (eds.), Non-canonical gender systems, Oxford: Oxford University Press, 9–34. Corbett, Greville G. & Normann M. Fraser 2000 Gender assignment: a typology and a model. In: Gunter Senft (ed.), Systems of Nominal Classification. Cambridge: Cambridge University Press. Couvreur, Walter 1938 La désinence -au et le morphème -oy-, -i- du koutchéen. Bulletin de la Société de Linguistique 39: 243–248. 1947a Hoofdzaken van de Tochaarse klank- en vormleer. Leuven: Beheer van Philologische Studiën. 1947b Zum Tocharischen I. Revue des études indo-européennes 4: 137–158. 1954 Koetsjische literaire fragmenten uit de Berlijnse verzameling (naar aanleiding van Sieg & Siegling’s Tocharische Sprachreste). Zuidnederlandse Maatschappij voor Taal- en Letterkunde en Geschiedenis, Handelingen 8: 97–117. 1964 Nieuwe Koetsjische fragmenten van het Araṇemijātaka. Orientalia Gandensia 1: 237–249. 1968 Zu einigen Sanskrit-Kutschischen Listen von Stichwörtern aus dem Catuṣpariṣatsūtra, Daśottarasūtra und Nidānasaṃyutka. In: Jan C. Heesterman, Godard H. Schokker & V. I. Subramoniam (eds.), PRATIDĀNAM. Indian, Iranian and Indo-European Studies. Presented to F. B. J. Kuiper on his Sixtieth Birthday. Janua Linguarum Series Maior 34. Den Haag, 275–282. Cowell, Edward B. & Robert A. Neil (eds.) 1886 The Divyāvadāna: A Collection of Early Buddhist Legends Now First Edited from the Nepalese Sanskrit mss. in Cambridge and Paris. Cambridge: Cambridge University Press. Cowgill, Warren C. 2006 The z-cases of Germanic pronouns. In: Jared S. Klein (ed.), The Collected Writings of Warren Cowgill. Ann Arbor: Beech Stave, 519–534. Croft, William 2013 Agreement as anaphora, anaphora as coreference. In: Dik Bakker & Martin Haspelmath (eds.), Languages across Boundaries: Studies in Memory of Anna Siwierska. Berlin: de Gruyter, 107–129. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

366

Bibliography

Das, Rahul P. & Ronald E. Emmerick (eds.) 1998 Aṣṭāṅgahr̥dayasaṃhitā: the romanised text accompanied by line and word indexes. Groningen: Forsten. De Decker, Filip 2011 Stang’s Law and the Indo-European word for “cow”. Indogermanische Forschungen 116: 42–59. Dedè, Francesco 2013 I nomi greci in -αρ e -ωρ. Eteroclisi e classi nominali. Roma: Il Calamo. Degener, Almuth 1989 Khotanische Suffixe. Stuttgart: Steiner. Del Tomba, Alessandro 2018 On the pronominal feminine plural in Tocharian. Indogermanische Forschungen 123: 339–356. 2019 The origin of the Tocharian A plural ending -äṃ. Indo-European Linguistics 7: 1–12. 2020a The Tocharian Gender System. A diachronic Study. Diss. Sapienza Università di Roma–Universiteit Leiden. 2020b The development of PIE *-oi̯ in Tocharian. Rivista di Studi Orientali 93: 21–34. 2020c Metathesis of PIE *-u̯ r̥ in Tocharian. Münchener Studien zur Sprachwissenschaft 74.1 (2020 [2021]): 51–85. 2022 Avestan -ā ̊ŋhō, Young Avestan -ā ̊, Old Khotanese -e and the development of the Old Iranian i- and u-stems in Khotanese. Studi e Saggi Linguistici 60.1: 111–171. Del Tomba, Alessandro & Mauro Maggi 2021 A Central Asian Buddhist Term. Remarks on Khotanese saña- and Tocharian B sāñ, A ṣāñ. Indo-Iranian Journal 64.3: 199–240. Demiraj, Bardhyl 1997 Albanische Etymologien Untersuchungen zum albanischen Erbwortschatz. Amsterdam–Atlanta: Rodopi. Derksen, Rick 2009 Etymological Dictionary of the Slavic Inherited Lexicon. Leiden–Boston: Brill. 2015 Etymological Dictionary of the Baltic Inherited Lexicon. Leiden–Boston: Brill. Di Garbo, Francesca 2014 Gender and its interaction with number and evaluative morphology. An intraand intergenealogical typological survey of Africa. Diss. Stockholm University. 2016 Exploring grammatical complexity crosslinguistically: The case of gender. Linguistic Discovery 14.1: 46–85. Di Garbo, Francesca & Matti Miestamo 2019 The evolving complexity of gender agreement systems. In: Francesca Di Garbo, Bruno Olsson & Bernhard Wälchli (eds.), Grammatical gender and

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

367

linguistic complexity. Volume II: World-wide comparative studies. Berlin: Language Science Press, 15–60. Dragoni, Federico 2021 Materia Medica Tocharo-Hvatanica. Bulletin of the School of Oriental and African Studies 84.2: 295–319. 2022 Watañi lāntaṃ. Khotanese and Tumshuqese Loanwords in Tocharian. Diss. Universiteit Leiden. Dressler, Wolfgang U. & Anna M. Thornton 1996 Italian nominal inflection. Wiener Linguistische Gazette 57–59: 1–26. Driessen, Michiel 2003 *h2é-h2us-o-, the Proto-Indo-European term for ‘gold’. Journal of Indo-European Studies 31: 347–362. dtb 2013 Douglas Q. Adams. A dictionary of Tocharian B. Revised and greatly enlarged. 2 vols. Amsterdam–New York: Rodopi. Duchesne-Guillemin, Jacques 1941 Tocharica. Bulletin de la Société de Linguistique de Paris 41: 140–183. Dunkel, George 2016 Proto-Indo-Iranian *stríH- and PIE *sór- ‘female, woman’. In: Dieter Gunkel, Joshua T. Katz, Brent Vine & Michael Weiss (eds.), Sahasram Ati Srajas. IndoIranian and Indo-European Studies in Honor of Stephanie W. Jamison. Ann Arbor: Beech Stave, 36–45. Emmerick, Ronald E. 1968a The Book of Zambasta. A Khotanese poem on Buddhism. London: Oxford University Press. 1968b Saka Grammatical Studies. London: Oxford University Press. 1969 Notes on “The Book of Zambasta”. Journal of the Royal Asiatic Society of Great Britain and Ireland 1: 59–74. 1980 r-/n-stems in Khotanese. In: Manfred Mayrhofer, Oskar E. Pfeiffer & Martin Peters (eds.), Lautgeschichte und Etymologie. Akten der VI. Fachtagung der Indogermanischen Gesellschaft. Wien, 24.–29. September 1978. Wiesbaden: Reichert, 166–72. 1982 The Siddhasāra of Ravigupta. Volume 2: The Tibetan version with facing English translation. Edited and translated by R. E. Emmerick. Wiesbaden: Franz Steiner Verlag. 1984 Some Remarks on the History of Leprosy in India. Indologica Taurinensia 12: 93–105. 1989 Khotanese and Tumshuqese. In: Rüdiger Schmitt (ed.), Compendium linguarum Iranicarum. Wiesbaden: Reichert, 204–229. 1990 Khotanese śśāman- ‘face’. Tocharian and Indo-European Studies 4: 33–34.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

368

Bibliography

1993

[Numerals in] Iranian. In: Jadranka Gvozdanović (ed.), Indo-European Numerals. Berlin–New York: de Gruyter, 289–345. 2009 Khotanese and Tumshuqese. In: Gernot Windfuhr (ed.), The Iranian Languages. London: Routledge, 377–415. Emmerick, Ronald E. & Prods O. Skjærvø 1982 Studies in the vocabulary of Khotanese. Vol. I. Wien: Verlag der Österreichischen Akademie der Wissenschaften. 1987 Studies in the vocabulary of Khotanese. Vol. II. Wien: Verlag der Österreichischen Akademie der Wissenschaften. 1997 Studies in the vocabulary of Khotanese. Vol. III. Edited by Ronald E. Emmerick. Contributed by G. Canevascini, R. E. Emmerick, H. Kumamoto, M. Maggi, N. Sims-Williams & P. O. Skjærvø. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Ernout, Alfred 1945 Morphologie historique du latin. Paris: Klincksieck. Ernout, Alfred & Antoine Meillet 1951 Dictionnaire étymologique de la langue latine. Histoire des mots. 3e éd. Paris: Librairie C. Klincksieck. Evangelisti, Enzo 1990 Studi sull’etimologia del tocario. In: Scritti tocarici e altri studi. Brescia: Paideia Editrice, 115–138 [Studi sull’etimologia del tocario. Brescia: Paideia Editrice, 1965]. EWAia 1986–96 Manfred Mayrhofer. Etymologisches Wörterbuch des Altindoarischen. Band I– III. Heidelberg: Winter. Faraoni, Vincenzo 2016 Manifestazioni del neutro italo-romanzo nella documentazione notarile altomedievale. In: Ev́a Buchi, Jean-Paul Chauveau & Jean-Marie Pierrel (eds.), Actes du XVIIe Congrès international de linguistique et de philologie romanes (Nancy, 15–20 juillet 2013). 3 vols. Strasbourg: Société de linguistique romane/ÉliPhi, 1: 381–95. Fedden, Sebastian & Greville G. Corbett 2017 Gender and classifiers in concurrent systems: Refining the typology of nominal classification. Glossa: a journal of general linguistics 2.1: 34. 1–47. Fellner, Hannes A. 2013 Studies in Tocharian Adjective Formation. Diss. Harvard University. 2014a PIE feminine *-eh2 in Tocharian. In: Sergio Neri & Roland Schumann (eds.), Studies on the Collective and Feminine in Indo-European from Diachronic and Typological Perspective. Leiden–Boston: Brill, 7–21. 2014b Das Femininum der thematischen Adjectiva im Tocharischen. In: Norbert Oettinger & Thomas Steer (eds.), Das Nomen im Indogermanischen: Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

369

Mor­phologie, Substantiv versus Adjektiv, Kollektivum. Akten der Arbeitstagung der Indogermanischen Gesellschaft vom 14. bis 16. September 2011 in Erlangen. Wiesbaden: Reichert, 65–77. 2014c Tocharian special agents: the nt-participles. Tocharian and Indo-European Studies 15: 53–67. 2014d The Tocharian adjectives in B -tse, A -ts. In: Stephanie W. Jamison, H. Craig Melchert & Brent Vine (eds.), Proceedings of the 25th Annual UCLA IndoEuropean Conference. October 25th and 26th, 2013. Bremen: Hempen, 49–56. 2017a The Tocharian gerundives in B -lle, A -l. In: Bjarne S. S. Hansen et al. (eds.), Usque ad Radices. Indo-European Studies in Honour of Birgit Anette Olsen. Copenhagen: Museum Tusculanum Press, 149–59. 2017b The syntax and semantics of agent formations in Tocharian. In: Claire Le Feuvre, Daniel Petit & Georges-Jean Pinault (eds.), Verbal Adjectives and Participles in Indo-European Languages. Proceedings of the conference of the Society for Indo-European Studies (Indogermanische Gesellschaft), Paris, 24th to 26th September 2014. Bremen: Hempen Verlag, 73–84. 2018 Notes on verbal governing compounds in Tocharian. In: David M. Goldstein, Stephanie W. Jamison & Brent Vine (eds.), Proceedings of the 26th Annual UCLA Indo-European Conference. November 11th and 12th, 2016. Bremen: Hempen, 53–69. Fellner, Hannes A. & Laura Grestenberger 2016 Greek and Latin verbal governing compounds in *-ā and their prehistory. In: Bjarne Simmelkjær Sandgaard Hansen, Benedicte Nielsen Whitehead, Thomas Olander & Birgit Anette Olsen (eds.), Etymology and the European Lexicon. Proceedings of the 14th Fachtagung der Indogermanischen Gesellschaft, 17–22 September 2012, Copenhagen. Wiesbaden: Reichert, 135–149. Filippone, Ela 2017 On the Meaning of Avestan nāuuiia- and Pahlavi *nāydāg. In: Enrico Morano, Elio Provasi & Adriano V. Rossi (eds.), Studia Philologica Iranica. Gherardo Gnoli Memorial Volume. Roma: ISMEO. Filliozat, Jean 1948 Fragments de textes koutchéens de médecine et de magie. Textes, parallèles sanskrits et tibétains, traduction et glossaire. Paris: Librairie d’Amérique et d’Orient Adrien Maisonneuve. Fraenkel, Ernst 1932 Zur tocharischen Grammatik. Indogermanische Forschungen 50: 220–231. 1962 Litauisches etymologisches Wörterbuch. Vol. I. Heidelberg: Winter. Friis, Louise S. 2021 Tocharian B agent nouns in -ntsa and their origin. Indo-European Linguistics 9: 1–25.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

370

Bibliography

Gąsiorowski, Piotr 2017 Cherchez la femme: Two Germanic suffixes, one etymology. Folia Lingui­ stica Historica 38: 125–147. Geng, Shimin, Jens-Peter Laut & Georges-Jean Pinault 2004a Neue Ergebnisse der Maitrisimit-Forschung (I). Zeitschrift der Deutschen Morgenländischen Gesellschaft 154: 347–369. 2004b Neue Ergebnisse der Maitrisimit-Forschung (II): Struktur und Inhalt des 26. Kapitels. Studies on the Inner Asian Languages 19: 29–94 + III. Gershevitch, Ilya 1961 A grammar of Manichean Sogdian. Oxford: Blackwell. gew 1954–72 Hjalmar Frisk. Griechisches Etymologisches Wörterbuch. 3 vols. Heidelberg: Winter. Gholami, Saloumeh 2009 Ergativity in Bactrian. Orientalia Suecana 58: 132–141. 2014 Selected Features of Bactrian Grammar. Wiesbaden: Harrassowitz. Giacalone Ramat, Anna 1998 Testing the boundaries of grammaticalization. In: Anna Giacalone Ramat & Paul J. Hopper (eds.), The Limits of Grammaticalization. Amsterdam: Benjamins, 107–128. Goetze, Albrecht 1937 Transfer of consonantal stems to the thematic declensions in Hittite. In: Louis Hjelmslev et al. (eds.), Mélanges linguistiques offerts à M. Holger Pedersen à l’occasion de son soixante-dixième anniversaire, 7 avril 1937. København: Universitetsforlaget i Aarhus, 488–495. 1960 Suffixes in “Kanishite” proper names. Revue hittite et asianique 18: 66–67. Gōto, Toshifumi 2013 Old Indo-Aryan Morphology and its Indo-Iranian Background. In co-operation with Jared S. Klein and Velizar Sadovski. Wien: Verlag der Österreichischen Akademie der Wissenschaften. Greenberg, Joseph H. 1963 Some universals of grammar with particular reference to the order of meaningful elements. In: Joseph H. Greenberg (ed.), Universals of Language. Cambridge: MIT Press, 73–113. Gusmani, Roberto 1985 Lydisch kãna- und luwisch wana-. In: Hermann M. Ölberg & Gernot Schmidt (eds.), Sprachwissenschaftliche Forschungen. Festschrift für Johann Knobloch zum 65. Geburtstag am 5. Januar 1984 dargebracht von Freunden und Kollegen. Innsbruck: Institut für Sprachwissenschaft, 127–132.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

371

Hackstein, Olav 1995 Untersuchungen zu den sigmatischen Präsensstammbildungen des Tochari­ schen. Göttingen: Vandenhoeck & Ruprecht. 2002 Uridg. *CH.CC > *C.CC. Historische Sprachforschung 155: 1–22. 2003 Reflexivpronomina, Präverbien und Lokalpartikel in den indogermanischen Sprachen. Tocharian and Indo-European Studies 10: 69–95. 2004 Rhetorical questions and the grammaticalization of interrogative pronouns as conjunctions in Indo-European. In: Adam Hyllested, Anders Richard Jørgensen, Jenny Helena Larsson & Thomas Olander (eds.), Per aspera ad asteriscos. Studia Indogermanica in honorem Jens Elmegård Rasmussen sexagenarii Idibus Martiis anno MMIV. Innsbruck: Institut für Sprachwissenschaft, 167–186. 2010 Apposition and nominal classification in Indo-European and beyond. Wien: Verlag der Österreichischen Akademie der Wissenschaften. 2012 Collective and feminine in Tocharian. In: Olav Hackstein & Ronald Kim (eds.), Linguistic developments along the Silkroad: Archaism and innovation in Tocharian. Wien: Verlag der Österreichischen Akademie der Wissenschaften, 143–177. 2013 Indogermanisch *h1k̑-u̯ -o-s, *h1ek̑-u̯ -o-s ‚Pferd, Hengst, Stute’: Genusindifferenz als morphologische Persistenz. In: Adam I. Copper, Jeremy Rau & Michael Weiss (eds.), Multi Nominis Grammaticus. Studies in Classical and IndoEuropean in honor of Alan J. Nussbaum on the occasion of his sixty-fifth birthday. Ann Arbor–New York: Beech Stave Press, 94–104. 2017 The phonology of Tocharian. In: Jared Klein, Brian Joseph, Matthias Fritz & Mark Wenthe (eds.), Handbook of Comparative and Historical Indo-European Linguistics. Vol. II. Berlin–Boston: de Gruyter, 1304–1335. Hackstein, Olav, Hiromi Habata & Christoph Bross 2015 Tocharische Texte zur Buddhalegende I. Die Geschichte von der Nonne Sundarī B15–18. Münchener Studien zur Sprachwissenschaft 68: 31–99. 2019 Tocharische Texte zur Buddhalegende. München: Verlag J. H. Röll. Hajnal, Ivo 1998 Mykenisches und homerisches Lexikon. Übereinstimmungen, Divergenzen und der Versuch einer Typologie. Innsbruck: Institut für Sprachwissenschaft. 2004 Zur Genese agglutinierender Flexionsmuster im Tocharischen: die Adjektive auf B -ṣṣe / A -ṣi. In: Peter Anreiter, Marialuise Haslinger & Heinz Dieter Pohl (eds.), Artes et Scientiae: Festschrift für Ralf-Peter Ritter zum 65. Geburtstag. Wien: Edition Praesens, 137–157. 2005 Die Flexion der ah2-Stämme im Tocharischen: ererbt oder geneuert? In: Günter Schweiger (ed.), Indogermanica. Festschrift Gert Klingenschmitt. Taimering: Schweiger VWT, 221–245.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

372

Bibliography

Hall, Robert A. 1965 The ‘Neuter’ in Romance: A pseudo-problem. Word 21: 421–427. Hamp, Eric P. 1959 The Germanic word for ‘tear’. Beiträge zur Geschichte der deutschen Sprachen und Literatur 81: 146–153. 1972 Latin dacrima, lacruma and Indo-European ‘tear’. Glotta 50: 291–299. 1998 Whose were the Tocharians? Linguistics subgrouping and diagnostic idiosyncrasy. In: Victor H. Mair (ed.), The Bronze Age and Early Iron Age peoples of Eastern Central Asia. Vol. 1. Archeology, migration and nomadism, linguistics. (JIES Monogr. 26). Washington, DC: Institute for the Study of Man, 307–346. Härtel, Herbert 1956 Karmavācanā: Formulare für den Gebrauch im buddhistischen Gemeindeleben aus ostturkistanischen Sanskrit-Handschriften. Sanskrittexte aus den Turfan­ funden III. Deutsche Akademie der Wissenschaften zu Berlin. Institut für Orientforschung, Veröffentlichung Nr. 30. Berlin: Akademie-Verlag. Harðarson, Jón Axel 1987 Zum urindogermanischen Kollektivum. Münchener Studien zur Sprach­ wissenschaft 48: 71–113. 2014 Das andere Wort für ‘Frau’ im Urindogermanischen. In: Sergio Neri & Roland Schumann (eds.), Studies on the Collective and Feminine in Indo-European from Diachronic and Typological Perspective. Leiden–Boston: Brill, 23–55. Hartmann, Markus 2013 Das Genussystem des Tocharischen. Hamburg: Baar. Haspelmath, Martin & Uri Tadmor (eds.) 2009 World Loanword Database. Leipzig: Max Planck Institute for Evolutionary Anthropology. Hegedűs, Irén 2002 Proto-Nuristani *kurka- ‘young of an animal’. In: Fabrice Cavoto (ed.), The Linguist’s Linguist. A Collection of Papers in Honour of Alexis Manaster Ramer. München: Lincom Europa, 189–195. Hilmarsson, Jörundur G. 1984a Reconstruction of a Tocharian paradigm: the numeral ‘one’. Zeitschrift für vergleichende Sprachforschung 97: 135–147. 1984b East Tocharian śorkäm ‘peg’ or ‘string’? Die Sprache 30: 16–28. 1985a Tocharian B kroriyai (obl. sg.), A kror ‘crescent, horn of the moon’ ~ Hitt. karawar ‘horn’ ~ Arm. ełǰewr ‘horn’ < I.-E. *ghrēu̯ r̥. Die Sprache 31: 40–47. 1985b Toch. A kāc, Lat. cutis, OIcel. húð < I.-E. *kuHtis ‘skin’. Zeitschrift für vergleichende Sprachforschung 98: 162–3. 1986a Studies in Tocharian Phonology, Morphology and Etymology with special emphasis on the o-vocalism. Diss. Universiteit Leiden.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

373

1986b Toch. A ṣtām, B stām, pl. stāna ‘tree’. Zeitschrift für vergleichende Sprach­ forschungen 99: 147–149. 1986c East Tocharian klop ‘misfortune, suffering, pain’. In: Eivind Kahrs (ed.), Layānamitrārāgaṇam. Essays in Honour of Nils Simonsson. Instituttet for sammenlignende kulturforskning. Series B: Skrifter LXX. Oslo, 105–114. 1986d A Tocharian-Greek isogloss: Toch. B ñor ‘below’ ~ Gk. νείαιρα ‘lower’. Glotta 64: 1–9. 1986e The etymology of Toch. A kur-, B kur- / kwr- ‘to grow old, decrepit’ and their derivatives A oṅkrac (indecl.), B oṅkrocce (obl.sg.m.) ‘immortal, eternal’. Münchener Studien zur Sprachwissenschaft 47: 87–98. 1987a The element -ai(-) in the Tocharian nominal flexion. Die Sprache 33: 34–55. 1987b On the History and Distribution of Suffixal -y-/-iy- in Tocharian. Die Sprache 33: 79–93. 1987c Reflexes of I.-E. *suH2n̥ to-/-ōn ‘sunny’ in Germanic and Tocharian. Die Sprache 33: 56–78. 1988a Tocharian B okt, A okät ‘8’ and the development of final Indo-European *-ō in Tocharian. In: Yoël L. Arbeitman (ed.), A Linguistic Happening in Memory of Ben Schwartz. Studies in Anatolian, Italic, and other Indo-European Languages. Louvain-la-neuve: Peeters, 505–519. 1988b Tocharian B yapoy, A ype ‘land’. Tocharian and Indo-European Studies 2: 31–51. 1988c West Tocharian lyauto ‘hole, opening’ and related forms. Historische Sprachforschung 101: 166–169. 1989a The Dual Forms of Nouns and Pronouns in Tocharian. Reykjavík: Málvís­inda­ stofnun Háskóla Íslands. 1989b West Tocharian särwāna ‘face’. Tocharian and Indo-European Studies 3: 77–90. 1991a The Nasal Prefixes in Tocharian: A Study in Word Formation. Reykjavík: Málvísindastofnun Háskóla Íslands. 1991b Tocharian etymological notes 1–13. Tocharian and Indo-European Studies 5: 137–183. 1991c The elements -ñ- and -ññ- in Tocharian present and subjunctive classes. Tocharian and Indo-European Studies 5: 61–122. 1991d The verb B sain-, A se-n-, in Tocharian”. In: Li Zheng et al. (eds.), Papers in Honour of Prof. Dr. Ji Xianlin on the Occasion of His 80th Birthday, Vol. I. Nanchang: Jiangxi renmin chubanshe, 67–90. 1996 Materials for a Tocharian Historical and Etymological Dictionary. Reykjavík: Málvísindastofnun Háskóla Íslands. von Hinüber, Oskar 2001 Das ältere Mittelindisch im Überblick. Wien: Verlag der Österreichischen Akademie der Wissenschaften.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

374

Bibliography

Hirt, Hermann Alfred 1912 Zur Bildung auf -ī im Indogermanischen. Indogermanische Forschungen 31: 1–23. 1927 Indogermanische Grammatik. Band III. Das Nomen. Heidelberg: Winter. Hitch, Doug 2016 The Old Khotanese Metanalysis. Diss. Harvard University. 2017 Old Khotanese type A stems in -a- and -e-. Bulletin of School of Oriental and African Studies 80: 491–523. Hock, Hans H. 1991 Principles of Historical Linguistics. Berlin–New York: de Gruyter. Hockett, Charles F. 1958 A Course in Modern Linguistics. New York: Macmillan. Hoernle, August Friedrich Rudolf 1902 Fac-simile reproduction of Weber MSS., Part IX and Macartney MSS., Set I. With Roman transliteration and indexes. Calcutta: Baptist Mission Press. Hoffmann, Roland 2016 Gender in Latin and in language typology. In: Paolo Poccetti (eds.), Latinitatis Rationes. Descriptive and Historical Accounts for the Latin Language. Berlin– Boston: de Gruyter, 820–839. Hoffmann, Karl & Forssman, Bernhard 2004 Avestische Laut- und Flexionslehre. Innsbruck: Institut für Sprachwissenschaft. Höfler, Stefan 2012 Untersuchungen zum Ablaut der neutralen s-Stämme des Indogermanischen. MA thesis, Universität Wien. 2015 Denominale Sekundärderivation im Indogermanischen: Eine Ochsentour. Münchener Studien zur Sprachwissenschaft 69: 219–244. Hübschmann, Helmut 1897 Armenische Etymologie. Lipsia: Breitkopf & Hartel. Huard, Athanaric 2020 On Tocharian B kents* and the origin of PIE *ĝʰans-. Wékwos 5: 215–262. 2022 Recherches sur les textes de méditation en tokharien. Thèse de doctorat, Université Paris Sciences et Lettres, École Pratique des Hautes Études. Humbach, Helmut 1974 Problems of Mihr Yašt in the Light of Philological Evidence. In: Richard N. Frye (ed.), Neue Methodologie in der Iranistik. Wiesbaden: Harrassowitz, 85–92. 1991 The Gāthās of Zarathushtra and the other Old Avestan Texts. Part I: Introduction, Text and Translation. Part II: Commentary. Heidelberg: Winter. iew 1959 Julius Pokorny, Indogermanisches Etymologisches Wörterbuch. Bern– München: Francke.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

375

Igartua, Iván 2006 Genus alternans in Indo-European. Indogermanische Forschungen 111: 56–70. Isebaert, Lambert 1980 De Indo-Iraanse bestanddelen in de Tocharische woordenschat. Vraagstukken van fonische productinterferentie, met bijzondere aandacht voor de IndoIraanser diafonen a, ā. Diss. Katholieke Universiteit te Leuven. 2004 La formation du hittite partāu̯ ar ‘aile, plume (d’oiseau)’. In: Michel Mazoyer & Olivier Casabonne (eds.), Antiquus Oriens. Mélanges offerts au Professeur René Lebrun (Volume I). Paris: L’Harmattan, 393–399. Itkin, Ilija B. 2019 Ukazatel’ slovoform k neopublikovannym toxarskim A tekstam iz sobranija Berlinskoj Biblioteki. Moskva: Institut Vostokovedenija, Rossijskaja Akademija Nauk. Jasanoff, Jay H. 1978 Stative and middle in Indo-European. Innsbruck: Institut für Sprachwissenschaft. 1989 Old Irish bé ‘woman’. Ériu 40: 135–141. 1989a Language and gender in the Tarim basin: the Tocharian 1sg. pronoun. Tocharian and Indo-European Studies 3: 125–147. 2003 Hittite and the Indo-European verb. Oxford: Oxford University Press. 2017 The impact of Hittite and Tocharian: Rethinking Indo-European in the 20th century and beyond. In: Jared Klein, Brian Joseph, Matthias Fritz & Mark Wenthe (eds.), Handbook of Comparative and Historical Indo-European Linguistics. Vol. I. Berlin–Boston: de Gruyter, 220–238. 2018 The phonology of Tocharian B okso ‘ox’. In: Lucien van Beek, Alwin Kloekhorst, Guus Kroonen, Michaël Peyrot, Tijmen Pronk & Michiel de Vaan (eds.), Farnah. Indo-Iranian and Indo-European Studies in Honor of Sasha Lubotsky. Ann Arbor: Beech Stave, 72–78. 2019 Some difficult Tocharian genitives. Indogermanische Forschungen 124: 47–59. Ji, Xianlin 1998 Fragments of the Tocharian A Maitreyasamiti-Nāṭaka of the Xinjiang Museum, China. Transliterated, translated and annotated by Ji Xianlin in collaboration with Werner Winter, Georges-Jean Pinault. Berlin–New York: de Gruyter. Johnsen, Sverre 2005 The Germanic (i)jō-stem declension. Origin and development. MA thesis, Universitetet i Oslo. Kahane, Henry & Renée Kahane 1949 The augmentative feminine in the Romance languages. Romance Philology 2: 135–137.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

376

Bibliography

Katz, Joshua T. 1997 Ein tocharisches Lautgesetz für Monosyllaba. Tocharian and Indo-European Studies 7: 61–87. Kellens, Jean 1974 Les noms-racines de l’Avesta. Wiesbaden: Reichert. 1994 Le pantheon de l’Avesta ancien. Wiesbaden: Reichert. 2009 Eaux stagnantes. Indo-Iranian Journal 52: 331–334. Kellens, Jean & Eric Pirart 1988–94 Les textes vieil-avestiques. 3 vols. Wiesbaden: Reichert. kewa 1956–80 Mayrhofer, Manfred. Kurzgefasstes etymologisches Wörterbuch des Altin­ doarischen. Heidelberg: Winter. Kibort, Anna 2010 Towards a typology of grammatical features. In: Anna Kibort & Greville G. Corbett (eds.), Features. Perspectives on a Key Notion in Linguistics. Oxford: Oxford University Press, 64–106. Kilarski, Marcin 2013 Nominal Classification: A history of its study from the classical period to the present. Amsterdam–Philadelphia: John Benjamins. Kim, Ronald I. 1999 The development of labiovelars in Tocharian: a closer look. Tocharian and Indo-European Studies 8: 139–187. 2005 Ossetic silæ/syl and the Indo-Iranian word for ‘female’. International Journal of Diachronic Linguistics and Linguistic Reconstruction 2: 123–168. 2006 Tocharian. In: Keith Brown (ed.), Encyclopedia of Language & Linguistics. Second Edition. Elsevier Ltd, 725–727. 2007a The Tocharian subjunctive in light of the h2e-conjugation model. In: Alan J. Nussbaum (ed.), Verba Docenti. Studies in historical and IndoEuropean linguistics presented to Jay. H. Jasanoff by students, colleagues, and friends. Ann Arbor–New York: Beech Stave Press, 185–200. 2007b Vowel weakening in Tocharian preterite participles and abstract nouns. Kyoto University Linguistic Research 26: 1–30. 2007c The duke of York comes to Xinjiang: Ablaut, analogy, and epenthesis in Tocharian nasal presents. Historische Sprachforschung 12o: 66–104. 2009a The feminine gender in Tocharian and Indo-European. In: Kazuhiko Yoshida & Brent Vine (eds.), East and West: Papers in Indo-European Studies. Bremen: Hempen, 69–87. 2009b Another Look at Tocharian B ṣarya. Historische Sprachforschung 122: 111–117.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

377

2012 The PIE thematic animate accusative plural revisited. In: Roman Sukač & Ondřej Šefčík (eds.), The sound of Indo-European 2. Papers on Indo-European Phonetics, Phonemics and Morphophonemics. Munich: LINCOM, 144–158. 2014 A tale of two suffixes: *-h2 and *-ih2, and the evolution of feminine gender in Indo-European. In: Sergio Neri & Roland Schumann (eds.), Studies on the Collective and Feminine in Indo-European from Diachronic and Typological Perspective. Leiden–Boston: Brill, 115–136. 2018a The Dual in Tocharian. From Typology to Auslautgesetz. München: Verlag J. H. Röll. 2018b One hundred years of re-construction: Hittite, Tocharian, and the continuing revision of Proto-Indo-European. In: Elisabeth Rieken, Ulrich Geupel & Theresa Maria Roth (eds.), 100 Jahre Entzifferung des Hethitischen. Morphosyntaktische Kategorien in Sprachgeschichte und Forschung. Akten der Arbeitstagung der Indogermanischen Gesellschaft vom 21. bis 23. September 2015 in Marburg. Wiesbaden: Reichert Verlag, 157–178. 2019a North Slavic -ě vs. South Slavic -ę: a problem of forward reconstruction. Journal of Slavic Linguistics 27: 1–26. 2019b The derivational history of Tocharian B war, A wär ‘water’. In: Dieter Gunkel, Stephanie W. Jamison, Angelo O. Mercado & Kazuhiko Yoshida (eds.), Vina Diem Celebrent. Studies in Linguistics and Philology in Honor of Brent Vine. Ann Arbor–New York: Beech Stave, 141–151. 2019c Adjectival suppletion in Tocharian. In: Ronald I. Kim (ed.), Diachronic Perspectives on Suppletion. Hamburg: Baar, 179–200. 2021 Remarks on linguistic variation and change in Tocharian A. In: Hannes A. Fellner, Melanie Malzahn & Michaël Peyrot (eds.), lyuke wmer ra. IndoEuropean Studies in Honor of Georges-Jean Pinault. Ann Arbor–New York: Beech Stave Press, 252–267. Klingenschmitt, Gert 1968 Farhang-i ōīm. Edition und Kommentar. Diss. Erlangen–Nürnberg. 1975 Tocharisch und Urindogermanisch. In: Helmut Rix (ed.), Flexion und Wortbildung. Akten der V. Fachtagung der Indogermanischen Gesellschaft, Regensburg, 9.–14. September 1973. Wiesbaden: Reichert, 148–163. 1994 Das Tocharische in indogermanistischer Sicht. In: Bernfried Schlerath (ed.), Tocharisch. Akten der Fachtagung der Indogermanischen Gesellschaft, Berlin, September 1990. Reykjavík: Málvísindastofnun Háskóla Íslands, 310–411. Kloekhorst, Alwin 2008 Etymological Dictionary of the Hittite Inherited Lexicon. Leiden: Brill. 2011 Weise’s Law: Depalatalization of palatovelars before *r in Sanskrit. In: Thomas Krisch & Thomas Lindner (eds.), Indogermanistik und Linguistik im Dialog.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

378

Bibliography

Akten der XIII. Fachtagung der Indogermanischen Gesellschaft vom 21. bis 27. September 2008 in Salzburg. Wiesbaden: Reichert, 261–270. 2013 Indo-European nominal ablaut patterns: The Anatolian evidence. In: Götz Keydana, Paul Widmer & Thomas Olander (eds.), Indo-European Accent and Ablaut. Copenhagen: Museum Tusculanum Press, 107–128. 2019 Kanišite Hittite. The Earliest Attested Record of Indo-European. Leiden–Boston: Brill. 2022 Anatolian. In: Thomas Olander (ed.), The Indo-European Language Family. A Phylogenetic Perspective. Cambridge: Cambridge University Press, 63–82. Kloekhorst, Alwin & Tijmen Pronk 2019 Introduction: Reconstructing Proto-Indo-Anatolian and Proto-Indo-Uralic. In: Alwin Kloekhorst & Tijmen Pronk (eds.), The Precursors of Proto-IndoEuropean. The Indo-Anatolian and the Indo-Uralic hypotheses. Boston: Brill– Rodopi, 1–14. Knoll, Gabriele 1996 Die Verwendungsweisen der Adjektive im Tocharischen. Diss. Universität Frankfurt-am-Main. Konow, Sten 1932 Saka Studies. Oslo Etnografiske Museum. Bulletin 5. Oslo. Kortlandt, Frederik H. H. 1983 Demonstrative pronouns in Balto-Slavic, Armenian, and Tocharian. In: André G. F. van Holk (ed.), Dutch contributions to the Ninth International Congress of Slavists. Kiev, September 6–14, 1983. Linguistics. Amsterdam: Rodopi, 311–322. 1985a Arm. artawsr ‘tear’. Annual of Armenian Linguistics 6: 59–61. 1985b Long vowels in Balto-Slavic. Baltistica 21: 112–124. 1988a On the development of PIE final syllables in Tocharian. Tocharian and Indo-European Studies 2: 80–88. 1988b The Tocharian word for ‘woman’. Tocharian and Indo-European Studies 2: 77–79. 2010 Studies in Germanic, Indo-European, and Indo-Uralic. Amsterdam–New York: Rodopi. 2013 The development of the Tocharian vowel system. Tocharian and Indo-European Studies 14: 95–104. 2016 What is Stang’s Law? Baltistica 52: 73–80. 2017 On the origin of grammatical gender. Tocharian and Indo-European Studies 18: 95–104. 2020 The origin of grammatical gender in Tocharian and Indo-European. Tocharian and Indo-European Studies 20: 113–124.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

379

Krause, Wolfgang 1951 Zur Frage nach dem nichtindogermanischen Substrat des Tocharischen. Zeitschrift für vergleichende Sprachforschung 69: 185–203. 1952 Westtocharische Grammatik. Band I: Das Verbum. Heidelberg: Winter. 1954 Das Numerussystem im Tocharischen. Nachrichten der Akademie der Wissen­ schaften in Göttingen aus dem Jahre 1954, Philologisch-Historische Klasse, Nr. 1. Göttingen: Vandenhoeck & Ruprecht. 1955 Tocharisch. Handbuch der Orientalistik, vierter Bd. Iranistik, dritter Abschnitt. Leiden: Brill. 1971 Tocharisch. Handbuch der Orientalistik, vierter Bd. Iranistik, dritter Abschnitt, Photomechanischer Nachdruck mit Zusätzen und Berichtigungen. Leiden– Köln: Brill. Kroonen, Guus 2011 The Proto-Germanic n-stems: A Study in Diachronic Morphophonology. Amsterdam–New York: Rodopi. 2013 Etymological Dictionary of Proto-Germanic. Leiden–Boston: Brill. Kudara, Kōgi 1974 Tokara-go B-ni yoru abidaruma ronsho kankei-no danpen-ni tsuite. I. Abhidharmāvatāra-prakaraṇa chū [Fragments of the Abhidharma texts in Tocharian B: I. Commentary on the Abhidharmāvatāra-prakaraṇa]. In: T. Inokuchi (ed.), Saiiki-ni okeru kyōten rutsū-no shomondai [Some aspects of the Buddhist sūtras from Central Asia], 21–36. Kuiper, Franciscus Bernardus Jacobus 1942 Notes on Vedic noun-inflection. Amsterdam. 1978 On Zarathustra’s Language. Amsterdam–London: Noord-Hollandsche Uitgeversmaatschappij. Kumamoto, Hiroshi 2000 The concluding Verses of a ‘Lyrical Poem in Khotanese’. In: Minoru Hara, Ryutaro Tsuchida & Albrecht Wezler (eds.), Harānandalaharī: volume in honour of Professor Minoru Hara on his seventieth birthday. Reinbeck: Dr. Inge Wezler Verlag für Orientalistische Fachpublikationen, 143–154. 2009 The Maitreya-samiti and Khotanese. Kuryłowicz, Jerzy 1927 Les effets du ə en indoiranien. Prace filologiczne 11: 201–243. Kümmel, Martin Joachim 2009 The range of Tocharian a-umlaut. In: Jens E. Rasmussen & Thomas Olander (eds.), Internal Reconstruction in Indo-European: Methods, Results, and Problems. Section Papers from the XVI International Conference on Historical Linguistics, University of Copenhagen, 11th–15th August 2003. Copenhagen: Museum Tusculanum Press, 171–179.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

380

Bibliography

2015

About the demonstrative system in Tocharian B. In: Melanie Malzahn, Michaël Peyrot, Hannes A. Fellner & Theresa S. Illés (eds.), Tocharian Texts in Context. International Conference on Tocharian Manuscripts and Silk Road Culture, June 25–29th, 2013. Bremen: Hempen, 109–116. Lane, George Sherman 1938 Problems of Tocharian Phonology. Language 14: 20–38. 1966 On the interrelationships of the Tocharian dialects. In: Henrik Birnbaum & Jaan Puhvel (eds.), Ancient Indo-European dialects. Proceedings of the conference on Indo-European linguistics, held at the University of California, Los Angeles, April 25–27, 1963. Berkeley: University of California, 213–233. 1976 Notes sur le sort des syllables finales i.e. en tokharien. Bulletin de la Société de Linguistique de Paris 71: 133–159. Lass, Roger 1990 How to do things with junk: Exaptation in language evolution. Journal of Linguistics 26: 79–102. Laut, Jens Peter & Jens Wilkens 2017 Alttürkische Handschriften Teil 3: Die Handschriftenfragmente der Maitrisimit aus Sängim und Murtuk in der Berliner Turfansammlung. Stuttgart: Franz Steiner. Lazzeroni, Romano 1968 Per una definizione dell’unità indoiranica. Studi e Saggi Linguistici 8: 131–159. 1992 Genere grammaticale e sistema nominale nel sanscrito: per una interpreta­ zione della deriva. Archivio Glottologico Italiano 77: 193–205. 1998a Il movimento come vita: una tassonomia vedica e iguvina. In: Romano Lazzeroni (ed.), La cultura indoeuropea. Laterza: Bari, 13–23. 1998b La tassonomia dimenticata: il nome greco dello schiavo. In: Romano Lazzeroni (ed.), La cultura indoeuropea. Laterza: Bari, 24–30. 1998c Le cose viventi: i nomi dell’acqua e del fuoco e il genere grammaticale. In: Romano Lazzeroni (ed.), La cultura indoeuropea. Laterza: Bari, 33–41. 2002 Il nome greco del sogno e il neutro indoeuropeo. Archivio Glottologico Italiano 87: 145–162. Ledo-Lemos, Francisco 2003 Feminine Genus. A study on the origins of the Indo-European feminine grammatical gender. München–New Castle: LINCOM Europa. Leukart, Alex 1987 po-ro-qa-ta-jo, to-sa-pe-mo, a-mo-ra-ma and others: further evidence for Proto-Greek collective formations in Mycenaean and early alphabetic Greek. Minos 20: 343–365. 2003 Die frühgriechischen Nomina auf -tās und -ās. Wien: Verlag der Österreichischen Akademie der Wissenschaften.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

381

Leumann, Manu 1977 Lateinische Laut- und Formenlehre. München. Lévi, Sylvain 1933 Fragments de textes koutchéens. Paris: Impremerie Nationale. Lidén, Evald 1916 Studien zur tocharischen Sprachgeschichte. Göttingen: Wetergren & Kerber. Lipp, Reiner 2009 Die indogermanischen und einzelsprachlichen Palatale im Indo-iranischen, 2 vols. Heidelberg: Winter. lipp 2014 Dunkel, George E. Lexikon der indogermanischen Partikeln und Pronominalstämme. 2 vols. Heidelberg: Winter. liv 2 2011 Lexikon der indogermanischen Verben. Die Wurzeln und ihre Primärstamm­ bildungen. Unter Leitung von Helmut Rix und der Mitarbeit vieler anderer bearbeitet von Marin Kümmel, Thomas Zehnder, Reiner Lipp, Brigitte Schirmer. Zweite, erweiterte und verbesserte Auflage bearbeitet von Martin Kümmel und Helmut Rix. Wiesbaden: Reichert. Lohmann, Johannes 1932 Genus und Sexus. Eine Morphologische Studie zum Ursprung der indogermanischen nominalen Genus-Unterscheidung. Göttingen: Vandenhoeck und Ruprecht. Loporcaro, Michele 2016 Gender. In: Adam Ledgeway & Martin Maiden (eds.), The Oxford Guide to the Romance languages. Oxford: Oxford University Press, 924–935. 2018 Gender from Latin to Romance: History, Geography, Typology. Oxford: Oxford University Press. Loporcaro, Michele & Tania Paciaroni 2011 Four gender system in Indo-European. Folia Linguistica 45: 389–433. Loporcaro, Michele, Vincenzo Faraoni & Francesco Gardani 2013 Manifestazioni del neutro nell’italo-romanzo medievale. In: Emili Casanova Herrero & Cesáreo Rigual Calvo (eds.), Actas del XXVI congreso internacional y de lingüística y de filología románicas. Valencia 2010. Berlin: de Gruyter, 171–182. 2014 The third gender of Old Italian. Diachronica 31: 1–22. Lubotsky, Alexander 1988 Tocharian A ṣurm, B ṣarm ‘cause’ and A ṣul, B ṣale ‘mountain’. Tocharian and Indo-European Studies 2: 89–95. 1994a The original paradigm of the Tocharian word for ‘king’. In: Bernfried Scherath (ed.), Tocharisch. Akten der Fachtagung der Indogermanischen Gesellschaft, Berlin, September 1990. Reykjavík: Málvísindastofnun Háskóla Íslands, 66–72.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

382

Bibliography

1994b Avestan θβōrəštar- and the Indo-European root √turḱ-. Die Sprache 36: 92–102. 2001 Reflexes of Proto-Indo-European *sk in Indo-Iranian. Incontri Linguistici 24: 25–57. 2013 The Vedic paradigm for ‘water’. In: Adam I. Copper, Jeremy Rau & Michael Weiss (eds.), Multi Nominis Grammaticus. Studies in Classical and Indo-European in honor of Alan J. Nussbaum on the occasion of his sixty-fifth birthday. Ann Arbor–New York: Beech Stave Press, 159–164. Lubotsky, Alexander & Sergei Starostin 2003 Turkic and Chinese loanwords in Tocharian. In: Brigitte L. M. Bauer & Georges-Jean Pinault (eds.), Language in time and space. A Festschrift for Werner Winter on the occasion of his 80th birthday. Berlin–New York: de Gruyter, 257–269. Luján, Eugenio Ramón 1996 Sobre la etimología de uxor. Cuadernos de Filología Clásica. Estudios latinos 10: 21–28. Lundquist, Jesse & Anthony Yates 2018 The morphology of Proto-Indo-European. In: Jared Klein, Brian Joseph, Matthias Fritz & Mark Wenthe (eds.), Handbook of Comparative and Historical Indo-European Linguistics. Vol. III. Berlin–Boston: de Gruyter, 2079–2175. Luraghi, Silvia 1998 The Anatolian Languages. In: Anna Giacalone Ramat & Paolo Ramat (eds.), The Indo-European languages. London: Routledge, 169–196. 2006 La nascita del genere femminile in indoeuropeo. In: Silvia Luraghi & Anna Olita (eds.), Linguaggio e genere. Grammatica e usi. Roma: Carocci, 89–106. 2009a The origin of the feminine gender in PIE: An old problem in a new perspective. In: Vit Bubenik, John Hewson & Sarah Rose (eds.), Grammatical change in Indo-European languages. Amsterdam–Philadelphia: John Benjamins, 3–13. 2009b Indo-European nominal classification: from abstract to feminine. In: Stephanie W. Jamison, H. Craig Melchert & Brent Vine (eds.), Proceedings of the 20th Annual UCLA Indo‑European Conference. October 31st and November 1st, 2008. Bremen: Hempen, 115–131. 2011 The origin of the Proto-Indo-European gender system: Typological considerations. Folia Linguistica 45: 435–464. 2014 Gender and word formation: The PIE gender system in cross-linguistic perspective. In: Sergio Neri & Roland Schumann (eds.), Studies on the Collective and Feminine in Indo-European from Diachronic and Typological Perspective. Leiden–Boston: Brill, 199–231. Macdonell, Arthur Anthony 1910 Vedic Grammar. Straßburg: Trübner.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

383

Machek, Václav 1957 Etymologický slovník jazyka českého. Praha: Lidové noviny. Maggi, Mauro 2016 Khotanese aysmua- and other souls. Faṣlnāme-ye taxaṣṣoṣi-ye zabān va katibe, Viženāme-ye Manṣur Šaki [Quarterly journal of language and inscription 1.1, Dedicated to Professor Mansour Shaki], 1–24. Maiden, Martin 2011 Morphological persistence. In: Martin Maiden, John Charles Smith & Adam Ledgeway (eds.), The Cambridge History of the Romance Languages, vol. 1. Structures. Cambridge: Cambridge University Press, 155–215, 699–706. 2016 The Romanian alternating gender in diachrony and synchrony. Folia Linguistica Historica 37: 111–144. Malandra, William 2002 Notes on the Avestan Grammatical Tradition. In: Madhav M. Deshpande & Peter E. Hook (eds.), Indian Linguistic Studies: Festschrift in Honor of George Cardona. Delhi: Motilal Banarsidass, 223–233. Mallory, James P. & Douglas Q. Adams 2006 The Oxford Introduction to Proto-Indo-European and the Proto-Indo-European world. Oxford: Oxford University Press. Malzahn, Melanie 2006 Westtocharische Substantive auf -au und einige Fortsetzer von idg. menStämmen im Tocharischen. In: Günter Schweiger (ed.), Indogermanica. Festschrift Gert Klingenschmitt. Indische, iranische und indogermanische Studien dem verehrten Jubilar dargebracht zu seinem fünfundsechzigsten Geburtstag. Taimering: Schweiger VWT, 389–407. 2007 Tocharian desire. In: Alan J. Nussbaum (ed.), Verba Docenti. Studies in Historical and Indo-European Linguistics presented to Jay. H. Jasanoff by Students, Colleagues, and Friends. Ann Arbor–New York: Beech Stave Press, 237–249. 2010 Tocharian Verbal System. Leiden–Boston: Brill. 2011 Speaking on tongue: the Tocharian B nouns with an oblique singular in -a. Tocharian and Indo-European Studies 12: 83–109. 2012a Now you see it, now you don’t  — Bewegliches -o in Tocharisch B. In: Olav Hackstein & Ronald Kim (eds.), Linguistic developments along the Silk Road: Archaism and Innovation in Tocharian. Wien: Verlag der Österreichischen Akademie der Wissenschaften, 33–82. 2012b Verbale Rektionskomposita im Tocharischen. In: Velizar Sadovski & David Stifter (eds.), Iranistische und indogermanistische Beiträge in memoriam Jochem Schindler (1944–1994). Wien: Verlag der Österreichischen Akademie der Wissenschaften, 113–35.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

384

Bibliography

2012c Review of Michaël Peyrot (2008). Kratylos 57: 176–182. 2013a Of demons and women TB yakṣa- and oppositional feminine forms in Tocharian. Tocharian and Indo-European Studies 14: 105–121. 2013b Cutting around “témos”: Evidence from Tocharian. In: Adam I. Copper, Jeremy Rau & Michael Weiss (eds.), Multi Nominis Grammaticus. Studies in Classical and Indo-European in honor of Alan J. Nussbaum on the occasion of his sixty-fifth birthday. Ann Arbor–New York: Beech Stave Press, 165–174. 2014a Tocharian A śorki ‘fear’ and two other TA scary words. Tocharian and Indo-European Studies 15: 87–94. 2014b Surprise at length of Tocharian nouns. Indogermanische Forschungen 119: 259–268. 2014c Das Kollektivum im Tocharischen. In: Norbert Oettinger & Thomas Steer (eds.), Das Nomen im Indogermanischen. Morphologie, Substantiv versus Adjektiv, Kollektivum. Akten der Arbeitstagung der Indogermanischen Gesellschaft vom 14. Bis 16. September 2011 in Erlangen. Wiesbaden: Reichert, 195–201. 2016 The second one to branch off? The Tocharian lexicon revisited. In: Bjarne Simmelkjær Sandgaard Hansen, Benedicte Nielsen Whitehead, Thomas Olander & Birgit Anette Olsen (eds.), Etymology and the European Lexicon. Proceedings of the 14th Fachtagung der Indogermanischen Gesellschaft, 17–22 September 2012, Copenhagen. Wiesbaden: Reichert, 281–292. 2017 Tocharian and Historical Sociolinguistics. Evidence from a fragmentary corpus. Open Linguistics 3: 157–177. Malzahn, Melanie & Hannes A. Fellner 2015 Lifting up the light: tläś and lkäś in Tocharian A. Tocharian and Indo-European Studies 16: 61–79. Marggraf, Jürgen 1970 Untersuchungen zum Akzent in Tocharisch B. Diss. Kiel. 1975 Bemerkungen zur historischen Phonologie und Morphologie einiger „primärer“ Kasusmorpheme im Tocharisch B. In: Helmut Rix (ed.), Flexion und Wortbildung: Akten der V. Fachtagung der Indogermanischen Gesellschaft. Regensburg 9.–14. September 1973. Wiesbaden: Reichert, 198–203. Martínez, Javier & Michiel de Vaan 2014 Introduction to Avestan. Leiden–Boston: Brill. Martinet, André 1956 Le genre féminin en indo-européen: examen fonctionnel du problème. Bulletin de la Société de Linguistique de Paris 52: 83–95. Martirosyan, Hrach K. 2010 Etymological Dictionary of the Armenian Inherited Lexicon. Leiden: Brill.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

385

Matasović, Ranko 2004 Gender in Indo-European. Heidelberg: Winter. 2009 Etymological Dictionary of Proto-Celtic. Leiden–Boston: Brill. 2014 Slavic Nominal Word-Formation. Heidelberg: Winter. Matzinger, Joachim 2006 Der altalbanische Text [E] Mbsuame e Krështerë (Dottrina Cristiana) des Lekë Matrënga von 1592, eine Einführung in die albanische Sprachwissenschaft. Dettelbach: Verlag J. H. Röll. Maue, Dieter 1990 Das Mahāvaidehikaṃ ghr̥ tam in Tocharisch B. Historische Sprachforschung 103.1: 159–165. Mayrhofer, Manfred 1986 Indogermanische Grammatik. Band I. Heidelberg: Winter. Meier-Brügger, Michael 2003 Indo-European Linguistics. Berlin: de Gruyter. Meillet, Antoine 1914 Le tokharien. Indogermanisches Jahrbuch 1: 1–19. 1921 La catégorie du genre et les conceptions indo-européennes. In: Antoine Meillet (ed.), Linguistique Historique et Linguistique Générale. Paris: Cham­ pion, 211–229. 1931 Essai de chronologie des langues indo-européennes. Bulletin de la Société de Linguistique de Paris 32: 1–28. 1937 Introduction à l’étude comparative des langues indo-européennes. Paris. Meillet, Antoine & Sylvain Lévi 1911 Étude des documents tokhariens de la mission Pelliot. Remarques linguistiques. Journal Asiatique 17: 431–464, 18: 119–150. Meissner, Torsten 2005 S-stem Nouns and Adjectives in Greek and Proto-Indo-European. Oxford: Oxford University Press. Melchert, H. Craig 1978 Tocharian verb stems in -tk-. Zeitschrift für vergleichende Sprachforschung 91: 93–130. 1994a Anatolian Historical Phonology. Amsterdam–Atlanta: Rodopi. 1994b The feminine gender in Anatolian. In: George E. Dunkel, Gisela Meyer, Salvatore Scarlata & Christian Seidl (eds.), Früh‐, Mittel‐, Spätindogermanisch. Akten der IX. Fachtagung der Indogermanischen Gesellschaft vom 5. bis 9. Oktober 1992 in Zürich. Wiesbaden: Reichert, 231–244. 2000 Tocharian plurals in -nt- and related phenomena. Tocharian and IndoEuropean Studies 9: 53–75.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

386

Bibliography

2011 The PIE Collective Plural and the τὰ ζῷα τρέχει rule. In: Thomas Krish & Thomas Lindner (eds.), Indogermanistik und Linguistik im Dialog. Akten der XIII. Fachtagung der Indogermanischen Gesellschaft vom 21 bis 27. September 2008 in Salzburg. Wiesbaden: Reichert, 395–400. 2014 PIE *-eh2 as an “individualizing” suffix and the feminine gender. In: Sergio Neri & Roland Schumann (eds.), Studies on the Collective and Feminine in Indo-European from Diachronic and Typological Perspective. Leiden–Boston: Brill, 257–272. 2021 Hittite neuter i-stems. In: Hannes A. Fellner, Melanie Malzahn & Michaël Peyrot (eds.), lyuke wmer ra. Indo-European Studies in Honor of Georges-Jean Pinault. Ann Arbor–New York: Beech Stave Press, 379–390. 2022 Hittite reflexes of the PIE root *welH-. In: Maria Kazanskaya, Petr Kocharov & Andrey Shatskov (eds.), Colloquia Classica et Indogermanica VII. Miscellanea in honorem Nikolai N. Kazansky Septuagenarii [Acta Linguistica Petropolitana 18.1], 246–266. forth. The position of Anatolian. In: Michael Weiss & Andrew Garrett (eds.), Handbook of Indo-European Studies. Meunier, Fanny 2013 Typologie des locutions en yām- du tokharien. Tocharian and Indo-European Studies 14: 125–185. 2015a Recherches sur le génitif en tokharien. Thèse de doctorat, École Pratique des Hautes Études, Paris. 2015b Some aspects of the translation of Sanskrit compounds into Tocharian. In: Melanie Malzahn, Michaël Peyrot, Hannes A. Fellner & Theresa S. Illés (eds.), Tocharian Text in Context. International Conference on Tocharian Manuscripts and Silk Road Culture, June 25–29th, 2013. Bremen: Hempen, 137–148. Miller, Jared L. 2008 Ein Ritual zur Reinigung eines Hauswesens durch eine Beschwörung an die Unterirdischen (CTH 446). In: Bernd Jankowski & Gernot Wilhelm (eds.), Omina, Orakel, Rituale und Beschwörungen, Texte aus der Umwelt des alten Testaments. (Neue Folge Band 4.). Gütersloh, 206–17. Müller, Friedrich W. K. 1907 Beitrag zur genaueren Bestimmung der unbekannten Sprachen Mittel­ asiens. Sitzungsberichte der Königlich Preußischen Akademie der Wissen­ schaften, 958–960. 1908 Uigurica I. Abhandlungen der Königlich Preußischen Akademie der Wissenschaften. Philosophisch — historische Klasse. Berlin: Reimer. mw 1995=1899 Monier Monier-Williams. A Sanskṛit-English Dictionary. Oxford: Clarendon Press. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

387

Neri, Sergio 2003 I sostantivi in -u del gotico. Morfologia e Preistoria. Innsbruck: Institut für Sprachwissenschaft. 2017 Elementi di morfologia flessiva nominale indoeuropea. Perugia: Università degli studi di Perugia. Neri, Sergio & Roland Schuhmann (eds.) 2014 Studies on the Collective and Feminine in Indo-European from a Diachronic and Typological Perspective. Leiden–Boston: Brill. nil 2008 Dagmar S. Wodtko, Britta Irslinger & Carolin Schneider, Nomina im Indo­ germanischen Lexikon. Heidelberg: Winter. Norbruis, Stefan 2020 Tocharian B ləka- ~ pəlka-, A läkā- ~ pälkā- ‘to see, look’. Tocharian and IndoEuropean Studies 20: 141–162. 2021 Indo-European Origins of Anatolian Morphology and Semantics. Innovations and Archaisms in Hittite, Luwian and Lycian. Diss. Universiteit Leiden. Amsterdam: LOT. Normier, Rudolf 1980 Tocharisch ñkät/ñakte ‘Gott’. Zeitschrift für vergleichende Sprachforschung 94: 251–281. Nussbaum, Alan J. 1986 Head and Horn in Indo-European. Berlin–New York: de Gruyter. 2014a Feminine, abstract, collective, neuter plural: some remarks on each (expanded handout). In: Sergio Neri & Roland Schuhmann (eds.), Studies on the Collective and Feminine in Indo-European from a Diachronic and Typological Perspective. Leiden–Boston: Brill, 273–306. 2014b Greek τέκμαρ ‘sign’ and τέκμωρ ‘sign: Why both? In: Norbert Oettinger & Thomas Steer (eds.), Das Nomen im Indogermanischen. Morphologie, Substantiv versus Adjektiv, Kollektivum. Akten der Arbeitstagung der Indogermanischen Gesellschaft vom 14. Bis 16. September 2011 in Erlangen. Wiesbaden: Reichert, 215–260. Oettinger, Norbert 1980 Die n-Stämme des Hethitischen und ihre indogermanischen Ausgangspunkte. Zeitschrift für vergleichende Sprachforschung 94, 44–63. 1983 Untersuchungen zur avestischen Sprache am Beispiel des Ardvīsur-Yašt. Unveröffentlichte Habilitationsschrift. 1987 Bemerkungen zur anatolischen i‑Motion und Genusfrage. Zeitschrift für vergleichende Sprachforschung 100: 35–43. 2014 Zum anatolischen und indogermanischen Kollektivum. In: Sergio Neri & Roland Schuhmann (eds.), Studies on the Collective and Feminine in IndoEuropean from a Diachronic and Typological Perspective. Leiden–Boston: Brill, 307–316. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

388

Bibliography

2016 Die Wechsel -Ø/n- und -i/n im Rahmen der indogermanischen Heteroklisie. In: Bjarne Simmelkjær Sandgaard Hansen, Benedicte Nielsen Whitehead, Thomas Olander & Birgit Anette Olsen (eds.), Etymology and the European Lexicon. Proceedings of the 14th Fachtagung der Indogermanischen Gesellschaft, 17–22 September 2012, Copenhagen. Wiesbaden: Reichert, 319–326. Ogihara, Hirotoshi 2009 Researches about Vinaya-texts in Tocharian A and B. Thèse de doctorat, École Pratique des Hautes Études, Paris. 2011 Notes on some Tocharian Vinaya fragments in the London and Paris collections. Tocharian and Indo-European Studies 12: 111–144. 2012a Tokarago B « Avadāna shahon » ni tsuite [The « Avadāna manuscript » in Tocharian]. Tokyo University Linguistic Papers 32: 109–243. 2012b A fragment of the Bhikṣu-prātimokṣasūtra in Tocharian B. Tocharian and Indo-European Studies 13: 163–179. 2013a On the Karmavācanā in Tocharian. In: Matteo De Chiara, Mauro Maggi & Giuliana Martini (eds.), Multilingualism and History of Knowledge. Vol. 1: Buddhism among the Iranian Peoples of Central Asia. Wien: Verlag der Österreichischen Akademie der Wissenschaften, 311–332. 2013b Tocharian fragment THT333 in the Berlin Collection. Tokyo University Linguistic Papers 33: 205–217. 2013c Agonkyōten ni kanren suru santen no tokarago B danpen ni tsuite [Three Tocharian B fragments related to the Āgama texts]. Tokyo University Linguistic Papers 34: 1–33. 2014 Fragments of secular documents in Tocharian A. Tocharian and Indo-European Studies 15: 103–129. 2015 Doitsu shozō Tokarago B 366 danpen THT1859–1860 ni tsuite [Kuchean Frag­ments THT1859 and THT1860 in the Berlin Collection]. Tokyo University Linguistics Papers 36: 103–129. Ogihara, Hirotoshi & Chao-jung Ching 2017 Some observations on the Tumshuqese documents newly published in China. In: Zur lichten Heimat. Studien zu Manichäismus, Iranistik und Zentralasienkunde im Gedenken an Werner Sundermann. Herausgegeben von einem Team „Turfanforschung“. Wiesbaden: Harrassowitz, 453–482. Olander, Thomas 2015 Proto-Slavic Inflectional Morphology. Leiden–Boston: Brill. 2019 Indo-European cladistic nomenclature. Indogermanische Forschungen 124: 231–244. Olsen, Birgit Anette 1999 The Noun in Biblical Armenian. Origin and Word-Formation. Berlin–New York: de Gruyter.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

389

́ In: Bernhard Forssman (ed.), On the origin of the types devī ́ and vr̥kīḥ. Indoarisch, Iranisch und die Indogermanistik. Wiesbaden: Reichert, 401–408. Opfermann, Andreas 2017 Griechische, Armenische und Albanische ,Frauen‘ und die wurzel *gu̯ en-. In: Harald Bichlmeier & Andreas Opfermann (eds.), Das Menschenbild bei den Indogermanen. Hamburg: Baar, 55–80. Otto, Claude 2007 Bemerkungen zum westtocharischen Lexem ṣarya „Geliebte“. Historische Sprachforschung 120: 110–114. Paciaroni, Tania, Graziella Nolè & Michele Loporcaro 2013 Persistenza del neutro nell’italo-romanzo centro-meridionale. Vox Romanica 72: 88–137. Pan Tao 2020 A new look at the Skt.-Toch. Bilingual Mahāparinirvāṇasūtra-Fragment THT 542. Münchener Studien zur Sprachwissenschaft 74.1 (2020 [2021]): 117–146. 2021 Untersuchungen zu Lexikon und Metrik des Tocharischen. Diss. Ludwig Maximilian Universität München. Panaino, Antonio C. D. 1990 Tištrya. Part I. The Avestan Hymn to Sirius. Roma: Istituto Italiano per il Medio ed Estremo Oriente. Pedersen, Holger M. 1913 Vergleichende Grammatik der keltischen Sprachen. Göttingen: Vandenhoeck & Ruprecht. 1925 Le groupement des dialectes indo-européens. Copenhagen: Høst. 1941 Tocharisch vom Gesichtspunkt der indoeuropäischen Sprachvergleichung. København: I Kommission hos Ejnar Munksgaard. Peters, Martin 1991 Ein tocharisches Auslautproblem. Die Sprache 34: 242–244. 1993 Ein weiterer Fall für das Rixsche Gesetz. In: Meiser von Gerhard (ed.), Indogermanica et Italica. Festschrift für Helmut Rix zum 65. Innsbruck: Institut für Sprachwissenschaft, 373–405. 2004 On some Greek nt-formation. In: John H. W. Penney (ed.), Indo-European Perspectives. Studies in Honour of Anna Morpurgo Davies. Oxford: Oxford University Press, 266–276. Petersen, Walter 1939 The Primary Cases of the Tocharian Nominal Declension. Language 15: 72–98. Peyrot, Michaël 2003 Matters of linguistics distribution: Tocharian B weśeña ~ weśeño ‘voice’ and pest ~ past ‘after, away’. In: Adam Hyllested, Anders R. Jørgensen & Thomas 2000

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

390

Bibliography

Olander (eds.), esse sīran sen labban quāitin, Festschrift for Jenny Helena Larsson on the occasion of her 30th birthday 18 October 2003. Copenhagen: Editiones Olander, 60–67. 2007 An edition of the Tocharian fragments IOL Toch 1 – IOL Toch 822 in the India Office Library, London. London: International Dunhuang Project, idp.bl.uk. 2008 Variation and Change in Tocharian B. Amsterdam–New York: Rodopi. 2010a Proto-Tocharian syntax and the status of Tocharian A. Journal of Indo-European Studies 38: 132–146. 2010b On the formation of the Tocharian preterite participle. Historische Sprach­ forschung 121: 69–83. 2011 Tocharian A māskā- ‘be difficult’. Tocharian and Indo-European Studies 12: 145–154. 2012a The Tocharian A match of the Tocharian B obl. sg. -ai. Tocharian and Indo-European Studies 13: 181–220. 2012b Tocharian ‘eat’ and the strong imperfect in Tocharian A. In: Olav Hackstein & Ronald I. Kim (eds.), Linguistic developments along the Silkroad: Archaism and innovation in Tocharian. Wien: Verlag der Österreichischen Akademie der Wissenschaften, 85–119. 2013a The Tocharian Subjunctive. A Study in syntax and verbal stem formation. Leiden–Boston: Brill. 2013b Review of Melanie Malzahn (2010). Tocharian and Indo-European Studies 14: 213–259. 2014 Notes on Tocharian glosses and colophons in Sanskrit manuscripts I. Tocharian and Indo-European Studies 15: 131–179. 2015 Review of Dieter Maue (2015), “Alttürkische Handschriften Teil 19: Dokumente in Brāhmī und tibetischer Schrift. Teil 2”. Tocharian and Indo-European Studies 16: 215–224. 2016a Language contact in Central Asia: On the etymology of Tocharian B yolo ‘bad’. In: Bjarne Simmelkjær Sandgaard Hansen, Benedicte Nielsen Whitehead, Thomas Olander & Birgit Anette Olsen (eds.), Etymology and the European Lexicon. Proceedings of the 14th Fachtagung der Indogermanischen Gesellschaft, 17–22 September 2012, Copenhagen. Wiesbaden: Reichert, 327–335. 2016b Further Sanskrit-Tocharian bilingual Udānavarga fragments. Tocharian and Indo-European Studies 17: 153–212. 2017 On the part of speech and the syntax of the Tocharian present participle. In: Claire Le Feuvre, Daniel Petit & Georges-Jean Pinault (eds.), Verbal Adjectives and Participles in Indo-European Languages. Proceedings of the conference of the Society for Indo-European Studies (Indogermanische Gesellschaft), Paris, 24th to 26th September 2014. Bremen: Hempen Verlag, 327–341.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

391

2018a

Tocharian agricultural terminology: Between inheritance and language contact. In: Guus Kroonen, James P. Mallory & Bernard Comrie (eds.), Talking Neolithic: Proceedings of the workshop on Indo-European origins held at the Max Planck Institute for Evolutionary Anthropology, Leipzig, December 2–3, 2013. Washington DC: Institute for the Study of Man, 242–277. 2018b Tocharian B etswe ‘mule’ and Eastern East Iranian. In: Lucien van Beek, Alwin Kloekhorst, Guus Kroonen, Michaël Peyrot, Tijmen Pronk & Michiel de Vaan (eds.), Farnah. Indo — Iranian and Indo-European Studies in Honor of Sasha Lubotsky. Ann Arbor: Beech Stave, 270–283. 2018c Interrogative stems in Hittite and Tocharian. Indogermanische Forschungen 123: 86–90. 2019a Indo-Uralic, Indo-Anatolian, Indo-Hittite. In: Alwin Kloekhorst & Tijmen Pronk (eds.), The Precursors of Proto-Indo-European. The Indo-Anatolian and the Indo-Uralic hypotheses. Boston: Brill–Rodopi, 186–202. 2019b The deviant typological profile of the Tocharian branch of Indo-European may be due to Uralic substrate influence. Indo-European Linguistics 7: 72–121. 2020 The simplification of the Archaic Tocharian B clusters ltk and rtk — with a note on the vowel assimilation in Late Tocharian B oṅkorño. Tocharian and Indo-European Studies 20: 191–200. 2021 On a festive occasion. The inflection of Tocharian B ekṣalye and morphologically related forms. In: Hannes A. Fellner, Melanie Malzahn & Michaël Peyrot (eds.), lyuke wmer ra. Indo-European Studies in Honor of Georges-Jean Pinault. Ann Arbor–New York: Beech Stave Press, 450–461. 2022 Tocharian. In: Thomas Olander (ed.), The Indo-European Language Family. A Phylogenetic Perspective. Cambridge: Cambridge University Press, 83–101. Peyrot, Michaël, Federico Dragoni & Chams Bernard 2022 The spread of iron in Central Asia: On the etymology of the word for “iron” in Iranian and Tocharian. Bulletin of the School of Oriental and African Studies. Peyrot, Michaël & Meng Xiaoqiang 2021 Tocharian B santse ‘daughter-in-law’. Indogermanische Forschungen 126: 405–409. Peyrot, Michaël, Georges-Jean Pinault & Jens Wilkens 2019 Vernaculars of the Silk Road — A Tocharian B-Old Uyghur Bilingual. Journal Asiatique 307.1: 65–90. Peyrot, Michaël & Ablet Semet 2016 A comparative study of the beginning of the 11th act of the Tocharian A Maitreyasamitināṭaka and the Old Uyghur Maitrisimit. Acta Orientalia Academiae Scientiarum Hungaricae 69: 355–378.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

392

Bibliography

Pinault, Georges-Jean 1987 Épigraphie koutchéenne, I. Laissez  — passer de caravanes, II. Graffites et inscriptions. In: Chao Huashan, Simone Gaulier, Monique Maillard & Georges-Jean Pinault (eds.), Sites divers de la région de Koutcha, Épigraphie koutchéenne. (Mission Paul Pelliot 8) Paris: Collège de France, 59–196 and plates 40–96. 1988 Le Pratītyasamutpāda en koutchéen. Tocharian and Indo-European Studies 2: 96–165. 1989a Introduction au tokharien. LALIES 7. Paris: École Normale Supérieure, 5–224. 1989b Une version koutchéenne de l’Aggañña-sutta. Tocharian and Indo-European Studies 3: 149–220. 1991 Un témoignage tokharien sur les premières nonnes bouddhistes. Bulletin d’Études Indiennes 9: 161–194. 1994 Une nouvelle inscription koutchéenne de Qumtura: Légende de scènes bouddhiques de Praṇidhi. Bulletin d’Études Indiennes 11–12 (1993–1994): 71–220. 1997a Remarque sur le pluriel tokh. B akrūna, A ākrunt. In: Alexander Lubotsky (ed.), Sound Law and Analogy. Papers in honor of Robert S. P. Beekes on the occasion of his 60th Birthday. Amsterdam–Atlanta: Rodopi, 219–233. 1997b Terminologie du petit bétail en tokharien. Studia Etymologica Cracoviensia 2: 175–218. 1999 Tokharien A kapśañi, B kektseñe. In: Heiner Eichner, Hans Christian Luschützky & Velizar Sadovski (eds.), Compositiones Indogermanicae in memoriam Jochem Schindler. Praha: Enigma, 457–478. 2001 Le type latin uorāgō: un reflet d’un suffixe indo-européen. Glotta 77: 85–109. 2002 Tocharian and Indo-Iranian: relations between two linguistic areas. In: Nicholas Sims-Williams (ed.), Indo-Iranian Languages and Peoples. Oxford: Oxford University Press, 243–284. 2003a Contacts linguistiques en Asie Centrale à la lumière des textes tokhariens. In: Sven Bretfeld & Jens Wilkens (eds.), Indien und Zentralasien, Sprach- und Kulturkontakt, Vorträge des Göttinger Symposions vom 7. Bis 10. Mai 2001. Wiesbaden: Harrassowitz, 45–83. 2003b On the tracks of the Tocharian Guru. In: Brigitte L. M. Bauer & Georges-Jean Pinault (eds.), Language in time and space. A Festschrift for Werner Winter on the occasion of his 80th birthday. Berlin–New York, 331–346. 2005 Tocharian ‘woman’. In: Nikolai N. Kazansky (ed.), Hr̥dā́ mánasā. Studies presented to Professor Leonard G. Herzenberg on the occasion of his 70th birthday. Saint Petersburg: Nauka, 201–214. 2006a Retour sur le numeral ‘un’ en tokharien. Indogermanische Forschungen 111: 71–96.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

393

2006b Further links between the Indo-Iranian substratum and the BMAC language. In: Bertil Tikkanen & Heinrich Hettrich (eds.), Themes and Tasks in Old and Middle Indo-Aryan Linguistics. Papers of the 12th World Sanskrit Conference Held in Helsinki, Finland, 13–18 July 2003, vol. 5. Delhi: Motilal Banarsidass, 167–196. 2008 Chrestomathie tokharienne. Textes et grammaire. Leuven–Paris: Peeters. 2009a On the formation of the Tocharian demonstratives. In: Elisabeth Rieken and Paul Widmer (eds.), Pragmatische Kategorien. Form, Funktion und Diachronie. Akten der Arbeitstagung der Indogermanischen Gesellschaft vom 24. bis 26. September 2007 in Marburg. Wiesbaden: Reichert, 221–245. 2009b Elephant man. Sur le nom de l’éléphant en tokharien. In: Nalini Balbir & Georges-Jean Pinault (eds.), Penser, dire et représenter l’animal dans le monde indien. Paris: Champion, 447–498. 2009c Philologie des textes bouddhiques d’Asie Centrale. Annuaire de l’École pratique des hautes études (EPHE), Section des sciences historiques et philologiques 140, 405–409. 2010 Review of Xavier Tremblay (2003), “La déclinaison des noms de parenté indo-européens en -ter-”. Kratylos 54: 24–36. 2011a Let us Now Praise Famous Gems. Tocharian and Indo-European Studies 12: 155–220. 2011b L’origine déictique du genre féminin en indo-européen. Bulletin de la Société de linguistique de Paris 106: 129–182. 2011c Some Tocharian abstract suffixes. In: Thomas Krisch & Thomas Lindner (eds.), Indogermanistik und Linguistik im Dialog. Akten der XIII. Fachtagung der Indogermanischen Gesellschaft vom 21. Bis 27. September 2008 in Salzburg. Wiesbaden: Reichert, 453–462. 2012a Tocharian -nt- participles and agent nouns. In: Olav Hackstein & Ronald Kim (eds.), Linguistic developments along the Silkroad: Archaism and innovation in Tocharian. Wien: Verlag der Österreichischen Akademie der Wissenschaften, 179–204. 2012b La parfaite générosité du roi Ambara (PK NS 32). Tocharian and Indo-European Studies 13: 221–243. 2013a The Lady (Almost) Vanishes. In: Adam I. Copper, Jeremy Rau & Michael Weiss (eds.), Multi Nominis Grammaticus. Studies in Classical and Indo-European in honor of Alan J. Nussbaum on the occasion of his sixty-fifth birthday. Ann Arbor–New York: Beech Stave Press, 240–254. 2013b Body and soul: the reflexive in Tocharian. Indogermanische Forschungen 118: 339–359. 2014a Distribution and origins of the PIE suffixes *-ih2. In: Norbert Oettinger & Thomas Steer (eds.), Das Nomen im Indogermanischen. Morphologie, Adjektiv

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

394

Bibliography

vs. Substantiv, Kollektivum. Akten der Arbeitstagung der Indogermanischen Gesellschaft von 14. Bis 16. September 2011 in Erlangen. Wiesbaden: Reichert, 273–306. 2014b The ‘one night-and-day observance’ of lay-followers in Tocharian Buddhism. Tocharian and Indo-European Studies 15: 187–215. 2015a The formation of Buddhist languages, as exemplified by the Tocharian evidence. In: Melanie Malzahn, Michaël Peyrot, Hannes A. Fellner & Theresa S. Illés (eds.), Tocharian Text in Context. International Conference on Tocharian Manuscripts and Silk Road Culture, June 25–29th, 2013. Bremen: Hempen, 159–185. 2015b Review of Markus Hartmann (2013). Bulletin de la Société de linguistique de Paris 110.2: 178–197. 2015c The legend of the unicorn in the Tocharian version. Journal of the Inter­ national Association of Buddhist Studies 38: 144–174. 2015d Tocharian nostalgia. Tocharian and Indo-European Studies 16: 131–179. 2015e The Tocharian background of Old Turkic yaŋı kün. In: Elisabetta Ragagnin & Jens Wilkens (eds.), Kutadgu Nom Bitig. Festschrift für Jens Peter Laut zum 60. Geburtstag. Wiesbaden: Harrassowitz, 377–405. 2017a The morphology of Tocharian. In: Jared Klein, Brian Joseph, Matthias Fritz & Mark Wenthe (eds.), Handbook of Comparative and Historical Indo-European Linguistics Vol. II. Berlin–Boston: de Gruyter, 1335–1352. 2017b In the heat of the day. In: Ivo Hajnal, Daniel Kölligan & Katharina Zipser (eds.), Miscellanea Indogermanica. Festschrift für José Luis García Ramón zum 65. Geburtstag. Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft, 641–659. 2017c Current issues in Tocharian etymology and phonology. Tocharian and Indo-European Studies 18: 127–164. 2019 Surveying the Tocharian B lexicon. Orientalistische Literaturzeitung 114: 91–97. 2020a The Dharma of the Tocharians. In: Vincent Tournier, Vincent Eltschinger & Marta Sernesi (eds.), Archaeologies of the Written: Indian, Tibetan, and Buddhist Studies in Honour of Cristina Scherrer-Schaub. UniorPress: Napoli, 461–492. 2020b Tocharian lexicon in the light of contact phenomena. In: Romain Garnier (ed.), Loanwords and Substrata. Proceedings of the Colloquium held in Limoges (5th–7th June, 2018). Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft, 367–401. 2021 Regard comparatif sur la dérivation nominale en tokharien. In: Alain Blanc & Isabelle Boehm (eds.), Dérivation nominale et innovations dans le langues indo-européennes anciennes. Actes du colloque international de l’université de Rouen (ÉRIAC), 11–12 octobre 2018. Lyon, Maison de l’Orient et de la Méditerranée–Jean Pouilloux, 113–132. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

395

Pirart, Eric 1986 Le traitement des laryngales intervocaliques en vieil-avestique métrique. Münchener Studien zur Sprachwissenschaft 47: 159–191. Pisani, Vittore 1941–42 Appunti di tocarico. Glottica Parerga 1. 1942–43 Etimologie tocariche. Glottica Parerga 5. 1954 Griech. γλῶσσα. Indogermanische Forschungen 61: 141–146. 1964 Le lingue dell’Italia antica oltre il latino. Torino: Rosenberg & Sellier. Pischel, Richard 1981 Grammatik der Prākrit-Sprachen. Strassburg. Engl. transl. A Grammar of the Prākrit Languages. Delhi. Piwowarczyk, Dariusz R. 2016 The origin of the Latin -iēs/-ia inflection. Historische Sprachforschung 129: 108–123. 2021 The Latin -iēs/-ia inflection. Synchronic evidence and diachronic origin. Jagiellonian University Press. Plank, Frans & Wolfgang Schellinger 1997 The uneven distribution of genders over numbers: Greenberg Nos. 37 and 45. Linguistic Typology 1: 53–101. Pott, August 1833–35 Etymologische Forschungen. 2 vols. Lemgo: Meyerʾscher Hof. Poucha, Pavel 1955 Thesaurus Linguae Tocharicae Dialecti A. Praha: Státní Pedagogické Nakladatelství. Priestly, Tom M. S. 1983 On ‘drift’ in Indo-European gender system. Journal of Indo-European Studies 11: 339–363. Pronk, Tijmen 2015a On the origin of the dual endings Tocharian A -ṃ, B -ne. In: Melanie Malzahn, Michaël Peyrot, Hannes A. Fellner & Theresa S. Illés (eds.), Tocharian Text in Context. International Conference on Tocharian Manuscripts and Silk Road Culture, June 25–29th, 2013. Bremen: Hempen, 199–214. 2015b Singulative n-stems in Indo-European. Transactions of the Philological Society 113: 327–348. 2016 Stang’s Law in Baltic, Greek and Indo-Iranian. Baltistica 60: 19–35. Pulleyblank, Edwin G. 1991 Lexicon of Reconstructed Pronunciation in Early Middle Chinese, Late Middle Chinese and Early Mandarin. Voncouver: UBC Press. Rasmussen, Jens E. 1989 Studien zur Morphophonematik der indogermanischen Grundsprache. Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

396

Bibliography

Rau, Jeremy 2007 The Origin of Indic and Iranian Feminines in -ānī-́ d. Journal of the American Oriental Society 127: 57–66. Redard, Georges 1949 Les noms grecs en -ΤΗΣ, -ΤΙΣ et principalement en -ΙΤΗΣ, -ΙΤΙΣ. Paris: Libraire C. Klincksieck. Reichelt, Hans 1900 Der Frahang I Oīm. Wiener Zeitschrift für die Kunde des Morgenlandes 14: 177–213. Rieken, Elizabeth 1999 Untersuchungen zur nominalen Stammbildung des Hethitischen. Harrassowitz: Wiesbaden. 2005 Neues zum Ursprung der anatolischen i-Mutation. Historische Sprachforschung 118: 48–74. 2009 Der Archaismus des Hethitischen: eine Bestandsaufnahme. Incontri Lingu­i­ stici 32: 37–52. Ringe, Don 1987 A closer look at Tocharian e and o and the Indo-European medio-passive. Tocharian and Indo-European studies 1: 98–138. 1989 Tocharian B ausu, auṣu, aultsu. Tocharian and Indo-European Studies 3: 35–50. 1991 Evidence for the position of Tocharian in the Indo-European family? Die Sprache 34: 59–123. 1996 On the Chronology of Sound Changes in Tocharian. Volume 1: From Proto-IndoEuropean to Proto-Tocharian. New Haven: American Oriental Society. 2000 Tocharian class II presents and subjunctives and the reconstruction of the Proto-Indo-European verb. Tocharian and Indo-European Studies 9: 121–142. 2006 A Linguistic History of English. Volume I: From Proto-Indo-European to ProtoGermanic. Oxford: Oxford University Press. 2017 Indo-European dialectology. In: Jared Klein, Brian Joseph, Matthias Fritz & Mark Wenthe (eds.), Handbook of Comparative and Historical Indo-European Linguistics. Vol. I. Berlin–Boston: de Gruyter, 62–75. Ringe, Don, Tandy Warnow & Ann Taylor 2002 Indo-European and computational cladistics. Transactions of the Philological Society 100: 59–129. Rix, Helmut 1985 Sūdor und sīdus. In: Hermann M. Ölberg, Gernot Schmidt & Heinz Bothien (eds.), Sprachwissenschaftliche Forschungen. Festschrift für Johann Knobloch zum 65. Geburtstag am 5. Januar 1984 dargebracht von Freunden und Kollegen. Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft, 339–350.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

397

1986

Die Endung des Akkusativ Plural commune im Oskischen. In: Annemarie Etter (ed.), o-o-pe-ro-si. Festschrift für Ernst Risch zum 75. Geburtstag. Berlin– New York: de Gruyter, 583–597. Rovai, Francesco 2012 Between feminine singular and neuter plural: re-analysis patterns. Transac­ tions of the Philological Society 110: 94–121. Sadovski, Velizar 2017 The lexicon of Iranian. In: Jared Klein, Brian Joseph, Matthias Fritz & Mark Wenthe (eds.), Handbook of Comparative and Historical Indo-European Linguistics. Vol. I. Berlin–Boston: de Gruyter, 566–599. Saito, Haruyuki 2006 Das Partizipium Präteriti im Tocharischen. Wiesbaden: Harrassowitz. Sapir, Edward 1936 Tibetan influences on Tocharian, I. Language 12: 259–271. Sasseville, David 2018 New evidence for the PIE common gender suffix -eh2 in Anatolian: Luwian -ašša- (c.) and Lycian B -asa- (c.). In: Elisabeth Rieken, Ulrich Geupel & Theresa Maria Roth (eds.), 100 Jahre Entzifferung des Hethitischen. Morphosyntaktische Kategorien in Sprachgeschichte und Forschung. Akten der Arbeitstagung der Indogermanischen Gesellschaft vom 21. bis 23. September 2015 in Marburg. Wiesbaden: Reichert Verlag, 303–318. Schaefer, Christiane 1997 waṣiḵ\ kalpaṣṣuki, Zu den westtocharischen Nominalbildungen auf -uki. Tocharian and Indo-European Studies 7: 163–176. Schindler, Jochem 1967 Tocharische Miszellen. Indogermanische Forschungen 72: 239–49. 1975 L’apophonie des thèmes indo-européens en -r/n. Bulletin de la Société de Linguistique de Paris 70: 1–10. 1980 Review of A. J. Van Windekens (1976). In: Die Sprache 26: 83–4. Schmidt, Claus-Peter 1972 Maskuline genuskongruenz beim Plural der Substantiva alternantia im Tocharischen. Diss. Frankfurt am Main. Schmidt, Johannes 1889 Die Pluralbildungen der indogermanischen Neutra. Weimar; Hermann Böhlau. Schmidt, Klaus H. 1987 Zur semantischen Differenzierung des indogermanischen Kasussystems im Tocharischen. Tocharian and Indo-European Studies 1: 139–151. 1990 The postulated Pre-Indo-European substrates in Insular Celtic and Tocharian. In: Thomas L. Markey & John C. Greppin (eds.), When Worlds Collide. Ann Arbor: Karoma, 179–202.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

398

Bibliography

Schmidt, Klaus T. 1974 Die Gebrauchsweisen des Mediums im Tocharischen. Diss. Universität Göttingen. 1980 Zu Stand und Aufgaben der etymologischen Forschung auf dem Gebiete des Tocharischen. In: Manfred Mayrhofer, Oskar E. Pfeiffer & Martin Peters (eds.), Lautgeschichte und Etymologie. Akten der VI. Fachtagung der Indogermanischen Gesellschaft. Wien, 24.–29. September 1978. Wiesbaden: Reichert, 394–411. 1982 Spuren tiefstufiger seṭ-Wurzeln im tocharischen Verbalsystem. In: Johann Tischler (ed.), Serta Indogermanica. Festschrift für Günter Neumann zum 60. Geburtstag. Innsbruck: Institut für Sprachwissenschaft, 363–372. 1986a Fragmente eines buddhistischen Ordinationsrituals in westtocharischer Sprache. Aus der Schule der Sarvāstivādins. Text, Übersetzung, Anmerkungen und Indizes. Unveröffentlichte Habilitationsschrift. 1986b Bemerkungen zur westtocharischen Umgangssprache. In: Annemarie Etter (ed.), o-o-pe-ro-si. Festschrift für Ernst Risch zum 75. Geburtstag. Berlin–New York: de Gruyter, 635–649. 1987 Zu einigen Archaismen in Flexion und Wortschatz des Tocharischen. In: Wolfgang Meid (ed.), Studien zum indogermanischen Wortschatz. Innsbruck: Institut für Sprachwissenschaft, 287–300. 1992 Archaismen des Tocharischen und ihre Bedeutung für Fragen der Rekonstruktion und der Ausgliederung. In: Robert Beekes, Alexander Lubotsky & Joseph Johannes Sicco Weitenberg (eds.), Rekonstruktion und relative Chronologie. Akten der VIII Fachtagung der Indogermanischen Gesellschaft, Leiden, 31. August–4. September 1987. Innsbruck: Institut für Sprachwissenschaft, 101–112. 1997 Liebe und Sexualität im Spiegel der tocharischen Sprachzeugnisse. In: Michaela Ofitsch (ed.), Eros, Liebe und Zuneigung in der Indogermania. Akten des Symposiums zur indogermanischen Kultur- und Altertumskunde in Graz (29.–30. September 1994). Graz: Leykam, 227–262. 2001 Die westtocharische Version des Araṇemi-Jātakas in deutscher Übersetzung. In: Louis Bazin & Peter Zieme (eds.), De Dunhuang à Istanbul. Hommage à James Russell Hamilton. Turnhout: Brepols, 299–327. 2002 Beobachtungen zur tocharischen Landwirtschaftsterminologie. Die Sprache 41: 1–23. 2008 THT 107. “Die Speisung des Bodhisattva vor der Erleuchtung”. Die westtocharische Version im Vergleich mit der Sanskritfassung der Mūlasarvāstivādins. In: Peter Zieme (ed.), Aspects of research into Central Asian Buddhism. In memoriam Kōgi Kudara. Turnhout: Brepols, 209–342.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

399

2018

Klaus T. Schmidt Nachgelassene Schriften. 1. Ein westtocharisches Ordinations­ ritual. 2. Eine dritte tocharische Sprache: Lolanisch. Bearbeitet und herausgegeben von Stefan Zimmer. Bremen: Hempen Verlag. Schrijver, Peter 1990 Latin festīnāre, Welsh brys. Münchener Studien zur Sprachwissenschaft 51: 243–247. Schulze, Wilhelm 1927 Zum Tocharischen. Ungarische Jahrbucher 7: 168–177. Schwartz, Martin 1974 Irano-Tocharica. In: Philippe Gignoux & Ahmad Tafazzoli (eds.), Mémorial Jean de Menasce. Louvain: Imprimerie Orientaliste, 399–411. Seebold, Elmar 1970 Vergleichendes und etymologisches Wörterbuch der germanischen starken Verben. The Hague–Paris: de Gruyter. Seržant, Ilja A. 2014 Das Kausativ im Tocharischen. München: Lincom. Sharma, Priyavrat 2014 Caraka Saṃhitā. Text with English Translation by Prof. Priyavrat Sharma. Chaukhambha Orientalia: Varanasi. Sieg, Emil 1938 Die Kutschischen Karmavibhaṅga-Texte der Bibliothèque Nationale in Paris (zu Prof. Sylvain Lévi’s Ausgabe und Übersetzung). Zeitschrift für vergleichende Sprachforschung 65: 1–54. 1944 Übersetzungen aus dem Tocharischen. Berlin: de Gruyter. 1952 Übersetzungen aus dem Tocharischen II. Berlin: Akademie-Verlag. 1955 Die medizinischen und tantrischen Texte der Pariser Sammlung in Tocharisch B. Bemerkungen zu Prof. Filliozats Textausgabe und Übersetzung (Nachgelassenes Manuskript, herausgegeben von Werner Thomas). Zeitschrift für vergleichende Sprachforschung 72: 63–82. Sieg, Emil & Wihelm Siegling 1908 Tocharisch, die Sprache der Indoskythen. Vorläufige Bemerkungen über eine bisher unbekannte indogermanische Literatursprache. Sitzungsberichte der Berliner Akademie der Wissenschaften zu Berlin, Jahresgang, 915–34. 1921 Tocharische Sprachreste. Band I: Die Texte. A. Transcription. Berlin–Lepizig: de Gruyter. 1949 Tocharische Sprachreste. Sprache B, Heft 1. Die Udānālaṅkāra-Fragmente. Text, Übersetzung und Glossar. Göttingen: Vandenhoeck & Ruprecht. 1953 Tocharische Sprachreste. Sprache B, Heft 2. Fragmente Nr. 71–633. Aus dem Nachlaß hg. von Werner Thomas, Thomas, Werner, ed. Göttingen: Vandenhoeck & Ruprecht.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

400

Bibliography

Sihler, Andrew L. 1977 The PIE origins of the Germanic feminine nomina agentis in *-stri(ō)n. Die Sprache 23: 36–47. 1995 New Comparative Grammar of Greek and Latin. New York–Oxford: Oxford University Press. Sims-Williams, Nicholas 1994 The triple system of deixis in Sogdian. Transactions of the Philological Society 92: 41–53. 2002 Ancient Afghanistan and its invaders: Linguistic evidence from the Bactrian documents and inscriptions. In: Nicholas Sims-Williams (ed.), Indo-Iranian Languages and Peoples. Oxford: Oxford University Press, 225–242. Simms, Douglas P. A. 2009 The words for ‘fire’ in Germanic. Journal of Germanic Linguistics 21: 297–333. Skjærvø, Prods Oktor 2007 Avestan and Old Persian Morphology. In: Alan B. Kaye (ed.), Morphologies of Asia and Africa. Winona Lake: Eisenbrauns, 953–940. 2022 Khotan IV. The Khotanese Language. In: Elton L. Daniel (ed.), Encyclopædia Iranica XVII.2. Leiden: Brill, 117–133. Stang, Christian S. 1945 Zum indoeuropäischen Kollektivum. Norsk Tidsskrift for Sprogvidenskap 13: 282–294. 1965 Indo-européen *gwōm, *diēm. In: Adam Heintz, Mieczysław Karaś, Tadeusz Milewski, Jan Safarewicz & Witold Taszycki (eds.), Symbolae Linguisticae in honorem Georgii Kuryłowicz. Wrocław: Zakład Narodowy Im. Ossolińskich, 292–296. Starke, Frank 1990 Untersuchung zur Stammbildung des keilschrift-luwischen Nomens. Wiesbaden: Harrassowitz. Steblin-Kamenskij, Ivan M. 1982 Očerki po istorii leksiki pamirskix jazykov. Nazvanija kul’turnyx rastenij [Essay on the history of the lexicon of the Pamir languages. Names of cultural plants]. Moskva: Nauka. Steele, Susan 1978 Word order variation: a typological study. In: Joseph H. Greenberg, Charles Albert Ferguson, Edith A. Moravcsik (eds.), Universals of Human Language. Vol. IV: Syntax. Stanford: Stanford University Press, 583–623. Steer, Thomas 2014 Zum Kontrastakzent und Wurzelablaut thematischer Kollektiva des Urindogermanischen. In: Sergio Neri & Roland Schumann (eds.), Studies on the Collective and Feminine in Indo-European from Diachronic and Typological Perspective. Leiden–Boston: Brill, 333–362. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

401

Stempel, Reinhard 1994 ī-Genitiv, ī-Motion und Femininum: Versuch einer Synthese. In: Roland Bielmeier & Reinhard [unter Mitarb. von René Lanszweert] (eds.), Indo­ germanica et Caucasica: Festschrift für Karl Horst Schmidt zum 65. Geburtstag. Berlin, New York: de Gruyter, 197–210. Stumpf, Peter 1971 Der Gebrauch der Demonstrativ-Pronomina im Tocharischen. Wiesbaden: Harrassowitz. 1974 Der Plural der westtocharischen Demonstrativ-Pronomina: zugleich ein Beitrag zur Dialekt-Gliederung des Westtocharischen. Orbis 23: 404–428. 1976 Westtocharisch se–seṃ: zwei Paradigmen oder nur eines? Zeitschrift für vergleichende Sprachforschung 90: 114–127. 1990 Die Erscheinungsformen des Westtocharischen, Ihre Beziehungen zueinander und ihre Funktionen. Reykjavík: Málvísindastofnun Háskóla Íslands. Sturtevant, Edgar H. 1949 An Indo-European word for ‘woman’. Language 25: 343–345. swtf 1972- Sanskrit-Wörterbuch der buddhistischen Texte aus den Turfan-funden und der kanonischen Literatur der Sarvāstivāda-Schule. Begonnen von Ernst Waldschmidt, edited by Georg von Simson, Michael Schmidt, Heinz Bechert and Jens-Uwe Hartmann. Göttingen: Vandenhoeck & Ruprecht. Szemerényi, Oswald 1960 Indo-European tālis, kwālis and the Greek -k-stems ἧλιξ, γυνή/γυναικ-. Annali dell’Istituto Universitario Orientale di Napoli — Sezione Linguistica 2: 1–30. 1979 Etyma Graeca IV. Homerica et Mycenaica. Studi Micenei ed Egeo-Anatolici 20: 207–225. 1989 An den Quellen des lateinischen Wortschatzes. Innsbruck: Institut für Sprachwissenschaft. Tamai, Tatsushi 2004 On Tocharian Document No. 174 in the Tokyo National Museum. Studies on the Inner Asian Languages 19: 95–107. 2009 The Tocharian Fragments in Or.15004, Or.15009 and Or.150010 of the Hoernle Collection. In: Seishi Karashima & Klaus Wille (eds.), Buddhist Manuscripts from Central Asia: The British Library Sanskrit Fragments. Volume II. Tokyo: IRIAB, Soka University, 659–665. 2011 Paläographische Untersuchungen zum B-Tocharischen. Innsbruck: Institut für Sprachwissenschaft. 2014 The Tocharian Karmavācānā. Annual Report of the International Research Institute for Advanced Buddhology (ARIRIAB) at Soka University for the Academic Year 2013. Vol. 17: 365–394.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

402 2018

teb 1960

Bibliography The Tocharian Araṇemi-Jātaka. Annual Report of the International Research Institute for Advanced Buddhology (ARIRIAB) at Soka University for the Academic Year 2017. Vol 21: 359–400.

Wolfgang Krause & Werner Thomas, Tocharisches Elementarbuch. Band I. Heidelberg: Winter. Tekin, Şinasi 1980 Maitrisimit nom bitig. Die uigurische Übersetzung eines Werkes der buddhisti­ schen Vaibhāṣika-Schule. 1. Teil: Transliteration, Übersetzung, Anmerkungen. – 2. Teil: Analytischer und rückläufiger Index. Berliner Turfantexte IX. Berlin: Akademie–Verlag. tg 1931 Emil Sieg, Wilhelm Siegling & Wilhelm Schulze, Tocharische Grammatik. Göttingen: Vandenhoeck & Ruprecht. Thomas, Werner 1952 Die tocharischen Verbaladjektive auf -l. Eine syntaktische Untersuchung. Berlin: Deutsche Akademie der Wissenschaften zu Berlin, Institut für Orientforschung. 1954 Die Infinitive im Tocharischen. In: Johannes Schubert & Ulrich Schneider (eds.), Asiatica. Festschrift Friedrich Weller, Zum 65. Geburtstag gewidmet von seinen Freunden, Kollegen und Schülern. Leipzig: Harrassowitz, 701–764. 1957 Der Gebrauch der Vergangenheitstempora im Tocharischen. Wiesbaden: Harrassowitz. 1964 Tocharische Elementarbuch. Band II: Texte und Glossar. Heidelberg: Winter. 1967 Besteht ein formaler Unterschied zwischen N. und Obl. Sg. M. bei den wtoch. Verbaladjektiven? Die Sprache 13: 20–30. 1968 Zur Verwendung von toch. A oki / B ramt und A mäṃtne / B mäkte in Vergleichen. Orbis 17: 198–231. 1977 Zum formalen Problem bei den Verbaladjektiven des Tocharischen. Zeitschrift für vergleichende Sprachforschung 91: 256–265. 1984 Review of A. J. Van Windekens (1979). Indogermanische Forschungen 88: 204–226. 1985 Die Erforschung des Tocharischen (1960–1984). Stuttgart. 1991 Probleme der Wortforschung im Tocharischen. In: Lambert Isebaert (ed.), Studia Etymologica Indoeuropaea. Memoriae A. J. Van Windekens (1915–1989) dicata. Leuven: Peeters, 295–304. 1994 Zur Frage der nichtindogermanischen Komponente im Tocharischen. In: Roland Bielmeier & Reinhard [unter Mitarb. von René Lanszweert] (eds.), Indogermanica et Caucasica: Festschrift für Karl Horst Schmidt zum 65. Geburtstag. Berlin, New York: de Gruyter, 223–233.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

403

1997

Interpretationsprobleme im Tocharischen. Unflektiertes A puk, B po „ganz, all, jeder“. Stuttgart: Steiner. Thomason, Sarah Grey & Terrence Kaufman 1988 Language Contact, Creolization and Genetic Linguistics. Berkeley: University of California Press. Thornton, Anna M. 2005 Morfologia. Roma: Carocci. 2009 Constraining gender assignment rules. Language Sciences 31.1: 14–32. 2011 La non canonicità del tipo it. braccio/braccia/bracci: sovrabbondanza, difet­ tività o iperdifferenziazione? Studi di Grammatica Italiana 29–30 (2010–11 [2013]): 429–477. Tichy, Eva 1993 Kollektiva, Genus femininum und relative Chronologie im Indogermanischen. Historische Sprachforschung 106: 1–19. Þórhallsdóttir, Guðrún 1988 Tocharian contraction across -w-. Tocharian and Indo-European Studies 2: 184–210. Trask, Robert Lawrence 200o Dictionary of Historical and Comparative Linguistics. Edinburgh: Edinburgh University Press. Tremblay, Xavier 2005 Irano-Tocharica et Tocharo-Iranica. Bulletin of the School of Oriental and African Studies 68: 421–449. UrIG 2021 Thomas Lindner. Urindogermanische Grammatik. Teil II: Flexionsparadigmen. Heidelberg: Winter. de Vaan, Michiel 2000 The Indo-Iranian animal suffix *-āćá-. Indo-Iranian Journal 43: 279–293. 2003 The Avestan Vowels. Amsterdam: Rodopi. 2008 Etymological dictionary of Latin and the other Italic Languages. Leiden–Boston: Brill. 2011 Review of Rainer Lipp (2009). Kratylos 56: 1–14. 2019 Proto-Indo-European *sm and *si ‘one’. In: Alwin Kloekhorst & Tijmen Pronk (eds.), The Precursors of Proto-Indo-European. The Indo-Anatolian and the Indo-Uralic hypotheses. Boston: Brill–Rodopi, 203–218. Vaillant, André 1958 Grammaire comparée des langues slaves. Tome II: Morphologie. Lyon: IAC. Van Brock, Nadia 1971 Le traitement des nasales voyelles en tokharien. Zeitschrift für vergleichende Sprachforschung 85: 280–295.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

404

Bibliography

Van de Velde, Freek & Muriel Norde 2016 Exaptation: Taking stock of a controversial notion in linguistics. In: Freek Van de Velde & Muriel Norde (eds.), Exaptation and Language Change. Amsterdam–Philadelphia: Benjamins, 1–35. Van Epps, Nadia & Gerd Carling 2019 Patterns of gender assignment in the Jamtlandic variety of Scandinavian. Nordic Journal of Linguistics 43: 93–125. Van Windekens, Albert 1941 Lexique étymologique des dialectes tokharien. Louvain. 1942 Contribution a l’étude de la morphologie grecque et latine. L’Antiquité Classique 11: 291–297. 1944 Morphologie comparée du tokharien. Louvain. 1976 Le tokharien, confronté avec les autres langues indo-européennes. Vol. I: La phonétique et le vocabulaire. Louvain: Centre International de Dialectologie Générale. 1979 Le tokharien, confronté avec les autres langues indo-européennes. Vol. II,1: La morphologie nominale. Louvain: Centre International de Dialectologie Générale. Vaux, Bert 2002 Stang’s Law and Szemerényi’s Law in nonlinear phonology. In: Mark Southern (ed.), Indo-European Perspectives. Washington: Institute for the Study of Man, 317–327. Vijūnas, Aurelijus 2009 The Indo-European primary t-stems. Innsbruck: Institut für Sprachwissenschaft. Wackernagel, Jacob 2009 Lectures on Syntax: With Special Reference to Greek, Latin, and Germanic. Translated and edited with notes and bibliography by David Langslow. Oxford: Oxford University Press [English translation of Wackernagel, Jacob. 1926–1928. Vorlesungen über Syntax: mit besonderer Berücksichtigung von Grieschisch, Lateinisch und Deutsch. 2nd ed. Basel: Birkhäuser. I (1926): Erste Reihe. II (1928): Zweite Reihe]. Wälchli, Bernhard & Francesca Di Garbo 2019 The dynamics of gender complexity. In: Francesca Di Garbo, Bruno Olsson & Bernhard Wälchli (eds.), Grammatical gender and linguistic complexity. Volume II: World-wide comparative studies. Berlin: Language Science Press, 201–354. Weber, Dieter 1985 Khotansakisch āṣaṇa- ‘wert, würdig’. In: Harold W. Bailey, Adrian D. H. Bivar, Jacques Duchesne-Guillemin & John R. Hinnells (eds.), Acta Iranica: Hommages et Opera Minora. Volume X. Papers in honour of Professor Mary Boyce. Leiden: Brill, 675–682. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

405

Weiss, Michael 2018 Tocharian and the West. In: Olav Hackstein & Andreas Opfermann (eds.), Priscis Libentius et Liberius Novis. Indogermanische und sprachwissenschaftliche Studien. Festschrift tur Gerhard Meiser zum 65. Mitarbeit von Harald Bichlmeier und Sabine Hausler. Hamburg: Baar, 373–381. 2020 Outline of the Historical and Comparative Grammar of Latin. Second edition. Ann Arbor: Beech Stave Press. Widmer, Paul 1997 Zwei keltische t-Stämme. Historische Sprachforschung 119: 122–127. ́ 2005 Der altindische vr̥kī-Typus und hethitisch nakki-: Der indogermanische Instrumental zwischen Syntax und Morphologie. Die Sprache 45: 190–208. Wilkens, Jens 2021 Handwörterbuch des Altuigurischen. Akademie der Wissenschaften zu Göttingen. Wilkens, Jens, Georges-Jean Pinault & Michaël Peyrot 2014 A Tocharian B parallel to the legend of Kalmāṣapāda and Sutasoma of the Old Uyghur Daśakarmapathāvadānamālā. Acta Orientalia Academiae Scientiarum Hungaricae 67: 1–18. Wilkens, Jens & Michaël Peyrot 2017 Weitere Parallelen in Tocharisch B zur altuigurischen Daśakarmapathā­ vadānamālā: Mahendrasena- und Ṣaḍdanta-Avadāna. In: Zur lichten Heimat. Studien zu Manichäismus, Iranistik und Zentralasiekunde im Gedenken an Werner Sundermann. Herausgegeben von einem Team „Turfanforschung“. Wiesbaden: Harrassowitz, 685–710. Willi, Andreas 2010 Hera, Eros, Iuno Sororia. Indogermanische Forschungen 115: 234–267. Willms, Lothar 2010 On the IE etymology of Greek (w)anax. Glotta 86: 232–271. Winter, Werner 1952 An Indo-European prefix *n̥ - ‘together with’. Language 28: 186–191. 1961 Lexical interchange between ‘Tocharian’ A and B. Journal of the American Oriental Society 81: 271–280. 1962a Nominal and pronominal dual in Tocharian. Language 38: 111–134. 1962b Die Vertretung indogermanischer Dentale im Tocharischen. Indogermanische Forschungen 67: 16–35. 1962c Review of Wolfgang Krause & Werner Thomas (1960). Deutsche Literaturzeitung für Kritik der internationalen Wissenschaft 83: 1066–1069. 1963 Tocharians and Turks. In: Denis Sinor (ed.), Aspects of Altaic civilization, Proceedings of the Fifth Meeting of the Permanent International Altaistic Conference Held at Indiana University, June 4–9, 1962 (Uralic and Altaic Series 23). Bloomington–The Hague: Indiana University, 239–251. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

406

Bibliography

1965

Tocharian evidence [for laryngeals]. In: Werner Winter (ed.), Evidence for laryngeals. The Hauge–London–Paris: Mouton, 190–211. 1968 Archaismen in der tocharischen Adjektiv- und Nominalbildung. In: Jan C. Heesterman, Godard H. Schokker & V. I. Subramoniam (eds.), PRATIDĀNAM. Indian, Iranian and Indo-European Studies presented to F. B. J. Kuiper on his sixtieth birthday. The Hague–Paris: Mouton, 60–64. 1970 Some widespread Indo-European titles. In: George Cardona, Henry M. Hoenigswald & Alfred Senn (eds.), Indo-European and Indo-Europeans: Papers Presented at the Third Indo-European Conference at the University of Pennsylvania. Philadelphia: University of Pennsylvania Press, 49–54. 1971 Baktrische Lehnwörter im Tocharischen. In: Robert Schmitt-Brandt (ed.), Donum Indogermanicum: Festgabe für Anton Scherer zum 70. Geburstag. Heidelberg: Winter, 217–223. 1972 Zur Vertretung von *w nach Konsonant in Tocharisch B. Orbis 21: 385–390. 1975 Review of Peter Stumpf (1971). Kratylos 18: 136–138. 1977 Internal structure and external relationship of two verbal paradigms: Tocharian B weñ-, A weñ- ‘say’. Tocharian and Indo-European Studies 5: 133–159. 1980 Zum Beitrag der tocharischen Sprachen zu Problemen der lautlichen Rekonstruktion des Indogermanischen. In: Manfred Mayrhofer, Martin Peters & Oskar E. Pfeiffer (eds.), Lautgeschichte und Etymologie. Akten der VII. Fachtagung der Indogermanischen Gesellschaft, Wien, 24.–29. September 1978. Wiesbaden: Reichert, 542–563. 1981 Review of A. J. Van Windekens (1976). Language 57: 935–941. 1983 A Tentative English-Walapai Dictionary und Etymologicum Tocharicum: Unterschiede und Gemeinsamkeiten. In: Alfred Bammesberger (ed.), Das etymologische Wörterbuch: Fragen der Konzeption und Gestaltung. Regensburg: Verlag Friedrich Pustet, 313–327. 1985a Tocharian B soy, A se and related forms. Journal of the American Oriental Society 105: 259–264. 1985b ‘Left’ or ‘right’. In: Jacez Fisiak (ed.), Historical semantics. Historical wordformations. Berlin–New York–Amsterdam: de Gruyter, 283–595. 1988 Cardinal points and other directions in Tocharian A and B. In: Mohammed Ali Jazayery & Werner Winter (eds.), Languages and Cultures: Studies in Honor of Edgar C. Polomé. Berlin: de Gruyter, 775–791. 1989 Tocharian B -aiñ : B -añ/-āñ and related problems. Tocharian and Indo-European Studies 3: 111–120. 1992 [Numerals in] Tocharian. In: Jadranka Gvozdanović (ed.), Indo-European Numerals. Berlin–New York: de Gruyter, 97–161. 1997 Lexical archaisms in the Tocharian languages. In: Hans H. Hock (ed.), Historical, Indo-European, and lexicographical studies: A Festschrift for Ladislav Zgusta on the occasion of his 70th birthday. Berlin–New York: de Gruyter, 183–193. Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Bibliography

407

1998 Tocharian. In: Anna Giacalone Ramat & Paolo Ramat (eds.), The Indo-European languages. London: Routledge, 154–168. 1999 Tocharian marginalia. Tocharian and Indo-European Studies 8: 247–274. 2003a A new look at a Tocharian B text. Tocharian and Indo-European Studies 10: 105–124. 2003b Review of Douglas Q. Adams (1999). Journal of the American Oriental Society 123: 202–208. 2011 Vowel lengthening before distinctively voiced consonants in Tocharian. Tocharian and Indo-European Studies 12: 221–238. Witczak, Krzysztof Tomasz 2017 The Germanic word for ‘fleshy part of the body, calf’: A new reconstruction and etymology. Linguistica Silesiana 38: 89–93. Yakubovich, Ilya 2oo2 Nugae Sogdica. Bulletin of the School of Oriental and African Studies 65: 544–549. Zair, Nicholas 2012 The Reflexes of the Proto-Indo-European Laryngeals in Celtic. Leiden–Boston: Brill. Zaliznjak, Andrey 1967 Russkoe imennoe slovoizmenenie [Russian noun inflection]. Moskva: Nauka. Zimmer, Stefan 1976 Die Satzstellung des finiten Verbs im Tocharischen. The Hague: Mouton. 1982 Die Funktion der tocharischen ṣi/ṣṣe-Adjektive. Zeitschrift für vergleichende Sprachforschung 96: 277–289.

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Index Verborum This index contains a list of words cited in the book. Included are members of compounds (°x or x°), deduced forms (x*), and ghost words (†x). Restored Tocharian forms are cited without brackets. The order of the languages is: (1) Tocharian A, (2) Tocharian B, (3) Vedic, Classical Sanskrit and Buddhist Sanskrit (4) Middle Indian languages, (5) Avestan, (6) Khotanese, (7) Sogdian, (8) Bactrian, (9) Khwarezmian, (10) Ossetic, (11) Old Persian, (12) Middle Persian, (13) Parthian, (14) New Persian, (15) Old Church Slavonic, (16) Lithuanian, (17) Latvian, (18) Old Prussian, (19) Hittite, (20) Armenian, (21) Ancient Greek (and Mycenaean Greek), (22) Albanian, (23) Latin, (24) Romanian, (25) Italian, (26) Gothic, (27) Old High German, (28) Old Norse, (29) Old English, (30) Celtic languages, (31) Old Uyghur. Tocharian A akṣar 39 añcu* 106, 128–130 añcwāṣi 106, 303 añcwāṣṣāñ 303 añcwāṣṣāṃ 303 atäṅkät 283, 286 anityāt 55n73 aptsar 41n47 amok 46 amokäts 46 amoktsāp 44 aräṃ 191, 213–214 arämpāt 191 arkämnāṣi 69, 236, 303 arkämnāṣṣās 303 aśi 76–77, 92, 98, 162 aśśāṃ* 77, 98 aśśe 77, 98, 162 aśśeṃ 298 asaṃkh(y)e 39 asaṃkh(y)es 39 āk 189 āk- 45 ākär 214, 219–220 ākälṣäl 48 āknats 150n169, 308 ākrunt 214, 219–220 ākläṣlyi 48 āti 51, 173, 178n217 āp 53–56 āmāś 40, 97 āmāśāñ 40

āmpuk 149n167 āy 38–39, 40n44, 51, 102, 113, 137–140, 143 āyäntu 51n66, 102, 139 ārānt 94 ārāntāñc* 94 ārki 311 ārkiṃ 297, 311 ārkyant 311 ārkyaṃś 311 ārwar 216 ārṣal 154, 156 ārṣlās 156 ālak 250, 270–273, 276, 290, 349 ālakäṃ 273 āleṃ 157 ālkont 271 ālyāk 271, 273 ālyakäṃ 271, 273 ālyäkyāṃ 271, 273 ālyek 271, 273 ālyekäs 273 ālykes 271, 273 āśand 45 āśäntāṃ 45 āṣāṃ 283n45 āṣānik 283n45 āṣträṃ 317, 349 ās 51 äntāṃ 251 äntsaṃ 250–251 upādhyā 44 upādhyāy 44 ekro 49n63 ekrorñe 49n63

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

409

Index Verborum enāk 283n45 eṣäk 320 es 150n169, 320 okāk 34, 272 okät 149n167 oko 21, 51, 203, 206 oṅkaṃ 41, 317, 348 oṅkaläm 72, 153, 156, 163 oṅkälmāñ 72, 156 oṅkälmāṣi 156 oṅkälmās 156 oṅkälme 163 oṅkriṃ 176n212 oñant 156 oñi 299 oñi-cmolṣāṃ 303 oñi-cmolṣi 299, 303 opänt 34n36, 299 opärkā 197 opäs* 147–148, 151 oppal 38–39 oppal-yokāṃ 115 opṣäly 39, 151 opsi 151 omäl 174 omlyi 174 or 186 olar 72 olariñ 72 oṣke 158 kaṃ 188, 229 kanweṃ 79n15, 157 kapśiññāṣi 303 kapśiṃñāṣās 303 kayurṣ 51 karak* 211 karmavācak* 94 karmavāckāñc* 94 kalp* 39 kaśśi 299, 301 kaśśiñ 301 kākmart 126n111 kāc* 133 kāckäl 97 kācke 175 kāccap* 97 kātak* 97n45 kātäk* 97n45

kātkāñ 97n45 kātkāśśi 97n45 kānikāñc 94 kāpā- 123–124 kāplune* 124 kārāś 212 kāwaṣi 103n56 kāwas 103n56 kāśyap 44 kāśyapi 44 kāsu 49, 313–314 kāswone 49 kāts 102, 130–134, 156, 163n192 kātsaṣi* 102, 156 kātse 163n192 kätwes 122n98 kän- 309 känt 187n233, 241n325 käntu 51, 102, 113–117, 121–122, 156 käntuyo 114–115 °käntwāṣi 102, 121 käntwāsyo 102, 115, 156 käntwis 122n98 käntwes* 122n98 °käṃṣe 309 kärpi 301 käryāñ 103 käryās 103 kälyme 158, 226, 230n319 käṣiyāp 44 käṣṣi 40 käṣṣiñ 40–41 käṣṣyāp 45 kinnarñā* 48n61, 93, 296n61 kinnarñāṃ 48n61 ku 149n167 Kutluk 92 kuryar 126, 216 kuryart 47n57, 126–127 kursär 69, 217, 241n325 kursärwā 217 kurtsru 217 kulewāñ 77–78 kulewās 77 kuli 77, 299 kus 125n107, 153, 250 ko 51 ko- 45

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

410 kotär 218n291 koṃ 20, 71, 228 koläm 147, 156 kowi 151 koṣant 45 kos 265 knānmune 29, 49 krañcäṃ 278 krant 278, 314, 341 krāṃ 214, 236 krāmärts 150n169, 284 kränolāñc 94 kräsā- 213n284 kräts 313–314 kri 103, 122, 156 kror 179–181, 237 klāypsā- 151 kläysā- 320–321 klop 193n250 klopant 70 klopantyo 70 klopäṃtwaṃ 70 klośäṃ 147n163 klots 147n163 °klyu 189n239 klyom 162, 278, 312 klyomäñcäs 278, 312 klyomänt 278, 312 klyomäṣ 278, 312 klyomiṃ 95, 297, 312 klyomināñ 312–313 klyomine 162 kwärā- 241n325 caṃ 21 cākkär 218n291 ce- 262, 268 ces- 262, 264, 268 ciñcri 38n38 ckācar 41, 71 cmolwāṣi 303 cmolwāṣṣās 303 ñäkci 284, 301, 303, 349 ñäkciṃ 301, 349 ñäkcyāñ 303 ñäkcyāṃ 303 ñäkcyās 303 ñäkteññā 48, 48n61, 93 ñäkteññānac 302n67

Index Verborum ñäkteññāy 44 ñuk 250 ñemiṣi 303 ñemiṣyāṃ 303 ñemiṣṣās 303 ñom 211, 232, 236 ñwi* 333 ta- 19, 32, 254, 256 tarp 39 tarpañ 39 tāṃ 102n52 tārśoṃ 214, 221, 236, 243 täpre 278 täm 33 tämne 34 tämnek 34 tämyo 34 tämyok 34 tärkär 217 tärkrunt 217, 281n39 tu 250 tuṅk 29 to- 84, 265 tom 29 tos- 264, 267–268, 276 tkaṃ 20 tpär 278 träk- 221 träyk- 115n86 tricäṃ 349 trice 318 trit 34, 283, 318, 349 tritaṣ 35 triśkaṃ 115 triśkās 115 tre 84 natäk 96 napeṃṣās 282, 303 napeṃṣi 303 napeṃṣinās 282 nas- 49 naslune 49 nākäm 50, 236 nākmant 69, 236 nācki 97 nātäk 48, 95–97 nāśi 48, 89n28, 94–97 nätk- 95–97

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

411

Index Verborum näṣ 250 nu 257 pañi 173, 179 paräṃ 314 parivrājak 48 parno 314 parnoreṣi 303 parnoreṣṣās 303 paloṃ 214, 221, 236, 243 palonās 221 paloṃṣi 221 pawtā- 197–198 pats 80 pāk 189n236, 191n244 pācar 41, 71, 349 pācri 44 °pāt 191n245, 195 pātär 148 °pāṣe 46, 309 pās- 46, 182, 309 pänt 34n36, 286, 299 päyk- 45 pärkā- 197 pärkär 286 pärs 200, 206–207 pärsant 200, 200–201n256 pärwāṃ 228 pälā- 231 pälk 166, 197, 200 pälkā- 197, 198n253 pälkäntu 200 pälkets 28n31 pälketsāñ 28n31 pält 135n137 pältwā 102, 135n137 pälmäs 229 pälskant 200 pälskā- 197 pältsäk 197, 198n253, 200 pikār 39 pikārās 39 puk 149n167, 273 pukäl 39 puklā 39 puttiśparṣās 303 puttiśparṣṣāṃ 303 puttiśparṣṣās 303 pe 51, 53, 157n183

pekant 45 peñ* 157n183 peṃ 157n183 pes 157n183 poke 51, 72, 147, 157 pokeñ 72 pokeṃ 72, 145n155, 157, 198, 297 pokes 157 poñcäṃ 313 poñś 278 poto 197 pont 273, 313, 341 pont- 265 pontsāṃ 313 por 41, 224, 227–229 poräṃ 41, 224 poṣi 173, 179, 182, pos* 182 posac 179, 182 posaṃ 179, 182 posā 182 pkänt 187n233 pñintu 29 pyāpi 51, 147, 156 pyāppyāñ 156 pyāppyās 156 pyāpyāñ 156 pyāpyās 156 pracar 41, 71, 117, 265, 320n88 pracre 321, 349 pracres 320n88, 321 pravārāpakāñc 94 praṣt 39, 178 praskañi 299 praski 146n157, 299 pratsak 154 pre 162 pret 94 pretāñc 94 pruccamo 49n63 pruccamñe 49n63 plantā- 198n252 plamäs 229 plāc 39, 41, 224, 231, 235 plācäṃ 224, 231, 235 plācänyo 235 plānto 197 pläskā- 198n252

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

412 plos 229 bodhisatvāp 45 brāmaṃ 94 brāmnāñc 94 bhādrāy 44 maku 238 mañ 71 malañ 238 mācar 41, 71 mācri 44 mätkont 273 mättak 250, 273 mäśkit 94, 124 mäśkitāñc 94 mok 265 yas 250 yāntär 218n291 yāmlaṃ 29 yärk 189, 191, 195 yuk 41, 52, 187, 320 yukañ 71, 320 yok- 49, 202 yokañi 299 yoke 147, 157, 186, 201, 299 yoktsi 49 yoñi 173, 179, 183 yoṃ 179, 183 ykorñe 49n63 ytār 39, 41, 150n169, 224–226 ytāräṃ 39, 41, 224 ynāñmune 29 ype 214, 218n292 ypeyu 214 ymār 150n169 yme 158 ysār 21, 41, 186, 224–226, 232 ysāräṃ 21, 41, 224, 232 rake 214, 222, 319 rakentu 214 rackiñ 97n47 rackisyo 97n47 ratäk 97n47, 318n82 rape 39 ri 72 riñ 72 riṣakyāp 45 rtär 286–287 lāk* 52

Index Verborum lāñci 301–302 lāñciṃ 301 lāñcināñ 303 lāñcinās 303 lāñcinäm 301 lānt 48, 84 lānts 48, 72, 76, 78, 84–86, 99 lāntsac 78 lāntsañ 78, 86 lāntsas 86 lāṃtsas 78 lāntsāṃ 72, 76, 78, 84–85 lāntsānac 78 lāntsānā 78 lu 51, 102, 134–135, 143 lukäśnu 175, 175n210 lukśone 175 lok 126 lokit 105, 124, 126–127 lokitā- 127 lkäś 175n210 lwā 69, 102, 135n137, 136 lweṃ 284, 298 lyalypu 42 lyā 102 lyäm 125n107 lyiyā 102 lyiyā-āpsā 28n31 lykäṃ 231n321, 348 lymeṃ 157 vidyādharñā* 48n61 vidyādharñāñ 48n61 wak 49, 174 wac 113, 127 waco 47n57, 127 wañi 242, 242n328 want 42–44, 187 wark 188 warpi 126, 201 warpiśke 126 wartsi 173, 178 waśirṣi 303 waśirṣṣāñ 303 waśeṃ 49, 173–174, 297–298 waṣt 117, 186 waṣti 302n67 waṣtiṃ 298, 302, 321 was 250

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

413

Index Verborum wasu 216–217n288 wākā- 228 wākäm 50, 236 wākmant 50, 69, 219, 236 wākmats 28n31 wākmtse 28n31 wātu* 52 wāpants* 45 wāpäṃtsune 45 wārpā- 201 wāsak 48 wāskāñc 48, 94 wäknant 188 wät 34, 125n107, 153 wäynās- 242 wär 53, 53–54n69, 56, 241n325 wärkänt 156 wärkäntā(s) 156 wärt 200, 205–206 wärtant 200 wärts 286 wäl 48, 73, 84 wäṣta-käntwāñ 116 wäs 192, 213, 213n281, 242 wu 149n167, 266–267n23 weñ- 45–46, 239n324, 298 wes 321 wesis 321 wkäṃ 153, 188, 194 wtaṣ 34 wtā 34 wtāk 34 wram 41, 224, 230, 232, 235 wramäṃ 41, 224, 230, 236 wramnāśśi 236 wrasañ 50 wrasas 50 wlaluneṣi 303 wlaluneṣṣās 303 wlāṃ-ñkät 164 wles 214, 218n292, 32o wlesant 214 wmār 243 wmāri 243 wrasom 50 wsār 51, 150n169 wsāṣi 303 wsāṣṣāṃ 303

wsok 49 wsokone 49 śanweṃ 79n15, 157 śāmpāṃ* 129 śäk 292n55 śäktālyi 51 śäṃ 76, 78–79, 81–84, 86, 245, 352 śoṃ* 147, 156 śom* 50, 94–95 śomäṃ 94–95 śomiṃ 50, 94–95, 297 śtwar 153 śnās 76, 78, 86, 99 śnu 78–79 śne 84 śmoññe 173, 175 śwā- 49 śwātsi 49 śwātsi-yoktsi 49 ṣar 71, 90 ṣāmner 94 ṣāmnerāñc 94 ṣäpnant 188, 195 ṣäpnas 188, 195 ṣäpnasaṃ 188 ṣärttw- 199n254 ṣälyp 189–190 ṣälypañ 190 ṣälypas 189 ṣik 197, 200 ṣikās 200 ṣurm 210–211 ṣurmant 211 ṣotre 214, 220 ṣotreyäntu 214 ṣomaṃ 258n15 ṣtām 186, 211 ṣtāmäntu 211 ṣpäṃ 188, 195 sa- 250, 256, 288 saṃ 102n52, 253, 253n6 saṃsārṣi 282, 303 saṃsārṣinās 282 saṃsārṣṣās 282, 303 salat 52 sas 259, 338n115 sā- 256, 331 sāṃtäk 97n45, 127n113

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

414

Index Verborum

sāṃtkenu 47n57, 127n113 sārmäntu 211, 236 säṃ 297, 338n115 säm 253, 256 säykā- 197 säs 253, 256 sukaṣinās 21 se 83, 218n292 sewāñ 78 spartwā- 199n254 swāñceṃ 157, 228 swār 228, 286 tsar 51, 229 tsäṅkär 69, 217 tsäṅkrunt 217 tsärk 113, 129 tspok- 46 tspokäts* 46 tsraṣi 301 tsraṣiñ 301 tsraṣis 301 Tocharian B akarte 127 akartte 127 aknātsa 150n169, 292n52, 308 akruna 215, 219 akrūna 214, 218, 218n293, 220, 237 akwatse 220n298 akṣār 39 aks- 46, 309 aksaṣṣuki 46 ajātaśatruñ 45 añcali 319n84 añcalyi 319n84 at 256 atiya 51, 173, 178n217 ate 193n251, 256 anityāt 55n73 anaiśai 162 apasmār 131 appakke 47, 170n204 aptsar 41n47 amāc 40, 40n45 amāskai 162 ampa- 176

amparwa 69n4, 217–218 ampoña* 154, 173, 176 ampoño 154, 176 ammakki 49 ay 138 ay- 45, 49 ayāṣṣe 138 ayk- 174, 309 arañce 103 araṇemiñ 45 ar(a)hānte 71 ariwe 51 arkañe* 173n207 arkwañña 128n117, 297, 310–311, 312n75 arkwiṃ 310, 310na arkwinäṃ 310–311 arśakärśa 173 arṣāklaṃ 160, 163, 168–169 arṣāklatstse* 145, 153 arṣāklo 154, 156 alaṃ-yritaññe 52 alaṃ-śrotaññe 52 alekak 273 alekä 273 alekk 273 aleksa 273 aletstse 274n29 alokälymi 274n29 aloṅkna 273 allek 246, 250, 266, 270–275, 276, 288, 290– 291, 329, 331, 349 allok 82n19, 271, 274n29, 275, 327, 332 alloṅk 271–272, 275–276, 336 alloṅk- 275, 337 alloṅkä 272 alloṅkna 142, 271–272, 275 alloṅtä 272 alloṅna 272 alloykna 272 allyeṅkä 291 alyāk 82n19, 271, 274–276, 327, 332 alyek 271, 275, 349 alyekepi 272 alyeṅk 271, 276 alyeṅkäṃ 271 alyaik 271, 275, 319n85 alyaiṅk 271 aśari 40

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

415

Index Verborum aśiya 20, 41, 44, 76–77, 92, 98, 162 298 aśiyaññe 298 aśiyana 41, 77 aśiyāntse 77, 98 aśiyai 77, 98, 162 aścemaṣṣeṃ 279n35, 281 aṣanīke 283n45 aṣāṃ 283n45 asaṃkhyainta 39 astaräññe 166 astariññe 166 astareṃ 279, 304, 317n79, 349 āu 51 āka 238–239 ākär* 220 āke 189 °ākṣi 309 ānandi 44 āntse 150n169, 320 āp 53–56 āpäṃ 54–55 āppantse 170n204 āppo* 47, 170n204 āmpär* 69, 217 āmma* 49 †āy 137 āya 39, 102, 137–139 āyo 38–39, 51, 101–102, 113, 137–139, 143, 229n318 āyor 49, 138 ārkwi 310–311, 314 ārwa 219 ārwer 216 āl 51 āl yrīye 52 āläṃ 274n29 ālek 273 āline 319n84 āllyaik 291 ālyine 319n84 āśce 51, 71 ās 51 āsta 39, 102, 137–139, 186, 229n318, 238 īke 189–190 ikenta 190 iñcuwaññe 106 iñcuwo 106, 129 indrinta 280–281

intsu 250 uttari 44 udāvarttäntse 127 upāsakāñca 48, 94 upāsake 48, 94 uppāl 38 ek 147, 239n324 ekamätte 286 ekita 104 ekita yam- 104 ekīta 104 ekītatsñe 104 ekītatstse 104 ekṣalye 151, 291 eṅkwaññe 299, 312n75 eṅkweṃ 41, 348 eñcuwa 106 eñcuwaññe 106 eñcuwo 106, 128–130, 145 etaṅkätte 283, 286 eneṅkāmeṃ 281 eṃṣke 43, 277, 320 epiṅkte 34n36, 299 epets ay- 81 †epetsa 80 epetso 80 emalle 174 emalya 44, 173–174 empele 291n51 eynāke 283n45 ere 188, 191, 194, 213–214 ereṃ 188, 213n280 ereṃtsa 213n280 erepate 191, 241 erkäntse 183 erkenma 69, 236 ersna 70, 213, 236, 238, 243 elauke 45, 81 eweta 81, 193n250 eśaiṃ 147n163 eṣe 81 eṣerña* 92 eṣerñāna 93 aināke 283n45 aiśamo 49 aiśamñe 49, 312n75 °aiśi 309 aiṣṣeñca 45

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

416 oko 18, 21, 51, 203, 206 okt 149n167 oksai 145, 161n189, 168 oksaiññe 145–146 oksaiṃ 160, 161n189, 163, 167–168 oksaine 145n155 okso 146–149, 151, 160, 206, 331 oṅkarño 176n212 oṅkolma 48, 93, 331–332 oṅkolmañ 72 oṅkolmaññe 137n140, 166, 298n64 oṅkolmāñ 168 oṅkolmai 163 oṅkolmaiññe 137n140, 166 oṅkolmo 48, 72, 93, 136, 153, 156, 163, 332 otkasa 216n288 onolme 50 onolmeṃntsä 321 onolmentso 321 omp 253, 256 ompalskoñeṣṣi 280 ompe 253, 256 or 186, 229–230n318 oraṣṣe 284 orotstse 286 olyi 46, 127 olyitau 46, 126 oskiye 158 ost 117, 186, 312n75 ostañña 48, 93 ostaññe 48, 93, 298, 312n75, 321 osta-ṣmeñca 45 ostūwa 219 auñentaṃ* 156 auñento 156 ausa 105 †ausane 105 ausu 216n288 kaka- 197, 198n252 kakāpau 123 kaccalya* 173–174 kaṭapūtanäñña 93 kaṭapūtane* 93 katk- 48, 174 katkauña 48, 173, 175 katkewña 175 kattāke 97n45 kante 187n233, 241n325

Index Verborum kantwa 102, 117, 121–122 kantwaṃ 111, 114 kaṃtwaṃ 114 kantwaṃtsa 114 kantwo 47, 51, 102, 113–122, 128, 156, 160 *kamarta- 126n111 kamartāññe 126–127 kamartīke 127, 127n111 kamarttāññe 127 kamarttīke 127 karak 212 karakna 212 karāk 210, 212 karākna 211–212 karāś 212 karttse 127 karyaṃ 103, 109 karyo* 103, 122, 144, 156 karyor 46, 49, 126, 181, 216 kartsa 313–314 kartse 127, 221–222, 313 kalpanma 55 kalymi 161n189 kaw- 45 kawa- 123–124 kawāñ 103 kawātse 103 kaswātse 130 kā 257 kāko 197, 198n252 kāñ 166n197 kārak 211–212 kālp 39, 55, 69 kāwa* 102–103 kāwalyñe 124 kāwo 102–103, 123–124, 144 kāswa 102 kāswasā 130 kāswo 102, 130–132, 144–145 kātsa 102, 133 kātso 102, 130–134, 156, 163n192 käntwāṃtsa 114 käntwāśke 47 kǝrya- 49 käryāñ 103, 109 käryor-pläṅṣiñña 46, 309 käryortaññe 127 käryorttau 46, 47n57, 126, 240

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

417

Index Verborum kǝrsa- 49 kärsor 49 kǝlp- 46, 241 kǝlpa- 46 kälpaṣṣuki* 46 kälpauca 46 kälymiṃ 161n189, 169 kälymiye 158, 226, 230n319 °kälywe 189n239 käṣṣi 40n45, 309 käṣṣī 40, 277 *kito 104, 129 kindarña 48n61 kinnare 93 kinnarña 48n61, 93 ku 149n167 kunti 47 kuntiśke 47 kurkamäṣṣi 108 kulāst 109n74 kuse 125n107, 153, 250 kuṣaiñä 167 keu 51 ket 256 keta 106, 113 ketāntse 106 kete 209n269, 256 keto 106, 129 kene 188, 209n269, 229 kerekauna 221, 236, 242n327 kercapaṃ 168 kercapo 136, 154–155 kerccapaṃ 168 kerccappaṃ 168 kerci 318 kercci 238, 318 kerte 127, 318 kerteṃ 318 kertte 127, 318 kertteṃ 318 keściye 301 kokaleṃ* 318 kokalyi 318–319 kokalyiśke* 319n83 kokleṃ 318–319 kotaiñ 167 koto* 147, 147n161 kottarwa 69, 218n291

kottär 69, 218n291 kompaino 167na korai 147n158, 168 koraiṃ 147n158, 167 koraiśke 147n158 koro* 147, 147n158 korraiṃ 168 kolmo* 147, 156 koṣko 147–148 koṣṭä 131 kos 265 kauṃ 71 kaumiye 158 kaumaiño 167 kaurṣe 51 kauṣeñca 45 kausenta 45 kraṅko 136, 147n159, 170 kramartse 150n169 krasa- 197–198 krāna 214, 236, 243 krāmär 150n169 krāso 197–198 kräṅkañe 170 kräṅkaiññe 147n159, 170 krəsta- 198n253 krent 221, 278 krenta 314, 341 krentaṃtso 321 krentäṃtsä 321 krenteuna 215, 222n304 krentewna 221 krentewnaṣṣe 215 krentauna 214, 221–222, 238 krentaunaṣṣi 280 krem° 183–184 kremīya 173, 183–184 kremot 183n229 kror° 180 kroriya* 173, 179, 181 klañtsa 170 klayksa- 151 klǝntsa- 321 kliye 77 klai 77 klai-yritaññe 52 klaiñ 77 klaiñtsa 170

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

418 klaiṃ 77, 209 klaiṃ-śrotaññe 52 klaina 71, 77 klaiyna śroñ 52 klautsa-pälṣi 147–148n163 klautso 147, 147n163 klyotañ 170 klyotaiñä 167, 170 klyomo 162, 223, 278, 312 klyomoñ 223, 278, 312 klyomoṃ 278, 312 klyomña 95, 297, 312 klyomñana 312–313, 341–342 klyomñai 162, 312 kwa- 197, 198n252 kwarsär 69, 217, 241, 241n325 kwəra- 241n325 kwärsarwa 217, 241 kweṃ 228 kṣai 131 ksa 240, 250, 257 ktsaitsñe 29 cake 189–190, 195, 280 cakkarwi 218n291 Cañca 92 cāk 69 cākkär 218n291 ceṃ 260, 260n17, 262, 264, 268–269 cenäṃ 260, 260n17 cempaṃts 259, 270 cey 84, 260–262, 268–269, 349 ceyna 260, 262 ceym 259, 269 cai 208, 260–262, 268–269, 349 caim 259, 269 caimp 259, 269 colormeṣṣeṃ 114 Cowaśke 47 ckācko* 154 ckenta 280 cpi 254n8 cwi 254n8 ñakte 48, 93, 296 ñatke 193n251 ñaś 166n198 ñasso 107–108 ñāssa 107–108 †ñāsso 107–108

Index Verborum ñäkteñña 44, 48, 93, 296 ñäś 250, 296 ñiññe 296 ñuwa* 331, 333 ñuwe 46 ñem 211, 236 ñemna 211, 232, 236 ñyās 108 ñyāssa ñäsk- 108 ñwetse 46 taññe 296 tanākkai 49n62, 128n117 tanākkaisa 49n62, 128n117 tanākko 49n62, 128n117, 154 tanāñ 109 tanaulykañä 168 tanaulykaṃ 168na tapre 278, 304 tarkāntsa* 45 tarkär 217 tarya 29 tarśauna 214, 221, 236, 243 tallāw 53, 127, 314 tā 257–258, 265 tākoṃ 164 tāna 102 tānaṃ 109 tāno 49n62, 51, 102, 109, 128–129, 150n169 tārśi* 221n300 tārśai 221n300 tǝrk- 221 tǝrka- 45 täpreñ 278, 304 tärkarwa 217–218 tärkänallya 291 tumeṃ 34 tuwe 250, 296 tusa 34 te 19, 32–33, 254, 256 teki 29 tekīta 104, 127 tekīto* 105, 124–127 tai 292n53, 349 †taine 208 taiwe 50 to 207 totka 209 ton 262

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

419

Index Verborum toṃ 21, 29, 131, 208, 262, 264–265, 267, 269, 321, 336–338 tonak 264 †tonta 209 toṃn 262 tonmeṃ 262 tonts 262 tontsa 262 toṃtsā 262 toy 84, 260, 263, 269 toyna 260, 263, 269 toym 259, 270, 270n26 tkacera 71 tkācer 41, 71, 80 tkātär 80 tkātärñ 71 †tkātre 80 traṅko 123, 197, 200–201n256, 240 träṅkänta 200–201n256 trənk- 197 trəywa- 197, 199 trici 318 trice 318, 349 trite 34–35, 283, 318, 349 triteṣṣe 35 tritesa 35 trey 84 traiwo 197, 199, 202n258 traiwoṣṣe 199 tvāṅkarañ 168 tvāṅkaro 128, 154 twe 250 tweri 189 twere* 189, 194 nakanma 50, 69, 236 °nakṣi 309 naśi 104–105 nāki 50, 236 nǝk- 105, 192n248, 309 nətka- 96n44 nekīta 104–105, 126 nekīto* 105, 124, 126 newiya 44, 173, 178n217 nes- 49 nesalñe 49 naine 207–208 no 257 nauntaiñä 167

nauntaiṃ 156, 167 nauntainä 167na nauntaino 167 naunto* 146–147, 156 naumikkane 47 naumikke* 47 naumiye 47 naumyenta 280 pacera 71 °pate 191n245, 195 pantañ 107n68, 170 pantaiṃ 156 pantaintsā 170 pantaitstse 170 parivrājakāñca* 48, 94 paruwa 240–241, 241n325 parwa 240 parwā 240 parwāne 240 parśaiṃ 167 parso 200, 203, 206–207 paleuna 215 palauna 214, 221, 236, 243 palsko 198n253, 200 pawta- 197–198 paṣṣallyi 291 patsa 108 patsaṃ 108–111 patso 108–109, 111 pāke 191n244 pācer 41, 71, 80, 330n104, 349 pātär 121, 349 pātärñ 71 pātri 44, 80 pātro 148 pānto 107n68, 146–147, 156, 170 pār 105, 240–241, 241n325 pərka- 197 pärkare 286 pärwāne 228 pärsanta 200–201n256 pärsonta 200 pǝyl- 309 pǝla- 221, 231 pǝlka- 46 pälkostau* 46, 126 pälkaucäkka 49 pälskonta 200

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

420 pälskoṣṣi 280 pälskauca 46 päst 256 pätsāñä 108 pikāränta 39 pikul 39–40 pikwala 39 piṅkte 34n36, 286, 299 pīta 203–204 †pītantse 203–204 †pitaiṃ 203–204 †pitaitse 204 pīto 147n162, 203–204 pitosa 203–204 pirko 197 pilko 166, 197, 200, 202n258 pilkonta 200 pilta 136–137 piltāṣ 137 piltāsa 69, 102, 135–136 °pilṣi 309 piśträ 131 puwar 224, 227–229 pūwar 227n315, 227–228, 281 peñiya 172n206, 173, 179 peñiyā 172n206 perne 314 pernew 53 pelkiñ 107, 169 pelkiṃ 169 pew 53 pest 256 †petso 80 paiñ 157n183 paiṃ 157n183 paine 157n183 paiyye 51, 53, 143, 157n183 paiyyeññe 296 po 83, 313 po-aiśi 309 pokai 51, 147, 157 pokaiññe 297 pokaine 145n155 pokaiyño 167 poñc 80, 278 poñco 80 ponta 273, 313, 341 poyśi 40, 83

Index Verborum poysiñña 309–310 poyśiññana 309 poyśiññe 310 poyśinta 309 porsno 147 poṣiya* 173, 179, 182 pauto 197 pkante 187n233 pkänte-yamiñña 309 pyapyantse 170 pyapyaiññe 296 pyapyaiṃ 167 pyapyaintse 170 pyāppyaiñ 167 pyāpyai 161 pyāpyo 51, 147, 156, 170, 183, 296 pratsāko 154 prəska- 202 pruknānträ 103n57 preṃtsa 85n23 preściya 173, 178 proksa 238–239 procer 41, 71, 80, 117, 265 protär 121 proskiye 146n157 prosko 146n157, 147, 153n176, 202 plaktukäñña 48 °plaṅṣi 46, 309 planta- 197, 198n252 plāce 230–231 plānto 197 pläṅkṣi 93 °pläṅkṣiñña* 93 plǝnk- 46, 309 plǝska- 46, 123, 198n252 ploriya* 169, 173, 181, 237 ploriyaṃ 169n199, 181 ploriyatstse* 181 pwāra 218, 224 ptsāñ 108–109 bram-ñikte 166 brahm-ñäkte 16 brahmaṇāñca 94 brāhmaṇe 94 bhājanta 281 ma 257 makte 250, 273 mañiya 48, 93, 331–332

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

421

Index Verborum mañiye 48, 93, 332 mantālaitse 170 malkwer 216 maśc 349 maśce 231, 349 maskwatstsai 102n53 mācer 41, 71, 80, 330n104 mātär 121 mātri 44, 80 māla 106 mālatsai 106 mālo 106, 130, 144–145 †māskwa 102n53 māskwä 102n53 †māskwo 102n53 mäksu 250, 256, 263 mäktoṃ 263 mäktoynas 263 mäñcuṣka 166 məsk- 197 miñcuṣka 166 mit 166 mittarwi 218n291 mittär* 218n291 Mitraśke 47 miśo 202, 206 mīsa 70, 238 misaiwenta 238 misko 197 mekwa 238–240 meñe 71, 123n100 meyya 141n149 meyyā 140, 141n147, 143n153 meli 238, 319n84 melyi 319n84 mewiya 48, 51 mewīyañä 168 mewiyo 48, 51, 136, 153–154 maiyya 102, 111n82, 113, 140–143, 145 maiyyañ 140–141 maiyyaṃ 140–142 maiyyana 140, 141n50, 142, 143n153 maiyyā 140 maiyyāu 127 maiyyo 102, 140–143, 145 maiwe 142, 311 moko 265 mokoṃśka 47n60

mokoṃśke 47n60 mokośka 47n60 mokośwañ 168 mokauśka 47n60 mcuṣka 93, 94n40 mcuṣke 40, 93, 124 mñcuṣka 93–94 mñcuṣke 93 mlyokotau 126n111 yakne 153, 188, 194 yakwi 71 yakwe 41, 52, 187, 220n298, 317, 354 yakṣañña 93 yantarwa 218n291 yapoy 214, 218n292 yam- 46, 49, 241, 309 yamaṣṣuki 46 °yamortstse 46 yari 349 yarke 189, 191, 195 yarpär 217 yarpo 197, 199 yal 47 Yaśodhara 92 yasa 213, 213n281, 242 yasa- 197–198 yasar 21, 186, 224–226 yasoñña 183 yasna 213, 238 yāntär 218n291 °yāmi 46, 309 yāmor 46, 49 yāso 183 yäktāñ 166n197 yäktāñm 166n197 yäkwaṣke 47n58 yäkwe-pläṅṣi 46 yǝta- 189n238, 204 yärparwa 217 yəsa- 198n252 yente 42–44, 187 yenme 47, 127, 193n250 yenmeu 47, 127 yerkwantaṃ 156 yerkwantane 145n155 yerkwantalañ 168, 272 yerkwantalo* 154 yerkwanto* 156

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

422 yerter 216 yes 250 yok- 49, 202 yoko 147, 157, 202 yoktsi 49 yoñiya 173, 179, 183, 208 yoñyeṣṣe 208 yotkolau 46, 127 yolo 307 ykāk 34, 272 yke-postäṃ 190–191n243 ykenta 190–191 ytārye 150n169, 224–226 ynamo* 52 ynamñana 52 ypauna 214, 218n292 ymiye 158 yriye 226, 349 ylaṃśke 47, 95 ylai-ñäkte 164 ysāra 218 ysāre 51, 150n169 raktsi 49 ratre 286, 342 rapaññe 166 ravaiññe 166 raso 197 rākṣatsa 48 rākṣatse 48 rǝk- 49 rǝyn- 46 rəs- 197, 198n252 riñña 93 riññe 93 riṃ 169 rīnätstse* 46 reki 214–220, 222, 319 rekewna 215 rekauna 214, 220 recci 318n82 rser 216 lakutse 49, 175 lakle 21, 193 laṃsūna 214 lantsona 41, 99 layka- 197 larekke* 47

Index Verborum lāñc 73, 107, 231n321 lānt 48, 84, 302n67 lāntäṃ 73, 107, 231n321, 320 lāṃs 39n42, 214, 218n292, 320 lāntsa 41, 48, 78, 84–86, 91, 99 lāre 47 läkutsewña 175 läkutsauña 48, 173, 175 läkle-lyakāñ 166n197 läkleñ 103n54 läkṣāññe 166 läksaiñe* 166 lipär 319n84 luwa 39, 102, 135–136 luwāñ 136n139, 166n197 luwo 39, 51, 101–102, 113, 134–138, 143, 166n197 laiko 197 laukaññe 124 laukīta* 104 laukīto 104–105, 124–127, 144 lauke 45, 81, 104, 126, 193n251 lwāññe 284, 298 lwāsa 39, 69, 102, 135–136, 167n197, 298n63 lyam 125n107 lyipär 319n84 lyiya* 135n136 lyiyo* 39n40, 135n136, 139, 140n146 lyekśiye 51 lyauto 147 lykaṃ 231n321, 348 lykaśke 166n198 lykiśke 166n198 lyyāsa 39n40, 69, 102, 135, 136, 139 wa 257 waka- 228 wakanma 50, 69, 236 wakse* 230 wace 35 wate 34–35, 125n107, 153, 318 watesa 34 wato 327 wapa- 45 wapāntsa 45, 308 wamer 39, 243 war 53, 53–54n69, 56, 241n325 war-katsa 102n51

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

423

Index Verborum warto 127 wartto 127, 200, 202, 205–206 warpa- 201 wartse 286 walo 48, 73, 84, 107, 231n321 waṣākane 130n122 wase 192 wastsi 49 wāki 50, 236 wāṣmo 223 wāṣmoñ 223 wāṣmoṃ 223 wəta- 124n105 wäntarwa 309 wǝynask- 242 wərpa- 197, 199 wärttonta 200 wǝs- 49, 106, 216n288 wi 84n21 wi-pew* 52–53 wicuko 154 wīna 241–242 winasāre 71 witsakaṃ 111 witsako 154 wek 49, 174 weñ- 45–46, 239n324, 298 weñenta 45 weta 47, 81, 112–113, 127 wetāu 47, 127 weto* 124n105 weṃts 321 werke 188 werpiśkatse 46 werpiśke 46, 126, 201 werpye* 126 werwiye 201 wertsiya 173, 178 wertsiyaṃ 169n199, 172 weśeñña 44, 49, 173–175, 297 weṣṣuki 46 wotk- 216n288 wcūkane 145n155 wcuko 154 wtentse 34 wtentsesa 35n37 wteṣṣe 35

wmera 218, 243 wrakaiṃ 170 wrākaññeṃ 170 wrākaiññeṃ 170 wrāko* 170 wreme 230 wrotstse 286 wrauña 173, 175–176n211 wṣeñña 172–175 wsenta 192 śak 292n55 śakāto* 154 śana 41, 76, 78, 81–84, 98–99, 352 śano 76, 82–84, 86, 98 †śaṃts 210n271 śamñāṃśka 49, 93, 95n41 śay- 95, 333 śaro* 106 śarko* 129 śalna 213n283, 236, 242 śaw- 95, 223 śāñcapo 203n259 śāñcapotse 203n259 śātre 51, 192 śāmña 94 śāmñe 94, 223 śāmna 50, 223 śāmpa 113 śāmpo* 129 śäktālye 51 śəw(a)- 49 śeśuwer 49 śoline 54n71, 319n84 śolyine 54n71, 319n84 śaumo 39, 50, 95, 223 śaumoṃ 223 śauwlo 80 śka-maiyya 141n148 ścono 147, 156 ścmoñña 173–175 śtwer-pew 52–53 śnoy 76, 84 śnona 41, 76, 99 śrānäṃ 107, 164, 338n115 śrāy 107, 107n67, 164, 165n196 śwātsi 49 śwātsi-yoktsi 49

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

424 ṣaḍvargi* 40, 40n45 ṣaḍvarginta 40n45 ṣamāne 20 ṣar 51, 89, 229 ṣarm 210 ṣarmire 47 ṣarmirśke 47 ṣarmna 211, 236 ṣarya 78, 86–91, 99, 352 ṣaryo* 78 ṣaryompa 78 ṣalype 189–190 ṣārtto* 199n254, 202 ṣəm- 45 ṣǝrt- 199n254 ṣərtt- 202 ṣärmana 211, 211n272 ṣärmanma 211, 211n272 ṣik- 199 ṣiko 197, 200 ṣikonta 200 ṣitaiṃ 167 ṣito 204 ṣe 258n15, 259, 338n115 ṣek 34, 272 ṣer 49, 71, 90 ṣertwe 199n254 ṣerśka 41, 49, 93 ṣerśkana 41 ṣewi 214, 222, 222n305 ṣewauna 214 ṣotarnma 220 ṣotarye 220 ṣotri 214, 220 ṣotrūna 214, 220, 237 ṣotrūni 220 ṣñor 216–217, 237 ṣñaura 216–218, 237 ṣpakiye 158 ṣpakaiṃ 167nc ṣpane 188 ṣlyamñana 52 sana 297, 338n115 saṃtkinau 47n57 saṃtkīnau 46, 127n113 santse 42, 61, 188, 354 samākane 154 samp 250, 252–253, 253n6, 256, 259, 270

Index Verborum sarmana 211 sā 254, 256–257 sāñ 69, 166n197 sāṃtke 46, 97n45, 127n113 sārm 211, 236 sārmna 211, 236 sǝyk- 199 səyka- 197 särwāna 211, 238, 242n327 sälyiye 158 säsuwa 71, 78, 319n83 säsuśkañ 169n199, 319n83 säswaśkañ 319n83 su 34, 253, 263 suwa 102, 135–136 suwo 102, 134–136, 143 se 21, 208, 250, 253, 256, 288 seṃ 253, 253n6 soṃśke 95 somo 258n15, 327, 338n115 somona 258n15 soy 83, 218n292, 319n83 skiyo 151–153, 228n316 stāna 211 stām 211 stǝma- 174 stmānma 238–239 snai-netke 96 snai-maiyyañ 141 spartta- 199n254 spertte 199n254, 205 srukalñe 29 swañcai 167 swañcaiñ 168 swañcaiṃ 167 swāñco 146, 157, 228 swāre 228, 286 tsaktsaiṃ 147, 167 tsaṅkär 69, 217 tsāktso* 147 tsāra 102 tsāro 102, 123–124, 202 tsäṅkacca 183 tsäṅkana 183n228, 210 tsäṅkantä 183n228 tsäṅkarwa 217 tsər- 202, 266–267n23 tseṃ 210n270, 310–311

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

425

Index Verborum tserekwa 238–239 tsereññ- 239n324 tsain 69 tsainwa 69 tskertane 154 Vedic, Sanskrit and Buddhist Sanskrit akṣara- 39 ákṣita- 286 ájñāta- 150n169 ajñāti- 274n29 anāryaiḥ 301 anityatā- 55n73 anyá- 270 áp- 53 apasmāra- 131 apsaras- 41n47 amā́tya- 40n45, 97 amr̥ t́ a- 286 ayám 253–254 ar(a)hant(a)- 71, 94 árāya- 177 ́ arāyī- 177 arká- 191 árjuna- 311 árjunī- 311 arya- 92n36 áśru- 219n296 áśva- 61, 187, 355 áśvāso vr̥ ṣ́ aṇo 61 ásr̥ -k, asnáḥ 186, 226 asaú 253 asnāvirá- 216 áhar/n- 219–220n297 áhāni 219–220n297 ācārya- 40n45 ānanda- 44 ā́yu- 139 ārya- 92n36 āryikā- 92n36 ukṣán- 148 utpala- 38 udakodara- 102n51 udán- 53, 53n69 udára- 102n51, 132n128

upasaṃpadā 131 upādhyāya- 44 upāsaka- 48, 94 upāsikā- 94 ū́dhar/n- 233–234 ūrdhvá- 178n218, 205 r̥ ṣvá- 213 ekānta 274n29 ekoti 274n29 ogha- 221 kacchapa- 97 kaṭapūtana- 93 kanyakā- 94 karaka- 212 kárṇa- 233 *karmavācaka- 94 kalpa- 39, 69 kāśyapa- 44 kiṃnara- 93 kiṃnarī- 48n61, 93 kilāsa- 131 kukkuṭa- 147 kulattha- 109n74 kuṣṭha- 130–131 kuṣṭhin- 130 kr̥kara- 147n159 kr̥kavā́ku- 147n159 kr̥ ṣṇá- 177 ́ kr̥ ṣṇī- 177 kṣaya- 131 khara- 147n158 gaṇa-nāthaka- 96 gaṇḍa- 131 gárgara- 221 gardabhá- 154–155 gahastha- 97n45 gr̥hástha- 97n45 gotra- 69, 218n291 govinda-nātha- 96 gnā ́- 81 grīṣmasya 179 cakra- 218n291 cátasras 90 cátuṣpad- 52–53 chāyā́- 152 jáni- 81 jánitar- 177 jánitrī- 177

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

426 jihvā́- 117, 120n96 jīvati 95 juhū́- 117 takṣaka- 45 tát 19, 32, 253–254 tā́n 262 tisrás 90 tr̥ ṇ́ a- 233 dakodara- 102n51 dā́ru- 186 devá- 332n90, 339 ́ devī- 332n90, 339 duhitúḥ 80 dvā́ra- 189 dvipád- 52–53 dhānā́ḥ 128 nāka-nātha- 96 nāka-nāthaka- 96 nāthá- 96–97 nāthaka- 97 náva- 322 návā- 322 pakṣín- 52 patatrín- 52 páti- 80 páttra- 240 padám 186 padmakesara- 111 padvát- 53 parivrājaka- 48, 94 parṇá- 240 pātra- 148 pitár- 61 pítriya- 295 ́ pīvan- 182 ́ pīvarī- 181 punaḥ-sará- 190 pr̥ ṣṭhá- 178 pratiṣṭha- 178 preta- 94 pretī- 94 br̥hánt- 286 brāhmaṇa- 94 brāhmaṇī- 94 bhága- 191n244 bhāgá- 191n244 bhāta- 191 bhā́ti 179

Index Verborum bhāti- 191 bhāna- 179 bhānú- 179 bhrū́- 228 maṇḍilya 173n207 mádhu- 109–110, 130, 144 mádhya- 295 mātár- 61, 330n103 māṃsá- 238 māṣa- 110–111 mitra- 218n291 muṣṭí- 231 mūtra- 206n266 mūlaka- 109–111 méhati 206 maireya- 106 yákr̥-t, yaknás 233, 235 yakṣa- 93 yakṣī- 93 yantra- 218n291 yásati 198n252 yugám 186 rátha- 177 ́ rathī- 177 rākṣasa- 48 loka-nātha- 96 vā́c- 174 vā́ta- 42–44, 187 vāta-nāśanaḥ 44 vāditra- 181 vā́r 53n69 vā́stu- 186 vinayadhara- 71 viṣá- 192 vr̥ḱ a- 177 ́ vr̥kī- 177 vr̥kṣa-nāthaka- 96 vr̥ ti- 205 śaśá- 132n125 śāntaka- 46, 97n45, 127n113 śrávas- 189–190n239 śrāmaṇera- 94 śrāmaṇerikā- 94 śrāmaṇerī- 94 śrómata- 312 sá 253, 256 sáḥ 253, 256 sákhā 222

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

427

Index Verborum samá- 338n115 sarpís- 190, 190n241 sarvajña- 309 sā́ 253, 256 sīm 257 sr̥kva- 211 sr̥kvaṇī 211 sŕ̥kvan- 211 ́ strī- 91n31 snā́van- 216 svápna- 188 svādú- 228, 286, 322 ́ svādvī- 322 hávate 198n252 hr̥d́ - 122 hr̥d́ aya- 122 Middle Indian acariya (Gāndh.) 40n45 ayariya (Gāndh.) 40n45 ayya (Pāli) 92n36 ācariya (Pāli) 40n45 āpa (Pāli) 55n73 upala (Gāndh.) 38 uppala (Pāli) 38 ṇāha (Pkt.) 96 nasa (Gāndh.) 96 nātha (Pāli) 96 ṣamanera (Gāndh.) 94 sāmaṇera (Pāli) 94 Avestan aiiārə̄ 219n297 ā̆p- 53–54n69, 178n217 aniia- 270 asaiia- 151 asti 186 aspa- 187 ā̊ŋhairī 91 ərəduua- 205 ərəδβa- 205 uxšan- 148 karǝna- 233 kasuuiš- 132

kahrka-tāt 147n159 kahrkāsa- 147n159 xara- 147n158 xvafna- 188 xvə̄ṇg 235 gaēθā- 104 gənā- 81 ɣənā- 82 jaini- 81 jə̄ni- 81 juuaiti 95 taka- 190 tat̰ 253 daēuua- 339 dānō.karš(a)- 128n114 pairiθna 224 parəna- 240 puxδa- 32 baēuuarə/n- 233 baga- 191n244 baɣa- 191n244 bāga- 191n244 bānu- 179 nāuuiia- 178n217 maēzaiti 206 maδu- 130 mātar- 61 mąθrān- 149 mušti- 231 yakarə 235 vaēti- 154 vāxš 174 vāta- 42, 187 ̆ vīša- 192 sāxvə̄nī 220n297 strī- 91n31 snāuuarə.bāzura- 216 sraoman- 312 zaēna- 69 zaēnuš 69 zərəd- 122 zərədaiia- 122 hā 253, 256 hāirišī- 91 hə̄ 253, 256 hitō.hizuuā̊ 119 hizuua 118, 120 hizuuā- 118–119, 120n96

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

428

Index Verborum

hizuuā̊ 118–120 hizuuā̊° 118–120 hizuuā̊.āuuərətō 119 hizuuā̊.uxδāiš 118 hizuuąm 118 hizuuō 118 hī 257 ̆ 235 huuarə̄  hō 253, 256 hū- 134 Khotanese aśiā- 92 āmāca- 40n45 āśiria- 40n45 āṣaṇa- 283n45 āstaa- 139n144 kamala- 126n111 karāśśä 212n278 karāśśi- 212 kavā- 296n61 kavīña- 296n61 kārra- 233 kindarīña- 296n61 kr̥ ṅga- 147n159 †kṣīʾa- 40n45 kṣundai hor- 81 kūlasta- 109n74 khara- 147n158 khāysāna- 133 gīha- 104 ggīhaa- 104 gūha- 147n161 graysa- 198 jsīr- 239 gyūna- 18 ttarra- 233 ttuṃgare 154 daha- 296n61 dahīña- 296n61 dānā- 128, 144 nāgīña- 296n61 padmakyesärä 111 paʾsa- 207 pīha- 203 bāga- 191n244 bāta- 43–44

bāti- 43–44 beta jinākä 44 bete jsa 43 beva 44 beva jsa 44 byūrru 233 mīysaa- 206 mau- 130, 144 mauya 154 muyi 154 *mūya- 154 yaulā̆- 307 śśaśvāna- 203n259 saña- 69, 166n197 saha- 132n12 strīyā- 91n31 haṃbūta- 154 haṃbva- 176 hīśśana- 129–130 Sogdian ʾynʾqwc (Chr.) 283n45 brywr (Chr.) 233 βrywr (Man.) 233 δʾn (Man.) 128 kʾrtʾk (Buddh.) 97n45 kʾrtk (Buddh.) 97n45 myw (Man.) 154 syʾk (Man.) 152 syʾq (Chr.) 152 Bactrian αβο 41n46, 317n80 βyźg (MBctr.) 130n122 βιζαγο 130n122 καμιρδο 126n111 *καμιρδιγο 126n111 μολο 130, 144 Khwarezmian ʾspny 129 hnčw 129

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

429

Index Verborum Ossetic æfsæn 129 biræ (I), be(w)ræ (D) 233 kark 147n159 kur 173n207 rad (I), radæ (D) 318n82 syl (I), silæ (D) 91n31 widag (I), wedagæ, jedagæ (D) 154 Old Persian ā̆p- 54n69, 178n217 duvara- 189 nāviya- 178n217 baga- 191n24 hama- 338n115 hạzān- 117 hạzzānam 117 Middle Persian āb 178n217 āhun 129 āsen 129 bēwar 233 galōg 221 ǰagar 235 kurrag 173n207 *nāydāg 178n217 radag 318n82 rōd 178n217 sāyag 152 uzwān 120 Parthian āsun 129 rdg (Man) 318n82 New Persian āhan 129 kark 147n159

kurra 173n207 rada 318n82 Old Church Slavonic agoda 206 bogъ 191n244 črěmošь 184 duxъ 201 dvorъ 189 językъ 117 nova 322 novъ 322 pasti 182 pero (cs) 240 prostъ 178 rěčь 222 sěnь 152, 152n171 sijati 152n171 sǫdii 177 sǫdъ 177 sъnъ 189 tokъ 190 žena 82 vrana 175n211 Lithuanian árklas 187 dukterès 80 dúona 128 dvãras 189 gývas 152n171 ̃ 184 kermušė  kiáutas 133 líesas 140n146 liežùvis 117 pùsė 182 spar̃nas 240 spindà 179 spingėt́ i 179 tãkas 190 úoga 206 ùpė 54n69 várna 175n211

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

430

Index Verborum Latvian

asinis 186, 226 auglis 206 augti 206 duõna 128 dzîvs 152n171 liẽss 140n146 puse 182 seja 152 Old Prussian ape 54n69 genna 82 insuwis 117 pausan 182 pauson 182 Hittite ēšḫar, išḫan- 186, 226 ḫalii̯e/a-zi 140n146 ḫarki- 311 ḫaršai- 213 ḫaršar, ḫaršn- 213, 236 ḫarši- 213 ḫaššu- 61 ḫaššuššara- 61 ḫaštāi- 222 išḫa- 61, 90 išḫaššara- 61, 90 †itar 225n310 iuka- 186 karāu̯ ar, karaun- 180 kard(i)- 122 kattan 132 kūrka- 173n207 lēši 140n146 paḫḫur, paḫḫuen- 227, 235 paḫš- 182 parkuš 286 partāu̯ ar, partaun- 240 pattar, pattan- 240 pēdan 186 šāḫ- 83

šakriaḫšu 326 šakriašue 326 šikriašue 326 šupiašu 326 šupiašue 326 tāru- 186 u̯ arpa- 201 u̯ ātar, u̯ iten- 235 watniaḫšu 326 watniašue 326 zaḫḫai- 222 Armenian arawr 187 ariwn 226 ayl 270 ayt 139 barjr 286 ełǰewr 180 ełungn 239 erg 191 howr 227 kanay-kʿ 82n18 kin 81 knaw 81 lezu 117 neard-kʿ 216 now 188, 354 Ancient Greek e-re-ta (Myc.) 126 i-qo (Myc.) 187 ku-na-ja (Myc.) 82 ku-na-ki-si (Myc.) 82 wa-na-ka (Myc.) 95 ἀγνώς 150n169 ἄγνωτος 150n169 ἀγών, -ῶνος 151 ἄδικος, -ον 61 ἄεται 83 αἰών, -ῶνος 151 ἀκοστή 239 ἄλλος 270, 290 ἄμβροτος 286

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

431

Index Verborum ἀμμά 49 ἄναξ, -κτος 95–96 ἄνασσα, -ης 95 ἀνδράποδα 53n68 ἀπειλέω 231 ἀργαίνω 311 ἄργιλλα 311 ἄργιλλος 311 ἀργινόεις 311 ἀργός 311 ἄργυρος 311 ἄργυφος 311 ἄροτρον 187 ἄρρενα 9 ἄρσενες ἵπποι 61 ἄρσην, -ενος 150 ἄστυ 186 αὔξω 206 ἄφθιτος 286 βανά (Beotian) 82 γάνυμαι 124 γηθέω 124 γλῶσσα 177 γλῶχες 177 γόργυρα 221 γραῦς 107n67 γυνά (Doric) 82 γυναικ- 82n18 γυνή 82, 82n18 δαίμων, -ονος 151 δάκρυ 218n293, 219n296 δάκρυμα 218n293 δῖα 339 δόρυ 186 δοῦλος 53n68 ἔαρ 226, 226n314 ἐέλδωρ 226n313 εἶθαρ 220n299 εἴρω 90 εἷς 338n115 ἔλπος 190 ἔλφος 190 ἐρέτας 126 ἐρυθρός 286 Ϝάναξ (Beotian) 95 ζέω 198n252 ζυγόν 186 ζῷον 152n171

ζώω 95 ἡ 253, 256 ἡ θεός 61 ἡ θήλεα ἵππος 61 ἦαρ 226 ἡδεῖα 322 ἡδύς 228, 286, 322 ἧπαρ 220n297 ἥπατα 220n297 Ἥρᾱ 91 ἧτορ 103 θήλεα 9 θυγάτηρ 61 θυγατρός 80 ἵ 257 ἰός 192 ἵππος 187, 355 hίππος (Laconian) 355 ἱππότᾰ (Hom) 126 ἱππότης 126 καρδίᾱ 122 κατά 132 κῆρ 103, 122 κλέος 189–190n239 κόραξ 175–176n211 κραδίη (Hom) 103, 122 κρέξ 147n159 κρόμβυον 184 κρόμμυον 184 κρόμυον (Hom) 184 κύων, κυνός 150–151 λέων 134–135n135 μέ(σ)σος 295 μέλι, -τος 177 μέλισσα 177 μήτηρ 330n104 μία 297, 338n115 νέᾱ 322 νέος 322 νεῦρον 216 νυός 42, 61, 188, 354 ὁ 253 ὁ θεός 61 ὁ θῆλυς ὀρεύς 61 (ϝ)οἶκος 190 ὄλπη 190n241 ὀμείχω 206 ὁμός 338

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

432

Index Verborum

ὄνυξ 239 ὀρθός 178n218, 205 ὀρός 190 ὄρος 188, 213 οὐδέτερον 10 οὖθαρ, -ατος 234 οὗτος 253 ὀφρῦς 228 *ὄψ 174 πάν 313 πᾶς 313 πᾶσα 313 πατέρα 121 πάτριος 295 πέδον 186 πειθώ 222 περικτίτης 125 πῇ 257 πῖαρ 182 πίειρα 181 πίων 182 πόσις 80 πτερόν 240 πῦρ 227 πῠρός 227 ῥῆμα, -ατος 230 σκαιός 152n171 σκεύη 9–10 σκιᾱ́ 152 τετράποδα 53n68 τό 19, 32, 254 τὸ μεταξύ 10 τούς 262 τρίτος 32 ὕδωρ, ὕδατος 235 υἱύς 61, 83 ὕπνος 188 ὗς 134 φαίνω 179 φεύγω 177 φύζα 177 χείρος 90 ὦμος 150n169, 320 Albanian gjalpë 190 gjumë 189

hē 152 hije 152 Latin acus, aceris 189, 239 aevum, -ī 193 aevus 193 agnus fēmina 61 agnus mās 61 alius 270 amma 49 arātrum 187 argentum 311 arma, -ōrum 201 assarātum 226 as(s)yr 226 botulus 132–133 caelum, -ī 193 caelus 193 campus, -ī 195 cānus 132n125 carō, carnis 214 cāseum 194 cāseus 194 catilla 194 catillus 194 comes, -itis 125 cordis 122 cor 122 corium, -ī 193 corius 193 corpus, -oris 194–195 corvus 175–176n211 cutis 133 dacruma 218n293 deus 339 dingua 117 dīvus 339 dorsum, -ī 193 dorsus 193 equa 187 equus 187, 355 equus fēmina 61 femina 219–220n297, 233 femur, feminis 220n297, 233 femur, femoris 233 focus, -ī 195

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

433

Index Verborum folium 202 forum 189 frāter 61 fulgō 198n252 gaudeō 124 genetrīx 46 genitrīx 46 gurges 221 homō, -inis 151 hortus, -ī 195 iānua 183 iānus 183 iecur, iocineris 233 ignōtus 150n169 immortālis, -e 61 invictus 286 is 72 iter, itineris 224, 233 iugum 186 lac, lactis 193 lactem 193 lingua 117 lupus fēmina 61 mās aquila 61 māter 330 medius 295 meiō 206 nāsum 194 nāsus 194 nauta 126 nervus 216 nova 322 ́ nŏvum 164 novus 322 nurus 42, 61, 188 obscūrus 153n175 ossa 186 pāne 194 pānis 194 patrem 121 patrius 295 ́ pĕdem 164 penna 240 porcus fēmina 61 ́ pŏrtum 164 postis 178 potis 80 praedō, -ōnis 151 prātum, -ī 195

puteum 194 puteus 194 quā 257 quārtus 32 ruber 286 sāl, -is 194 sāle 194 sanguen 226 sanguīs 226 satis 83 scaevus 152n171 sermō, -ōnis 151 serō 90 serum 190 somnus 188 soror 61 splendeō 198n252 stāmen 186 sūs 134 tristis, -e 61 ūber, -ris 233–234 unda 53–54n69 unguis 239 ungula 239 ūrīna 53–54n69 uterum 194 uterus 194 uxor 91 venia 242 ventus 42, 187 verbum 187 vīcus 190 vīnum, -ī 193 vīnus 193 vīrus 192 vīvus 152n171 vorāgō 221 vorāx, -cis 221 vōx 174 Romanian băiat 17 bun 17 fată 17 scaun 17

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

434

Index Verborum Italian

bracci 25 braccio 25, 27 brutti 27 cambio 195 cambiora 195 campo 195 campora 195 carcere 25n25 carceri 25n25 dito 27 fuoco 195 fuocora 195 gregge 25n25 greggi 25n25 lenzuoli 25 lenzuolo 25 orto 195 ortora 195 piede 164 prato 195 pratora 195 questi 27 rotti 27 Gothic akran 206 aljis 270 auhsa 148 dius 135, 201 fon 227, 235 funins 227 liþus 140n146 mimz 238 qino 82 qiþus 132–133 salba 190 sama 338n115 sauil 235 si 257n14 skeinan 152n171 sunnin 235 sunno 235 tuggo 117 unkunþs 150–151n169

wato 235 waúrd 187 winds 42, 188 þius  322n90 þiwi 322n90 Old High German amma 49 bellan 231 fast 178 fedara 240 frist 178 fuir 227, 235 haso 132n125 hliumunt 312 hūt 133 lid 140n146 lūs 134 nagal 239 ohso 148–149 quiti 133 salb 190 salba 190 scuwo 153n175 wagan 188 wazzar 235 wint 42, 188 Old Norse dýr 201 frest 178 funi 235 fúr 227 heri 132n125 hǫss 132n125 hrang 147n159 húð 133 liðr 140n146 limr 140n146 rauðr 177 reyðr 177 vænn 242 vatn 235

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

435

Index Verborum Old English cwiða 133 dēor 135, 201 fearn 240 feðer 240 first 178 fȳr 235 hara 132n125 haso 132n125 hramsan 184 hringan 147n159 līra 140n146 lows 134 oxa 148–149 snoru 61, 355 swefn 189 wæter 235 worþ 205

clú (OIrish) 189–190n239 craf (W) 184 crem (MIrish) 184 cride (OIrish) 122 crim (MIrish) 184 dér (OIrish) 218n293, 219n296 fén (OIrish) 188 fine (OIrish) 242 gwain (W) 188 lleu (MW) 134 lou (MBret.) 134 mná (OIrish) 81 ohen (OBret.) 149 oss (MIrish) 148 ros (MIrish) 178 sí (OIrish) 257, 257n14 taman (MIrish) 186 téoir (OIrish) 90 ych (MW) 148–149

Celtic Languages

Old Uyghur

aile (OIrish) 270 arathar (OIrish) 187 bán (OIrish) 179 bé (OIrish) 81 cercc (MIrish) 147n159 cethéoir (OIrish) 90

ičägüläri 103 k(ä)ši 40n45 Kutlug 92n35 pretanč 94 sekriyü sučıyu 103n57 yavlak 307

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access

Alessandro Del Tomba - 978-90-04-53289-2 Downloaded from Brill.com04/26/2023 07:34:45PM via free access